|
|
The burden of disease and injury in Australia - 2003 The Australian Institute of Health and Welfare is Australias national health and welfare statistics and information agency. The Institutes mission is better health and wellbeing for Australians through better health and welfare statistics and information.Please note that as with all statistical reports there is the potential for minor revisions of data in this report over its life. Please refer to the online version at <www.aihw.gov.au/bod>. 1 School of Population Health, University of Queensland, Brisbane2 Australian Institute of Health and Welfare, CanberraAIHW cat. no. PHE 82 Stephen Begg 1, Theo Vos1, Bridget Barker1,Chris Stevenson 2, Lucy Stanley1 and Alan D Lopez1May 2007 © Australian Institute of Health and Welfare 2007 This work is copyright. Apart from any use as permitted under the Copyright Act 1968, no part may bereproduced without prior written permission from the Australian Institute of Health and Welfare. Requests and enquiries concerning reproduction and rights should be directed to the Head, Business Promotion and Media Unit, Australian Institute of Health and Welfare, GPO Box 570, Canberra ACT 2601. A complete list of the Institutes publications is available from the Institutes website <www.aihw.gov.au>. ISBN 978 1 74024 648 4 Suggested citation Begg S, Vos T, Barker B, Stevenson C, Stanley L, Lopez AD, 2007. The burden of disease and injury in Australia 2003. PHE 82. Canberra: AIHW. Australian Institute of Health and Welfare Board Chair Hon. Peter Collins, AM, QC Director Penny Allbon Any enquiries about or comments on this publication should be directed to: John Goss Australian Institute of Health and Welfare GPO Box 570 Canberra ACT 2601 Phone: (02) 6244 1151 Email: burdenofdisease@aihw.gov.au Published by the Australian Institute of Health and Welfare P Prriinntteedd bbyy Canprint Communicationsv Foreword Exactly a decade ago, the results of the first Global Burden of Disease (GBD) Study were published by Harvard University on behalf of the World Health Organization and the World Bank. These organisations and several countries then became interested in applying the GBD approach to better inform health policy, leading to a series of country studies on all continents. Probably the most technically competent and comprehensive of these were the Australian studies, led by Colin Mathers, for Australia as a whole, and Theo Vos for the state of Victoria. These analyses were based around 1996 data and have been widely used to inform priority setting and health policy debates in Australia. As a result of these initial studies, governments across Australia have become interested in using the burden of disease framework to help quantify health needs. There have also been advances in methods over the past ten years and greater interest among the health policy community in information about the burden of disease in population subgroups. This has all stimulated the need for a revised Australian burden of disease and injury study to update and extend the initial efforts. This report responds to that need. Some of the worlds leading researchers in burden of disease studies, with extensive experience in national applications of the methods, have joined the University of Queensland to create the great focus of expertise reflected in this study. Building on the analytical framework of previous studies, the report includes a number of important extensions of the framework that are highly relevant for policy. They include disease projections, small area analyses and state-level burden of disease results. Also, better methods around comorbidity and risk factor assessment have much improved the scientific basis of the findings reported here. This comprehensive study will undoubtedly meet the need for detailed information about the burden of disease and injury in Australia and its jurisdictions, about the principal causes of that burden, and how it is changing. But it alone is not enough. With rising pressure on health budgets, governments will increasingly rely not only on information about the burden of disease and injury, but also on cost-effective ways of reducing that burden. This study is a critical and fundamental step in that policy process and we expect it to be used widely to help improve the health of all Australians. Alan Lopez Penny Allbon Professor of Medical Statistics and Director Population Health, Australian Institute of Health and Welfare The University of Queensland vii Contents Foreword .............................................................................................................................................. v Executive summary ..............................................................................................................................1 Introduction ...................................................................................................................................1 Key findings ...................................................................................................................................2 Total burden of disease and injury.......................................................................................2 Health risks ..............................................................................................................................5 Differentials in burden across Australia..............................................................................6 Trendspast, present and future.........................................................................................7 Key implications.....................................................................................................................8 1 Introduction...................................................................................................................................9 1.1 Purpose....................................................................................................................................9 1.2 Background.............................................................................................................................9 1.3 Summary measures of population health .........................................................................10 1.4 Disability-Adjusted Life Years ............................................................................................11 1.5 Burden of disease analysis in Australia.............................................................................12 1.6 Burden in Aboriginal and Torres Strait Islander peoples ...............................................13 1.7 Structure of report.................................................................................................................13 2 Methodological developments .................................................................................................15 2.1 Social value choices...............................................................................................................15 2.2 Causal attribution .................................................................................................................17 Categorising deaths ..............................................................................................................18 Redistributing non-specific causes of death......................................................................19 Alternative categories...........................................................................................................25 2.3 Comorbidity and health.......................................................................................................25 2.4 Risks to health........................................................................................................................27 Explicit counterfactuals......................................................................................................28 Joint risk attribution..............................................................................................................29 2.5 Past, present and future burden .........................................................................................30 Mortality trends and projections ........................................................................................32 Incidence and case-fatality...................................................................................................32 Non-fatal conditions .............................................................................................................33 2.6 Differentials in burden ..........................................................................................................34 Categorising geographic areas ............................................................................................34 Estimating burden for subpopulations..............................................................................35 Subpopulation comparisons in this report........................................................................36 viii 3 Burden of disease and injury in Australia .............................................................................37 3.1 Disability-adjusted life years...............................................................................................37 3.2 Years of life lost .....................................................................................................................40 3.3 Years lost due to disability...................................................................................................43 Incident YLD..........................................................................................................................43 Prevalent YLD .......................................................................................................................46 3.4 Age and sex patterns ............................................................................................................47 Children aged 014 years.....................................................................................................47 Older children and adults aged 1544 years.....................................................................49 Adults aged 4564 years ......................................................................................................50 Adults aged 6574 years ......................................................................................................51 Older people aged 75 years and over.................................................................................53 3.5 Specific disease and injury categories ................................................................................54 Cancers ..................................................................................................................................55 Cardiovascular disease.........................................................................................................58 Mental disorders ...................................................................................................................59 Neurological and sense disorders.......................................................................................62 Chronic respiratory diseases ...............................................................................................64 Injuries ...................................................................................................................................65 Diabetes .................................................................................................................................67 Musculoskeletal diseases .....................................................................................................69 Alternative categories for selected conditions ..................................................................71 4 Risks to health in Australia.......................................................................................................72 4.1 Overview...............................................................................................................................72 4.2 Combined effect of 14 selected risks to health..................................................................73 4.3 Individual contribution of 14 selected risks to health......................................................76 Tobacco..................................................................................................................................76 High blood pressure .............................................................................................................77 High body mass ....................................................................................................................79 Physical inactivity .................................................................................................................81 High blood cholesterol .........................................................................................................83 Alcohol...................................................................................................................................84 Low fruit and vegetable consumption...............................................................................87 Illicit drugs .............................................................................................................................88 Occupational exposures and hazards ................................................................................90 Intimate partner violence.....................................................................................................92 Child sexual abuse ................................................................................................................93 Urban air pollution ...............................................................................................................95 ix Unsafe sex ..............................................................................................................................98 Osteoporosis ..........................................................................................................................99 5 Differentials in burden of disease and injury across Australia .......................................101 5.1 Overview.............................................................................................................................101 5.2 Health-adjusted life expectancy........................................................................................102 5.3 State and territory differentials .........................................................................................105 5.4 Differentials by socioeconomic status..............................................................................108 5.5 Differentials by remoteness ...............................................................................................111 6 Past, present and future burden of disease and injury in Australia ...............................114 6.1 Overview..............................................................................................................................114 6.2 Health-adjusted life expectancy.........................................................................................115 6.3 Burden .................................................................................................................................122 7 Discussion and conclusions ....................................................................................................129 7.1 Potential applications .........................................................................................................129 7.2 Policy implications..............................................................................................................130 7.3 Precision of estimates .........................................................................................................131 Fatal burden.........................................................................................................................131 Non-fatal burden.................................................................................................................131 7.4 Access to data ......................................................................................................................134 7.5 Future directions .................................................................................................................134 Appendix 1: Methods for estimating disability burden ...........................................................137 1A Infectious and parasitic diseases........................................................................................138 Tuberculosis........................................................................................................................138 Sexually transmitted diseases (excluding HIV/AIDS)..................................................138 HIV/AIDS...........................................................................................................................138 Diarrhoeal diseases .............................................................................................................139 Childhood immunisable diseases.....................................................................................139 Pertussis...............................................................................................................................139 Tetanus ................................................................................................................................139 Measles ................................................................................................................................140 Rubella.................................................................................................................................140 Haemophilus influenzae type b.............................................................................................140Meningitis ............................................................................................................................140 Septicaemia ..........................................................................................................................141 Arbovirus infections ...........................................................................................................141 Hepatitis ...............................................................................................................................141 Malaria.................................................................................................................................143 Trachoma.............................................................................................................................143 ix Unsafe sex ..............................................................................................................................98 Osteoporosis ..........................................................................................................................99 5 Differentials in burden of disease and injury across Australia .......................................101 5.1 Overview.............................................................................................................................101 5.2 Health-adjusted life expectancy........................................................................................102 5.3 State and territory differentials .........................................................................................105 5.4 Differentials by socioeconomic status..............................................................................108 5.5 Differentials by remoteness ...............................................................................................111 6 Past, present and future burden of disease and injury in Australia ...............................114 6.1 Overview..............................................................................................................................114 6.2 Health-adjusted life expectancy.........................................................................................115 6.3 Burden .................................................................................................................................122 7 Discussion and conclusions ....................................................................................................129 7.1 Potential applications .........................................................................................................129 7.2 Policy implications..............................................................................................................130 7.3 Precision of estimates .........................................................................................................131 Fatal burden.........................................................................................................................131 Non-fatal burden.................................................................................................................131 7.4 Access to data ......................................................................................................................134 7.5 Future directions .................................................................................................................134 Appendix 1: Methods for estimating disability burden ...........................................................137 1A Infectious and parasitic diseases........................................................................................138 Tuberculosis........................................................................................................................138 Sexually transmitted diseases (excluding HIV/AIDS)..................................................138 HIV/AIDS...........................................................................................................................138 Diarrhoeal diseases .............................................................................................................139 Childhood immunisable diseases.....................................................................................139 Pertussis...............................................................................................................................139 Tetanus ................................................................................................................................139 Measles ................................................................................................................................140 Rubella.................................................................................................................................140 Haemophilus influenzae type b.............................................................................................140Meningitis ............................................................................................................................140 Septicaemia ..........................................................................................................................141 Arbovirus infections ...........................................................................................................141 Hepatitis ...............................................................................................................................141 Malaria.................................................................................................................................143 Trachoma.............................................................................................................................143 x 1B Acute respiratory infections ................................................................................................144 Lower respiratory tract infections ....................................................................................144 Upper respiratory tract infections ....................................................................................144 Otitis media..........................................................................................................................145 1C Maternal conditions .............................................................................................................145 1D Neonatal causes....................................................................................................................146 Birth trauma and asphyxia ................................................................................................146 Low birth weight.................................................................................................................146 Neonatal infections .............................................................................................................147 Other conditions arising in the perinatal period ............................................................147 1E Nutritional deficiencies........................................................................................................147 Iron deficiency anaemia .....................................................................................................147 2F Malignant neoplasms ...........................................................................................................147 Disease incidence data........................................................................................................148 Cure rate and mean survival time ....................................................................................148 Long-term sequelae of cancer............................................................................................149 2G Other neoplasms ..................................................................................................................150 2H Diabetes ................................................................................................................................150 Diabetes cases ......................................................................................................................150 Retinopathy.........................................................................................................................151 Cataract and glaucoma.......................................................................................................151 Renal failure........................................................................................................................152 Neuropathy.........................................................................................................................152 Peripheral vascular disease ...............................................................................................152 Amputation and diabetic foot ...........................................................................................152 Ischaemic heart disease and stroke ..................................................................................153 2I Endocrine and metabolic disorders.....................................................................................153 Haemolytic anaemia...........................................................................................................153 Other non-deficiency anaemia ..........................................................................................153 Cystic fibrosis ......................................................................................................................153 Haemophilia ........................................................................................................................154 2J Mental disorders ....................................................................................................................154 Depression & anxiety, substance abuse (excluding heroin and stimulant dependence), borderline personality disorder and bipolar disorder ..........................154 Heroin dependence and harmful use...............................................................................156 Stimulant dependence........................................................................................................156 Psychotic disorders .............................................................................................................157 Eating disorders ..................................................................................................................158 Childhood disorders...........................................................................................................158 xi 2K Nervous system and sense organ disorders.....................................................................159 Dementia ..............................................................................................................................159 Epilepsy ...............................................................................................................................159 Parkinsons disease.............................................................................................................159 Motor neurone disease .......................................................................................................160 Multiple sclerosis ................................................................................................................160 Huntingtons chorea ...........................................................................................................161 Muscular dystrophy ...........................................................................................................161 Vision loss ............................................................................................................................161 Hearing loss .........................................................................................................................162 Intellectual disability ..........................................................................................................162 Migraine ...............................................................................................................................163 2L Cardiovascular disease ........................................................................................................163 Ischaemic heart disease ......................................................................................................163 Heart diseases resulting in heart failure ..........................................................................164 Stroke ...................................................................................................................................165 Other cardiovascular disease ............................................................................................166 2M Chronic respiratory diseases..............................................................................................167 Chronic obstructive pulmonary disease ..........................................................................167 Asthma.................................................................................................................................167 2N Diseases of the digestive system........................................................................................168 Peptic ulcer disease.............................................................................................................168 Cirrhosis of the liver ...........................................................................................................168 Inflammatory bowel disease .............................................................................................168 Other diseases of the digestive system............................................................................169 2O Genitourinary diseases........................................................................................................169 Nephritis & nephrosis ........................................................................................................169 Benign prostatic hypertrophy ...........................................................................................169 Urinary incontinence ..........................................................................................................170 Infertility..............................................................................................................................170 Other genitourinary diseases ............................................................................................171 2P Skin diseases.........................................................................................................................171 Eczema, acne and psoriasis................................................................................................171 Other skin diseases .............................................................................................................171 2Q Musculoskeletal diseases ....................................................................................................172 Rheumatoid arthritis ..........................................................................................................172 Osteoarthritis .......................................................................................................................172 Back pain ..............................................................................................................................173 xii Slipped disc..........................................................................................................................173 Occupational overuse syndrome ......................................................................................173 Gout......................................................................................................................................174 Other musculoskeletal disorders ......................................................................................174 2R Congenital anomalies...........................................................................................................175 Congenital heart disease ....................................................................................................175 Digestive system malformations.......................................................................................175 Renal agenesis .....................................................................................................................175 Other urogenital tract malformations ..............................................................................175 Other congenital anomalies...............................................................................................176 2S Oral conditions ......................................................................................................................176 Caries ...................................................................................................................................176 Edentulism..........................................................................................................................177 Periodontal disease .............................................................................................................178 Pulpitis.................................................................................................................................178 2Z Chronic fatigue syndrome...................................................................................................178 3 Injuries .....................................................................................................................................179 Appendix 2: Methods for attributing risk ...................................................................................180 Estimating population attributable fractions .........................................................................180 Choice of theoretical minimum................................................................................................181 Estimating attributable burden ................................................................................................181 Tobacco................................................................................................................................182 High blood pressure ...........................................................................................................182 High body mass ..................................................................................................................183 Physical inactivity ...............................................................................................................183 High blood cholesterol .......................................................................................................184 Alcohol.................................................................................................................................184 Low fruit and vegetable consumption.............................................................................185 Illicit drugs ...........................................................................................................................185 Occupational exposures and hazards ..............................................................................186 Child sexual abuse and intimate partner violence .........................................................187 Urban air pollution .............................................................................................................188 Unsafe sex ............................................................................................................................191 Osteoporosis ........................................................................................................................191 Annex tables .....................................................................................................................................201 Acknowledgments...........................................................................................................................285 Advisory committee ..................................................................................................................285 Expert advisors ...........................................................................................................................286 xiii Abbreviations and symbols............................................................................................................288 References.........................................................................................................................................291 List of tables .....................................................................................................................................315 List of figures ...................................................................................................................................319 1 Executive summary Introduction This report is the first complete assessment of the health of Australians to be released in the new millennium. The findings in this report - (i) identify the extent and distribution of health problems in Australia, and (ii) quantify the contribution of key health risk factors to these problems. Levels of death and disability from a comprehensive set of diseases, injuries and risks to health are combined to measure the total health burden. This report is the second of this type in Australia, the first having been released in 1999. It expands the scope of that previous report and also presents for the first time: the differentials of health burden across areas and population groups in Australia the joint contribution of key health risksincluding combined lifestyle, physiological, social and environmental factorson health an analysis of past trends of health burden and the likely health of Australians in 20 years from now should those trends continue. The findings of this report describe the health loss due to disease and injury that is not ameliorated by current treatment, rehabilitative and preventive efforts of the health system and society generally. Thus they represent the unmet challenges of the health system and are best interpreted as opportunities for health gain. By providing a comprehensive database of all relevant epidemiological and burden parameters through time, the report will benefit health policy development and research in relation to -
The study upon which the report is based was funded by the Australian Government Department of Health and Ageing. A report specifically examining the burden of disease and injury in Aboriginal and Torres Strait Islander people will be published separately. 2 Key findings Total burden of disease and injury The key measure used in this report to measure the total burden of disease and injury is the disability-adjusted life year (DALY). It describes the amount of time lost due to both fatal and non-fatal events, that is, years of life lost due to premature death coupled with years of healthy life lost due to disability. Cancer Cardiovascular Neuro- Mental logical Chronic respiratory Injuries Diabetes Musculoskeletal Other 19% 18% 12% 13% 7% 7% 5% 4% 14% Total Injuries Diabetes Cardiovascular Chronic respiratory Cancer Mental Neurological Musculoskeletal 52% 70% 54% 53% 53% 53% 47% 47% 42% 48% 30% 46% 47% 47% 47% 53% 53% 58% Males Females Total Cancer Cardiovascular Injuries Chronic respiratory Diabetes Neurological Musculoskeletal Mental 49% 82% 78% 76% 38% 22% 17% 7% 7% 51% 18% 22% 24% 62% 78% 83% 93% 93% Fatal Non-fatal Burden (DALYs) by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 In 2003, more than 2.63 million years of healthy life (that is, DALYs) were lost due to the burden of disease and injury in Australia. Cancers (19%) and cardiovascular disease (18%) were the leading causes of the burden of disease and injury in Australia in 2003, accounting for 37% of the total burden. Four-fifths of that burden was from premature deaths. For the first time, cancer has overtaken cardiovascular disease as the greatest cause of burden in Australia. Lung, colorectal and breast cancer were the leading specific causes of the burden of cancer. Ischaemic heart disease, stroke, and peripheral vascular disease were the leading specific causes of cardiovascular burden. Mental disorders and neurological & sense disorders were the next largest contributors, together accounting for a further 25% of the total health burden. Less than one-fifth of that burden was from premature deaths. Anxiety & depression, alcohol abuse, and personality disorders dominated the burden of mental disorders. Dementia, adult-onset hearing loss, and vision loss were the leading causes of burden due to neurological & sense disorders. Anxiety & depression also carries a risk of ischaemic heart disease and suicide, increasing the total burden due to the combined category of anxiety & depression from 7.3% to 8.2%. Diabetes also carries a risk of ischaemic heart disease and stroke, increasing the total burden of diabetes from 5.5% to 8.3%, and making it the fourth largest contributor to overall burden after cancer, CVD and mental disorders. The eight national health priority conditionsasthma, cancer, cardiovascular disease, diabetes mellitus, injuries, mental health, arthritis and musculoskeletal conditions, and dementiaaccounted for 72.8% of the total burden in 2003. Distribution of the burden between the sexes was roughly equal except for injuries (70% of the burden in males) and musculoskeletal (58% of the burden in females). The five leading specific causes of burden in men were ischaemic heart disease (11.1%), Type 2 diabetes (5.2%), anxiety & depression (4.8%), lung cancer (4.0%) and stroke (3.9%). The five leading specific causes of burden in women were anxiety & depression (10.0%), ischaemic heart disease (8.9%), stroke (5.1%), Type 2 diabetes (4.9%) and dementia (4.8%). Disability from all diseases and injuries resulted in a loss of 1.5% of healthy time lived by children, increasing with age to 14.7% in those aged 65 to 69 years, to 41.5% in the very aged. Fatal burden Life expectancy estimates the average years of life that a person can expect to live given current risks of mortality. In 2003 in Australia, life expectancy at birth was 80.7 years (78.3 years for males and 83.2 years for females). Fatal burdenmeasured in years of life lost (YLL)accounted for 49% of the total burden of disease and injury in Australia in 2003. Cancers (32.0%), cardiovascular disease (29.0%) and injuries (11.0%) were responsible for almost three-quarters of the fatal burden. Males experienced 55% of total fatal burden. The five leading specific causes of mortality burden among men were ischaemic heart disease (18.2%), lung cancer (7.3%), suicide & self-inflicted injury (5.4%), stroke (5.1%) and colorectal cancer (3.9%). Females experienced 45% of total fatal burden. The five leading specific causes of mortality burden among women were ischaemic heart disease (15.7%), stroke (8.5%), breast cancer (7.0%), lung cancer (5.5%) and colorectal cancer (4.2%). Non-fatal burden Health adjusted life expectancy (HALE) estimates the average years of equivalent healthy life that a person can expect to live. In 2003 in Australia, the average HALE was 72.9 years (70.6 years for males and 75.2 years for females), with 9.7% of life expectancy at birth lost due to disability. Non-fatal burdenmeasured in years of healthy life lost due to disability (YLD)accounted for 51% of the total burden of disease and injury in Australia in 2003. 4 Mental disorders (24%) and neurological & sense disorders (19%) contributed most to non-fatal burden. The five leading specific causes of non-fatal burden among men were anxiety & depression (10.0%), Type 2 diabetes (8.5%), adult-onset hearing loss (6.5%), asthma (4.2%) and dementia (3.9%). The five leading specific causes of non-fatal burden among women were anxiety & depression (18.1%), Type 2 diabetes (7.2%), dementia (6.4%), asthma (4.5%) and ischaemic heart disease (3.3%). Age patterns and total burden Distribution of population and burden (DALYs) by five broad age groups, Australia, 2003 Age group Population(a) Per cent of total DALYs Per cent of total 014 years 3,979,410 20.0 221,536 8.4 1544 years 8,622,610 43.4 633,260 24.1 4564 years 4,733,808 23.8 681,566 25.9 6574 years 1,349,949 6.8 428,904 16.3 75 years and over 1,195,692 6.0 667,504 25.4 Total 19,881,469 100.0 2,632,770 100.0 (a) Estimated resident population figures as at 30 June 2003 (ABS cat. no. 3201.0). Adults aged 45 to 64 years comprised 23.8% of the population in 2003 and experienced the largest proportion (25.9%) of disease and injury burden across key age groups. Cancer (28%), cardiovascular disease (16%) and neurological disorders (10%) accounted for more than half the total burden in this age group. Almost half of the burden was due to mortality. Adults aged over 75 years comprised 6.0% of the population but experienced the second highest proportion of burden (25.4%). Cardiovascular disease (34%) and cancer (19%) accounted for more than half of the burden. Overall, 68% of the burden was due to mortality. Adults aged 15 to 44 years represented the largest age group (43.4% of the population) and experienced 24.1% of the burden. Mental disorders (36%) and injuries (17%) accounted for more than half of the total burden in this age group. Mortality contributed 29% to the burden in this age group. Children aged 0-14 years comprised 20.0% of the population and experienced 8.4% of the total burden of disease and injury (DALY) in Australia in 2003. Twenty-three per cent of this burden was due to mental disorders, 18% to chronic respiratory disorders, and 16% to neonatal conditions. About one-quarter of the burden was due to mortality. Mental disorders (24%) and neurological & sense disorders (19%) contributed most to non-fatal burden. The five leading specific causes of non-fatal burden among men were anxiety & depression (10.0%), Type 2 diabetes (8.5%), adult-onset hearing loss (6.5%), asthma (4.2%) and dementia (3.9%). The five leading specific causes of non-fatal burden among women were anxiety & depression (18.1%), Type 2 diabetes (7.2%), dementia (6.4%), asthma (4.5%) and ischaemic heart disease (3.3%). Age patterns and total burden Distribution of population and burden (DALYs) by five broad age groups, Australia, 2003 Age group Population(a) Per cent of total DALYs Per cent of total 014 years 3,979,410 20.0 221,536 8.4 1544 years 8,622,610 43.4 633,260 24.1 4564 years 4,733,808 23.8 681,566 25.9 6574 years 1,349,949 6.8 428,904 16.3 75 years and over 1,195,692 6.0 667,504 25.4 Total 19,881,469 100.0 2,632,770 100.0 (a) Estimated resident population figures as at 30 June 2003 (ABS cat. no. 3201.0). Adults aged 45 to 64 years comprised 23.8% of the population in 2003 and experienced the largest proportion (25.9%) of disease and injury burden across key age groups. Cancer (28%), cardiovascular disease (16%) and neurological disorders (10%) accounted for more than half the total burden in this age group. Almost half of the burden was due to mortality. Adults aged over 75 years comprised 6.0% of the population but experienced the second highest proportion of burden (25.4%). Cardiovascular disease (34%) and cancer (19%) accounted for more than half of the burden. Overall, 68% of the burden was due to mortality. Adults aged 15 to 44 years represented the largest age group (43.4% of the population) and experienced 24.1% of the burden. Mental disorders (36%) and injuries (17%) accounted for more than half of the total burden in this age group. Mortality contributed 29% to the burden in this age group. Children aged 0-14 years comprised 20.0% of the population and experienced 8.4% of the total burden of disease and injury (DALY) in Australia in 2003. Twenty-three per cent of his burden was due to mental disorders, 18% to chronic respiratory disorders, and 16% to neonatal conditions. About one-quarter of the burden was due to mortality. 5 Health risks Individual and joint burden (DALYs) attributable to 14 selected risk factors by broad cause group, Australia, 2003 Broad cause group Cancer CVD Mental Neurological Injury Diabetes Other All causes Total burden ( 000) 499.4 473.8 350.5 312.8 185.1 143.8 667.4 2,632.8Attributable burden (%)(a) Tobacco 20.1 9.7 0.6 0.5 8.9 7.8 High blood pressure 42.1 7.6 High body mass 3.9 19.5 54.7 1.1 7.5 Physical inactivity 5.6 23.7 23.7 >0.1 6.6 High blood cholesterol 34.5 6.2 Alcohol Harmful effects 3.1 0.9 9.7 18.1 <0.1 3.3 Beneficial effects 5.6 >0.1 1.0 Net effects 3.1 4.7 9.7 18.1 <0.1 2.3 Low fruit & vegetable consumption 2.0 9.6 >0.1 2.1 Illicit drugs <0.1 8.0 3.6 2.5 2.0 Occupational exposures & hazards 3.1 0.4 0.8 4.7 3.4 2.0 Intimate partner violence 0.5 0.3 5.5 0.1 2.5 0.2 1.1 Child sexual abuse <0.1 <0.1 5.8 1.4 <0.1 0.9 Urban air pollution 0.8 2.7 0.4 0.7 Unsafe sex 1.0 1.4 0.6 Osteoporosis 2.4 0.2 Joint effect(b) 32.9 69.3 26.9 0.2 31.7 60.1 17.2 32.2 (a) Attributable burden within each column is expressed as a percentage of total burden for that column. (b) Figures for joint effects are not column totals. See Section 4.1 for further details. Findings on the amount of burden in 2003 that was attributable to current and past exposures to risks to health considered the following: Lifestyle behaviours (tobacco smoking, physical inactivity, alcohol consumption, low fruit and vegetable consumption, use of illicit drugs, and unsafe sex) Physiological states (high body mass, high blood pressure, high cholesterol, and osteoporosis) Social and environmental factors (occupational exposures and hazards, intimate partner violence, child sexual abuse, and urban air pollution). The 14 risks together explained 32.2% of the total burden of disease and injury in Australia in 2003. Tobacco was responsible for the greatest disease burden in Australia (7.8% of total burden), followed by high blood pressure (7.6%), high body mass (7.5%), physical +inactivity (6.6%), and high blood cholesterol (6.2%). 6 The five leading risks in males in 2003 were tobacco (9.6%), high blood pressure (7.8%), high body mass (7.7%), high blood cholesterol (6.6%) and physical inactivity (6.4%). Among women the leading risks were high blood pressure (7.3%), high body mass (7.3%), physical inactivity (6.8%), high blood cholesterol (5.8%) and tobacco (5.8%). This report sets out for the first time the combined or joint effect of these risks on health, accounting for the fact that many risks share complex causal pathways. It is difficult to quantify the exact contribution of each risk to the combined totals, but the proportion of totalburden that is explained by multiple risks within each disease and injury category can be reported with sufficient accuracy. Ten risks were associated with cancer and together explained 32.9% of the cancer burden. The majority was explained by tobacco but also included the effect of physical inactivity, high body mass, and alcohol consumption. Twelve risks were associated with cardiovascular disease and together explained 69.3% of the disease burden; for ischaemic heart disease this figure was 85.2%. High blood pressure and high blood cholesterol were the largest contributors. Three risks were associated with neurological and sensory disorders and together explained 0.2% of the burden from these disorders. This reflects a lack of knowledge about causation in this group. Two risks were associated with Type 2 diabetes and together explained 60.1% of the total burden. High body mass was by far the largest contributor (54.7%) followed by physical inactivity (23.7%). The burden associated with harmful alcohol consumption (3.2%) was partially offset by the cardiovascular disease prevented by safe levels of alcohol consumption (0.9%). This protective factor only becomes apparent after 45 years of age, whereas the harmful effects of alcohol are apparent at all ages. Differentials in burden across Australia This report shows for the first time that there are differentials across Australia in the proportion of life expectancy lost due to disability. There was a strong socioeconomic gradient in this measure, and differentials with respect to remoteness were also apparent but not as large. Health-adjusted life expectancy (HALE) in 2003 in Australia was 72.9 years (70.6 for males and 75.2 for females), with an average 9.7% of life expectancy at birth lost due to disability. Across states and territories, the proportion of life expectancy lost due to disability ranged from 7.7% in the ACT to 10.6% in South Australia. The NT had almost twice the rate of total burden of the ACT due to a higher rate of burden for most causes, but particularly cardiovascular disease, diabetes, and injury. Across socioeconomic quintiles, the proportion of life expectancy lost due to disability ranged from 8.7% in the highest quintile to 10.6% in the lowest. The 31.7% greater burden for the most disadvantaged population compared to the highest was due to higher rates of burden for most causes, but particularly mental disorders and cardiovascular disease. 7 Health-adjusted life expectancy (HALE) and life expectancy at birth lost due to disability by area and sex, Australia, 2003 Health-adjusted life expectancy (HALE) (years) At birth At age 60 Life expectancy at birth lost due to disability (%) Area Males Females Persons Males Females Persons Males Females Persons Jurisdiction NSW 70.5 75.3 72.9 17.1 20.6 18.9 9.8 9.5 9.6 Vic 71.1 75.4 73.2 17.5 20.8 19.2 9.6 9.4 9.5 Qld 70.5 75.3 72.8 17.0 20.4 18.7 10.1 9.7 9.9 WA 71.5 75.6 73.5 17.5 20.6 19.1 9.6 9.6 9.6 SA 69.3 74.2 71.7 16.4 20.0 18.3 10.8 10.5 10.6 Tas 68.8 73.7 71.3 16.3 19.7 18.1 10.2 9.8 10.0 NT 65.8 70.2 67.7 12.6 15.1 13.6 10.0 10.6 10.3 ACT 73.9 77.8 75.9 18.9 21.9 20.5 7.8 7.5 7.7 Socioeconomic quintile Low 68.7 73.8 71.2 16.1 19.7 17.9 10.7 10.4 10.6 Moderately low 69.5 74.6 72.0 16.4 20.1 18.2 10.2 9.9 10.1 Average 69.9 74.6 72.2 16.6 20.1 18.4 10.0 9.8 9.9 Moderately high 71.4 75.9 73.6 17.6 20.8 19.3 9.7 9.1 9.4 High 73.8 77.2 75.5 19.2 21.9 20.6 8.7 8.7 8.7 Remoteness Major cities 71.3 75.6 73.5 17.5 20.8 19.2 9.6 9.4 9.5 Regional 69.6 74.5 72.0 16.5 20.1 18.3 10.3 9.8 10.1 Remote 67.3 72.3 69.5 15.4 18.5 16.8 10.8 11.3 11.0 Australia 70.6 75.2 72.9 17.1 20.5 18.9 9.8 9.6 9.7 Based on remoteness, the proportion of life expectancy lost due to disability ranged from 9.5% in major cities to 11.0% in remote areas. The 26.5% greater burden for remote areas compared to major cities reflected a higher burden per person from most causes but particularly injuries. Trendspast, present and futureThis report presents an analysis of health trends over a 30-year period, based on the past decade and projected trends of health burden if these trends continued over the next 20 years. The average years of healthy life a person can expect to live (HALE) will grow at a slower rate than life expectancy over the next 20 years. If past trends in morbidity and mortality continue, HALE will increase 0.22% annually and life expectancy 0.24% annually. This is partly because declines in mortality will be accompanied by a somewhat smaller decline in time that is lost to disability. 8 The rate of disability will actually decline in most age groups, except for those 80 years and over, where it is expected to increase and thereby offset some of the gains for younger age groups. The growing rate of disability in the oldest age group mostly comes from expected increases in diabetes and neurological conditions. Key implications Key implications of the reports findings include: Ageing of Australias population will result in increasing numbers of people with disability from diseases more common in older ages such as dementia, Parkinsons disease, hearing and vision loss, and osteoarthritis. This will increase demand for services in the home, community care, residential aged care and palliative care sectors. Cardiovascular disease has been overtaken by cancer as the major cause of burden in Australia. This has been largely as a result of programs which have reduced smoking and facilitated the use of therapies to lower cholesterol and blood pressure levels, as well as better treatment of existing cardiovascular disease. It is likely that additional gains could be made through increasing the coverage of these interventions. Cancer is expected to retain its share of total health burden. Age-standardised rates of death and disability are expected to fall, but cancer will remain the largest contributor to the health burden in 20 years time. There is likely to be strong growth in the burden of diabetes over the next 20 years, mostly as a direct consequence of increasing levels of obesity. The disability consequences of increasing obesity will be magnified as fatality rates for people with diabetes continue to decline. This increased survival will mean an increase in the risk of people developing other largely non-fatal but disabling consequences of diabetes such as renal failure and vision loss. Australia is likely to benefit from further efforts towards expanding the range of effective prevention and treatment strategies for all causes of burden, while recognising that the returns for these efforts can take time to be realised. Only in recent years, for example, have smoking-related cancers started to decline as the result of several decades of successful tobacco control programs. 9 1 Introduction 1.1 Purpose The study upon which this report is based is the first complete assessment of the health of Australians in the new millennium and the second study in this country with comparable objectives. The original study, the results of which continue to be used widely in policy and research environments, was conducted by the Australian Institute of Health and Welfare (AIHW: Mathers et al. 1999) and provided a comprehensive overview of disease and injury burden for the year 1996. Increasing demand for a contemporary picture of health status in Australia led, in 2003, to an Australian Government-funded collaboration between the University of Queensland and the Australian Institute of Health and Welfare (AIHW), the aim of which was to update and expand the original work. The objectives of this collaboration were to report on the following: full burden of disease and injury results for the year 2003 by age group, sex and cause projections of disease and injury burden 20 years into the future improved models for attributing disease and injury burdens to risk factors sub-national estimates of burden for state and territory jurisdictions, socioeconomic quintiles, remoteness categories and small areas the burden of disease and injury in Aboriginal and Torres Strait Islander populations. This report presents the main findings of this collaboration and meets the above objectives, except the last which is covered in a separate report. 1.2 Background Changes in demography and technology are placing increasing pressure on the health budgets of developed countries around the world. Mortality and fertility rates have decreased consistently over recent decades, resulting in increases in life expectancy and the proportion of total population alive at old and very old ages (AIHW 2006). In addition, developments in knowledge and medical technology are contributing to a growing demand for health services and, in many cases, to higher costs of providing these services. In Australia and elsewhere, these factors have brought into focus the need for more rigorous debate about how health systems can achieve their dual objectives of maximising health gains for given levels of expenditure and maintaining fair and equitable access to health services. Improving the evidence base that informs this debate is critical if health systems are to be meaningfully held to account. Such an agenda requires contributions from a number of areas, including: detailed assessments of the size and impact of health problems in a population, including information on the causes of loss of health in the population (in terms of both diseases and injury, and risk factors or broader determinants) 10 information on inequalities in health status, health determinants, and access to and use of health services (including prevention and treatment services) information on health expenditure and health infrastructure (a national system of health accounts) detailing the availability of resources for health improvement and the current use of these resources information on the cost-effectiveness of available technologies and strategies for improving health information on current levels of investment in health research and development, and on the opportunities for investment with the greatest likelihood of developing new or improved interventions that best remedy major health problems. This report contributes to the development of such an agenda in Australia by providing a detailed and internally consistent assessment of the incidence, prevalence, duration, mortality and burden for an exhaustive and mutually exclusive set of major diseases and injuries experienced in this country. The burden from these causes is quantified for various subpopulations, risks to health and points in time using a summary measure of population health that combines both fatal and non-fatal health outcomes, and includes comorbidity adjustments to account for individuals who simultaneously experience multiple conditions. This assessment provides an unprecedented level of detail on the magnitude and distribution of health problems in contemporary Australia. Although solutions to these problems are not addressed explicitly in the following chapters, the analyses described encompass a methodology that is increasingly being used in Australia and elsewhere to assess health outcomes both for descriptive purposes and in comparative analyses of the costs and effectiveness of particular health interventions. The report can be regarded, therefore, as an important foundation for further work on improving health system performance in Australia. 1.3 Summary measures of population health Summary measures of population health are measures that combine information on mortality and non-fatal health outcomes into a single number to represent one or more dimensions of health at a population level (Field & Gold 1998). In the past 15 years, there has been a marked increase in interest in the development, calculation and use of summary measures. The range of potential applications includes: comparing health conditions or overall health status between two populations or the same population over time quantifying health inequalities ensuring that non-fatal health outcomes receive appropriate policy attention measuring the magnitude of different health problems using a common currency analysing the benefits of health interventions for use in cost-effectiveness studies providing information to help set priorities for health planning, public health programs, research and development, and professional training (Murray et al. 1999b). Most summary measures fall into one of two broad groups: health expectancies and health gaps. Both groups use time (either lived in health states or lost through premature death and illness) as the unifying currency for combining the impact of mortality and non-fatal health outcomes. Another common feature is the requirement for explicit or implicit choices 11 in their application: mortality-based indicators, for example, exclude considerations regarding non-fatal loss of health; indicators of potential years of life lost ignore deaths beyond an arbitrary age (for example 65 years); and indicators of disability-free life expectancy do not place any positive value on years lived with disability. Health gap measures, in particular, quantify the gap between a populations actual health status and some ideal or reference status. The most widely known example of such a measure, and the one used in this report, is the disability-adjusted life year or DALY. Another measure commonly used in economic evaluations but not in population health status assessments is the Quality Adjusted Life Year (QALY). 1.4 Disability-Adjusted Life Years The DALY was first developed to provide information to support health policy and priority setting at a global level. The concept was developed as part of a comprehensive assessment of global health for the year 1990 in what became known as the Global Burden of Disease or GBD study (Murray & Lopez 1996a, 1996b; World Bank 1993). It has since become synonymous with burden of disease and the terms tend to be used interchangeably. The DALY was originally intended to: allow estimates of health effects to be mapped to causes, either in terms of disease and injury, or risk factors and broader social determinants provide a common measure for estimating population health effects and cost-effectiveness of interventions use common values and health standards for all regions of the world provide a common measure for fatal and non-fatal health outcomes. In this way, the DALY extends the concept of potential years of life lost due to premature death (PYLL) by including equivalent years of healthy life lost by virtue of being in states of poor health or disability. A DALY for a disease or health condition is calculated as the sum of the years of life lost due to premature mortality (YLL) in the population and the equivalent healthy years lost due to disability (YLD) for incident cases of the health condition: DALY = YLL + YLD where YLL = number of deaths x standard life expectancy at age of death and YLD = incidence x duration x severity weight. The loss of healthy life due to health conditions (YLD) requires estimation of the incidence of the disabling health condition (disease or injury) in the specified time period. For each new case, the number of years of healthy life lost is obtained by multiplying the average duration of the condition (to remission or death) by a severity weight that quantifies the equivalent loss of healthy years of life due to living with the health condition or its sequelae. The YLD is as an incidence-based measure, therefore, which captures the future health consequences of new cases of disease and injury that occur in the baseline year (2003 in this study). Such a measure, when combined with YLL, enables the full health loss of different diseases and injuries to be compared and has most application in planning. 12 Alternatively, health loss can be measured by counting it at the age it is lived. This is the prevalent burden or prevalent years lost due to disability (PYLD) and is calculated thus: PYLD = prevalence x severity weight Prevalent burden is useful from a service utilisation or expenditure perspective and measures the amount of disability (but not the fatal burden) being experienced in a population at a point in time. From the perspective of the International Classification of Functioning, Disability and Health (ICF) (see <www3.who.int/icf/icftemplate.cfm>) the YLD measures the impact of a health condition on an individuals functioning, now and into the future. Functioning includes the functional and structural integrity of the human body as well as activities undertaken by people and participation in life situations. Interpreting the DALY The DALY methodology provides a way to link information on disease causes and occurrence to information on both short-term and long-term health outcomes, including activity limitations and restrictions in participation in usual roles, and death. The burden of disease methodology is designed to inform health policy about the prevention and treatment (cure or reduction in severity) of adverse health outcomes. It is not designed to inform policy for the provision of social support or welfare services for people with long-term disability. When using the DALY for the first time, Murray and Lopez sought to make explicit the value choices that they had to make in their application of a summary measure at a global level. For example, they chose to use the same life expectancy ideal standard for all population subgroups across the globe, whether or not their current life expectancy was lower than that of other groups. They also excluded all non-health characteristics (such as race, socioeconomic status or occupation), apart from age and sex, from consideration in calculating lost years of healthy life. Most importantly, they used the same severity weight for everyone living a year in a specified health state. These and other aspects of the DALY are described in further detail in Chapter 2. 1.5 Burden of disease analysis in Australia Since its introduction, burden of disease analysis has been applied in an increasing number of international and national settings; for example, it was used for a period by the World Health Organization (WHO) to inform global health planning (WHO 2002). Burden of disease analysis has a particularly strong history in Australia. The first study by the AIHW assessed the burden of disease and injury in Australia for the year 1996 (AIHW: Mathers et al. 1999). Starting in June 1998, the first study was partly funded by the then Commonwealth Department of Health and Aged Care and was conducted in parallel with a state-level analysis for Victoria by the Victorian Department of Human Services (DHS 1999a, 1999b). Both project teams worked together closely on methods and analyses. This work represented the first attempt to carry out a systematic and comprehensive analysis of over 170 disease and injury categories in this country. It also substantially extended the 13 international work on burden of disease in many areas, as shown by the fact that a number of its methodological advancements were subsequently picked up in the GBD 2000 work at WHO (Mathers et al. 2004). Since then, burden of disease analysis has been undertaken in most jurisdictions throughout Australia, at varying levels of detail. The update of the Victorian Burden of Disease study for the year 2001 (DHS 2005) deserves special mention as a number of disability models and data sources were shared between the researchers working on that project and those working on the present study. This study was conducted in close consultation with relevant jurisdictional stakeholders, and the national and jurisdictional estimates in this report are intended to complement existing estimates from individual State and Territory based burden of disease studies. Because of somewhat different estimating methods and data sources, the jurisdictional estimates in this report may differ somewhat from State and Territory based estimates. This does not mean that one estimate is more correct than the other, but reflects the uncertainties inherent in any analysis which attempts to estimate burden for over 170 conditions. 1.6 Burden in Aboriginal and Torres Strait Islander peoples Findings about Aboriginal and Torres Strait Islander peoples are not covered in this report, the primary focus of which is on the health status of Australians as a whole. This is not a problem for most of the comparisons presented, although special caution should be taken when interpreting the results of Chapter 5 on health differentials, particularly the estimates for remote areas and the Northern Territory. The higher proportion of Indigenous people in these areas explains most of the greater health loss in these areas compared with those where the proportion of Indigenous people is lower. However, the contribution of Indigenous populations to this loss has not been quantified in this report. Readers seeking to know such comparisons are referred to the companion report on the burden of disease and injury in Aboriginal and Torres Strait Islander peoples. 1.7 Structure of report Details of the specific methodological developments of this study are presented in Chapter 2. Chapter 3 provides an overview of the total burden of disease and injury in Australia, by cause, age and sex. Chapter 4 provides estimates of the burden of disease and injury attributable to selected risk factors in Australia. Chapter 5 shows how the burden of disease and injury across Australia varies according to where people live and their socioeconomic status. Chapter 6 presents the past, present and projected burden of disease and injury in Australia and Chapter 7 provides a general discussion of the major findings. Technical notes on the methods used for estimating non-fatal health outcomes and attributing risk are presented in Appendixes 1 and 2, respectively. Annex table 1 summarises the disease and injury categories used and their respective International Classification of Diseases codes. Annex table 2 summarises the primary data sources used to construct the core set of results. Tabulations of the core results are included in Annex tables 3 to 9. More detailed tabulations of the core results are available in Annex tables 10 to 25, which are available on the web at <www.aihw.gov.au/bod>. 14 Readers should note that every attempt was made to identify the best available information in the preparation of this report, and to consult as widely as possible on decisions about methods, assumptions and data sources. For some aspects of the study, however, it was not possible with the resources available to go beyond simple models and assumptions about some key parameters. For many disease models, not all required information was available and analyses drew on information from overseas or expert opinion. In the projections work, trends in disease occurrence were nonexistent for many conditions. The results presented in the following chapters, therefore, represent a complex synthesis of information, judgment and, in some cases, even speculation. It is hoped that further improvements over time in methods, models and data will result in increasing accuracy and certainty in estimates of burden of disease and injury in Australia. The authors at the University of Queensland and the Australian Institute of Health and Welfare welcome suggestions for such improvements. 15 2 Methodological developments This chapter discusses the key methodological considerations that underpin the findings presented throughout the report. Readers who are only interested in these findings can skip to chapters 3 through 7 and return to this chapter at a later time. Those wishing to understand the ways in which the methodological challenges were resolved are encouraged to read on. The chapter begins by outlining some solutions to various methodological issues that are unavoidable in the application of the burden of disease and injury framework, including social value choices, causal attribution and comorbidity. It concludes with a description of the specific methods that were adopted to derive the findings on risks to health, burden across time and differentials in burden. 2.1 Social value choices The burden of disease and injury framework encompasses some obviously normative characteristics (that is, it incorporates certain value judgments about how things ought to be). This is because its main measure (the disability-adjusted life year or DALY) comprises only a selection of all possible parameters that could be used to characterise health, and the numerical weighting given to each parameter implies a judgment about its relative importance to the total measure. These judgments have come to be known collectively as social value choices. While the implications of certain choices over others are important and sometimes contested, as reflected by the growing literature in this area (Anand & Hanson 1997; Reidpath et al. 2003; Williams 1999), such considerations are beyond the scope of this chapter. The purpose here is to provide a brief discussion of the key choice that differs from the previous study. Readers are referred elsewhere for a more in-depth discussion on the merits of the other social value choices (Murray et al. 2002). As mentioned in the previous chapter, the DALY is a health gap measure that requires an ideal against which to quantify the gap between current patterns of mortality and a counterfactual scenario in which all mortality is averted until very old age. The steering committee of the previous Australian Burden of Disease and Injury Study requested that projected life expectancy, based on a cohort life table (which takes into account past trends in mortality) for Australia, be used to define the mortality gap for the purposes of calculating the years of life lost due to premature mortality (YLL). Until then, the standard that had been used in all burden of disease studies was based on the Coale and Demeny West level 26 model life table (Coale & Guo 1989), chosen after observing the highest life expectancy recorded for any nation (82.5 years for women in Japan at the time). It was then assumed that the minimum malefemale biological difference in survival potential was in the order of 2.5 years, but because there was no male schedule with a life expectancy of 80 years, the standard for males was based on the Coale and Demeny West level 25 schedule for females (Murray & Lopez 1996a). The cohort life tables for Australia used in the 1996 study and the standard life tables used in other studies are very similar, and the substitution of one for the other would have had little effect on the final results. This is particularly true for discounted YLL, where the small differences in time lost would have been even further reduced by a time discount rate of 3%, although some differences were observable if undiscounted YLL were compared. For the current study, however, the situation is complicated by the fact that life expectancy in 16 Australia has changed since 1996 (an increase of 0.25 years and 0.3 years per annum for females and males, respectively). If the projected cohort life expectancy were to be used again, the mortality gap would be somewhat different because the projected cohort life expectancy based on changes in mortality rates to 2003 would be different from the old cohort life expectancy, which was based on changes in mortality rates to 1996. While the difference is not great, it does not aid comparisons to have a standard that is continually changing. Thus the current advisory committee has recommended a return to the internationally recognised standard used in most other burden of disease studies. It is worth noting here that the life table for a population that actually achieves the ideal standard (that is, no mortality until age 82.5 in females and 80 in males) would be very different from the standard life table. It is best to view the choice of the standard life table, therefore, as a weighting for age at death, without reference to the properties of the life table used to derive these weights. All other social value choices remain as they were in the previous study (Table 2.1): uniform age weights and a discount rate of 3% were applied, and a combination of disability weights from the original GBD study (Murray & Lopez 1996a) and the Disability Weights for Diseases in the Netherlands study (Stouthard et al. 1997) were used. For some health states, there was no equivalent in either the Dutch or GBD set of weights, or the weights that appear in the published material seemed implausible. In these instances, the weights that were specifically derived for the previous Australian studies were applied. Unfortunately, a study to determine local weights for the range of health states most relevant to Australia was not able to be done. The complete list of weights is available at <www.aihw.gov.au/bod>. Table 2.1: Social value choices used in the calculation of DALYs, 1996 study and present study Choice 1996 study Present study Mortality counterfactual Projected life expectancy based on cohort life tables for Australia in 1996 International standard first reported in Murray & Lopez 1996a Age weighting Uniform Uniform Discount rate 3% 3% Source of disability weights Murray & Lopez 1996a, Stouthard et al. 1997 and locally derived Murray & Lopez 1996a, Stouthard et al. 1997 and locally derived 17 Box 2.1: Interpreting a disability weight To place a value on the time lived in non-fatal health states, health state weights are used to formalise and quantify social preferences for different states of health. Depending on how these weights are derived, they are referred to as disability weights, quality-adjusted life year (QALY) weights, health state valuations, health state preferences or health state utilities. QALY weights are measured as a number on a scale of 0 1, where 0 is assigned to a state comparable to death and 1 is assigned to a state of ideal health. This assignment for the DALY (where 0 = perfect health and 1 = death) is the complement to 1, compared to that used for the QALY, because the QALY measures equivalent healthy years lived, whereas the DALY measures loss of health. Although the disability weights used in DALY calculations quantify societal preferences for different health states, the weights do not represent the lived experience of any disability or health state, or imply any societal value for the person in a disability or health state. Rather, they quantify societal preferences for health states in relation to the societal ideal of good health. Thus, a weight for paraplegia of 0.57 does not mean that a person in this health state is half-dead, that they experience their life as halfway between life and death, or that society values them less as a person compared with healthy people. It means that, on average, society judges a year with blindness (weight 0.43) to be preferable to a year with paraplegia (weight 0.57), and a year with paraplegia to be preferable to a year with unremitting unipolar major depression (weight 0.76). It also means that, on average, society would prefer a person to have a year in good health followed by death than a year with paraplegia followed by death. Society would also prefer to restore a person with paraplegia to good health rather than restore a persons sight if the costs of cure are the same for the two interventions. 2.2 Causal attribution There are two traditions for causal attribution of health outcomes or states: categorical attribution and counterfactual analysis (Mathers et al. 2001). In categorical attribution, an event such as death is attributed to a single cause (such as a disease or risk factor) or group of causes according to a defined set of rules, such as the International Classification of Disease (ICD) system for attributing causes of death (WHO 1992). In counterfactual analysis, the contribution of one or a group of risk factors to disease or mortality is estimated by comparing the current or future disease burden with the levels that would be expected under some alternative hypothetical scenario (referred to as the counterfactual). This study uses both approaches: categorical attribution for attributing burden to diseases and injuries, which is discussed below, and counterfactual analysis for attributing burden to more distal risks to health, which is discussed in a subsequent section. Estimates of burden are typically attributed to a comprehensive set of disease and injury entities (for example ischaemic heart disease or falls). These entities represent the smallest unit of disaggregation in the analysis and are referred to in this report as specific causes or conditions. Each entity is mutually exclusive and belongs to one of a number of broad cause groups, most of which correspond to chapter-level headings of the ICD (for example cardiovascular disease or intentional injuries). Each broad cause group, in turn, belongs to one of three broad clusters: Group I: Communicable, maternal, neonatal and nutritional conditions Group II: Non-communicable diseases Group III: Injuries. 18 Annex Table 1 defines the classifications used in this study in terms of ICD-10 codes, most of which are consistent with the classifications used by WHO in the GBD2000 project (Mathers et al. 2004). A comparison of the ICD-10 list and the one based on ICD-9 used in the previous study is available at <www.aihw.gov.au/bod>. Categorising deaths The ICD has its origins in the preparation of mortality statistics, and standard death statistics use the categorical approach to causal attribution. While any number of conditions may be recorded on a death certificate, the ICD allows for only one to be selected for primary tabulation purposes. This single cause is referred to as the underlying cause of death and is intended to represent the condition, event or circumstances without the occurrence of which the person would not have died. The concept of underlying cause has been central to mortality coding and comparable international mortality reporting over the 100-year period that the ICD has been used for such purposes. Box 2.2: Death registration in Australia Registration of deaths in Australia is the responsibility of the state and territory Registrars of Births, Deaths and Marriages. Information on the cause of death is supplied by the medical practitioner certifying the death or a coroner. Other information about the deceased is supplied by a relative or other person acquainted with the deceased or by an official of the institution where the death occurred. Registration of death is a legal requirement in Australia, and compliance is almost complete. The information is provided by the Registrars to the Health and Vitals Unit at the Queensland office of the Australian Bureau of Statistics (ABS) for coding and compilation into national statistics. The ABS began automated coding of death certificates using software known as the Mortality Medical Data System (MMDS) in 1997 and has made available multiple causes of death data coded in ICD-10 for all years since that time. Before 1997, only underlying cause of death data are available. The MMDS was developed by the National Center for Health Statistics in the United States of America to facilitate the coding of all causes of death reported on death certificates, and the designation of the underlying cause of death according to ICD criteria. The availability of an unambiguous set of rules, such as can be found in the ICD, does not alter the fact that the accuracy of the information to which these rules are applied is dependent on several factors: the availability and quality of the clinical evidence at the time of certification; the thoroughness and diligence with which physicians and coroners record this information on the death certificate; and the quality of the system used to transcribe information from death certificates and translate this information to ICD codes. Australia is regarded as having a high-quality system of registration by international standards and this is reflected by one measure of quality, the proportion of total deaths coded to non-specific underlying causes of death. The small amount of non-specific coding that does occur is confined mainly to the ill-defined sections of the cardiovascular disease, cancer and injury chapters, with only a very small proportion of deaths being coded to the general signs and symptoms chapter. However, with the exception of a few studies on sensitivity and specificity in relation to specific conditions, relatively little is known about the frequency with which Australian doctors attribute the correct underlying cause to the majority of deaths. It is likely that accuracy varies with the location of the death (for example in an 19 institutional setting versus at home), but the assumption that inaccuracies tend to cancel each other out at the population level is largely speculative and is an area deserving of further research. While this study largely followed the ICD concept of underlying cause in the categorisation of deaths, in some cases deaths were reallocated to more specific or different categories to ensure consistency with the estimates for years lost due to disability (YLD). For example, the proportion of liver cancer and liver cirrhosis mortality that is attributable to hepatitis was redistributed to the hepatitis B and hepatitis C categories in the core results. Similarly, data on the underlying cause of renal failure from the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA) was used to redistribute renal failure deaths to nephritis & nephrosis, diabetes mellitus, injuries, congenital conditions, cancers and infectious diseases. It is important to note that for many conditions there is a difference between the number of deaths attributed to the disease and amount of excess mortality that occurs in prevalent cases of the disease. This is often due to comorbidity and the fact that diseases may cluster in people exposed to the same risk factors that also affect the risk of dying from other causes. Examples of this are schizophrenia, where part of the excess risk is due to the high prevalence of smoking and diseases associated with the usually lower socioeconomic status of people with chronic and severe mental disease; and cardiovascular disease, where the main lifestyle risk factors also increase the risk of dying from diabetes and some cancers. For the overall cause of death structure presented in this report, recorded underlying causes of death were used, subject to the redistribution algorithms discussed below. In the disease modelling discussed in Appendix 1, however, best available estimates of excess mortality were used in order to derive the most accurate estimates of disease duration. Redistributing non-specific causes of death In keeping with established burden of disease methods, attempts were made to remove possible distortions to the reported overall cause of death structure by reallocating deaths with certain codes known to be problematic to valid and specific underlying causes of death. The rationale for not taking reported causes of death at face value is that policy objectives are best served by information that is corrected for possible sources of systematic bias. By world standards, the extent of distortions in cause of death information in Australia is small (around 610%, depending on what codes are included in this definition). In some areas, however, there are obvious anomalies that require specific attention. Murray and Lopez (1996a) were the first to provide convincing evidence that a significant and varying proportion of ischaemic heart disease deaths are coded in many countries to ill-defined codes such as heart failure. They argued that this, in part, helps to explain the French paradox in which mortality from ischaemic heart disease in France is comparatively low despite high levels of exposure in the French population to risks known to be associated with this disease. In fact, many ischaemic heart disease deaths are most probably being coded to heart failure or other equally non-specific cardiovascular causes. Policy is better served by correcting this misclassification error. Various redistribution algorithms to correct non-specific cause of death coding have been developed in response to these considerations throughout the world. In the previous Australian Burden of Disease and Injury Study, for example, a number of decisions were made about what to do with problem coding based on local considerations regarding the cause of death collection system at the time. One of the guiding principles of the present 20 study was not to change past decisions such as these unnecessarily, unless there were compelling reasons to do so, such as new evidence. In the period since the completion of the previous study, the vital registration system in Australia has changed in two significant respects. First, the ABS moved from the coding of mortality using version 9 of the ICD to version 10 in 1997. Second, at the same time, the ABS implemented automated coding of mortality statistics using software developed in the United States. The use of this system allows multiple cause of death coding (that is, coding of the underlying cause of death as well as all other associated causes recorded on the death certificate by the certifying medical practitioner), significantly enhancing the amount of information on official mortality files (see Box 2.2). To facilitate an assessment of the impact of these changes, the ABS retained the old system of coding for a period of two years, thus providing an invaluable resource for researchers trying to assemble comparable data on causes of death in Australia over time. The availability of this additional information has allowed known problematic codes to be examined in much greater detail than has been possible in the past. It has also allowed the identification of some areas where possible new coding anomalies are emerging. The most glaring of these is the much greater number of deaths being coded to pneumonia under the new system. In the seven years to 1997, there were around 1,700 deaths from this condition annually. With the advent of automated coding, this number has risen to around 3,300 deaths annually. Such dramatic shifts are not due to changes in underlying disease frequency, but are rather an artefact of a greater preference under the new system to code deaths to this category (manual coders, on the other hand, were probably more likely to attribute an underlying chronic condition). Rather than correcting for this large discontinuity, which would then need to be repeated in the future to ensure comparability, the coding for these deaths was left unchanged. This explains the rapid rise in lower respiratory tract infections from 1993 to 2003 described in Chapter 6. The other area where a discontinuity of this magnitude is apparent is the greater preponderance under the new system to code deaths due to external causes to exposure to unspecified factor (ICD-10 code X59). Analysis of the dual-coded data revealed that the majority of these deaths in the elderly were in fact coded to falls under the old system. In this instance, an additional allocation algorithm was applied whereby deaths coded to this category (around 0.6% of all deaths) were reallocated to falls if they also had a fracture code in the multiple cause of death data (AIHW: Cripps & Carman 2001). This approach was also used for unspecified septicaemia (ICD-10 code A419), whereby deaths in this category (again, around 0.6% of all deaths) were reallocated to nephritis & nephrosis if they also had an acute renal failure code (ICD-10 code N17). Another area where the new system may be in error is in the assigning of inappropriate underlying causes where another code would have been more informative. For example, in the 7-year period to 2003, 548 deaths were coded to tobacco dependence as an underlying cause. Likewise, 885 deaths were coded to obesity and 2,072 to hypercholesterolaemia and dyslipidaemia over the same period. These codes are most appropriately regarded as risk factors for more specific underlying disease processes and preferably should not be used in primary underlying cause of death tabulations. The number of deaths coded to these categories is likely to substantially underestimate the true mortality attributable to these risks (which is estimated in this report using very different methods, as discussed in Appendix 2). Deaths coded to tobacco dependence were therefore redistributed across lower respiratory tract infections, mouth and oropharynx cancers, lung cancer, ischaemic heart disease, stroke, other cardiovascular disease, chronic obstructive pulmonary disease (COPD) 21 and other chronic respiratory diseases based on a probability analysis of multiple-cause information over the period 1997 to 2003. Obesity was allocated to other endocrine & metabolic disorders and the other two codes (about 300 deaths per year) to ill-defined cardiovascular disease, which was ultimately reapportioned to specific cardiovascular diseases (largely ischaemic heart disease). The probability approach using multiple causes of death information was also applied to two other categories: ill-defined nutritional (ICD-10 codes E64 and E639) and essential hypertension (ICD-10 code I10). The first (representing 0.1% of all deaths) was redistributed across lower respiratory tract infections, other endocrine & metabolic disorders, dementia, other chronic respiratory diseases, and nephritis & nephrosis. The second (accounting for 0.2% of all deaths) was redistributed across all specific cardiovascular diseases. Useful though it is, multiple cause of death information provides no new insights about three known problematic areas: ill-defined cancer, ill-defined injury and ill-defined non-injury deaths. It turns out that these deaths are assigned non-specific codes precisely because there is very little other information of relevance either on the death certificate or through coronial investigations (in the case of external causes) to make a more accurate determination. In the previous study these causes were allocated to specific cause groupings on a pro-rata basis on the assumption that the proportional distribution within these groupings reflected the most likely probabilities for causal attribution to a specific cause. There is no new evidence to alter these decisions. These causes and the cause groupings to which they were proportionately redistributed are listed in Table 2.2. Table 2.2: Ill-defined causes of death and specific cause groupings to which they were allocated on a pro rata basis Ill-defined cause(a) Per cent of all causes Specific cause groupings(a) Ill-defined malignant neoplasms(b) 1.92 All specific cancer sites Uterus cancerunspecified(b) 0.04 Cervix cancer Corpus uteri cancer Other anaemias 0.06 Haemolytic anaemia Other non-deficiency anaemia Ill-defined non-injuries (i.e. diseases)(a) 0.39 All specific non-injury causes Ill-defined unintentional accidents (no fracture)(b) 0.11 All specific unintentional injury causes (a) Refer to Annex Table 1 for the ICD-10 codes that correspond to these cause categories. (b) Denotes a redistribution decision derived from the previous Australian Burden of Disease and Injury Study. Based on an assessment of cause of death statistics in Australia over a 25-year period, including the seven years of multiple causes of death information to 2003, a number of redistribution decisions were retained from the previous study, largely because there was no compelling reason to do otherwise. The list of these causes and the corresponding specific causes to which they are proportionately redistributed is outlined in Table 2.3. 22 Table 2.3: Ill-defined causes of death and percentage allocation to specific causes Allocation to specific causes (%)(a) Ill-defined cause(a) Deaths in ill-defined causes as % of all causes Chlamydia Other STD Hepatitis B Hepatitis C Birth trauma & asphyxia Low birth weight Neonatal infections Other perinatal Type 1 diabetes Type 2 diabetes Nephritis & nephrosis Peptic ulcer disease Cirrhosis of the liver Ischaemic heart disease Inflammatory heart disease Hypertensive heart disease Other cardiovascular disease Suicide &self-inflicted injuries Homicide & violence Road traffic accidents Poisoning Falls Fires/burns/scalds Drowning Other unintentional injuries Pelvic inflammatory disease(b) 0.01 60 40 Hepatitis sequelae(b) 0.07 50 50 Neonatal deaths coded to maternal condition(c) 0.26 16 56 6.6 21 Unspecified diabetes mellitus(b) 1.32 Males 014 years 100 Males 1524 years 89 11 Males 2534 years 79 21 Males 3544 years 33 67 Males 4554 years 9.8 90 Males 5564 years 5.5 95 Males 65+ years 2.8 97 Females 014 years 100 Females 1524 years 75 25 Females 2534 years 56 45 Females 3544 years 42 58 Females 4554 years 16 84 (continued) 23 Table 2.3 (continued): Ill-defined causes of death and percentage allocation to specific causes Allocation to specific causes (%)(a) Ill-defined cause(a) Deaths in ill-defined causes as % of all causes Chlamydia Other STD Hepatitis B Hepatitis C Birth trauma & asphyxia Low birth weight Neonatal infections Other perinatal Type 1 diabetes Type 2 diabetes Nephritis & nephrosis Peptic ulcer disease Cirrhosis of the liver Ischaemic heart disease Inflammatory heart disease Hypertensive heart disease Other cardiovascular disease Suicide & self-inflicted injuries Homicide & violence Road traffic accidents Poisoning Falls Fires/burns/scalds Drowning Other unintentional injuries Females 5564 years 8.1 92 Females 65+ years 4.7 95 Hypertensive heart and renal disease(b) 0.10 50 50 Heart failure(b) 2.10 Persons 04 years 100 Persons 529 years 75 25 Persons 3044 years 70 25 5 Persons 4559 years 70 5 25 Persons 60+ years 60 10 30 Ill-defined cardiovascular conditions(b) 0.94 Persons 059 years 75 25 Persons 60+ years 80 20 Gastric haemorrhage(b) 0.24 50 50 Road traffic accidentsintentundetermined(b) 0.00 Persons 014 years 100 (continued) 24 Table 2.3 (continued): Ill-defined causes of death and percentage allocation to specific causes Allocation to specific causes (%)(a) Ill-defined cause(a) Deaths in ill-defined causes as % of all causes Chlamydia Other STD Hepatitis B Hepatitis C Birth trauma & asphyxia Low birth weight Neonatal infections Other perinatal Type 1 diabetes Type 2 diabetes Nephritis & nephrosis Peptic ulcer disease Cirrhosis of the liver Ischaemic heart disease Inflammatory heart disease Hypertensive heart disease Other cardiovascular disease Suicide & self-inflicted injuries Homicide & violence Road traffic accidents Poisoning Falls Fires/burns/scalds Drowning Other unintentional injuries Persons 15+ years 90 10 Fallsintent undetermined (b) 0.01Persons 014 years 100 Persons 15+ years 90 10 Poisonin gintent 0.04Persons 014 years 100 Persons 15+ years 90 10 Burnsintent undetermined (b) 0.00Persons 014 years 100 Persons 15+ years 90 10 Drownin gintent 0.01Persons 014 years 100 Persons 15+ years 90 10 Other accidentsintent 0.00Persons 014 years 100 Persons 15+ years 90 10 (a) Refer to Annex Table 1 for the ICD-10 codes that correspond to these cause categories. (b) Denotes a redistribution decision derived from the previous Australian Burden of Disease and Injury Study (c) Denotes neonatal deaths coded to maternal conditions (ICD-10 codes P00P02) and subsequently redistributed back to neonatal causes based on an analysis of dual-coded data. 25 Alternative categories In order to present the burden for mutually exclusive categories, decisions had to be made on how to classify sometimes closely linked conditions while still adhering to ICD rules. Chapter 3, however, presents alternative calculations of the burden (Table 3.20) due to certain disease entities that otherwise are split across a number of categories in the main disease and injury tabulations. The three entities are intellectual disability, renal failure and vision disorders, although other groupings are also possible (for example heart failure). Underlying causes of intellectual disability are various and include Down syndrome, central nervous system defects, birth trauma, low birth weight, infection, injury, brain tumours, chromosomal causes, epilepsy and autism. Renal failure can be attributed to diabetes, some cancers, congenital conditions and injury. Alternative calculations are also presented for diabetes and depression & anxiety because these conditions are themselves risk factors for other causes of disability. The alternative estimate for diabetes includes the proportion of burden from ischaemic heart disease and stroke that is due to this disease. Likewise, for depression & anxiety the proportions of ischaemic heart disease and suicide caused by this condition are attributed. A new approach in this study, also, is that suicide is attributed to a range of mental and substance use disorders rather than to depression alone. These alternative calculations appear under the relevant disease or injury group subheading. 2.3 Comorbidity and health It is not uncommon for two or more conditions to occur simultaneously in a person, either by chance or because the conditions are related to each other. This is referred to as comorbidity. Independent comorbidity is the situation where the probability of having two or more conditions simultaneously equals the product of the probabilities for having each of the conditions. Dependent comorbidity, on the other hand, refers to the situation where the probability of having two or more diseases is greater than the product of the probabilities for each disease, reflecting common causal pathways (for example common risk factors causing both diabetes and heart disease) and also that one disease may increase the risk of another. Both types of comorbidity are problematic for burden of disease estimation because the available disability weights are almost exclusively derived for a condition as it exists independently from other conditions. Little attention has been directed towards estimating weights for comorbid (or coexisting) conditions due to the enormity of the task. The severity of health states associated with two or more conditions in combination may not simply be the sum of the disability weights for each of the conditions. In many cases it is likely to be less than the sum, but in some cases there may be exacerbating effects on health of having the combined set of conditions. For example, the experience of symptomatic grade 2 osteoarthritis of the hip and severe vision loss together is probably not as disabling as the addition of the two weights for these health states (0.14 and 0.43, respectively). The experience of the latter with profound deafness, however, may be equal to or even more disabling than the summation approach would suggest. In contrast to the GBD 1990 study, an attempt was made in the original Australian studies to accommodate this phenomenon by adjusting the disability weights for the 21 most common non-fatal conditions of older age (for example hearing loss, osteoarthritis, heart conditions, and diabetes). A multiplicative model was used to estimate weights for comorbid conditions, 26 and the change in total weight deducted from the weight for the milder of the conditions (see Box 2.3). Mental health problems are less prevalent at older ages, apart from dementia, and no attempt was made to adjust for mentalphysical comorbidities, although comorbidity between mental disorders was accounted for. A key assumption in the implementation of this adjustment procedure was that the prevalence of a set of comorbid conditions is equal to the product of the individual prevalences of these conditions. In other words, dependent comorbidity was not considered. More recent work as part of the GBD 2000 study, however, suggests that dependence is important and has a non-trivial impact on final results (Mathers et al. 2006). As a result, it was decided to incorporate the empirical evidence, limited though it is, on disease dependence into the overall corrections for comorbidity. Box 2.3: Combining disability weights The simplest approach to estimating the disability weight for the combined conditions 1 and 2 is to assume that the health state valuations (1 disability weight) are multiplicative, so that the combined weight is more severe than the weight for either condition on its own but less than if they were simply added together, and remains bounded by 0 and 1. The disability weight for the combined conditions 1 and 2 is given by: DW1 +2 = 1 − (1 −DW1)Χ (1 −DW2 )This formula can be generalised to deal with more than two causes as follows: = − Π( − )i total i DW 1 1 DWwhere Π denotes the product operator.In the original Australian studies, this method was used to derive a composite weight for comorbid conditions. In the case of two conditions, the weight for the most severe condition remained unchanged, while the weight for the milder condition was deemed to be the balance of the composite weight minus the weight for the more severe condition. For example, if a person has symptomatic grade 2 osteoarthritis of the hip or knee (0.14) and severe vision loss (0.43), the composite weight for both conditions is 0.51 and the adjusted weight for the osteoarthritis is 0.08. In the current study the disability weights are proportionately reduced for each comorbid state. The approach taken in this study was to determine the numbers of people for every combination of causes of ill-health measured by the major Australian health surveys and in the National Hospital Morbidity Database. While none of these data sources contained information on every cause of interest, each overlapped in the causes they did provide information on, at least to some degree. This allowed comorbidity to be simulated across the full range of causes by deriving conditional probabilities on causes common to two or more surveys and generating an artificial cohort of people based on these probabilities. The assumption was that the correlations observed in self-report surveys and hospital diagnoses are reasonable proxies for the co-occurrence of disability in these samples, even though these data sources may not accurately reflect the actual levels of disease at the population level. Unlike the previous study, this study did not incorporate a severity hierarchy of the disability weights by causes. Instead, a proportional downward adjustment was made to the disability weight of each coexisting cause. The proportion used to deflate individual 27 disability weights was the total adjusted disability weight divided by the total unadjusted disability weight for each cause and all possible combinations. A further consideration that has not been explicitly addressed in previous work is that when a disability weight changes with advancing age (due to comorbidity corrections or for some other reason), incident YLD should be calculated to incorporate these changes. In other words, if the duration of a condition is 20 years, incident YLD should be calculated using the disability weight that is relevant to each age above the age of incidence until the 20-year duration has been reached, rather than using the weight at the age of incidence for the whole 20-year period. This correction was implemented in the present study. 2.4 Risks to health Reliable and comparable assessments of the impact on population of exposure to health risks are fundamental to prevention and health promotion activities. Until relatively recently, health risk assessment has been conducted in the context of the methodological traditions of individual risk factors, with little regard to achieving consistency between these traditions when combining results. In the original Australian study, for example, the criteria for evaluating the scientific evidence on prevalence, causality and hazard size varied greatly among the 10 health risks assessed, resulting in lack of comparability between the estimated population health impacts of these risks. Techniques for attributing outcomes to health risks have advanced considerably in recent times, particularly through the contribution of the Comparative Risk Assessment (CRA) project. This was a large-scale effort by international panels of experts under the direction of the World Health Organization (WHO) to collect the most up-to-date information on the prevalence of exposure to health risks and the relationship between these exposures and health outcomes. WHO dedicated its 2002 World Health Report to describing the results of this effort (WHO 2002), and subsequently published a two-volume book containing detailed information on each of the 22 health risks covered by the project (Ezzati et al. 2004a, 2004b). The key advances of the CRA approach over previous attempts to attribute burden to health risks are: 1. A consistent theoretical framework that uses the hypothetical minimum as the counterfactual against which burden due to a risk is calculated. 2. Inclusion of continuous risk variables that previously were categorical in nature, that is, taking into account the full range of risk from elevated blood pressure, serum cholesterol, body mass index (BMI) or inadequate fruit and vegetable intake rather than defining thresholds for hypertension, hypercholesterolaemia, underweight/obesity and low fruit and vegetable consumption. 3. A more systematic review of the international literature on the impact of risk factors on health outcomes, including estimates of relative risk for a unit of increase in continuous risk factors. 4. A theoretical framework and provisional methods for estimating the joint effects of multiple risks to health. 28 Explicit counterfactualsEstimating the health risks associated with exposure to a particular hazard in a population is typically undertaken with reference to an alternative or counterfactual distribution of exposure (for example exposed versus not exposed). While different counterfactual distributions may be used for different purposes (Murray and Lopez (1999) identify at least four of potential interest), an important contribution of the CRA project was to seek consistency in the definition and use of this distribution across each of the 22 risks analysed. In burden of disease and injury studies, the counterfactual of greatest relevance to the question How much of this health outcome is due to that exposure? is the theoretical minimum risk distribution. This is defined as the distribution of exposure that would yield the lowest possible risk in a population (for example zero tobacco use) and is useful for determining how much of current burden is due to past exposure to a particular hazard (the light grey area of Figure 2.1). This is distinct from intervention analyses, which are typically interested in how much future burden could realistically be avoided by shifting current exposure through the implementation of a particular intervention (various scenarios depicted in the dark grey area of Figure 2.1). Figure 2.1: Conceptual model for health risk assessment which identifies unavoidable burden due to past exposures (light grey), avoidable burden due to current and future exposure (dark grey) and burden unrelated to risk (mid-grey at bottom). c d Unavoidable Disease Burden Time T 0 Past Future T0% 25% 50% 75% 100% (Theoretical minimum) a b Exposure reduction at T0 29 While simple enough to operationalise in the context of hazards for which absence of exposure is indeed the lowest possible risk, the concept of no exposure is problematic when lack of exposure is not meaningful, as is the case for blood pressure, cholesterol and body mass. Before the CRA project, this issue was avoided by the categorisation of these hazards into normal and abnormal (for example hypercholesterolaemia, hypertension, overweight or obesity). Although relevant from a clinical management perspective, this approach is likely to underestimate the population-level attributable burden; even though the elevation in risk at levels of exposure below these cut-points may be small, the large numbers of people at these levels contribute substantially to total population-level risk. The approach advocated by the CRA researchers was to respect the continuous nature of these hazards by assessing risk across the full distribution of exposure experienced by a population. This meant defining theoretical minimum distributions even for hazards for which lack of exposure is not meaningful, which they did by drawing on evidence from very low-risk populations in the literature (Ezzati et al. 2003). Joint risk attribution Another area where the CRA project made an important contribution was the joint attribution of risks. Health risk assessment before this project typically provided information about burden attributable to a hazard in isolation from other hazards. The difficulty with this approach is that if several analyses are added together it can appear as if more than 100% of total burden for any one disease or injury is being accounted for by the hazards in combination. This is not an error in the individual risk attribution method itself but rather it is an issue of interpretation. Individual risk attribution analyses should not be added together, although this can be a difficult message to convey, particularly when they are presented together. Estimating the joint effects of multiple risks is complex in practice for several reasons. First, some of the effects of the more distal factors (for example physical inactivity) are mediated through more proximal factors (for example via high BMI and from BMI via high blood pressure). Estimating the joint effects of more distal and proximal factors requires knowledge of independent hazards of the distal ones and the amount of risk mediated through proximal risk factors. Second, the hazard due to a risk factor may depend on the presence of other risk factors (effect modification). Third, there may be correlation between exposures to various risk factors, because they are affected by the same distal factors and social dynamics. The approach used to estimate joint population attributable fractions (PAFs) in this study is based on methods developed for the CRA, in which the assumption is made that health risks are biologically independent and uncorrelated. This is, of course, an over-simplification, as some risks are not biologically independent (for example physical inactivity and BMI), and various exposures are highly correlated (for example smokers also tend to be drinkers). However, it allows the joint PAF for n number of risks to be expressed as:joint PAF 1 (1 ) 1 i n i = − Π − PAF= where PAF i is the PAF of individual risk factors.The second term in the right-hand side of this equation (that is, the product of all [1 PAF i]terms) is the fraction of burden not attributable to any of the n risk factors. One minus thisterm is the fraction attributable to the combined effects of the n risk factors.30 For instance, inadequate intake of fruit and vegetables and high BMI increase the risk of colon cancer. Assuming there is no dependence or correlation between these two risks, if the PAF for fruit and vegetable intake is 0.20 and the PAF for BMI is 0.10, the burden attributable to the two risks equals 1 ( 1 0.2) x ( 1 0.1 ) = 1 0.8 x 0.9 = 0.28. Epidemiological studies on the effects of high BMI, physical inactivity, and low fruit and vegetable consumption on cardiovascular disease risk have illustrated some attenuation of the effects after adjustment for more proximal factors (for example blood pressure or cholesterol) (Berlin & Colditz 1990; Blair et al. 2001; Eaton 1992; Gaziano et al. 1995; Jarrett et al. 1982; Jousilahti et al. 1999; Khaw & Barrett-Connor 1987; Liu & Manson 2001; Manson et al. 1990; Rosengren et al. 1999; Tate et al. 1998). This attenuation confirms that some of the hazard of the more distal factors operates by increasing levels of risk in factors closer in the causal pathway to the disease. The attenuation varies among studies but is consistently less than one-half of the excess risk (that is, RR - 1) of the more distal factors. An upper bound of 50% is used in this study as the proportion of the excess risk from BMI, physical activity and fruit and vegetable intake that is mediated through proximal factors that are themselves among the risks being analysed. For example, if the relative risk of BMI for diabetes is 4 for a particular level of BMI exposure, for the joint effects calculation a relative risk of (4 1) x 0.5 + 1 = 2.5 is used to calculate the PAF that eventually feeds into the equation on the previous page. Joint risk factor estimates for cardiovascular disease are not very sensitive even to large variations in this assumption of attenuation (Ezzati et al. 2004a). The burden attributable to both child sexual assault and intimate partner violence is estimated in this study for the first time. Evidence suggests that girls who experience child sexual abuse are more likely than non-abused girls to experience intimate partner violence (Mouzos & Makkai 2004). In the joint effects analysis for these exposures, the burden due to child sexual abuse and intimate partner violence is calculated as the sum of the PAFs for exposure to child sexual abuse only, exposure to intimate partner violence only, and the combined state of exposure to both risks. 2.5 Past, present and future burden Forecasts about the future play an important role in shaping public policy. For example, an important consequence of economic development has been improvements in health, particularly among the elderly. Better health, in turn, has led to greater economic development and more people surviving to old age. Together with decreasing fertility, this has contributed to population ageing. There is increasing analysis being undertaken in relation to the long-term sustainability of public finances in the context of these widespread demographic trends across the developed world. Under the Charter of Budget Honesty Act 1998, the Australian Government is required to prepare an Intergenerational Report (IGR) that assesses the long-term sustainability of current Government policies over the next 40 years, and to take account of the financial implications of demographic change. The first IGR was released on 14 May 2002 as part of the 200203 Federal Budget (Budget Paper No. 5) and considered future health care costs based on expected demographic trends and projected Australian Government expenditure on health services, represented as a proportion of gross domestic product (GDP), for the period 2002 to 2041. Likely trends in disease occurrence were not explicitly accounted for in the IGR as the analytical projections were based on historical trends in major health expenditure program 31 groupings (medical benefits, pharmaceutical benefits and hospitals) at selected ages. It is optimistic to assume that simply because underlying changes in disease occurrence were embedded within the historical data on expenditure that they are therefore plausibly reflected in these analyses. An analysis that explicitly takes account of changes in both disease occurrence and per unit expenditure at the level of individual diseases is likely to provide much firmer ground upon which to base estimates of future health expenditure. Common to both approaches, of course, is the assumption that the rate of change in policy responses to emerging problems in the future is consistent with the rate observed in the historical period upon which the projections are based (that is, business as usual). If these dynamics change, expectations with regard to the future will consequently change. One objective of the present study was to address the need for comprehensive health projections in Australia by analysing the most likely changes in burden of disease and injury to the year 2023. The past is a good (but far from perfect) predictor of the future and an important by-product of such work is a comprehensive analysis of past trends in disease occurrence. To pre-empt the inevitable requests for information on the past, this part of the study was extended to include back-casting of disease burden as well. This has the logical appeal of ensuring consistency between estimates of past, present and future disease burden. More importantly, it may limit the potential for misinterpretation should people compare these current and future burden estimates with results based on alternative methods. The inevitable comparison that people will make between the results presented in this report and those of the previous Australian Burden of Disease and Injury Study should be regarded in this light. Australia has an excellent vital registration system by international standards and, with few exceptions (for example pneumonia), observable trends in vital events over time are arguably the most reliable and consistently recorded information on changes in the frequency of diseases and injuries that result in death. Previous work (Barendregt et al. 2003) has shown that the complete epidemiology of a disease is ultimately a function of only three parameters: incidence (the hazard of getting the disease), remission (the hazard of being cured from having the disease) and case-fatality (the hazard of dying as a consequence of having the disease). For most chronic diseases, cause-specific mortality is influenced by only two of theseincidence and case-fatalitywith remission having little if any role. It follows, therefore, that any epidemiological parameter of interest for a chronic disease can be back-cast from a point in time for which the complete epidemiology of that disease is known simply by making assumptions about the relative contribution of incidence and casefatality to the observed changes in mortality. This idea also applies to projections, providing one is willing to make predictions about cause-specific mortality into the future. Since it has already been argued that cause-specific mortality is a reliable and consistently recorded source of information on changes in disease frequency in many cases, cause-specific mortality is a sound starting point for projecting the epidemiology of a disease. Other approaches that are based on predicting incidence from risk factors may have more intuitive appeal but are more tenuous as they involve multiple assumptions about diseaseexposure relationships and future exposure trajectories. The methods used in this study involved a number of separate analytical or computational steps. A brief outline of the overall approach is presented below. More complete details are provided in subsequent sections of the report as indicated. 1. Baseline models for over 170 diseases and injuries for Australia in 2003 were developed as part of the core set of analyses for the present study. Appendix 1 discusses each of these models in detail. 32 2. Trends in observed cause-specific mortality over the period 1979 to 2003 were analysed and projected into the future using a combination of regression techniques. 3. For mostly fatal conditions, each baseline disease model was extrapolated backwards and forwards in time based on assumptions about the relative contribution of incidence and case-fatality to changes in mortality. Baseline models for mostly non-fatal conditions were extrapolated based on assumptions about changes in incidence only. The complete epidemiology of each was then estimated separately in a fully dynamic model that accounted for changes in all-cause mortality as well as changes in incidence and case-fatality (where appropriate) so that incidence, prevalence and duration by age, sex and cause was described over the past as well as into the future. 4. Absolute numbers of incident and prevalent cases were derived by applying the rates from the above analyses to the ABS Series 8 projection series population estimates (ABS 2003d). This series assumes a high net overseas migration of 125,000 annually, constant improvements in life expectancy (low mortality assumption), and a total fertility rate declining to 1.6 by 2011 and then remaining constant. Incident and prevalent YLD for each disease were calculated for non-baseline periods by applying durations and extrapolated numbers of incident and prevalent cases from the dynamic model to disability weights that were corrected for probabilities of comorbidity in 2003. Years of life lost (YLL) for non-baseline periods were calculated directly from observed deaths in the past and projected deaths into the future. Mortality trends and projections Observed all-cause mortality rates for the period 1979 to 2003 were extrapolated into the future using simple log-linear Poisson regression. Cause-specific mortality data for the same period were then collapsed into 51 clinically meaningful conditions, or groups of conditions. Multinomial logistic regression was used to model changes in the contribution of each group as a proportion of all-cause mortality, with changes in absolute levels of all-cause mortality expressed as the natural log of the rate per unit of population. These models were used to predict the future cause-specific structure of mortality based on projected all-cause mortality rates. Separate analyses were done for each age group and sex. Among the causes analysed, cardiovascular disease, cancers, chronic obstructive pulmonary disease (COPD), diabetes, alcohol-related conditions, road traffic accidents, falls, suicide and homicide showed significant mortality trends. The apparent trend in dementia mortality was ignored because: (a) there has been a shift in coding practices with more deaths being attributed to dementia; (b) the prevalence data from international epidemiological studies showed no clear change over time; (c) the case-fatality was unlikely to have changed much over time as there are no effective life-saving interventions. Incidence and case-fatality Mortality trends for cancers, COPD, diabetes, alcohol-related conditions, road traffic accidents, falls, suicide and homicide were assumed to be fully due to changes in incidence. Incidence trends for these causes were therefore adjusted to reflect changes in mortality over the projection period, with case-fatality being held constant. Findings from Unal et al. (2004) suggest that 58% of the drop in cardiovascular mortality observed in England and Wales was due to a drop in incidence and the remaining 42% due to a reduction in case-fatality. The 33 same proportions were assumed to apply in this study to all cardiovascular disease over the projection period. Changes in the diagnostic criteria for Type 2 diabetes in surveys and a paucity of representative survey data meant that there was no direct measurement of trends of Type 2 diabetes in Australia from which to project the incidence of this disease. Body mass index (BMI, defined as body weight in kilograms divided by the square of height in metres), overwhelmingly the main risk factor for Type 2 diabetes, however, has been measured consistently at various points over recent time. The approach taken in this study, therefore, was to translate historical trends in BMI into expected changes in diabetes incidence following the risk attribution methods described in the WHO Comparative Risk Assessment project. Haby and colleagues (2006) analysed trends in BMI using data from five measurement surveys: the three National Heart Foundation Risk Factor Prevalence studies in the 1980s, the National Nutrition Survey of 1995 and the AusDiab study in 1999 and 2000. Projected mean BMI by age group and sex was derived from Haby and colleagues regression model of the mean of log-transformed BMI values on age, birth cohort and sex. Similar techniques were applied to the standard deviations of BMI values so as to fully describe the expected change in the distribution of this risk into the future (a change which can be characterised as a broadening of the distribution in the tail towards the highest BMI values rather than at the other end of the distribution with low values). The population-level risk of diabetes is simply the area under the curve represented by the distribution of BMI multiplied by the relevant relative risk of developing diabetes at each level of BMI. This is easiest to derive using integration techniques. Proportional changes in the size of this area over time represent changes in the incidence of diabetes resulting from changes in BMI. Ni Mhurchu and colleagues (2006) undertook a meta-analysis of results from the Asia-Pacific Cohort Study collaboration and report the relative risk of developing diabetes for each unit increase in BMI by age and sex. Using these relative risks and the predicted BMI distributions derived above, changes in the incidence of diabetes were estimated over the projection period. For consistency with CRA methods, a theoretical minimum distribution of BMI (mean of 21 and standard deviation of 1) was incorporated into the calculations, below which no excess risk of diabetes was assumed. Information on trends in case-fatality rates amongst people with diabetes is scarce. In the absence of such information, an assumption was made that at least half the mortality in these people is due to vascular causes and is subject to the same factors that influence cardiovascular disease mortality more generally. Changes in case-fatality for diabetes, therefore, were assumed to reflect half the trends in case-fatality for cardiovascular disease, which were estimated to be decreasing over the projection period. The combined effect of increasing BMI and decreasing case-fatality was a considerable increase in the incidence of Type 2 diabetes, and an even greater increase in future prevalence. Non-fatal conditions Mortality trend data are not relevant for conditions that are largely non-fatal. These include mental, sense organ and musculoskeletal disorders. The only mental health survey in Australia was carried out in 1997 and hence there are no trend data. Internationally there is no clear evidence of trends due to a paucity of mental health survey data collected using comparable diagnostic tools and criteria. Therefore no trends were assumed. Similarly, no 34 disease trends were applied to hearing loss (only one community survey), and the various causes of vision loss and musculoskeletal disorders (no evidence for trends). 2.6 Differentials in burden The high demand for information on health differentials, both between and within populations, is one measure of the obvious public policy implications of such information. For example, knowing that the gap in life expectancy at birth between Aboriginal and Torres Strait Islander Australians and other Australians is demonstrably very large is a sound basis for new initiatives to improve Indigenous health. One of the aims of the original study was to develop estimates of disease burden for different groups within the Australian population. To this end, the final report presented preliminary analyses of inequalities in disease burden by level of socioeconomic disadvantage, although it was not possible to complete a comprehensive analysis of non-fatal burden within the time available. An objective of the current project was to extend these analyses by providing a more complete picture of disease burden for a much greater range of subgroups within the Australian population. The methods used in this study build on the first comprehensive attempt to describe small area variability in health status across Victoria (DHS 2006), and are in the methodological tradition of describing differences in health across population subgroups. Murray and colleagues (1999a) differentiate this from descriptions of health inequalities, a term they reserve for analysis of the variation in health status across individuals in a population (analogous to analyses of income inequality, which measure the distribution of income at the level of individuals). While health inequalities are sometimes regarded as synonymous with subgroup differences in health in the literature, analyses of the latter are based on subgroup averages and as such can mask the true extent of inequalities between individuals. Categorising geographic areas The most disaggregated geographic information on place of usual residence for most Australian health data is the Statistical Local Area (SLA), and this geographic entity is used as the unit of analysis for this component of the study. For various reasons, SLA names and boundaries are revised over time, the most substantial revision occurring as a result of local government amalgamations in the early 1990s. To achieve geographic consistency, all data, regardless of year, were analysed in terms of ASGC definitions for the year 2001 (ASGC, or Australian Standard Geographical Classification, being the reference used to define SLAs) (ABS 2001a). Data defined in terms of SLAs fragmented as a result of boundary revisions were reapportioned using information from the 2001 Census on the proportion of each old SLA population residing in each current SLA after the redrawing of the boundaries. Irregular coding in data arising from such revisions was resolved on a case-by-case basis using historic documentation provided by the ABS. Estimated mid-year resident population figures for each SLA by year (1999 to 2003), 5-year age groups (0, 5 85+) and sex were obtained from the ABS. The ASGC 2001 provides for the classification of SLAs in terms of both socioeconomic status and remoteness. Socioeconomic status can be determined from one of four socioeconomic indexes for areas (SEIFA indexes) developed by the ABS from the 2001 Census using principal component methods on attributes such as low income, low educational attainment, 35 high levels of public sector housing, high unemployment, and jobs in relatively unskilled occupations (ABS 2001b). This study uses the index of disadvantage that is functionally equivalent to the Index of Relative Socioeconomic Disadvantage derived from the 1996 Census. This index is estimated at a collector-district level to be normally distributed at a national level, and can be population-weighted to derive values for ASGC 2001 SLAs. Socioeconomic quintiles were derived by ranking SLAs in order of disadvantage index then grouping them into five categories such that each category contains approximately 20% of the total Australian population. Remoteness can be determined from the Accessibility/Remoteness Index of Australia (ARIA+) developed by the Australian Government Department of Heath and Ageing and the National Centre for Social Applications of Geographic Information Systems (GISCA), and subsequently incorporated into ASGC 2001 (ABS 2001a). ARIA+ is a continuous varying index with values ranging from 0 (high accessibility) to 15 (high remoteness), and is based on road distance measurements from 11,879 populated localities to the nearest service centre. Index values for each locality have been interpolated to a 1 km grid so that all areas of Australia have an index value and scores for larger areas such as SLAs can be derived. Each SLA was classified into one of three groups based on the following standard cut-points as defined in ASGC 2001: Major cities (00.20), Regional (>0.205.92) and Remote (>5.92). Estimating burden for subpopulations One category of information readily available for disaggregating national estimates of burden to subpopulations is data on observed variations in event frequency for any aggregation from the level of the SLA and above. This includes the National Mortality dataset, the National Hospital Morbidity Database and the National Cancer Statistics Clearing House dataset. The other category comprises information that can be tabulated by state or territory jurisdiction, disadvantage quintile or remoteness category, but cannot be disaggregated below these strata. Most surveys (for example the National Health Survey, the Survey on Disability, Ageing and Carers, the National Survey of Mental Health and Wellbeing, and the Australian Diabetes Obesity and Lifestyle Study (AusDiab)) and published data tabulations fit this description. The primary objective with either category was to derive relativities between whatever level of disaggregation was possible, and to ensure that these relativities were as accurate as possible and not simply an artefact of small numbers. Of less concern was the absolute level of disease occurrence being reported, because these would be constrained by national estimates. The adopted strategy was intended to ensure consistency in the use of the available information and to ensure sufficient numbers at each level of the analysis. First, all sources were assessed for whether they could provide simple state/territory jurisdiction proportions (preferably by sex, but not necessarily by age) for any condition in the studys list of diseases and injuries. Most sources could provide this information. Next, they were assessed for plausibility as a valid proxy for variability in disease occurrence across a 15-cell matrix comprising five SEIFA categories and three remoteness categories. Not as many sources could provide this information and, of these, a few could provide information on only one dimension (that is, either SEIFA or remoteness, but not both). Age-standardised rates were then calculated for each cell of observed data, and these were divided by the crude rate for the whole matrix to derive 15 cell-specific standardised rate ratios. In matrices with only one dimension, ratios for the observed dimension were held constant across the missing dimension. 36 This estimation process means that the estimates of deaths of cancer cases in a particular SLA are not the same as the actual deaths or cancer cases in that SLA, but are synthetic estimates which reflect the rates of deaths and cancer in SLAs of similar type. Having determined possible sources for two pieces of information (state/territory proportions and matrices of rate ratios), an assessment was made for each disease and injury category as to whether there was agreement between sources (if there were more than one) and which information seemed sufficiently robust in terms of underlying numbers. For conditions with a predominantly fatal burden, preference was given to information derived from mortality data. For other conditions, preference was given to the data source upon which the national disability model was based. Each condition was then assigned a single source to be used to derive the proportion of national incidence cases that would be expected to occur in each state and territory. If no source could be identified, the number of incident cases was unconstrained at this step in the disaggregation. The implied jurisdiction-specific rate (or national rate where jurisdiction numbers were unconstrained) was then distributed to subpopulations within the jurisdiction using one of the matrices of rate ratios derived in the previous step. If no matrix was available, rates were held constant across subpopulations within jurisdictions. Derived incident cases were then rescaled to be consistent with jurisdiction totals where applicable, and ultimately national totals. Deaths were treated in the same way as incident cases. The final step was to derive prevalent cases and duration for each condition and its sequelae for each subpopulation within jurisdictions. An automated implementation of the equations underlying DisMod (an epidemiological modelling software package) was applied to subpopulation-specific incidence rates and national assumptions regarding remission and case-fatality to derive these parameters. In order to derive accurate durations, one of 15 sets of all-cause mortality rates was used according to the SEIFA and remoteness category of the subpopulation. All subpopulation-specific prevalent cases and YLD (both incident and prevalent) were then rescaled to be consistent with national totals. Subpopulation comparisons in this report This report is limited to the following subpopulation comparisons: 1. state and territory jurisdictions 2. remoteness categories 3. socioeconomic quintiles. While the analyses were aimed at disaggregating national burden estimates to the level of the SLA, there was no intention to disseminate results at this level of detail. In addition to the potential privacy considerations of the data providers, the release of such information may be misleading given the methods used. Rather, the authors and various jurisdictional stakeholders are working to regroup the data into meaningful aggregations of SLAs for specific health policy and planning purposes. 37 3 Burden of disease and injury in Australia This chapter discusses the burden of disease and injury in Australia in 2003 by fatal and nonfatal burden, sex, age and leading broad cause group. The numbers presented in this chapter should not be compared with those presented in the previous report (AIHW: Mathers et al. 1999) due to substantially different methods for many of the disability models. Readers who are interested in gaining an understanding of changes in burden over time are referred to Chapter 6 which discusses trends in population health over a 30-year period. 3.1 Disability-adjusted life years Cancer, cardiovascular disease and mental disorders were the leading causes of total burden of disease and injury in Australia in 2003 (Figure 3.1). Cancer and cardiovascular disease accounted for 37% of the total burden; for both causes, four-fifths of this burden was from mortality. Mental disorders and neurological & sense disorders were the next largest contributors, together accounting for a further quarter of the total burden. The contribution of mortality to the burden from these two groups was small, highlighting the importance of including non-fatal health outcomes in population health measurement. Overall, half the total burden (49%) was due to mortality and the distribution between the sexes was roughly equal for most causes, with injuries (70% of the burden in males) and musculoskeletal disorders (58% of the burden in females) the exceptions. Cancer Cardiovascular Neuro- Mental logical Chronic respiratory Injuries Diabetes Musculoskeletal Other 19% 18% 12% 13% 7% 7% 5% 4% 14% Total Injuries Diabetes Cardiovascular Chronic respiratory Cancer Mental Neurological Musculoskeletal 52% 70% 54% 53% 53% 53% 47% 47% 42% 48% 30% 46% 47% 47% 47% 53% 53% 58% Males Females Total Cancer Cardiovascular Injuries Chronic respiratory Diabetes Neurological Musculoskeletal Mental 49% 82% 78% 76% 38% 22% 17% 7% 7% 51% 18% 22% 24% 62% 78% 83% 93% 93% Fatal Non-fatal Figure 3.1: Burden (DALYs) by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 38 Total burden in absolute terms increased at a relatively constant rate until age 75 (Figure 3.2b), while the burden per head of population continued to rise exponentially, with small but significant peaks in childhood and early adulthood (Figure 3.2a). Injuries in males and mental disorders were the main cause groups until middle age and accounted for the majority of total burden in early adulthood, after which cancer, cardiovascular disease and neurological & sense disorders were more prominent. The contribution from cancer peaked at age 70 then declined, leaving cardiovascular disease as the major cause of burden in the elderly (Figure 3.2b). 0 500 1,000 Rate 1,000 0 20 40 60 80 100 Age Males Females 0 100 200 300 (thousands) 0 20 40 60 80 100 Age Other Musculoskeletal Diabetes Injuries Chronic respiratory Neurological Mental Cardiovascular Cancer Figure 3.2: Burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by broad cause group, Australia, 2003 The burden due to specific disease and injury categories reflected the more general picture at the broad cause group level. Ischaemic heart disease was the largest single cause in males, accounting for 11.1% of the total male burden (Table 3.1). For females, anxiety & depression was the leading cause, accounting for 10.0% of the total female burden. Ischaemic heart disease, stroke, Type 2 diabetes and dementia were the next four leading causes of DALYs in females. In males, Type 2 diabetes, anxiety & depression, lung cancer and stroke were the next four leading causes. Seven health areas have been identified by the Commonwealth, state and territory governments for priority attention as National Health Priority Areas: asthma, cancer, cardiovascular disease, diabetes mellitus, injuries, mental health, and arthritis and musculoskeletal conditions. In addition, dementia is an Australian Government health priority. In 2003, these eight health groupings accounted for 72.8% of the total burden, 17 of the 20 leading conditions for males and 15 of the 20 leading conditions for females. per Number 39 Table 3.1: Leading causes of burden (DALYs) by sex, Australia, 2003 Rank Males DALYs Per cent of total Females DALYs Per cent of total 1 Ischaemic heart disease 151,107 11.1 Anxiety & depression 126,464 10.0 2 Type 2 diabetes 71,176 5.2 Ischaemic heart disease 112,390 8.9 3 Anxiety & depression 65,321 4.8 Stroke 65,166 5.1 4 Lung cancer 55,028 4.0 Type 2 diabetes 61,763 4.9 5 Stroke 53,296 3.9 Dementia 60,747 4.8 6 COPD 49,201 3.6 Breast cancer 60,520 4.8 7 Adult-onset hearing loss 42,653 3.1 COPD 37,550 3.0 8 Suicide & self-inflicted injuries 38,717 2.8 Lung cancer 33,876 2.7 9 Prostate cancer 36,547 2.7 Asthma 33,828 2.7 10 Colorectal cancer 34,643 2.5 Colorectal cancer 28,962 2.3 11 Dementia 33,653 2.5 Adult-onset hearing loss 22,200 1.8 12 Road traffic accidents 31,028 2.3 Osteoarthritis 20,083 1.6 13 Asthma 29,271 2.1 Personality disorders 16,339 1.3 14 Alcohol abuse 27,225 2.0 Migraine 15,875 1.3 15 Personality disorders 16,248 1.2 Back pain 15,188 1.2 16 Schizophrenia 14,785 1.1 Lower respiratory tract infections 14,233 1.1 17 Osteoarthritis 14,495 1.1 Falls 13,269 1.0 18 Back pain 14,470 1.1 Parkinsons disease 13,189 1.0 19 Melanoma 13,734 1.0 Schizophrenia 12,717 1.0 20 Parkinsons disease 13,664 1.0 Rheumatoid arthritis 12,062 1.0 Table 3.2 compares burden by broad cause groups in 2003 with total health system expenditures in 200001. This table is included to illustrate a misunderstanding about the relationship between health expenditure and health outcomes. It is sometimes argued, for example, that the proportion of total expenditure that is committed to a particular health problem should be commensurate in some way to its contribution to total burden. This is not necessarily the case. Burden estimates describe the health problems that remain in a population in spite of all currently implemented prevention and treatment strategies. Large expenditure for a cause with a small burden is money well spent if that expenditure reflects an efficient health service response to what otherwise would have been a much larger problem. Oral conditions, for example, account for only 0.9% of total burden but consume 6.7% of total expenditure ($3.4 billion). This commitment of resources may well represent a good investment if it keeps the burden from oral conditions at low levels and that without it, the burden would be much higher. If, on the other hand, some of this expenditure is not impacting on the burden, because it is being directed towards cosmetic or ineffective services, for example, or there are inefficiencies in the delivery of oral health services, then the conclusion may be less sanguine. The real test of whether an investment has been worthwhile depends on the change in burden resulting from the expenditure as well as the opportunity cost of that expenditure to investments in other areas of the health sector. Exploring this requires information on the 40 effectiveness and costs associated with all current prevention and treatment strategies. Such analyses are beyond the scope of this report. The proportion of burden for a particular health problem vis-a-vis expenditure, therefore, is more appropriately used as one argument amongst others for prioritising research into the development of new treatment and preventive interventions, and into assessing the effectiveness of these interventions. It should not be used to prioritise existing treatment and preventive activities. Table 3.2: Burden (DALYs) in 2003 and expenditure in 200001 by broad cause group, Australia DALYs in 2003 Expenditure in 200001(a)Cause No. (thousands) Per cent $ (millions) Per cent Neoplasms(b) 510.3 19.4 2,918 5.8 Cardiovascular disease 473.8 18.0 5,479 10.9 Mental disorders 350.5 13.3 3,741 7.5 Neurological & sense disorders 312.8 11.9 4,942 9.9 Respiratory disease(c) 222.2 8.4 3,742 7.5 Injuries 185.1 7.0 4,013 8.0 Diabetes mellitus 143.8 5.5 812 1.6 Musculoskeletal diseases 105.5 4.0 4,634 9.2 Genitourinary diseases 65.2 2.5 2,076 4.1 Diseases of the digestive system 58.0 2.2 2,811 5.6 Infectious & parasitic diseases 44.7 1.7 1,224 2.4 Neonatal causes 34.6 1.3 358 0.7 Congenital anomalies 33.2 1.3 221 0.4 Endocrine & metabolic disorders 28.6 1.1 1,587 3.2 Oral conditions 24.5 0.9 3,372 6.7 Skin diseases 20.3 0.8 1,370 2.7 Maternal conditions 2.2 0.1 1,315 2.6 Other(d) 17.5 0.7 5,530 11.0 All causes 2,632.8 100.0 50,146 100.0(a) Total health system expenditures from AIHW 2005c. (b) Includes cancers (malignant neoplasms) and other (non-malignant) neoplasms. (c) Includes chronic respiratory disease and acute respiratory infections. (d) Includes Signs, symptoms and ill-defined conditions which includes expenditure on diagnostic and other services for signs, symptoms and ill-defined conditions where the cause of the problem is unknown and includes other contact with the health system such as fertility control, reproduction and development, elective plastic surgery, general prevention, screening and health examination; and treatment and aftercare for unspecified disease. 3.2 Years of life lost Years of life lost (YLL), or fatal burden, accounted for 49% of the total burden of disease and injury in Australia in 2003 (Figure 3.1c). Cancers, cardiovascular disease and injuries were responsible for almost three-quarters of this burden (Figure 3.3a). Since the 1996 study, cancer has overtaken cardiovascular disease as the greatest cause of fatal burden as 41 cardiovascular mortality has declined much more than cancer mortality over the past three to four decades. Males experienced 55% of the total fatal burden. Cancer Cardiovascular Injuries Chronic respiratory Neurological Diabetes Other 32% 29% 11% 6% 4% 3% 16% Total Injuries Diabetes Chronic respiratory Cardiovascular Cancer Neurological 55% 72% 56% 54% 54% 53% 45% 45% 28% 44% 46% 46% 47% 55% Males Females Figure 3.3: Fatal burden (YLL) expressed as: (a) proportion by broad cause group, and (b) proportion by sex for each broad cause group, Australia, 2003 As with total burden, fatal burden increased in absolute terms at a relatively constant rate until age 75 (Figure 3.4b), while the burden per head of population continued to increase exponentially, with small but important peaks in childhood and, in males, early adulthood (Figure 3.4a). Injury was the main cause of fatal burden until age 35 and accounted for the majority of fatal burden in early life, after which cancer and cardiovascular disease were more prominent. The contribution from cancer peaked at age 70 then declined, leaving cardiovascular disease as the major cause of fatal burden in the elderly (Figure 3.4b). 42 0 200 400 600 0 20 40 60 80 100 Age Males Females 0 50 100 150 Number (0 20 40 60 80 100 Age Other Chronic respiratory Injuries Cardiovascular Cancer Figure 3.4: Fatal burden (YLL) by age expressed as: (a) rates by sex, and (b) numbers by broad cause group, Australia, 2003 Again, the fatal burden due to specific disease and injury categories reflected the more general picture at the broad cause group level. Ischaemic heart disease was the disease contributing most to YLL in both males and females. Stroke was the second largest disease causing YLL in females, followed by breast cancer and lung cancer. In males, lung cancer ranked second, followed by suicide & self-inflicted injuries, stroke, and colorectal cancer (Table 3.3). Rate per 1,000 thousands) 43 Table 3.3: Leading causes of mortality burden (YLL) by sex, Australia, 2003 Rank Males YLL Per cent of total Females YLL Per cent of total 1 Ischaemic heart disease 128,991 18.2 Ischaemic heart disease 89,152 15.7 2 Lung cancer 51,505 7.3 Stroke 48,548 8.5 3 Suicide & self-inflicted injuries 38,434 5.4 Breast cancer 40,080 7.0 4 Stroke 36,152 5.1 Lung cancer 31,551 5.5 5 Colorectal cancer 27,997 3.9 Colorectal cancer 23,735 4.2 6 Road traffic accidents 26,674 3.8 COPD 21,025 3.7 7 COPD 26,183 3.7 Dementia 16,009 2.8 8 Prostate cancer 23,175 3.3 Lower respiratory tract infections 12,309 2.2 9 Type 2 diabetes 15,273 2.2 Type 2 diabetes 11,751 2.1 10 Hepatitis 12,524 1.8 Pancreas cancer 10,984 1.9 11 Alcohol abuse 11,449 1.6 Ovary cancer 10,946 1.9 12 Lower respiratory tract infections 11,221 1.6 Suicide & self-inflicted injuries 10,945 1.9 13 Pancreas cancer 11,136 1.6 Road traffic accidents 9,678 1.7 14 Brain cancer 10,718 1.5 Nephritis & nephrosis 9,521 1.7 15 Lymphoma 10,474 1.5 Lymphoma 8,324 1.5 16 Melanoma 10,108 1.4 Brain cancer 7,809 1.4 17 Leukaemia 10,039 1.4 Leukaemia 7,468 1.3 18 Oesophagus cancer 9,427 1.3 Hepatitis 6,534 1.1 19 Nephritis & nephrosis 9,336 1.3 Falls 5,845 1.0 20 Stomach cancer 8,209 1.2 Stomach cancer 5,609 1.0 3.3 Years lost due to disability Years lost due to disability (YLD), or non-fatal burden, are typically calculated from incidence cases in a base year and as such are to be interpreted as the number of healthy years lost due to disability that will accrue into the future from new cases of disease and injury in that base year. Incident non-fatal burden is added to fatal burden (YLL) to derive total burden (DALYs). An alternative way of calculating non-fatal burden uses prevalent cases as the basis. Prevalent non-fatal burden (PYLD) is to be interpreted as the number of healthy years lost due to disability currently experienced by a population. This cannot be added to fatal burden to derive total burden in the same way as incident non-fatal burden. Both methods of calculating non-fatal burden are presented below. For all the other sections of this report, references to non-fatal burden reflect incident non-fatal burden unless otherwise specified. Incident YLD Incident non-fatal burden accounted for 51% of the total burden of disease and injury in Australia in 2003 (Figure 3.1c). Mental, neurological and sense disorders contributed most, 44 together accounting for 43% of this burden (Figure 3.5a). While cancer, cardiovascular disease and injuries contributed 72% to the total fatal burden (Figure 3.3a), these causes did not make a similar contribution to incident non-fatal burden. Incident non-fatal burden was distributed equally between the sexes (Figure 3.5b). Mental Neurological Chronic respiratory Diabetes Cardiovascular Musculoskeletal Other 24% 19% 8% 9% 8% 7% 25% Total Diabetes Chronic respiratory Cardiovascular Neurological Mental Musculoskeletal 48% 53% 52% 51% 47% 45% 42% 52% 47% 48% 49% 53% 55% 58% Males Females Figure 3.5: Non-fatal burden (YLD) expressed as: (a) proportion by broad cause group, and (b) proportion by sex for each broad cause group, Australia, 2003 Incident non-fatal burden increased rapidly in absolute terms until early adulthood then levelled out, while the rate per head of population continued increasing, but at a slower rate than for fatal burden (Figure 3.6). Mental disorders were the main causes of incident nonfatal burden until middle age and accounted for the majority of fatal burden in early life, after which neurological & sense disorders were more prominent, accounting for the majority of non-fatal burden in the elderly. Chronic respiratory conditions accounted for a small but consistent proportion of incident non-fatal burden, with peaks in childhood due to asthma and at older ages from chronic obstructive pulmonary disease (Figure 3.6). 45 Figure 3.6: Incident non-fatal burden (YLD) by age expressed as: (a) rates by sex, and (b) numbers by broad cause group, Australia, 2003 Anxiety & depression and Type 2 diabetes were the leading causes of incident non-fatal burden in males and females (Table 3.4). Dementia was the third leading cause in females, followed by asthma and ischaemic heart disease. In males, adult-onset hearing loss ranked third, followed by asthma. Mental disorders accounted for six of the 20 leading causes of incident non-fatal burden in males and three in females. 0 100 200 Rate per 1,000 0 20 40 60 80 100 Age Males Females 0 50 100 150 Number (thousands) 0 20 40 60 80 100 Age Other Diabetes Chronic respiratory Neurological Mental 46 Table 3.4: Leading causes of incident non-fatal burden (YLD) by sex, Australia, 2003 Rank Males YLD Per cent of total Females YLD Per cent of total 1 Anxiety & depression 65,208 10.0 Anxiety & depression 126,244 18.1 2 Type 2 diabetes 55,903 8.5 Type 2 diabetes 50,012 7.2 3 Adult-onset hearing loss 42,653 6.5 Dementia 44,738 6.4 4 Asthma 27,649 4.2 Asthma 31,405 4.5 5 Dementia 25,558 3.9 Ischaemic heart disease 23,238 3.3 6 COPD 23,018 3.5 Adult-onset hearing loss 22,200 3.2 7 Ischaemic heart disease 22,116 3.4 Breast cancer 20,440 2.9 8 Stroke 17,144 2.6 Osteoarthritis 19,775 2.8 9 Personality disorders 16,248 2.5 Stroke 16,619 2.4 10 Alcohol abuse 15,775 2.4 COPD 16,525 2.4 11 Schizophrenia 14,673 2.2 Personality disorders 16,339 2.3 12 Osteoarthritis 14,429 2.2 Migraine 15,868 2.3 13 Back pain 14,355 2.2 Back pain 15,129 2.2 14 Prostate cancer 13,372 2.0 Schizophrenia 12,577 1.8 15 Autism spectrum disorders 11,702 1.8 Rheumatoid arthritis 10,918 1.6 16 Parkinsons disease 10,623 1.6 Parkinsons disease 10,534 1.5 17 Refractive errors 8,241 1.3 Refractive errors 10,520 1.5 18 Peripheral vascular disease 7,965 1.2 Infertility 8,076 1.2 19 Heroin or polydrug abuse 7,498 1.1 Falls 7,424 1.1 20 Benign prostatic hypertrophy 7,378 1.1 Macular degeneration 7,259 1.0 Prevalent YLD Figure 3.7 illustrates the prevalent non-fatal burden by age. The difference between prevalent and incident non-fatal burden is most apparent for childhood conditions, such as asthma and congenital disorders, and for chronic mental disorders, the incidence of which peaks in childhood and early adulthood. Incident non-fatal burden at these life stages is much larger compared to prevalent non-fatal burden because most incident cases of chronic conditions at young ages are expected to remain prevalent cases at older ages. This explains the shift to the right in the picture of prevalent non-fatal burden (Figure 3.7) compared to incident non-fatal burden (Figure 3.6). The rate of prevalent burden was lowest in children between 1 and 4 years of age (15 PYLD per thousand) and increased to 147 in people aged 65 to 69 years and then to 415 per thousand in people over the age of 95. In other words, disability from all diseases and injuries resulted in a loss of 1.5% of healthy time lived by young children, increasing with age to 14.7% in those 65 to 69 years and 41.5% in the very old. 47 Figure 3.7: Prevalent non-fatal burden (PYLD) by age expressed as: (a) rates by sex, and (b) numbers by broad cause group, Australia, 2003 3.4 Age and sex patterns In this section, the size and composition of burden is reported by five broad age groups (Table 3.5). Table 3.5: Distribution of population and burden (DALYs) by five broad age groups, Australia, 2003 Age group Population (a) Per cent of total DALYs Per cent of total 014 years 3,979,410 20.0 221,536 8.4 1544 years 8,622,610 43.4 633,260 24.1 4564 years 4,733,808 23.8 681,566 25.9 6574 years 1,349,949 6.8 428,904 16.3 75 years and over 1,195,692 6.0 667,504 25.4 Total 19,881,469 100.0 2,632,770 100.0 (a) Estimated resident population figures as at 30 June 2003 (ABS cat. no. 3201.0). Children aged 014 yearsChildren aged 014 years comprised 20.0% of the total population and experienced 8.4% of the total burden of disease and injury in Australia in 2003 (Table 3.5). Twenty-three per cent of this burden was due to mental disorders (that is anxiety & depression, attention-deficit hyperactivity disorder and autism spectrum disorders), 18% due to chronic respiratory conditions (mostly asthma) and 16% due to neonatal conditions. Less than a quarter of the 0 200 400 Rate per 1,000 0 20 40 60 80 100 Age Males Females 0 50 100 150 Number (thousands) 0 20 40 60 80 100 Age Other Cardiovascular Chronic respiratory Neurological Mental 48 burden was due to mortality (Figure 3.8). Males experienced 56% of the burden in this age group. Mental Chronic respiratory Neonatal Congenital Injuries Neurological Other 23% 18% 16% 12% 7% 7% 18% Total Chronic respiratory Congenital Mental Injuries Neurological Neonatal 56% 58% 58% 57% 57% 57% 55% 44% 42% 42% 43% 43% 43% 45% Males Females Total Neonatal Injuries Congenital Neurological Chronic respiratory Mental 24% 55% 50% 41% 17% 3% 0% 76% 45% 50% 59% 83% 97% 100% Fatal Non-fatal Figure 3.8: Burden (DALYs) in 014 year olds by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Asthma was the leading cause of burden for both males and females (Table 3.6). This was followed by autism spectrum disorders, anxiety & depression, and low birth weight in males. In females, anxiety & depression, low birth weight and birth trauma & asphyxia were the next leading causes. The leading 10 causes of burden accounted for 58.7% of the total burden in this age group. Table 3.6: Leading causes of DALYs in 014 year olds by sex, Australia, 2003 Rank Males DALYs Per cent of total Females DALYs Per cent of total 1 Asthma 21,953 17.6 Asthma 16,490 17.0 2 Autism spectrum disorders 11,703 9.4 Anxiety & depression 15,507 16.0 3 Anxiety & depression 9,554 7.7 Low birth weight 7,142 7.4 4 Low birth weight 8,281 6.6 Birth trauma & asphyxia 4,221 4.4 5 Attention-deficit hyperactivity disorder 7,082 5.7 Attention-deficit hyperactivity disorder 2,840 2.9 6 Birth trauma and asphyxia 5,086 4.1 Epilepsy 2,446 2.5 7 Congenital heart disease 3,434 2.8 Congenital heart disease 2,202 2.3 8 Epilepsy 3,249 2.6 Autism spectrum disorders 2,056 2.1 9 Neonatal infections 2,156 1.7 Otitis media 1,377 1.4 10 Road traffic accidents 1,991 1.6 Road traffic accidents 1,336 1.4 49 Older children and adults aged 1544 yearsOlder children and adults aged 1544 years comprised 43.4% of the total population and experienced 24.1% of the total burden of disease and injury in Australia in 2003 (Table 3.5). Over a third of the total burden in this age group was attributable to mental disorders, and another 17% was due to injuries (Figure 3.9). There were considerable sex differences in this age group, with females experiencing a greater share of the burden from neurological disorders, chronic respiratory diseases, cancers and mental disorders than males. Males, on the other hand, experienced more than three-quarters of the injury burden, partly because of their greater inclination for risk taking. Overall, 29% of the burden in this age group was due to mortality. Mental Cancer Injuries Neurological Cardiovascular Chronic respiratory Other 36% 7% 17% 7% 5% 5% 23% Total Injuries Cardiovascular Mental Cancer Chronic respiratory Neurological 51% 78% 64% 45% 43% 43% 41% 49% 22% 36% 55% 57% 57% 59% Males Females Total Injuries Cancer Cardiovascular Neurological Chronic respiratory Mental 29% 80% 77% 63% 16% 10% 4% 71% 20% 23% 37% 84% 90% 96% Fatal Non-fatal Figure 3.9: Burden (DALYs) in 1544 year olds by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Anxiety & depression was by far the leading single cause of burden in both males and females, followed by suicide & self-inflicted injuries and road traffic accidents in males, and migraine and Type 2 diabetes in females (Table 3.7). Mental disorders made up half of the top 10 leading causes of burden in males and three of the top 10 leading causes of burden in females. The leading 10 ranked conditions accounted for 54.8% of the burden in this age group. 50 Table 3.7: Leading causes of DALYs in 1544 year olds by sex, Australia, 2003 Rank Males DALYs Per cent of total Females DALYs Per cent of total 1 Anxiety & depression 42,237 13.0 Anxiety & depression 84,717 27.4 2 Suicide & self-inflicted injuries 27,592 8.5 Migraine 14,105 4.6 3 Road traffic accidents 22,845 7.1 Type 2 diabetes 12,487 4.0 4 Schizophrenia 14,376 4.4 Asthma 11,311 3.7 5 Alcohol abuse 13,953 4.3 Schizophrenia 11,064 3.6 6 Type 2 diabetes 12,868 4.0 Personality disorders 9,389 3.0 7 Heroin abuse 11,882 3.7 Breast cancer 9,068 2.9 8 Personality disorders 10,526 3.2 Infertility 8,057 2.6 9 Ischaemic heart disease 9,750 3.0 Suicide & self-inflicted injuries 7,174 2.3 10 COPD 6,840 2.1 Road traffic accidents 6,751 2.2 Adults aged 4564 yearsAdults aged 4564 years comprised 23.8% of the total population and experienced 25.9% of the total burden of disease and injury in Australia in 2003 (Table 3.5). Cancer, cardiovascular disease and neurological disorders accounted for more than half of the total burden in this age group. Males experienced a greater share of the burden than females for all causes except mental disorders and musculoskeletal disorders (Figure 3.10). Overall, 49% of the burden in this age group was due to mortality. Cancer Cardiovascular Neurological Mental Diabetes Musculoskeletal Other 28% 16% 10% 9% 8% 6% 21% Total Cardiovascular Diabetes Neurological Cancer Mental Musculoskeletal 56% 68% 60% 59% 51% 44% 44% 44% 32% 40% 41% 49% 56% 56% Males Females Total Cancer Cardiovascular Mental Diabetes Neurological Musculoskeletal 49% 81% 69% 14% 14% 13% 4% 51% 19% 31% 86% 86% 87% 96% Fatal Non-fatal Figure 3.10: Burden (DALYs) in 4564 year olds by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 51 Ischaemic heart disease was the leading cause of burden in males, followed by Type 2 diabetes and lung cancer (Table 3.8). In females, the top three causes were breast cancer, anxiety & depression and Type 2 diabetes. The top 10 conditions accounted for 52.0% of total burden in this age group. Table 3.8: Leading causes of DALYs in 4564 year olds by sex, Australia, 2003 Rank Males DALYs Per cent of total Females DALYs Per cent of total 1 Ischaemic heart disease 47,782 12.5 Breast cancer 32,012 10.7 2 Type 2 diabetes 32,741 8.6 Anxiety & depression 25,744 8.6 3 Lung cancer 20,861 5.5 Type 2 diabetes 22,299 7.5 4 Adult-onset hearing loss 20,847 5.5 Ischaemic heart disease 17,489 5.8 5 COPD 15,389 4.0 Lung cancer 13,475 4.5 6 Colorectal cancer 14,130 3.7 Adult-onset hearing loss 10,576 3.5 7 Stroke 13,800 3.6 COPD 10,422 3.5 8 Anxiety & depression 11,757 3.1 Colorectal cancer 9,808 3.3 9 Alcohol abuse 10,077 2.6 Stroke 9,693 3.2 10 Prostate cancer 8,953 2.3 Back pain 6,620 2.2 Adults aged 6574 yearsAdults aged 6574 years comprised 6.8% of the total population and experienced 16.3% of the total burden of disease and injury in Australia in 2003 (Table 3.5). Cancer and cardiovascular disease accounted for over half of the total burden in this age group (Figure 3.11). Females experienced a greater share of the burden than males from musculoskeletal conditions, while the reverse was true for all other broad cause groups. Overall, 60% of the burden in this age group was due to mortality. 52 Total Cardiovascular Cancer Chronic respiratory Diabetes Neurological Musculoskeletal 57% 61% 59% 56% 56% 52% 42% 43% 39% 41% 44% 44% 48% 58% Males Females Total Cancer Cardiovascular Chronic respiratory Diabetes Neurological 60% 83% 76% 61% 35% 40% 17% 24% 39% 65% Fatal Non-fatal Figure 3.11: Burden (DALYs) in 6574 year olds by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Ischaemic heart disease, lung cancer and Type 2 diabetes were the leading causes of burden in males, together accounting for 29% of total male burden (Table 3.9). In females, ischaemic heart disease, Type 2 diabetes and breast cancer were the leading causes, accounting for 23% of total burden. The top 10 conditions accounted for 56.3% of total burden in this age group. Table 3.9: Leading causes of DALYs in 6574 year olds by sex, Australia, 2003 Rank Males DALYs Per cent of total Females DALYs Per cent of total 1 Ischaemic heart disease 37,860 15.5 Ischaemic heart disease 21,052 11.4 2 Lung cancer 19,258 7.9 Type 2 diabetes 11,517 6.2 3 Type 2 diabetes 14,203 5.8 Breast cancer 10,445 5.7 4 Prostate cancer 11,950 4.9 Dementia 10,236 5.5 5 Adult-onset hearing loss 11,920 4.9 Lung cancer 9,937 5.4 6 COPD 11,693 4.8 Stroke 9,635 5.2 7 Stroke 10,938 4.5 COPD 8,855 4.8 8 Colorectal cancer 10,531 4.3 Colorectal cancer 7,513 4.1 9 Dementia 7,872 3.2 Osteoarthritis 6,088 3.3 10 Parkinsons disease 3,958 1.6 Adult-onset hearing loss 5,834 3.2 Cancer Cardiovascular Neurological Chronic respiratory Diabetes Musculoskeletal Other 31% 15% 23% 7% 6% 5% 14% Musculoskeletal 13% 9% 87% 91% 53 Older people aged 75 years and over Older people aged 75 years and over comprised 6.0% of the total population and experienced 25.4% of the total burden of disease and injury in Australia in 2003 (Table 3.5). Cardiovascular disease and cancer accounted for over half of the total burden in this age group (Figure 3.12). Females experienced a greater share of the burden than males overall and for all broad cause groups except chronic respiratory diseases and cancer. Overall, 68% of the burden in this age group was due to mortality. Cardiovascular Cancer Neurological Chronic respiratory Diabetes Genitourinary Other 34% 19% 18% 7% 4% 3% 14% Total Cancer Chronic respiratory Genitourinary Diabetes Cardiovascular Neurological 43% 52% 51% 48% 42% 41% 38% 57% 48% 49% 52% 58% 59% 62% Males Females Total Cancer Cardiovascular Genitourinary Chronic respiratory Diabetes Neurological 68% 86% 86% 82% 70% 45% 23% 32% 14% 14% 18% 30% 55% 77% Fatal Non-fatal Figure 3.12: Burden (DALYs) in those aged 75 years and over by broad cause group expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and nonfatal outcomes, Australia, 2003 Ischaemic heart disease, stroke and dementia were the leading causes of burden in males, together accounting for 34% of total male burden (Table 3.10). In females, ischaemic heart disease, dementia and stroke were the leading causes, accounting for 42% of total burden. The top 10 conditions account for 60.9% of the total burden in this age group. 54 Table 3.10: Leading causes of DALYs in those aged 75 years and over by sex, Australia, 2003 Rank Males DALYs Per cent of total Females DALYs Per cent of total 1 Ischaemic heart disease 55,680 19.3 Ischaemic heart disease 70,853 18.7 2 Stroke 21,834 7.5 Dementia 46,984 12.4 3 Dementia 21,095 7.3 Stroke 39,830 10.5 4 Prostate cancer 15,484 5.4 Type 2 diabetes 15,330 4.1 5 COPD 14,900 5.2 COPD 13,318 3.5 6 Lung cancer 13,533 4.7 Colorectal cancer 9,703 2.6 7 Type 2 diabetes 11,262 3.9 Lower respiratory tract infections 9,137 2.4 8 Colorectal cancer 8,442 2.9 Lung cancer 9,059 2.4 9 Adult-onset hearing loss 7,052 2.4 Breast cancer 8,995 2.4 10 Lower respiratory tract infections 6,395 2.2 Falls 7,814 2.1 3.5 Specific disease and injury categories This section presents burden by 22 broad cause groupings (Table 3.11) and discusses the eight largest of these in greater detail. The section ends with a short discussion on three conditions (renal failure, vision loss and intellectual disability), the burden from which in the previous sections is split across multiple subheadings depending on aetiology. However, from a health service planning perspective there is value in presenting the aggregates for these conditions. 55 Table 3.11: Burden (YLD, YLL and DALYs) by broad cause group, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Cancers 87,463 6.5 411,953 32.2 499,416 19.0 Cardiovascular disease 104,429 7.7 369,365 28.9 473,794 18.0 Mental disorders 327,391 24.2 23,154 1.8 350,545 13.3 Neurological & sense disorders 258,638 19.1 54,127 4.2 312,766 11.9 Chronic respiratory diseases 115,398 8.5 71,339 5.6 186,737 7.1 Diabetes mellitus 111,536 8.2 32,295 2.5 143,831 5.5 Unintentional injuries 41,263 3.0 84,599 6.6 125,862 4.8 Musculoskeletal diseases 98,481 7.3 7,027 0.5 105,508 4.0 Genitourinary diseases 41,161 3.0 24,087 1.9 65,249 2.5 Intentional injuries 3,139 0.2 56,050 4.4 59,189 2.2 Diseases of the digestive system 30,246 2.2 27,710 2.2 57,957 2.2 Infectious & parasitic diseases 14,021 1.0 30,665 2.4 44,685 1.7 Acute respiratory infections 11,752 0.9 23,750 1.9 35,502 1.3 Neonatal causes 15,584 1.2 18,974 1.5 34,558 1.3 Congenital anomalies 16,331 1.2 16,897 1.3 33,228 1.3 Endocrine & metabolic disorders 14,968 1.1 13,598 1.1 28,565 1.1 Oral conditions 24,406 1.8 102 0.0 24,507 0.9 Skin diseases 18,130 1.3 2,173 0.2 20,302 0.8 Ill-defined conditions 8,781 0.6 2,536 0.2 11,317 0.4 Non-malignant neoplasms 3,209 0.2 7,694 0.6 10,903 0.4 Nutritional deficiencies 5,739 0.4 458 0.0 6,197 0.2 Maternal conditions 1,926 0.1 226 0.0 2,152 0.1 Total burden 1,353,992 100.0 1,278,778 100.0 2,632,770 100.0 Cancers Cancer was responsible for 19.0% of the total burden of disease and injury in Australia in 2003 (Table 3.11), with lung, colorectal, breast and prostate cancer accounting for half of this burden (Figure 3.13). Apart from the sex-specific cancers (that is, breast, cervical and uterine cancers in females and prostate cancer in males), there were considerable sex differences in the experience of cancer burden, with males having a greater share of the burden from melanoma, colorectal cancer, lymphomas and lung cancer than females. The difference in lung cancer between males and females was largely due to the higher prevalence of smoking in males than females two or more decades ago. More than four-fifths of the total cancer burden was due to mortality. 56 Total Prostate cancer Lung cancer Lymphoma Colorectal cancer Pancreas cancer Breast cancer 53% 100% 62% 56% 54% 50% 0% 47% 0% 38% 44% 46% 50% 100% Males Females Total Pancreas cancer Lung cancer Lymphoma Colorectal cancer Breast cancer Prostate cancer 82% 98% 93% 84% 81% 66% 63% 18% 2% 7% 16% 19% 34% 37% Fatal Non-fatal Figure 3.13: Cancer burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Total cancer burden, both in absolute terms and when expressed as a rate per head of population, increased exponentially until age 75, then declined (Figure 3.14). The contribution from lung cancer was greatest at this age after which it declined in proportion to other cancers. In males, the contribution from prostate cancer increased until old age, whereas in females the contribution from breast cancer increased until age 60, then declined in proportion to other cancers. The contribution from colorectal cancer was important across all ages. Lung cancer Colorectal cancer Breast cancer Prostate Pancreas cancer cancer Lymphoma Other 18% 13% 12% 4% 5% 7% 41% 57 0 50 100 150 0 20 40 60 80 100 Age Males Females Figure 3.14: Cancer burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Lung cancer was the fourth leading cause of overall burden in males, while prostate and colorectal cancers were the ninth and tenth, respectively. Breast cancer was the sixth leading cause of overall burden in females, while lung cancer and colorectal cancer were eighth and tenth, respectively (Table 3.1). Although cancer of the cervix was not a leading cause of death in females, it is a cancer priority area because it is one of the few cancers where precancerous lesions can and have been cost-effectively detected and treated through an organised screening program. Moreover, a vaccine against human papilloma virus has become available that has the potential to further reduce the burden of cervical cancer over time. This illustrates that burden information should be used with other evidence to determine health service priorities. Table 3.12: Cancer burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Lung cancer 5,848 0.4 83,056 6.5 88,904 3.4 Colorectal cancer 11,873 0.9 51,732 4.0 63,605 2.4 Breast cancer 20,440 1.5 40,214 3.1 60,654 2.3 Prostate cancer 13,372 1.0 23,175 1.8 36,547 1.4 Pancreas cancer 561 0.0 22,119 1.7 22,680 0.9 Lymphoma 3,465 0.3 18,798 1.5 22,263 0.8 Other 31,905 2.4 172,859 13.5 204,763 7.8 Total cancer burden 87,463 6.5 411,953 32.2 499,416 19.0 Rate per 1,000 0 20 40 60 Number (thousands) 0 20 40 60 80 100 Age Other Lymphoma Pancreas cancer Prostate cancer Breast cancer Colorectal cancer Lung cancer 58 Cardiovascular disease Cardiovascular disease were responsible for 18.0% the total burden of disease and injury in Australia in 2003 (Table 3.11), with ischaemic heart disease and stroke accounting for over four-fifths of this burden (Figure 3.15). These diseases were also in the five leading causes of overall burden (Table 3.1). The contribution from ischaemic heart disease was greater in males than in females, while the reverse was the case for stroke. Nearly four-fifths of total cardiovascular burden was due to mortality. Ischaemic heart disease Stroke Peripheral vascular Other 56% 25% 4% 15% Total IHD Peripheral vascular Stroke 53% 57% 57% 45% 47% 43% 43% 55% Males Females Total IHD Stroke Peripheral vascular 78% 83% 71% 31% 22% 17% 29% 69% Fatal Non-fatal Figure 3.15: Cardiovascular burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 In contrast to cancer, the total cardiovascular burden per head of population continued increasing until old age. This resulted in a larger proportion of the absolute burden at older ages than for cancer (Figure 3.16). Ischaemic heart disease dominated across all ages. 59 0 200 400 Rate 1,000 0 20 40 60 80 100 Age Males Females Figure 3.16: Cardiovascular burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Table 3.13: Cardiovascular burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Ischaemic heart disease 45,354 3.3 218,143 17.1 263,497 10.0 Stroke 33,763 2.5 84,699 6.6 118,462 4.5 Peripheral vascular disease 12,888 1.0 5,718 0.4 18,606 0.7 Inflammatory heart disease 3,689 0.3 12,215 1.0 15,904 0.6 Aortic aneurysm 209 0.0 11,129 0.9 11,338 0.4 Hypertensive heart disease 678 0.1 8,303 0.6 8,982 0.3 Other 7,848 0.6 29,157 2.3 37,005 1.4 Total cardiovascular burden 104,429 7.7 369,365 28.9 473,794 18.0 Mental disorders Mental disorders were responsible for 13.3% of the total burden of disease and injury in Australia in 2003 (Table 3.11), with anxiety & depression, alcohol abuse and personality disorders accounting for almost three-quarters of this burden (Figure 3.17). There were marked sex differences in the mental illness burden for particular disorders. The burden from anxiety & depression was twice as high for females as for males. Conversely, the burden from substance abuse was more than three times as high in males as in females. Eating disorders occurred mainly in females. Autism spectrum disorders were much more common in males, with females having just 15% of the total burden from these conditions. per Number (thousands) 0 50 100 0 20 40 60 80 100 Age Other Inflammatory heart disease Peripheral vascular disease Stroke Ischaemic heart disease 60 Seven per cent of the burden from mental disorders was due to mortality, most of which was accounted for by fatal outcomes associated with substance abuse. Heroin dependence Total Autism Alcohol dependence Heroin dependence Schizophrenia Personality disorders Anxiety & depression 47% 85% 80% 74% 54% 50% 34% 53% 15% 20% 26% 46% 50% 66% Males Females Total Alcohol dependence Heroin dependence Schizophrenia Autism Anxiety & depression Personality disorders 7% 42% 39% 1% 1% 0% 0% 93% 58% 61% 99% 99% 100% 100% Fatal Non-fatal Figure 3.17: Mental disorder burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 The burden from mental disorders both in absolute terms and when expressed as a rate per head of population was greater in early adulthood than at other ages (Figure 3.18). This was partly due to the peak in new cases of chronic mental illnesses at this life stage, the burden of which was experienced throughout adult life. Anxiety & depression contributed most until age 60, after which the contribution from alcohol abuse and personality disorders was more prominent. Anxiety & depression Alcohol dependence Personality disorders Schizophrenia Heroin dependence Autism Other 55% 10% 9% 8% 5% 4% 10% 61 0 20 0 20 40 60 80 100 Age Males Females thousands) Figure 3.18: Mental disorder burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 In males, anxiety & depression was the third leading cause of overall male burden, while alcohol abuse was the fourteenth. In females, anxiety & depression was the leading cause of overall female burden, while isolated personality disorders was the thirteenth (Table 3.1). Anxiety & depression also carries with it an increased risk of ischaemic heart disease and suicide. When this risk was accounted for, the burden attributable to anxiety & depression increased from 7.3% to 8.2% of total burden (Table 3.14). The contribution of other mental disorders to the burden of suicide follows in the section on injuries. Table 3.14: Mental disorder burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Anxiety & depression 191,452 14.1 334 0.0 191,786 7.3 Alcohol abuse 19,861 1.5 14,255 1.1 34,116 1.3 Personality disorders 32,587 2.4 0.0 32,587 1.2 Schizophrenia 27,250 2.0 252 0.0 27,502 1.0 Heroin or polydrug abuse 10,287 0.8 6,552 0.5 16,839 0.6 Autism spectrum disorders 13,756 1.0 110 0.0 13,866 0.5 Other 32,198 2.4 1,652 0.1 33,850 1.3 Total mental disorder burden 327,391 24.2 23,154 1.8 350,545 13.3 Ischaemic heart disease attributable to anxiety & depression 1,399 0.1% 5,689 0.4% 7,088 0.3% Suicide attributable to anxiety & depression 200 0.0% 16,708 1.3% 16,908 0.6% Total burden attributable to anxiety & depression 193,051 14.3% 22,731 1.8% 215,783 8.2% 40 Rate per 1,000 0 50 Number (0 20 40 60 80 100 Age Other Schizophrenia Personality disorders Alcohol dependence Anxiety & depression 62 Neurological and sense disorders Neurological & sense disorders were responsible for 11.9% of the total burden of disease and injury in Australia in 2003 (Table 3.11), with dementia, adult-onset hearing loss and vision loss accounting for two-thirds of this burden (Figure 3.19). There were marked sex differentials in the attribution of the neurological & sense disorder burden to particular conditions. Females contributed three times as much to migraine and twice as much to dementia than males. Conversely, the burden from hearing loss was twice as high in males as in females. The greater preponderance of burden from dementia and vision disorders in females was largely due to higher life expectancy in females than males. Only 17% of the burden from neurological & sense disorders was due to mortality. Total Hearing loss Epilepsy Parkinson's disease Vision loss Dementia Migraine 47% 66% 57% 51% 43% 36% 27% 53% 34% 43% 49% 57% 64% 73% Males Females Total Epilepsy Dementia Parkinson's disease Migraine Vision loss Hearing loss 17% 42% 26% 21% 0% 0% 0% 83% 58% 74% 79% 100% 100% 100% Fatal Non-fatal Figure 3.19: Neurological & sense disorder burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 The burden from neurological & sense disorders, both in absolute terms and when expressed as a rate per head of population, increased with age, with a small but important peak in early adulthood due to the contribution of migraine (Figure 3.20). The contribution from dementia increased from middle age to more than half the burden in the elderly and was more pronounced in females than in males. Conversely, the contribution from hearing loss decreased with age and was more pronounced in males than in females. Vision loss made a smaller but important contribution across all ages. Dementia Adult-onset hearing loss Vision loss Parkinson's disease Migraine Epilepsy Other 30% 15% 21% 9% 7% 5% 13% 63 0 50 100 150 0 20 40 60 80 100 Age Males Females Figure 3.20: Neurological & sense disorder burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Adult-onset hearing loss, dementia and Parkinsons disease were the seventh, eleventh and twentieth leading causes of overall male burden. In females, dementia was ranked the fifth leading cause of overall female burden, with hearing loss, migraine and Parkinsons disease ranked eleventh, fourteenth and eighteenth (Table 3.1). Table 3.15: Neurological & sense disorder burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Dementia 70,296 5.2 24,103 1.9 94,399 3.6 Adult-onset hearing loss 64,853 4.8 0 0.0 64,853 2.5 Vision loss 47,865 3.5 9 0.0 47,875 1.8 Parkinsons disease 21,157 1.6 5,695 0.4 26,852 1.0 Migraine 21,841 1.6 7 0.0 21,848 0.8 Epilepsy 8,601 0.6 6,220 0.5 14,821 0.6 Other 24,025 1.8 18,092 1.4 42,118 1.6 Total neurological & sense disorder burden 258,638 19.1 54,127 4.2 312,766 11.9 Rate per 1,000 0 20 40 Number (thousands) 0 20 40 60 80 100 Age Other Migraine Parkinson's disease Vision loss Adult-onset hearing loss Dementia 64 Chronic respiratory diseases Chronic respiratory diseases were responsible for 7.1% of total burden of disease and injury in Australia in 2003 (Table 3.11), with chronic obstructive pulmonary disease and asthma accounting for 46% and 34% of this burden, respectively (Figure 3.21). Males had a greater share of chronic obstructive pulmonary disease burden than females because of the greater prevalence of smoking in males 20 to 30 years ago. Fifty-four per cent of the burden from chronic obstructive pulmonary disease was due to mortality, whereas only 6% of the asthma burden was due to mortality. Total COPD Asthma 53% 57% 46% 47% 43% 54% Males Females Total COPD Asthma 38% 54% 6% 62% 46% 94% Fatal Non-fatal Figure 3.21: Chronic respiratory disease burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and nonfatal outcomes, Australia, 2003 The burden from chronic respiratory diseases per head of population in both sexes was higher in childhood than in middle age, after which it increased exponentially (Figure 3.22). This was due to the high incidence of asthma in childhood, particularly in boys, and then remission as the body matures. Chronic obstructive pulmonary disease, largely from smoking, was the leading cause of chronic respiratory disease burden from middle age onwards. COPD Asthma Other 46% 34% 20% 65 0 20 40 60 Rate 1,000 0 20 40 60 80 100 Age Males Females Figure 3.22: Chronic respiratory disease burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Chronic obstructive pulmonary disease was the sixth leading cause of overall male burden, with asthma at thirteenth. In females, these conditions were ranked seventh and ninth in the leading causes of overall female burden, respectively (Table 3.1). Table 3.16: Chronic respiratory disease burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total COPD 39,543 2.9 47,208 3.7 86,751 3.3 Asthma 59,054 4.4 4,045 0.3 63,100 2.4 Other 16,801 1.2 20,086 1.6 36,887 1.4 Total chronic respiratory disease burden 115,398 8.5 71,339 5.6 186,737 7.1 Injuries Injuries were responsible for 7.0% of the total burden of disease and injury in Australia in 2003 (Table 3.11), with suicide & self-inflicted injuries, road traffic accidents and falls accounting for nearly two-thirds of this burden (Figure 3.23). The burden in males was greater than females for most causes of injury. Males accounted for 73% of the burden due to road traffic accidents and 78% for suicide & self-inflicted injuries. The burden from falls, on the other hand, was equally distributed amongst males and females. Seventy-six per cent of the overall injury burden was due to mortality. per 0 10 20 30 Number (thousands) 0 20 40 60 80 100 Age Other Asthma COPD 66 Total Suicide & self-inflicted Road traffic accidents Homicide & violence Poisoning Falls 70% 78% 73% 71% 57% 50% 30% 22% 27% 29% 43% 50% Males Females Total Suicide & self-inflicted Poisoning Road traffic accidents Homicide & violence Falls 76% 99% 97% 86% 72% 47% 24% 1% 3% 14% 28% 53% Fatal Non-fatal Figure 3.23: Injury burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 The injury burden in males was greater in early adulthood than at other ages, both in absolute terms and when expressed as a rate per head of population (Figure 3.24). For females, the absolute burden was greatest in the very young, while the rate increased with age from a third of the male rate at early adulthood to similar levels at old age. The peak in absolute burden in early adulthood was due to the high mortality from road traffic accidents and suicide at this life stage. The distribution of injury burden by cause was similar between males and females at all ages. In males, suicide & self-inflicted injuries and road traffic accidents were the eighth and twelfth leading cause of overall male burden. In females, only falls were in the 20 leading causes of overall female burden at seventeenth (Table 3.1). 0 10 20 30 Rate 1,000 0 20 40 60 80 100 Age Males Females Figure 3.24: Injury burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 per Suicide & self-inflicted Road traffic Falls accidents Poisoning Homicide & violence Other 27% 23% 14% 7% 5% 24% 0 10 20 Number (thousands) 0 20 40 60 80 100 Age Other Homicide & violence Poisoning Falls Road traffic accidents Suicide & self-inflicted 67 Suicide & self-inflicted injuries were responsible for 27% of the total injury burden in Australia in 2003 (Figure 3.23), with anxiety & depression and alcohol abuse accounting for nearly three-quarters of this burden (Figure 3.25). The burden in males was greater than in females for all major causes of suicide & self-inflicted injury. Anxiety & depression Alcohol dependence Bipolar disorder Schizophrenia Personality disorders Heroin dependence Other 48% 25% 6% 5% 4% 4% 9% Total Heroin dependence Bipolar disorder Schizophrenia Personality disorders Alcohol dependence Anxiety & depression 78% 92% 91% 87% 84% 78% 73% 22% 8% 9% 13% 16% 22% 27% Males Females Total Schizophrenia Heroin dependence Bipolar disorder Personality disorders Anxiety & depression Alcohol dependence 99% 99% 99% 99% 99% 99% 99% 1% 1% 1% 1% 1% 1% 1% Fatal Non-fatal Figure 3.25: Suicide burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Table 3.17: Injury burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Suicide & self-inflicted injuries 537 0.0 49,379 3.9 49,916 1.9 Road traffic accidents 6,073 0.4 36,352 2.8 42,425 1.6 Falls 13,995 1.0 12,391 1.0 26,386 1.0 Poisoning 326 0.0 11,720 0.9 12,046 0.5 Homicide & violence 2,597 0.2 6,624 0.5 9,221 0.4 Other transport accidents 2,873 0.2 5,728 0.4 8,601 0.3 Other 18,001 1.3 18,454 1.4 36,454 1.4 Total injury burden 44,402 3.3 140,648 11.0 185,050 7.0 Diabetes Diabetes was responsible for 5.5% of the total burden of disease and injury in Australia in 2003 (Table 3.11), with Type 2 diabetes accounting for 92% of this burden. Eighty-five per cent of the total diabetes burden was due to diabetes per se (that is, the experience of being diabetic regardless of complications), with the remainder being due to complications such as 68 neuropathy, peripheral vascular disease (PVD), and diabetic foot (Figure 3.26). Twenty-two per cent of the total diabetes burden was due to mortality. Figure 3.26: Diabetes burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes The risk of burden from diabetes in both sexes increased linearly until age 85 then declined (Figure 3.27). The contribution from diabetes per se dominated at all ages. Diabetes ranked second and fourth in the 20 leading causes of burden for males and females, respectively (Table 3.1). Figure 3.27: Diabetes burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Diabetes per se Neuropathy PVD Diabetic foot Other 85% 5% 4%3%4% Total Diabetic foot Diabetes per se Neuropathy PVD 54% 67% 54% 52% 49% 46% 33% 46% 48% 51% Males Females Total Diabetes per se Neuropathy PVD Diabetic foot 22% 27% 0% 0% 0% 78% 73% 100% 100% 100% Fatal Non-fatal 0 50 100 150 Rate per 1,000 0 20 40 60 80 100 Age Males Females 0 50 100 Number (thousands) 0 20 40 60 80 100 Age Other Diabetic foot PVD Neuropathy Diabetes per se 69 Diabetes also carries with it an increased risk of ischaemic heart disease and stroke. When this risk was accounted for, the burden attributable to diabetes increased to 8.3% of total burden (Table 3.18). Table 3.18: Diabetes burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Diabetes per se 89,252 6.6 32,295 2.5 121,547 4.6 Neuropathy 6,500 0.5 0.0 6,500 0.2 Peripheral vascular disease 5,917 0.4 0.0 5,917 0.2 Diabetic foot 3,672 0.3 0.0 3,672 0.1 Amputation 2,455 0.2 0.0 2,455 0.1 Retinopathy 1,258 0.1 0.0 1,258 0.0 Other(a) 2,483 0.2 0.0 2,483 0.1 Total diabetes burden 111,536 8.2 32,295 2.5 143,831 5.5 Ischaemic heart disease attributable to diabetes 8,494 0.6 45,948 3.6 54,442 2.1 Stroke attributable to diabetes 3,985 0.3 16,260 1.3 20,245 0.8 Total burden attributable to diabetes 124,015 9.2 94,503 7.4 218,518 8.3 (a) Includes renal failure. Musculoskeletal diseases Musculoskeletal diseases were responsible for 4.0% of the total burden of burden and injury in Australia in 2003 (Table 3.11), with osteoarthritis, back pain and rheumatoid arthritis accounting for over three-quarters of this burden (Figure 3.28). Only 7% of the musculoskeletal burden was due to mortality. The sex difference evident in the burden due to osteoarthritis and rheumatoid arthritis was mainly a result of the higher female life expectancy, which in turn allowed for more incident cases of this disease. The lack of a plausible physiological or occupational explanation for the large sex difference in burden from occupational overuse syndrome (OOS) provides support to the notion that this syndrome is not a single entity. The risk of burden from musculoskeletal diseases in both sexes increased until age 80 then declined (Figure 3.29). The contribution of back pain was relatively constant until age 70 then declined in proportion to the contribution from osteoarthritis. Osteoarthritis and back pain conditions ranked seventeenth and eighteenth in the 20 leading causes of burden for males, and twelfth and fifteenth for females (Table 3.1). 70 Osteoarthritis Back pain Rheumatoid arthritis Slipped disc Occupational overuse syndrome Other 33% 28% 16% 6% 5% 13% Total Slipped disc Back pain Osteoarthritis Rheumatoid arthritis OOS 42% 56% 49% 42% 28% 14% 58% 44% 51% 58% 72% 86% Males Females Total Rheumatoid arthritis Osteoarthritis Back pain Slipped disc OOS 7% 10% 1% 1% 1% 0% 93% 90% 99% 99% 99% 100% Fatal Non-fatal Figure 3.28: Musculoskeletal disease burden (DALYs) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 10 20 Rate 1,000 0 20 40 60 80 100 Age Males Females 0 5 10 15 (0 20 40 60 80 100 Age Other Occupational overuse syndrome Slipped disc Rheumatoid arthritis Back pain Osteoarthritis Figure 3.29: Musculoskeletal disease burden (DALYs) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 per Number thousands) 71 Table 3.19: Musculoskeletal disease burden (YLD, YLL and DALYs) by specific cause, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Osteoarthritis 34,204 2.5 374 0.0 34,578 1.3 Back pain 29,484 2.2 173 0.0 29,658 1.1 Rheumatoid arthritis 15,215 1.1 1,626 0.1 16,841 0.6 Slipped disc 6,089 0.4 31 0.0 6,120 0.2 Occupational overuse syndrome 4,953 0.4 0.0 4,953 0.2 Gout 1,813 0.1 175 0.0 1,988 0.1 Other 6,722 0.5 4,647 0.4 11,369 0.4 Total musculoskeletal disease burden 98,481 7.3 7,027 0.5 105,508 4.0 Alternative categories for selected conditions The burden from intellectual disability, renal failure and vision disorders was attributed to multiple underlying causes in the primary listing of diseases and injuries and is therefore not discussed explicitly in the above sections. The burden from intellectual disability, apart from congenital conditions (for example Down syndrome), was divided among epilepsy, autism spectrum disorders, infectious diseases, injuries, and perinatal conditions. The burden from renal failure was divided among diabetic nephropathy, the injury category of medical misadventure (analgesic nephropathy), and congenital conditions (dysplasia, polycystic kidneys). The burden from total vision loss was divided among diabetic retinopathy, glaucoma, cataract, refraction errors, age-related macular degeneration and other causes of vision loss. The total burden from intellectual disability, renal failure and total vision loss after re-aggregation was 1.7%, 2.6% and 2.1%, respectively (Table 3.20). Table 3.20: Aggregated burden (YLD, YLL and DALYs) for selected conditions, Australia, 2003 Cause YLD Per cent of total YLL Per cent of total DALYs Per cent of total Intellectual disability 20,999 1.6 23,189 1.8 44,187 1.7 Renal failure 3,809 0.3 64,912 5.1 68,721 2.6 Total vision loss 50,671 3.7 4,868 0.4 55,539 2.1 72 4 Risks to health in Australia 4.1 Overview This chapter discusses the contribution of a number of health risks to the burden of disease and injury in Australia for 2003. The analyses are not meant to be comprehensive since choices had to be made about which risks to include on the basis of the availability of the following: 1. Good evidence of a causal association between the exposure to the risk and the health outcomes 2. Current estimates from reputable epidemiological studies of the relative risk involved 3. Reliable estimates of exposure in the Australian population to the health risk. The outcome of these considerations was a set of 14 selected health risks as outlined in Table 4.1. Several important dietary factors were considered for inclusion (for example sodium and saturated fat) as part of these deliberations, but were ultimately excluded due to inadequate data on exposure. With the exception of low fruit and vegetable consumption, therefore, the impact of poor diet is measured indirectly through the assessments for high body mass, blood cholesterol and blood pressure. Similarly, lack of data on prevalence and outcome prevented estimation of the burden of intimate partner violence in males. Table 4.1: Fourteen selected risks to health discussed in this report Lifestyle behaviours Physiological states Social and environmental factors 1. Tobacco 7. High body mass 11. Urban air pollution 2. Alcohol 8. High blood pressure 12. Intimate partner violence 3. Physical inactivity 9. High blood cholesterol 13. Child sexual abuse 4. Illicit drugs 10. Osteoporosis 14. Occupational exposures & hazards 5. Low fruit and vegetable consumption 6. Unsafe sex It is important to remember several points when interpreting the results in the following sections. Firstly, health risks tend to cluster around high risk individuals who experience more than one exposure (for example smokers tend to be drinkers). This combination of exposures may produce higher or lower levels of overall risk as a result of complex interaction effects. The analyses presented in this chapter do not explicitly account for these interactions, except to the extent to which confounding was controlled for in the studies from which the exposure outcome relationships were derived. Secondly, the causal paths between a number of related health risks and their eventual health outcomes can be complicated. For example, physical inactivity can lead to obesity, which can cause hypertension or high blood cholesterol, which can ultimately lead to cardiovascular disease. Most of the analyses presented in this chapter only measure the effect of a risk independent of the other exposures and irrespective of the risks place in a causal 73 path. The important implication here is that such analyses are not additive. Using the example above, the burden attributable to physical inactivity is estimated to be 23.7% of total cardiovascular disease burden, while that for high body mass, high blood cholesterol and high blood pressure was 19.5%, 34.5% and 42.1% of cardiovascular disease, respectively (Table 4.2). The burden attributable to these health risks in combination, however, is not the sum of burden from each risk (that is, the combined burden is not 119.9%). This is because the combined effect of these risks has to be expressly calculated rather than derived from the addition of their individual effects. Ignoring shared causal paths in this example leads to obvious over-estimation of the combined effect. To illustrate the total explanatory power of the 14 risk factors, the chapter begins with an analysis that accounts for many of the overlaps between risks that share causal paths. This is done using the joint effects method developed for the WHO Comparative Risk Assessment project (Ezzati et al. 2004b). Sensitivity analyses indicate that overall results based on this approach are relatively robust to the underlying assumptions; apportioning the combined overall risk back to each contributing risk factor is more difficult, however, and is much more sensitive to assumptions. Therefore, only the former analyses are presented in this report. Further details on the methods used for estimating joint effects are provided in Chapter 2. 4.2 Combined effect of 14 selected risks to health The 14 selected risk factors presented in this chapter together explained 32.2% of the total burden of disease and injury in Australia in 2003 (Table 4.2). These risk factors explained 35.1% and 29.1% of the total burden in males and females respectively (Table 4.3). This indicates that there is considerable potential to further reduce burden in Australia through interventions that target these health risks, each of which contribute to more than one health outcome. Additional evidence on the (cost-) effectiveness of such interventions may guide the setting of health service priorities to meet this objective. Key findings about broad cause groups were: Ten of the risks were associated with cancer and together explained 32.9% of the total burden from this cause. The majority was explained by tobacco. The contributions of the other risk factors (physical inactivity, high body mass, alcohol, occupational exposure, low fruit and vegetable consumption, air pollution and unsafe sex (through the link between the human papilloma virus and cancer of the cervix)) were comparatively much smaller. Twelve of the risks were associated with cardiovascular disease and together explained 69.3% of the burden from this group of causes; for ischaemic heart disease, this figure was 85.2%. High blood pressure and high blood cholesterol were the largest contributors, followed by physical inactivity, high body mass, tobacco, and low fruit and vegetable consumption. The very low prevalence of smoking in elderly Australians, who are most affected by cardiovascular disease, explains the relatively small contribution of tobacco to this disease. Four of the risks were associated with mental disorders and together explained 26.9% of the burden from this cause. Alcohol and illicit drugs contributed in roughly equal proportions. Intimate partner violence and child sexual abuse contributed less but were the only risks implicated in the large burden from anxiety and depression. 74 Three of the risks were associated with neurological and sense disorders, and together explained only 0.2% of the burden from these disorders. This reflects lack of knowledge about causation in this group. Ultraviolet light, causing cataract, is probably the most obvious omitted risk factor in this disease category but the burden of cataract is small because surgical treatment is widely available. Seven of the risks were associated with injury and together explained 31.7% of the burden from this cause. Alcohol was by far the largest contributor, followed by occupational exposures and hazards, illicit drugs, intimate partner violence, osteoporosis, child sexual abuse, and tobacco. Two of the risks were associated with Type 2 diabetes (including the proportion of cardiovascular disease caused by diabetes) and together explained 60.1% of the burden from this cause. High body mass was by far the largest contributor to this disease. Table 4.2: Individual and joint burden (DALYs) attributable to 14 selected risk factors by broad cause group, Australia, 2003 Broad cause group Cancer CVD Mental Neurological Injury Diabetes Other All causes Total burden ( 000) 499.4 473.8 350.5 312.8 185.1 143.8 667.4 2,632.8Attributable burden (%)(a) Tobacco 20.1 9.7 0.6 0.5 8.9 7.8 High blood pressure 42.1 7.6 High body mass 3.9 19.5 54.7 1.1 7.5 Physical inactivity 5.6 23.7 23.7 >0.1 6.6 High blood cholesterol 34.5 6.2 Alcohol Harmful effects 3.1 0.9 9.7 18.1 <0.1 3.3 Beneficial effects 5.6 >0.1 1.0 Net effects 3.1 4.7 9.7 18.1 <0.1 2.3 Low fruit & vegetable consumption 2.0 9.6 >0.1 2.1 Illicit drugs <0.1 8.0 3.6 2.5 2.0 Occupational exposures & hazards 3.1 0.4 0.8 4.7 3.4 2.0 Intimate partner violence 0.5 0.3 5.5 0.1 2.5 0.2 1.1 Child sexual abuse <0.1 <0.1 5.8 1.4 <0.1 0.9 Urban air pollution 0.8 2.7 0.4 0.7 Unsafe sex 1.0 1.4 0.6 Osteoporosis 2.4 0.2 Joint effect(b) 32.9 69.3 26.9 0.2 31.7 60.1 17.2 32.2 (a) Attributable burden within each column is expressed as a percentage of total burden for that column. (b) Figures for joint effects are not column totals. See Section 4.1 for further details. The 14 selected risk factors presented in this chapter had a differential impact on health in terms of both sex and age (Table 4.3). In the 044 year-old age group, alcohol and illicit drugs 75 were the leading causes of burden in males, mental disorders (alcohol abuse, and heroin and polydrug abuse) and injuries (suicide and self-inflicted injuries, and road traffic accidents) being the predominant health outcomes from these risks. In this age group, 23.6% of total male burden and 17.9% of total female burden was explained by the 14 risks in combination. In females, intimate partner violence and child sexual abuse were the leading causes in this age group, anxiety and depression and suicide and self-inflicted injuries being the predominant health outcomes from these risks. In the 4564 year-old age group, high body mass and tobacco were the leading causes in both sexes, Type 2 diabetes, ischaemic heart disease, stroke, lung cancer and chronic obstructive pulmonary disease (COPD) being the predominant health outcomes from these risks. The proportion of total burden in this age group that is explained by the 14 risks in combination was 43.8% in males and 33.6% in females. In the 65 years and over age group, high blood pressure was the leading cause in both sexes, followed by tobacco in males and high blood cholesterol in females. The predominant health outcomes from both high blood pressure and high blood cholesterol are ischaemic heart disease and stroke. For tobacco, the predominant health outcomes are lung cancer and COPD. The proportion of total burden in this age group that is explained by the 14 risks in combination was 38.4% and 34.8% in males and females, respectively. Table 4.3: Individual and joint burden (DALYs) attributable to 14 selected risk factors by sex and age group, Australia, 2003 Males Females 044 4564 65+ All ages 044 4564 65+ All agesTotal burden ( 000) 448.8 382.5 533.4 1,364.6 406.0 299.1 563.0 1,268.2Attributable burden (%)(a) Tobacco 1.9 14.7 12.5 9.6 1.1 8.7 7.6 5.8 High blood pressure 0.8 7.8 13.8 7.8 <0.1 4.0 14.2 7.3 High body mass 3.3 13.3 7.5 7.7 2.6 12.1 8.1 7.3 Physical inactivity 1.8 9.0 8.5 6.4 1.9 8.4 9.6 6.8 High blood cholesterol 1.9 9.6 8.3 6.6 0.7 5.1 9.9 5.8 Alcohol Harmful effects 8.1 5.5 1.8 4.9 2.2 2.4 0.8 1.6 Beneficial effects 0.3 1.5 1.5 1.1 0.2 0.9 1.5 0.9 Net effects 7.8 4.0 0.3 3.8 2.0 1.4 0.6 0.7 Low fruit & vegetable consumption 0.8 4.1 3.3 2.7 0.3 1.7 2.2 1.5 Illicit drugs 5.7 1.9 0.6 2.7 2.4 1.1 0.4 1.2 Occupational exposures & hazards 2.7 4.2 1.4 2.6 1.6 2.4 0.4 1.3 Intimate partner violence 4.8 2.8 0.3 2.3 Child sexual abuse 0.6 0.3 <0.1 0.3 3.4 1.7 <0.1 1.5 Urban air pollution 0.2 0.9 1.2 0.8 0.1 0.6 1.2 0.7 Unsafe sex 0.8 0.4 0.2 0.5 1.0 0.9 0.4 0.7 Osteoporosis <0.1 0.2 <0.1 <0.1 0.6 0.3 Joint effect(b) 23.6 43.8 38.4 35.1 17.9 33.6 34.8 29.1 (a) Attributable burden within each column is expressed as a percentage of total burden for that column. (b) Figures for joint effects are not column totals. See Section 4.1 for further details. 76 4.3 Individual contribution of 14 selected risks to health Tobacco Tobacco was responsible for 7.8% of the total burden of disease and injury in Australia in 2003 (Table 4.4), with lung cancer, COPD and ischaemic heart disease accounting for more than three-quarters of this burden (Figure 4.1). Of the 14 risk factors examined, tobacco was responsible for the largest amount of burden across all ages in males (Table 4.3). Almost two-thirds of the burden from tobacco was experienced by males due to the higher prevalence 20 to 30 years ago of smoking in males compared with females. More than threequarters of the burden from tobacco was due to mortality (Figure 4.1). Because of the long lag time between smoking and many of its ill effects on health, the health benefits of recent favourable trends in smoking prevalence will not be fully realised until many years in the future. The rate of burden from tobacco per head of population increased with age until 75 and the absolute burden was concentrated between the ages of 55 and 75. The contribution from lung cancer dominated at most ages but was overtaken by contributions from COPD and ischaemic heart disease in the elderly (Figure 4.2). Table 4.4: Deaths and burden (DALYs) attributable to tobacco by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Lung cancer 6,309 4.8 72,213 2.7 COPD 4,175 3.2 54,492 2.1 Ischaemic heart disease 1,962 1.5 31,435 1.2 Stroke 577 0.4 11,812 0.4 Oesophagus cancer 572 0.4 6,248 0.2 Other 1,916 1.4 28,588 1.1 Total attributable 15,511 11.7 204,788 7.8 77 Total IHD Oesophagus cancer Lung cancer Stroke COPD 64% 73% 71% 65% 61% 58% 36% 27% 29% 35% 39% 42% Males Females Total Oesophagus cancer Lung cancer IHD COPD Stroke 78% 94% 93% 80% 62% 57% 22% 6% 7% 20% 38% 43% Fatal Non-fatal Figure 4.1: Burden (DALYs) attributable to tobacco by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 20 40 60 Rate per 1,000 0 20 40 60 80 100 Age Males Females Number thousands) Figure 4.2: Burden (DALYs) attributable to tobacco by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 High blood pressure High blood pressure was responsible for 7.6% of the total burden of disease and injury in Australia in 2003 (Table 4.5), with ischaemic heart disease and stroke accounting for 93% of Lung cancer COPD Ischaemic heart disease Stroke Oesophagus cancer Other 35% 27% 15% 6% 3% 14% 0 10 20 30 (0 20 40 60 80 100 Age Other Oesophagus cancer Stroke Ischaemic heart disease COPD Lung cancer 78 this burden (Figure 4.3). Of the 14 risk factors examined, high blood pressure was responsible for the greatest amount of burden in the 65 years or over age group in both sexes (Table 4.3). Overall, the burden from high blood pressure was somewhat greater in males and 81% was due to mortality. The rate of burden from high blood pressure per head of population increased with age and the absolute burden was concentrated around old age (Figure 4.4). The contributions from ischaemic heart disease and stroke dominated across all ages. Table 4.5: Deaths and burden (DALYs) attributable to high blood pressure by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Ischaemic heart disease 14,089 10.7 125,461 4.8 Stroke 6,603 5.0 59,962 2.3 Other 1,812 1.4 13,893 0.5 Total attributable 22,504 17.0 199,315 7.6 Ischaemic heart disease Stroke Other 63% 30% 7% Total IHD Stroke 54% 58% 47% 46% 42% 53% Males Females Total IHD Stroke 81% 83% 76% 19% 17% 24% Fatal Non-fatal Figure 4.3: Burden (DALYs) attributable to high blood pressure by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 79 Rate 1,000 Figure 4.4: Burden (DALYs) attributable to high blood pressure by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 High body mass The body mass index (BMI) is a measure of weight in kilograms over height in metres squared and is typically categorised into under weight (BMI<20), normal weight (20BMI<25), over weight (25BMI<30) and obese (BMI30). Rather than use these categories, the health effects of high body mass in the following analyses were estimated using new methods in which BMI is measured on a continuous scale and risk is assessed against a minimum counterfactual distribution with a mean of 21 and a SD of 1 (see Appendix 2). This means that risk is attributed to all people in the population with a BMI of greater than 21, with the degree of risk increasing exponentially above this value. The consequence of this approach is that some of the attributable risk from high body mass comes from the large proportion of the population that is not over weight or obese in the conventional sense, but whose risk of disease is elevated, at least to some degree. High body mass was responsible for 7.5%of the total burden of disease and injury in Australia in 2003 (Table 4.6), with Type 2 diabetes and ischaemic heart disease (IHD) accounting for almost three-quarters of this burden (Figure 4.5). Of the 14 risk factors examined, high body mass accounted for the greatest amount of burden in the 4564 year age group in females (Table 4.3). The burden from high body mass was greater in males due to the higher incidence of Type 2 diabetes itself and the associated cardiovascular complications. Half of the burden from high body mass was due to mortality (Figure 4.5). The rate of burden from high body mass per head of population increased with age; the absolute burden was concentrated between the ages of 55 and 75. The contributions from Type 2 diabetes and ischaemic heart disease dominate across all ages (Figure 4.6). 0 100 200 per 0 20 40 60 80 100 Age Males Females 0 20 40 Number (thousands) 0 20 40 60 80 100 Age Other Stroke Ischaemic heart disease 80 Table 4.6: Deaths and burden (DALYs) attributable to high body mass by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Type 2 diabetes 1,381 1.0 78,688 3.0 Ischaemic heart disease 4,914 3.7 66,533 2.5 Stroke 1,528 1.2 22,218 0.8 Colorectal cancer 721 0.5 9,920 0.4 Breast cancer 379 0.3 7,125 0.3 Other 602 0.5 13,148 0.5 Total attributable 9,525 7.2 197,632 7.5 Total IHD Colorectal cancer Type 2 diabetes Stroke Breast cancer 53% 64% 54% 54% 50% 0% 47% 36% 46% 46% 50% 100% Males Females Total Colorectal cancer IHD Breast cancer Stroke Type 2 diabetes 50% 81% 80% 67% 61% 17% 50% 19% 20% 33% 39% 83% Fatal Non-fatal Figure 4.5: Burden (DALYs) attributable to high body mass by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Type 2 diabetes Ischaemic heart disease Stroke Colorectal cancer Breast cancer Other 40% 34% 11% 5% 4% 7% 81 0 20 40 Rate 1,000 0 20 40 60 80 100 Age Males Females Figure 4.6: Burden (DALYs) attributable to high body mass by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Physical inactivity Physical inactivity was responsible for 6.6% of the total burden of disease and injury in Australia in 2003 (Table 4.7), with ischaemic heart disease, Type 2 diabetes and stroke accounting for more than four-fifths of this burden. Overall, the burden from physical inactivity was shared equally between the sexes. With the exception of diabetes, most of the conditions attributable to physical inactivity were associated with high mortality (Figure 4.7). The rate of burden from physical inactivity per head of population increased with age and the absolute burden was concentrated around old age. The contributions from ischaemic heart disease and Type 2 diabetes dominated across all ages (Figure 4.8). Table 4.7: Deaths and burden (DALYs) attributable to physical inactivity by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Ischaemic heart disease 8,739 6.6 88,617 3.4 Type 2 diabetes 704 0.5 34,132 1.3 Stroke 2,390 1.8 23,742 0.9 Colorectal cancer 1,074 0.8 14,978 0.6 Breast cancer 584 0.4 12,962 0.5 Total attributable 13,491 10.2 174,431 6.6 per 0 10 20 30 Number (thousands) 0 20 40 60 80 100 Age Other Breast cancer Colorectal cancer Stroke Ischaemic heart disease Type 2 diabetes 82 Total IHD Colorectal cancer Type 2 diabetes Stroke Breast cancer 50% 58% 54% 53% 43% 0% 50% 42% 46% 47% 57% 100% Males Females Total IHD Colorectal cancer Stroke Breast cancer Type 2 diabetes 67% 82% 81% 69% 66% 19% 33% 18% 19% 31% 34% 81% Fatal Non-fatal Figure 4.7: Burden (DALYs) attributable to physical inactivity by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 50 100 1,000 0 20 40 60 80 100 Age Males Females Figure 4.8: Burden (DALYs) attributable to physical inactivity by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Rate per Ischaemic heart disease Type 2 diabetes Stroke Colorectal cancer Breast cancer 51% 20% 14% 9% 7% 0 10 20 Number (thousands) 0 20 40 60 80 100 Age Breast cancer Colorectal cancer Stroke Type 2 diabetes Ischaemic heart disease 83 High blood cholesterol High blood cholesterol was responsible for 6.2% the total burden of disease and injury in Australia in 2003 (Table 4.8), with ischaemic heart disease and stroke accounting for this entire burden. Both ischaemic heart disease and stroke were associated with high mortality. Overall, males experienced a slightly higher burden from high blood cholesterol than females (Figure 4.9). The rate of burden from high blood cholesterol per head of population increased with age and the absolute burden was concentrated around old age. The contribution from ischaemic heart disease dominated across all ages (Figure 4.10). Table 4.8: Deaths and burden (DALYs) attributable to high blood cholesterol by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Ischaemic heart disease 13,371 10.1 138,605 5.3 Stroke 1,980 1.5 24,986 0.9 Total attributable 15,351 11.6 163,591 6.2 Ischaemic heart disease Stroke 85% 15% Total IHD Stroke 55% 56% 46% 45% 44% 54% Males Females Total IHD Stroke 79% 82% 64% 21% 18% 36% Fatal Non-fatal Figure 4.9: Burden (DALYs) attributable to high blood cholesterol by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 84 Number Figure 4.10: Burden (DALYs) attributable to high blood cholesterol by age expressed as: (a) ratesby sex, and (b) numbers by specific cause, Australia, 2003 Alcohol Alcohol has both hazardous and protective effects on health, and the age and sex distribution of these effects varies in important ways. Of the 14 risk factors examined, alcohol was responsible for the greatest amount of burden in males under the age of 45 (Table 4.3). Alcohol harm was responsible for 3.2% of the total burden of disease and injury in Australia in 2003. Alcohol also prevented 0.9% per cent of the total burden in 2003 (Table 4.9). The benefits of alcohol consumption outweigh its harmful effects only in females over the age of 65. Given that the net impact of alcohol was to contribute to 2.3% of total burden, it is important to understand that, even though moderate intake of alcohol may have beneficial effects at middle and older ages, alcohol is harmful when taken in excess at all ages. Alcohol abuse, road traffic accidents and suicide contributed two-thirds of the harm attributed to alcohol (Figure 4.11). This study reports a substantially lower health benefit due to alcohol compared to the previous Australian burden study (AIHW: Mathers et al. 1999, AIHW: Ridolfo & Stevenson 2001) with only an estimated 2,346 deaths being prevented in 2003 compared to 7,157 deaths in 1996. This is due to the previous study underestimating the number of people who abstain from alcohol or drink less than 0.25 drinks per day. 0 50 100 Rate per 1,000 0 20 40 60 80 100 Age Males Females 0 10 20 (thousands) 0 20 40 60 80 100 Age Stroke Ischaemic heart disease 85 Table 4.9: Deaths and burden (DALYs) attributable to alcohol by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Harm Alcohol abuse 918 0.7 34,116 1.3 Suicide & self-inflicted injuries 553 0.4 12,245 0.5 Road traffic accidents 396 0.3 11,121 0.4 Oesophagus cancer 368 0.3 4,594 0.2 Breast cancer 184 0.1 4,152 0.2 Other 1,012 0.8 19,207 0.7 Total attributable harm 3,430 2.6 85,435 3.2 Benefit Ischaemic heart disease 1,950 1.5 20,659 0.8 Stroke 380 0.3 3,451 0.1 Other 16 0.0 233 0.0 Total attributable benefit 2,346 1.8 24,343 0.9Total attributable 1,084 0.8 61,091 2.3 Alcohol dependence Suicide & self-inflicted Road traffic accidents Oesophagus cancer Breast cancer Other 39% 13% 14% 5% 5% 25% Total Road traffic accidents Suicide & self-inflicted Alcohol dependence Oesophagus cancer Breast cancer 76% 89% 83% 80% 79% 0% 24% 11% 17% 20% 21% 100% Males Females Total Suicide & self-inflicted Oesophagus cancer Road traffic accidents Breast cancer Alcohol dependence 66% 99% 94% 88% 65% 42% 34% 1% 6% 12% 35% 58% Fatal Non-fatal Figure 4.11: Burden (DALYs) attributable to alcohol (alcohol harm) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 86 0 5 10 0 20 40 60 80 100 Age Males Females 0 5 10 Number 0 20 40 60 80 100 Age Other Breast cancer Oesophagus cancer Road traffic accidents Suicide & self-inflicted Alcohol dependence Figure 4.12: Burden (DALYs) attributable to alcohol (alcohol harm) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Ischaemic heart disease Stroke Other 77% 22% 1% Total IHD Stroke 56% 72% 0% 44% 28% 100% Males Females Total IHD Stroke 79% 83% 66% 21% 17% 34% Fatal Non-fatal Figure 4.13: Burden (DALYs) prevented due to alcohol (alcohol benefit) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 Rate per 1,000 (thousands) 87 Rate per 1,000 Figure 4.14: Burden (DALYs) attributable to alcohol (alcohol benefit) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Low fruit and vegetable consumption Low fruit and vegetable consumption was responsible for 2.1% of the total burden of disease and injury in Australia in 2003 (Table 4.10). Eating enough fruit and vegetables helps to prevent cancers, ischaemic heart disease and, to a lesser extent, stroke. Sixty-nine per cent of the burden from low fruit and vegetable consumption was due to ischaemic heart disease and two-thirds was experienced by males, partly because males tend to eat less fruit and vegetables than females, but also because males have a higher burden from ischaemic heart disease than females. Overall, 81%of the burden from low fruit and vegetable consumption was due to mortality. The absolute burden from low fruit and vegetable consumption peaked between the age of 60 and 80 while the rate per head of population continued to increase until old age. The contribution from ischaemic heart disease dominated at all ages (Figure 4.16). Table 4.10: Deaths and burden (DALYs) attributable to low fruit and vegetable consumption by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Ischaemic heart disease 3,219 2.4 37,981 1.4 Stroke 605 0.5 7,346 0.3 Lung cancer 463 0.3 5,956 0.2 Other 281 0.2 3,977 0.2 Total attributable 4,568 3.5 55,259 2.1 0 10 20 0 20 40 60 80 100 Age Males Females 0 2 4 Number (thousands) 0 20 40 60 80 100 Age Other Stroke Ischaemic heart disease 88 Ischaemic heart disease Stroke Lung cancer Stomach cancer Other 69% 13% 11% 3%4% Total Lung cancer IHD Stomach cancer Stroke 66% 68% 68% 66% 54% 34% 32% 32% 34% 46% Males Females Total Lung cancer Stomach cancer IHD Stroke 81% 93% 91% 82% 64% 19% 7% 9% 18% 36% Fatal Non-fatal Figure 4.15: Burden (DALYs) attributable to low fruit and vegetable consumption by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 10 20 Rate per 1,000 0 20 40 60 80 100 Age Males Females 0 5 10 Number (thousands) 0 20 40 60 80 100 Age Other Stomach cancer Lung cancer Stroke Ischaemic heart disease Figure 4.16: Burden (DALYs) attributable to low fruit and vegetable consumption by age expressed as: (a) rates by sex , and (b) numbers by specific cause, Australia, 2003 Illicit drugs Illicit drugs were responsible for 2.0% of the total burden of disease and injury in Australia in 2003 (Table 4.11). Illicit drugs are a direct cause of death and disability as well as being 89 risk factors for conditions such as HIV/AIDS, hepatitis, low birth weight, inflammatory heart disease, poisoning, and suicide and self-inflicted injuries. Almost three-quarters of the burden from illicit drugs was experienced by males because males are more likely to both use illicit drugs and adopt drug habits that put them at risk of dying. Overall, fifty-seven per cent of the burden from illicit drugs was due to mortality (Figure 4.17). The burden from illicit drugs, both in terms of rate per head of population and in absolute terms, peaked in early adulthood when drug addiction usually begins. The contribution from heroin dominated at this age but was overtaken by contributions from hepatitis B and C with increasing age as the long-term effects of drug use begin to manifest (Figure 4.18). Table 4.11: Deaths and burden (DALYs) attributable to illicit drugs by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Heroin & polydrug abuse 263 0.2 16,758 0.6 Hepatitis C 759 0.6 11,709 0.4 Cannabis abuse 0 0.0 5,206 0.2 Suicide & self-inflicted injuries 204 0.2 4,458 0.2 Hepatitis B 329 0.2 3,637 0.1 Benzodiazepine abuse 1 0.0 2,656 0.1 Other 149 0.1 7,040 0.3 Total attributable 1,705 1.3 51,463 2.0 Total Cannabis Suicide & self-inflicted Heroin Hepatitis C Hepatitis B Benzodiazepine 71% 78% 78% 74% 68% 64% 42% 29% 22% 22% 26% 32% 36% 58% Males Females Total Suicide & self-inflicted Hepatitis C Hepatitis B Heroin Benzodiazepine Cannabis 57% 99% 97% 97% 39% 1% 0% 43% 1% 3% 3% 61% 99% 100% Fatal Non-fatal Figure 4.17: Burden (DALYs) attributable to illicit drugs by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and nonfatal outcomes, Australia, 2003 Heroin Hepatitis C Cannabis Suicide & self-inflicted Hepatitis B Benzodiazepine Other 33% 23% 10% 9% 7% 5% 14% 90 0 5 10 0 20 40 60 80 100 Age Males Females Number thousands) Figure 4.18: Burden (DALYs) attributable to illicit drugs by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Occupational exposures and hazards Occupational exposures and hazards were responsible for 2.0% of the total burden of disease and injury in Australia in 2003 (Table 4.12). More than two-thirds of this burden was experienced by males, mostly because occupational exposures and hazards occur in industries dominated by male employment. Females, however, experienced 86% of the burden from occupational overuse syndrome (OOS). Overall, 43% of the burden from occupational exposures and hazards was due to mortality (Figure 4.19). The burden from occupational exposures and hazards was concentrated in the working ages and peaked in middle age, both in terms of rate per head of population and in absolute terms (Figure 4.20). Table 4.12: Deaths and burden (DALYs) attributable to occupational exposures and hazards by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Cancer 1,154 0.9 15,559 0.6 Back pain 1 0.0 7,806 0.3 Occupational overuse syndrome 4,944 0.2 COPD 111 0.1 4,563 0.2 Road traffic accidents 124 0.1 2,975 0.1 Other 264 0.2 15,515 0.6 Total attributable 1,654 1.3 51,362 2.0 Rate per 1,000 0 5 10 (0 20 40 60 80 100 Age Other Benzodiazepine dependence Hepatitis B Suicide & self-inflicted Cannabis dependence Hepatitis C Heroin dependence 91 Cancer Back pain Occupational overuse syndrome COPD Road traffic accidents Other 30% 15% 10% 9% 6% 30% Total COPD Road traffic accidents Cancer Back pain OOS 69% 91% 89% 76% 58% 14% 31% 9% 11% 24% 42% 86% Males Females Total Road traffic accidents Cancer COPD Back pain OOS 43% 98% 83% 27% 0% 0% 57% 2% 17% 73% 100% 100% Fatal Non-fatal Figure 4.19: Burden (DALYs) attributable to occupational exposures and hazards by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 5 10 0 20 40 60 80 100 Age Males Females 0 5 10 Number (thousands) 0 20 40 60 80 100 Age Other Road traffic accidents COPD Occupational overuse syndrome Back pain Cancer Figure 4.20: Burden (DALYs) attributable to occupational exposures and hazards by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Rate per 1,000 92 Intimate partner violence The attribution of burden to intimate partner violence was attempted only for females due to insufficient evidence on prevalence and risk among males. While this risk is unlikely to be zero, it is probably small in comparison with the risk experienced by females. Intimate partner violence was responsible for 1.1% of the total burden of disease and injury in Australia in 2003 (Table 4.13). Of the 14 risk factors examined, intimate partner violence contributed most to the burden in females under the age of 45 (Table 4.3). Most of the burden from intimate partner violence was due to anxiety and depression, and conditions arising due to the associated increased use of tobacco, alcohol and illicit substances (Figure 4.21). The burden from intimate partner violence, both in terms of rate per head of population and in absolute terms, peaked at around age 30 then declined with age (Figure 4.22). The contribution from anxiety and depression dominated throughout adulthood but was overtaken by contributions from tobacco-related disease with increasing age as the effects of higher smoking rates begin to manifest. Table 4.13: Deaths and burden (DALYs) for females attributable to intimate partner violence by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Anxiety & depression 3 0.0 18,358 0.7 Suicide & self-inflicted injuries 131 0.1 3,099 0.1 Lung cancer 89 0.1 1,477 0.1 Homicide & violence 35 0.0 1,260 0.0 COPD 49 0.0 1,114 0.0 Other 128 0.1 4,051 0.2 Total attributable 435 0.3 29,360 1.1 93 Total Anxiety & depression Suicide & self-inflicted Lung cancer Homicide & violence COPD 0% 0% 0% 0% 0% 0% 100% 100% 100% 100% 100% 100% Males Females Total Suicide & self-inflicted Lung cancer Homicide & violence COPD Anxiety & depression 27% 98% 93% 63% 49% 0% 73% 2% 7% 37% 51% 100% Fatal Non-fatal Figure 4.21: Burden (DALYs) attributable to intimate partner violence by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 2 4 6 per 1,000 0 20 40 60 80 100 Age Males Females (thousands) 0 20 40 60 80 100 Age Other COPD Homicide & violence Lung cancer Suicide & self-inflicted Anxiety & depression Figure 4.22: Burden (DALYs) attributable to intimate partner violence by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Child sexual abuse Child sexual abuse was responsible for 0.9% of the total burden of disease and injury in Australia in 2003 (Table 4.14). Ninety-four per cent of this burden was due to anxiety and depression Anxiety & Suicide & depression self-inflicted Lung cancer Homicide & violence COPD Other 11% 63% 5% 4% 4% 14% Rate 0 2 4 Number 94 depression, suicide and self-inflicted injuries, and alcohol abuse. Of the 14 risk factors examined, child sexual abuse was the second leading cause of burden in females under the age of 45 (Table 4.3). Just over four-fifths of the burden from child sexual abuse was experienced by females and 14% was due to mortality (Figure 4.23). The burden from child sexual abuse, both in terms of rate per head of population and in absolute terms, peaked at around 40 years-old then declined with age. The contribution from anxiety and depression dominated at this age after which contributions from suicide and self-inflicted injuries and alcohol abuse became increasingly important (Figure 4.24). Table 4.14: Deaths and burden (DALYs) attributable to child sexual abuse by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Anxiety and depression 7 0.0 19,133 0.7 Suicide & self-inflicted injuries 103 0.1 2,258 0.1 Alcohol abuse 24 0.0 730 0.0 Other 62 0.0 1,392 0.1 Total attributable 196 0.1 23,513 0.9 Total Alcohol dependence Suicide & self-inflicted Anxiety & depression 18% 59% 45% 11% 82% 41% 55% 89% Males Females Total Suicide & self-inflicted Alcohol dependence Anxiety & depression 15% 98% 54% 0% 85% 2% 46% 100% Fatal Non-fatal Figure 4.23: Burden (DALYs) attributable to child sexual abuse by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and nonfatal outcomes, Australia, 2003 Anxiety & depression Suicide & self-inflicted Alcohol dependence Other 81% 10% 3% 6% 95 0 2 4 0 20 40 60 80 100 Age Males Females Figure 4.24: Burden (DALYs) attributable to child sexual abuse by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Urban air pollution The health effects of urban air pollution are largely chronic conditions (such as ischaemic heart disease, lung cancer and stroke) resulting from long-term exposure to this risk. There may also be an additional burden from short-term exposure to abnormally high levels of urban air pollution, although this risk is more controversial. Table 4.15 provides estimates for both long-term and short-term effects; all other figures in this section reflect the long-term effects only. Urban air pollution was responsible for 1.0% of the total burden of disease and injury in Australia in 2003 (Table 4.15). Sixty-two per cent of the burden from urban air pollution was due to cardiovascular disease (ischaemic heart disease and stroke) and 53% of the burden from urban air pollution was experienced by males. Overall, 80% of the burden from urban air pollution was due to mortality (Figure 4.25). The absolute burden from urban air pollution peaked at age 80 while the rate per head of population continued to increase until old age. The contribution from cardiovascular disease dominated at all ages (Figure 4.26). Rate per 1,000 0 2 4 Number (thousands) 0 20 40 60 80 100 Age Other Alcohol dependence Suicide & self-inflicted Anxiety & depression 96 Table 4.15: Deaths and burden (DALYs) attributable to urban air pollution by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Long-term Ischaemic heart disease 959 0.7 8,483 0.3 Lung cancer 351 0.3 4,115 0.2 Stroke 432 0.3 3,738 0.1 COPD 184 0.1 2,654 0.1 Other 83 0.1 748 0.0 Total attributable to long-term exposure 2,009 1.5 19,738 0.7 Short-term Total attributable to short-term exposure 1,046 0.8 7,781 0.3 Total attributable 3,056 2.3 27,519 1.0 97 Ischaemic heart disease Lung cancer Stroke COPD Other 43% 21% 19% 13% 4% Total Lung cancer COPD IHD Stroke 53% 59% 55% 54% 41% 47% 41% 45% 46% 59% Males Females Total Lung cancer IHD Stroke COPD 80% 93% 83% 75% 55% 20% 7% 17% 25% 45% Fatal Non-fatal Figure 4.25: Burden (DALYs) attributable to urban air pollution (long-term effects) by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and non-fatal outcomes, Australia, 2003 0 5 10 15 0 20 40 60 80 100 Age Males Females 0 1 2 3 Number (thousands) 0 20 40 60 80 100 Age Other COPD Stroke Lung cancer Ischaemic heart disease Figure 4.26: Burden (DALYs) attributable to urban air pollution (long-term effects) by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Rate per 1,000 98 Unsafe sex Unsafe sex was responsible for 0.6% of the total burden of disease and injury in Australia in 2003 (Table 4.16). Over two-thirds of this burden was due to cervix cancer and HIV/AIDS. Sixty-three per cent of the burden from unsafe sex was due to mortality (Figure 4.27). The burden from unsafe sex in males peaked in early adulthood due to the impact of HIV infection, after which it declined and the long-term effects of hepatitis B infection began to manifest. In females, the rate per head of population continued to increase with age and the absolute burden was concentrated around middle age when the contribution from cervix cancer dominated (Figure 4.28). Table 4.16: Deaths and burden (DALYs) attributable to unsafe sex by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Cervix cancer 298 0.2 5,231 0.2 HIV/AIDS 105 0.1 4,873 0.2 Hepatitis B 225 0.2 2,499 0.1 Other 26 0.0 2,293 0.1 Total attributable 655 0.5 14,897 0.6 Cervix cancer HIV/AIDS Hepatitis B Other 35% 33% 17% 15% Total HIV/AIDS Hepatitis B Cervix cancer 42% 93% 61% 0% 58% 7% 39% 100% Males Females Total Hepatitis B Cervix cancer HIV/AIDS 63% 96% 83% 45% 37% 4% 17% 55% Fatal Non-fatal Figure 4.27: Burden (DALYs) attributable to unsafe sex by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and nonfatal outcomes, Australia, 2003 99 0 1 2 Rate per 0 20 40 60 80 100 Age Males Females 0 1 2 0 20 40 60 80 100 Age Other Hepatitis B HIV/AIDS Cervix cancer Figure 4.28: Burden (DALYs) attributable to unsafe sex by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Osteoporosis Osteoporosis was responsible for 0.2% of the total burden of disease and injury in Australia in 2003 (Table 4.17). Almost all of this burden was due to falls and more than three-quarters was experienced by females. More than half of the burden from osteoporosis was due to mortality (Figure 4.29). The burden from osteoporosis was experienced from age 60 onwards. The contribution from falls dominated at all ages (Figure 4.30). Table 4.17: Deaths and burden (DALYs) attributable to osteoporosis by specific cause, Australia, 2003 Deaths DALYs Specific cause Number Per cent of total Number Per cent of total Falls 534 0.4 4,329 0.2 Other 10 0.0 58 0.0 Total attributable 545 0.4 4,386 0.2 1,000 Number (thousands) 100 Falls Other 99% 1% Total Falls 23% 23% 77% 77% Males Females Total Falls 57% 56% 43% 44% Fatal Non-fatal Figure 4.29: Burden (DALYs) attributable to osteoporosis by specific cause expressed as: (a) proportions of total, (b) proportions by sex, and (c) proportions due to fatal and nonfatal outcomes, Australia, 2003 0 5 10 15 per 0 20 40 60 80 100 Age Males Females 0 1 1 2 0 20 40 60 80 100 Age Other Falls Figure 4.30: Burden (DALYs) attributable to osteoporosis by age expressed as: (a) rates by sex, and (b) numbers by specific cause, Australia, 2003 Rate 1,000 Number (thousands) 101 5 Differentials in burden of disease and injury across Australia 5.1 Overview This chapter describes differentials in burden of disease and injury across Australia in terms of the following stratifications of the population: state and territory jurisdictions, socioeconomic quintiles and remoteness categories (major cities, regional and remote). The chapter begins by comparing life expectancy and health-adjusted life expectancy (HALE) across each subpopulation within these strata. It then discusses the main differentials between subpopulations by leading causes of burden. Table 5.1 summarises for each subpopulation the important demographic characteristics that influence these differentials. Table 5.1: Selected demographic characteristics by area, Australia, 2003 Population(a) Per cent of population for area Age group (years) Area (000)Per cent of Australia <15 1559 6079 80+ Males Indigenous(b) Low SES(c)Jurisdiction NSW 6,687.5 33.6 19.9 62.4 14.2 3.5 49.7 2.0 19.0 Vic 4,918.0 24.7 19.5 62.9 14.1 3.4 49.3 0.6 16.0 Qld 3,797.3 19.1 20.8 62.9 13.3 3.0 49.9 3.3 23.9 WA 1,952.5 9.8 20.4 64.0 12.8 2.8 50.0 3.3 22.5 SA 1,527.6 7.7 18.8 61.7 15.4 4.1 49.5 1.6 14.9 Tas 477.2 2.4 20.4 60.6 15.4 3.7 49.3 3.6 55.4 ACT 322.9 1.6 19.8 67.4 10.7 2.2 49.4 1.2 0.3 NT 198.4 1.0 25.4 67.4 6.5 0.7 52.5 28.8 28.4 Socioeconomic quintile Low 3,917.1 19.7 21.9 61.5 13.7 2.9 50.0 n.a. n.a. Mod. low 3,973.8 20.0 21.2 60.6 14.9 3.3 49.8 n.a. n.a. Average 3,747.8 18.8 20.3 61.9 14.3 3.4 49.8 n.a. n.a. Mod. high 4,097.3 20.6 19.4 64.5 13.0 3.1 49.6 n.a. n.a. High 4,145.4 20.8 17.4 65.4 13.5 3.7 49.1 n.a. n.a. Remoteness Major cities 13,347.9 66.8 19.1 64.4 13.5 3.3 49.6 1.0 17.2 Regional 6,050.5 30.4 21.7 59.7 15.3 3.4 50.0 3.2 23.7 Remote 483.1 2.4 25.6 63.4 9.2 1.8 53.2 25.7 39.2 Australia 19,894.7 100.0 20.0 62.8 13.9 3.3 49.6 2.3 19.7(a) Estimated resident population figures as at 30 June 2003 (ABS cat. no. 3201 .0).(b) Based on people identifying as Indigenous in the 2001 Census (ABS cat. no. 2019.0 2019.8). (c) Based on Socio-Economic Indexes for Areas (SEIFA) (ABS cat. no. 2039.0.55.001). 102 5.2 Health-adjusted life expectancy HALE provides an estimate of the average years of equivalent healthy life that a person can expect to live at various ages. HALE is related to life expectancy, which provides an estimate of the average years of life a person can expect to live at various ages given current risks of mortality. HALE extends this concept by reducing the estimated duration by the proportion of time spend at each age in states less than perfect health, adjusted for the relative severity of those health states. The sum of prevalent years lost due to disability (PYLD) across all causes is used to derive this severity-weighted proportion for each age. Since the starting point for HALE is a life table, life expectancy at birth for the various subpopulations discussed in this chapter is presented first in Table 5.2. Table 5.2: Life expectancy at birth by area and sex, Australia, 2003 Life expectancy at birth (years) Area Males Females Persons Jurisdiction NSW 78.2 (78.078.3) 83.1 (83.083.3) 80.6 (80.580.8) Vic 78.6 (78.478.8) 83.2 (83.083.4) 80.9 (80.881.0) Qld 78.4 (78.278.6) 83.3 (83.183.5) 80.8 (80.781.0) WA 79.0 (78.779.3) 83.7 (83.484.0) 81.3 (81.181.5) SA 77.7 (77.378.0) 82.9 (82.583.2) 80.3 (80.080.5) Tas 76.7 (76.177.3) 81.7 (81.182.2) 79.2 (78.879.6) ACT 80.2 (79.480.9) 84.2 (83.484.9) 82.3 (81.782.8) NT 73.1 (72.274.0) 78.6 (77.679.6) 75.5 (74.876.1) Socioeconomic quintile Low 76.9 (76.777.1) 82.3 (82.182.5) 79.6 (79.479.7) Moderately low 77.4 (77.277.6) 82.8 (82.683.0) 80.0 (79.980.2) Average 77.7 (77.577.9) 82.7 (82.582.9) 80.2 (80.080.3) Moderately high 79.0 (78.879.2) 83.5 (83.383.7) 81.2 (81.181.4) High 80.9 (80.681.1) 84.5 (84.384.7) 82.7 (82.582.8) Remoteness Major cities 78.8 (78.778.9) 83.5 (83.483.6) 81.2 (81.181.2) Regional 77.5 (77.477.7) 82.7 (82.582.8) 80.0 (79.980.1) Remote 75.4 (74.876.1) 81.5 (80.982.2) 78.1 (77.678.6) Australia 78.3 (78.278.4) 83.2 (83.183.3) 80.7 (80.780.8)When interpreting the results presented in this chapter it is important to keep in mind that Indigenous people are a much greater proportion of the total population in the Northern Territory and remote areas of Australia. This accounts for the much greater health loss in these areas, although the contribution of Indigenous populations to this loss is not quantified in this report. Readers seeking such comparisons are referred to the separate report on the Indigenous component of this study. Once the Indigenous results are available separate small area comparisons can be made for non-Indigenous people. This is relevant to health 103 policy in that there is a raft of Indigenous health issues that is distinct from the health issues of the general population living in remote areas. HALE was calculated for subpopulations using the PYLD estimated for each population separately, as discussed in Chapter 2. Total HALE at birth across Australia in 2003 was 70.6 years for males, 75.2 years for females and 72.9 years for both sexes combined (Table 5.3). The figures for both sexes ranged from 67.7 to 75.9 years across state and territory jurisdictions, 71.2 to 75.5 years across socioeconomic quintiles and 69.5 to 73.5 years across remoteness categories. Table 5.3: Health-adjusted life expectancy (HALE) and life expectancy at birth lost due to disability by area and sex, Australia, 2003 Health-adjusted life expectancy (HALE) (years) At birth At age 60 Life expectancy at birth lost due to disability (%) Area Males Females Persons Males Females Persons Males Females Persons Jurisdiction NSW 70.5 75.3 72.9 17.1 20.6 18.9 9.8 9.5 9.6 Vic 71.1 75.4 73.2 17.5 20.8 19.2 9.6 9.4 9.5 Qld 70.5 75.3 72.8 17.0 20.4 18.7 10.1 9.7 9.9 WA 71.5 75.6 73.5 17.5 20.6 19.1 9.6 9.6 9.6 SA 69.3 74.2 71.7 16.4 20.0 18.3 10.8 10.5 10.6 Tas 68.8 73.7 71.3 16.3 19.7 18.1 10.2 9.8 10.0 NT 65.8 70.2 67.7 12.6 15.1 13.6 10.0 10.6 10.3 ACT 73.9 77.8 75.9 18.9 21.9 20.5 7.8 7.5 7.7 Socioeconomic quintile Low 68.7 73.8 71.2 16.1 19.7 17.9 10.7 10.4 10.6 Moderately low 69.5 74.6 72.0 16.4 20.1 18.2 10.2 9.9 10.1 Average 69.9 74.6 72.2 16.6 20.1 18.4 10.0 9.8 9.9 Moderately high 71.4 75.9 73.6 17.6 20.8 19.3 9.7 9.1 9.4 High 73.8 77.2 75.5 19.2 21.9 20.6 8.7 8.7 8.7 Remoteness Major cities 71.3 75.6 73.5 17.5 20.8 19.2 9.6 9.4 9.5 Regional 69.6 74.5 72.0 16.5 20.1 18.3 10.3 9.8 10.1 Remote 67.3 72.3 69.5 15.4 18.5 16.8 10.8 11.3 11.0 Australia 70.6 75.2 72.9 17.1 20.5 18.9 9.8 9.6 9.7 104 When the difference between life expectancy and HALE is expressed as a proportion of life expectancy, this represents the proportion of remaining life that is lost due to disability. Hereafter this is referred to as PLD (proportion of life expectancy lost due to disability). PLD at birth is the most commonly reported figure, although it can be calculated at any age and increases with age (Figure 5.1). 0 50 100 0 50 100 0 50 100 Males Females Remaining years lost due to disability Remaining years of healthy life Percent Age Figure 5.1: Proportion of life expectancy lost due to disability (%) by age and sex, Australia, 2003 This report shows for the first time that there are differentials in PLD at birth across Australia (Table 5.3). There was a strong socioeconomic gradient in this measure, with the lowest socioeconomic quintile losing 10.6% of life expectancy at birth through disability and the highest losing only 8.7%. Differentials with respect to remoteness category were also apparent but not as large, with remote areas losing 11.0% and major cities losing 9.5%. With respect to state and territory jurisdictions, the Australian Capital Territory had the lowest PLD at birth at 7.7% and South Australia had the highest at 10.6%. 105 NSW Vic Qld SA WA Tas NT ACT Low Average Mod low Mod high High Major cities Regional Remote 76 78 80 82 76 78 80 82 76 78 80 82 8 9 10 11 8 9 10 11 8 9 10 11 State/Territory SES quintile Remoteness Life expectancy at birth (years) Proportion of life expectancy at birth lost due to disability (%) Figure 5.2: Life expectancy at birth (years) versus proportion of life expectancy at birth lost due to disability (%) by area, Australia, 2003 Figure 5.2 shows the inverse relationship between life expectancy at birth and PLD, with subpopulations experiencing the highest life expectancy also having the lowest PLD. In other words, longevity is associated with lower average levels of disability throughout the life span. The remainder of this chapter presents health differentials across these subpopulations using the standard burden metric of disability-adjusted life years(DALYs). All rates per head of population were standardised to remove the effect of different age structures between populations. This standard technique is used when comparing populations whereby the agespecific rates of the populations of interest are applied to the age structure of a reference population before comparisons are made. 5.3 State and territory differentials The proportion of burden experienced by each state and territory jurisdiction was roughly proportional to the population size, with New South Wales accounting for the largest proportion (34.0%), followed by Victoria (24.8%) and Queensland (18.6%) (Table 5.4). Males experienced more of this burden than females in all jurisdictions except the Australian Capital Territory where it was more equally distributed between the sexes. In all jurisdictions except Tasmania, slightly more of total burden was due to non-fatal causes. 106 Table 5.4: Burden (DALYs) for state/territory jurisdictions by proportions of total, proportions by sex and proportions due to mortality, Australia, 2003 Area DALYs (000) Per cent of total Per cent male Per cent fatal burdenNSW 895.8 34.0 51.9 49.5 Vic 651.6 24.8 50.9 48.9 Qld 488.5 18.6 53.0 46.9 SA 234.3 8.9 51.5 48.7 WA 236.8 9.0 51.7 46.6 Tas 73.4 2.8 51.6 51.4 NT 22.9 0.9 58.5 46.6 ACT 29.5 1.1 50.4 47.5 Australia 2,632.8 100.0 51.8 48.6There were important differentials in burden experienced per head of population between jurisdictions. After age standardisation, the Northern Territory had almost twice the rate of total burden of the Australian Capital Territory for both males and females. This was due to higher rates of burden for most causes, but particularly for cardiovascular disease, diabetes and injuries (Figure 5.3). 0 50 100 150 200 NT Tas SA Qld NSW Vic WA ACT NT Tas SA Vic NSW Qld WA ACT Males Females Cancer Cardiovascular Chronic respiratory Diabetes Injuries Mental Neurological Other Rate per 1,000 Figure 5.3: Age-standardised DALY rates per 1,000 by state/territory jurisdiction, broad cause group and sex, Australia, 2003 Table 5.5 provides a comparison between burden rates for jurisdictions and the national average for the 10 leading broad causes of burden in Australia for 2003. Of these causes, the greatest difference between jurisdictions with the lowest and highest rates occurred (in order of magnitude of difference) in diabetes, injuries, genitourinary conditions and chronic respiratory diseases. The causes that contributed most in terms of the absolute difference 107 observed between jurisdictions were cardiovascular disease (19.7%), diabetes (15.5%) and injuries (13.6% for intentional and unintentional combined). Table 5.5: Differentials in burden (DALYs) by state/territory jurisdiction for the 10 leading broad cause groups, Australia, 2003 Rate Standardised rate ratio(b) % diff. % of Broad cause group Aust.(a) NSW Vic Qld SA WA Tas NT ACT high/low(c) total diff.(d) Cancer 25.1 1.00 1.02 0.98 1.06 0.96 1.13 1.15 0.87 31.3 6.9 Cardiovascular 23.8 1.04 0.94 1.02 1.09 0.88 1.13 1.61 0.78 104.8 19.7 Mental 17.6 1.03 0.97 1.01 1.06 0.94 1.15 1.05 0.76 52.1 7.0 Neurological 15.7 0.99 0.97 1.00 1.13 1.04 1.02 0.97 0.78 44.3 5.5 Chronic respiratory 9.4 0.99 0.98 0.99 1.20 0.94 1.18 1.72 0.81 111.6 8.6 Diabetes 7.2 0.88 1.15 0.95 1.14 1.00 1.15 2.71 0.57 371.8 15.5 Unintentional injuries 6.3 0.96 0.95 1.08 1.01 1.06 1.13 2.03 0.68 196.7 8.6 Musculoskeletal 5.3 0.96 1.00 1.05 1.02 1.05 1.18 0.96 0.87 35.5 1.7 Genitourinary 3.3 1.01 1.04 0.93 1.06 0.94 1.01 1.76 0.77 127.9 3.3 Intentional injuries 3.0 0.94 0.89 1.11 1.10 1.05 1.18 2.46 0.79 210.3 5.0 All causes 132.4 1.00 0.99 1.00 1.09 0.96 1.12 1.50 0.79 88.7 100.0 (a) DALY rate for Australia per 1,000. (b) Ratio of age-standardised DALYs per 1,000 population for area to DALYs per 1,000 population for Australia. (c) Calculated for each cause as the greatest difference in DALY rates between areas as a proportion of lowest rate for that cause. (d) Calculated for each cause as the greatest difference in DALY rates between areas as a proportion of greatest difference for all causes. Table 5.6 lists the 10 leading specific causes of burden for Australia and summarises for each jurisdiction these causes in terms of rank order and percentage of total burden. Diseases of old age, such as ischaemic heart disease and dementia, contributed less to the total burden in jurisdictions with younger populations (for example the Northern Territory and the Australian Capital Territory). 108 Table 5.6: Differentials in burden (DALYs) by state/territory jurisdiction for the 10 leading specific causes, Australia, 2003 Rank Per cent of total Specific cause(a) NSW Vic Qld SA WA Tas NT ACT NSW Vic Qld SA WA Tas NT ACT Ischaemic heart disease 1 1 1 1 1 1 3 2 10.4 9.6 10.2 10.8 8.8 10.7 6.5 8.1 Anxiety & depression 2 2 2 2 2 2 1 1 7.2 7.1 7.9 6.0 8.0 7.3 8.4 9.3 Type 2 diabetes 4 3 3 3 3 3 2 4 4.4 5.9 4.8 5.3 5.4 5.0 7.9 3.5 Stroke 3 4 4 4 5 4 11 3 5.0 4.3 4.4 4.5 3.9 4.4 2.0 3.9 Dementia 5 5 7 5 4 8 15 10 3.8 3.4 3.2 4.0 4.3 2.5 1.2 2.5 Lung cancer 7 6 5 7 6 6 10 7 3.4 3.4 3.3 3.2 3.5 3.8 2.3 2.8 COPD 6 7 6 6 7 5 7 8 3.4 3.1 3.3 3.7 2.8 4.0 3.3 2.6 Adult-onset hearing loss 10 10 8 8 10 9 35 12 2.2 2.5 2.9 2.6 2.4 2.4 0.7 2.3 Colorectal cancer 8 8 10 9 9 7 22 11 2.3 2.6 2.3 2.4 2.5 2.7 0.9 2.4 Asthma 11 9 9 11 8 10 8 5 2.2 2.5 2.5 2.3 2.7 2.4 2.3 3.3 (a) Sorted according to the leading specific causes for Australia. 5.4 Differentials by socioeconomic status Populations in areas with lower socioeconomic status experienced proportionally more burden than populations in areas with higher socioeconomic status (Table 5.7). Females experienced slightly more burden than males in areas with the highest socioeconomic status. Conversely, males experienced more burden than females in areas with the lowest socioeconomic status. The highest proportion of burden that was fatal was in the moderately low and average socioeconomic areas, and the lowest (47.6%) was in the low socioeconomic area. Table 5.7: Burden (DALYs) for socioeconomic quintiles by proportions of total, proportions by sex, and proportions due to mortality, Australia, 2003 Area DALYs ( 000) Per cent of total Per cent male Per cent fatal burdenLow SES 562.5 21.4 52.8 47.6 Moderately low SES 564.2 21.4 52.7 49.3 Average SES 523.6 19.9 52.1 49.5 Moderately high SES 507.7 19.3 52.0 48.0 High SES 474.8 18.0 49.1 48.4 Australia 2,632.8 100.0 51.8 48.6 Total burden per head of population increased with decreasing socioeconomic status, with the most disadvantaged populations having 31.7% greater burden than the most advantaged populations. Again, this was due to higher rates of burden for most causes, but particularly for mental disorders and cardiovascular disease (Figure 5.4). 109 0 50 100 150 Low Mod low Average Mod high High Low Mod low Average Mod high High Males Females Cancer Cardiovascular Chronic respiratory Diabetes Injuries Mental Neurological Other Rate per 1,000 Figure 5.4: Age-standardised DALY rates per 1,000 by socioeconomic quintile, broad cause group and sex, Australia, 2003 Table 5.8 provides a comparison between burden rates for areas by socioeconomic category and the national average for the 10 leading broad causes of burden in Australia for 2003. Of these causes, the greatest difference between areas with the lowest and highest rates occurred (in order of magnitude of difference) in diabetes, injuries, mental disorders and chronic respiratory diseases. The causes that contributed most in terms of the absolute difference observed between socioeconomic quintiles were mental disorders (20.9%), cardiovascular disease (17.6%) and diabetes (12.2%). Lifestyle-related (that is behavioural) risk factors are important underlying risks for these conditions; the much greater burden from these causes in lower socioeconomic areas is likely to be due to the greater prevalence of lifestyle risk factors in these areas compared with higher socioeconomic areas. Limited data availability on exposures by socioeconomic status, however, prevented further exploration of this association. 110 Table 5.8: Differentials in burden (DALY rates) by socioeconomic quintile for the 10 leading broad cause groups, Australia, 2003 Rate Standardised rate ratio(b) % diff. % of Broad cause group Aust.(a) Low Mod. low Average Mod. high High high/low(c) total diff.(d) Cancer 25.1 1.05 1.05 1.05 0.97 0.88 19.3 12.0 Cardiovascular 23.8 1.10 1.08 1.05 0.95 0.84 31.8 17.6 Mental 17.6 1.22 1.05 1.02 0.92 0.80 53.5 20.9 Neurological 15.7 1.02 1.02 1.03 1.00 0.93 10.2 4.2 Chronic respiratory 9.4 1.15 1.07 1.01 0.95 0.83 38.8 8.4 Diabetes 7.2 1.30 1.05 1.09 0.91 0.70 87.2 12.2 Unintentional injuries 6.3 1.14 1.12 1.12 0.93 0.72 57.8 7.3 Musculoskeletal 5.3 1.08 1.02 1.05 0.97 0.89 20.5 2.7 Genitourinary 3.3 1.07 1.02 1.04 0.97 0.92 16.0 1.4 Intentional injuries 3.0 1.28 1.11 1.00 0.91 0.73 75.1 4.6 All causes 132.4 1.12 1.05 1.04 0.96 0.85 31.7 100.0 (a) DALY rate for Australia per 1,000. (b) Ratio of age-standardised DALYs per 1,000 population for area to DALYs per 1,000 population for Australia. (c) Calculated for each cause as the greatest difference in DALY rates between areas as a proportion of lowest rate for that cause. (d) Calculated for each cause as the greatest difference in DALY rates between areas as a proportion of greatest difference for all causes. Table 5.9 lists the 10 leading specific causes of burden for Australia and summarises for each socioeconomic quintile these causes in terms of rank order and percentage of total burden. Ischaemic heart disease and anxiety & depression were the leading causes of burden across all socioeconomic quintiles. Table 5.9: Differentials in burden (DALYs) by socioeconomic quintile for the 10 leading specific causes, Australia, 2003 Rank Per cent of total Specific cause(a) Low Mod. low Average Mod. high High Low Mod. low Average Mod. high High Ischaemic heart disease 1 1 1 1 1 9.8 10.5 10.2 9.6 9.8 Anxiety & depression 2 2 2 2 2 8.1 7.3 6.8 7.5 6.6 Type 2 diabetes 3 3 3 3 5 5.9 5.0 5.3 4.8 4.2 Stroke 4 4 4 4 3 4.0 4.6 4.6 4.5 4.9 Dementia 7 6 5 5 4 2.9 3.5 3.7 3.7 4.2 Lung cancer 6 5 6 6 6 3.5 3.6 3.5 3.2 3.0 COPD 5 7 7 7 7 3.7 3.5 3.3 3.1 2.8 Adult-onset hearing loss 9 8 9 8 11 2.4 2.5 2.5 2.6 2.4 Colorectal cancer 10 9 8 9 9 2.1 2.4 2.5 2.5 2.6 Asthma 8 10 11 11 10 2.5 2.4 2.2 2.5 2.5 (a) Sorted according to the leading specific causes for Australia. 111 5.5 Differentials by remoteness The majority (64.5%) of the burden was experienced by people in the major cities as they account for 67% of the population. Regional areas accounted for 33.1% of the burden and remote areas 2.5% (Table 5.10). Males experienced more of this burden than females in all areas, but particularly in remote areas. Remote areas experienced proportionately slightly less fatal burden than other areas. Table 5.10: Burden (DALYs) for remoteness categories by proportions of total, proportions by sex, and proportions due to mortality, Australia, 2003 Area DALYs (000) Per cent of total Per cent male Per cent fatal burdenMajor cities 1,698.0 64.5 51.0 48.2 Regional 870.1 33.1 53.1 49.6 Remote 64.6 2.5 57.5 46.2 Australia 2,632.8 100.0 51.8 48.6Total burden per head of population increased with remoteness, with remote populations having 26.5% greater burden than populations in major cities. Again, this is due to higher rates of burden for most causes, but particularly for injuries (Figure 5.5). 0 50 100 150 200 Remote Regional Major cities Remote Regional Major cities Males Females Cancer Cardiovascular Chronic respiratory Diabetes Injuries Mental Neurological Other Rate per 1,000 Figure 5.5: Age-standardised DALY rates per 1,000 by remoteness category, broad cause group and sex, Australia, 2003 112 Table 5.11: Differentials in burden (DALY rates) by remoteness category for the 10 leading broad cause groups, Australia, 2003 Rate Standardised rate ratio(b) % diff. % of Broad cause group Aust.(a) Major cities Regional Remote high/low(c) total diff.(d) Cancer 25.1 0.98 1.04 0.98 7.0 4.6 Cardiovascular 23.8 0.96 1.07 1.10 14.6 9.1 Mental 17.6 0.98 1.05 1.06 8.5 4.0 Neurological 15.7 0.99 1.03 1.03 4.2 1.8 Chronic respiratory 9.4 0.97 1.04 1.30 33.6 8.3 Diabetes 7.2 0.94 1.08 1.93 105.6 19.5 Unintentional injuries 6.3 0.87 1.24 1.92 121.3 18.1 Musculoskeletal 5.3 0.95 1.10 0.99 16.0 2.2 Genitourinary 3.3 1.00 0.99 1.11 12.3 1.1 Intentional injuries 3.0 0.90 1.13 2.26 151.5 11.0 All causes 132.4 0.97 1.06 1.22 26.5 100.0 (a) DALY rate for Australia per 1,000. (b) Ratio of age-standardised DALYs per 1,000 population for area to DALYs per 1,000 population for Australia. (c) Calculated for each cause as the greatest difference in DALY rates between areas as a proportion of lowest rate for that cause. (d) Calculated for each cause as the greatest difference in DALY rates between areas as a proportion of greatest difference for all causes. Table 5.11 provides a comparison between burden rates for areas by remoteness category and the national average for the 10 leading broad causes of burden in Australia for 2003. Of these causes, the greatest difference between areas with the lowest and highest rates occurred (in order of magnitude of difference) in injuries, diabetes, chronic respiratory diseases, musculoskeletal disorders and cardiovascular disease. The cause that contributed by far the greatest proportion in terms of the absolute difference observed between remoteness categories was injuries (29.1% for intentional and unintentional combined), followed by diabetes (19.5%) and cardiovascular disease (9.1%). Table 5.12 lists the 10 leading specific causes of burden for Australia and summarises for each remoteness category these causes in terms of rank order and percentage of total burden. Type 2 diabetes was the leading cause of burden in remote areas whereas dementia was ranked twelfth, reflecting the younger age structure and higher proportion of Indigenous people in these areas compared with the rest of Australia. 113 Table 5.12: Differentials in burden (DALYs) by remoteness category for the 10 leading specific causes, Australia, 2003 Rank Per cent of total Specific cause(a) Major cities Regional Remote Major cities Regional Remote Ischaemic heart disease 1 1 2 9.8 10.6 7.3 Anxiety & depression 2 2 3 7.4 7.1 6.4 Type 2 diabetes 3 3 1 4.9 5.1 7.7 Stroke 4 4 8 4.6 4.4 2.8 Dementia 5 7 12 3.8 3.3 2.0 Lung cancer 6 6 10 3.4 3.5 2.6 COPD 7 5 4 3.1 3.6 3.8 Adult-onset hearing loss 11 8 11 2.3 2.7 2.1 Colorectal cancer 9 9 15 2.4 2.5 1.4 Asthma 8 10 9 2.5 2.3 2.7 (a) Sorted according to the leading specific causes for Australia. 114 6 Past, present and future burden of disease and injury in Australia 6.1 Overview This chapter presents trends in population health dynamics over a thirty-year period. The analyses involved consideration of health statistics over the last 25 years or more, although the discussion about the past is linked to health trends over the last decade. Also presented are the projected levels of the burden of disease and injury if these trends were to continue 20 years into the future. Since mortality is the starting point for many of these analyses, observed and projected trends in mortality by broad cause group are summarised in Table 6.1. The methods underlying all analyses presented in this chapter are described in detail in Chapter 2. Table 6.1: Changes in mortality by broad cause group and sex, Australia, 1993 to 2023 Standardised rate ratio(a) Rate per 100,000 for 2003 Males Females Broad cause group Males Females 1993 2003 2013 2023 1993 2003 2013 2023 Infectious 14.8 9.5 0.93 1.00 0.97 0.92 0.95 1.00 0.92 0.83 Acute respiratory(b) 16.5 20.9 0.43 1.00 1.00 1.00 0.40 1.00 1.00 1.00 Maternal 0.1 1.87 1.00 1.26 1.19 Neonatal 3.6 2.7 1.74 1.00 0.63 0.41 1.01 1.00 0.66 0.46 Nutritional 0.2 0.6 2.83 1.00 1.24 1.09 1.72 1.00 0.74 0.65 Cancer 211.1 163.7 1.20 1.00 0.89 0.75 1.12 1.00 0.92 0.82 Other neoplasms 4.3 4.1 1.06 1.00 0.88 0.74 0.92 1.00 0.96 0.91 Diabetes 19.5 16.6 1.01 1.00 0.96 0.90 1.13 1.00 0.88 0.76 Endocrine 5.6 7.0 1.73 1.00 1.00 0.89 0.95 1.00 1.07 1.06 Mental 10.4 3.4 1.18 1.00 0.90 0.75 1.10 1.00 0.91 0.79 Neurological 27.8 41.7 1.10 1.00 0.99 0.91 0.93 1.00 1.04 1.05 Cardiovascular 237.9 252.6 1.61 1.00 0.71 0.47 1.52 1.00 0.76 0.52 Chronic respiratory 48.1 37.7 1.38 1.00 0.82 0.69 1.00 1.00 1.04 1.07 Digestive 14.7 19.4 1.18 1.00 0.74 0.55 1.13 1.00 0.80 0.63 Genitourinary 16.7 20.1 1.01 1.00 0.95 0.87 0.86 1.00 0.98 0.93 Skin 1.2 2.0 1.06 1.00 0.96 0.83 0.88 1.00 1.03 1.04 Musculoskeletal 2.6 5.1 1.29 1.00 0.94 0.79 1.10 1.00 0.95 0.90 Congenital 4.2 3.4 1.17 1.00 0.75 0.60 1.30 1.00 0.75 0.58 Oral 0.0 0.1 1.61 1.00 1.29 1.11 0.39 1.00 0.73 0.74 Ill defined 0.5 0.6 3.13 1.00 0.54 0.25 2.01 1.00 0.69 0.52 Injuries 52.4 27.7 1.13 1.00 0.87 0.73 1.00 1.00 0.87 0.74 All causes 692.1 639.0 1.32 1.00 0.83 0.67 1.22 1.00 0.87 0.73 (a) Ratio of age-standardised mortality rates for year to mortality rates for 2003. (b) Age-specific rates for pneumonia post-2003 held at 2003 rates due to coding discontinuities between ICD-9 and ICD-10 for this cause. 115 6.2 Health-adjusted life expectancy This section begins, as did Chapter 5, by presenting life expectancy and health-adjusted life expectancy, but this time with a temporal dimension rather than with a focus on differentials between subpopulations. Over the last decade, total life expectancy in Australia improved from 78.0 years in 1993 to 80.7 years in 2003. This was an annual growth of 0.35% (or 0.28 of a year per year). If past mortality trends continue into the future as projected (that is, at an exponentially declining rate), life expectancy will increase to 82.6 years in 2013 and 84.6 years in 2023, an increase of 3.9 years from 2003. This represents an annual growth of 0.24% (or 0.20 of a year per year) over the 20-year period (Table 6.2). Table 6.2: Life expectancy and health-adjusted life expectancy by sex, Australia, 1993 to 2023 1993 2003 2013 2023 Males Females Both Males Females Both Males Females Both Males Females Both Population(a) (millions) 8.8 8.9 17.7 9.9 10.0 19.9 11.0 11.2 22.2 12.1 12.3 24.5 Proportion of population at selected ages (%) 059 years 85.8 82.8 84.3 84.1 81.6 82.8 79.8 77.6 78.7 75.1 72.7 73.9 6079 years 12.5 14.1 13.3 13.5 14.2 13.9 16.9 17.3 17.1 20.3 21.1 20.7 80+ years 1.6 3.2 2.4 2.4 4.2 3.3 3.4 5.1 4.2 4.6 6.1 5.4 Life expectancy (years) At birth 75.0 81.0 78.0 78.3 83.2 80.7 80.6 84.8 82.6 83.3 86.5 84.6 At age 60 19.5 23.8 21.7 22.1 25.6 23.9 23.7 26.8 25.2 25.9 28.1 26.8 At age 80 7.2 9.1 8.4 8.4 9.9 9.3 9.1 10.5 9.8 10.6 11.3 10.7 Health-adjusted life expectancy (years) At birth 68.0 73.5 70.7 70.6 75.2 72.9 72.5 76.6 74.5 74.7 78.0 76.2 At age 60 15.2 19.2 17.3 17.1 20.5 18.9 18.4 21.5 19.9 20.1 22.5 21.2 At age 80 4.7 6.3 5.6 5.4 6.8 6.2 5.9 7.2 6.6 6.8 7.7 7.1 Healthy life expectancy lost due to disability (years) At birth 7.0 7.5 7.3 7.7 7.9 7.8 8.0 8.2 8.1 8.6 8.5 8.5 At age 60 4.4 4.6 4.5 5.0 5.1 5.0 5.3 5.3 5.2 5.8 5.6 5.6 At age 80 2.5 2.8 2.7 3.0 3.2 3.1 3.2 3.3 3.3 3.7 3.6 3.5 Healthy life expectancy lost due to disability as proportion of total life expectancy (%) At birth 9.4 9.2 9.3 9.8 9.6 9.7 9.9 9.6 9.8 10.3 9.9 10.0 At age 60 22.3 19.3 20.6 22.6 19.8 21.1 22.2 19.7 20.8 22.6 19.9 21.0 At age 80 35.2 31.1 32.5 35.8 31.8 33.2 35.3 31.8 33.1 35.4 31.9 33.3 (a) Estimated resident population figures as at 30 June 1993 and 2003 (ABS 2006, Cat. no. 3201.0, Table 9) and ABS population projections series 8 (ABS 2003a, Cat. no. 3222.0). Health-adjusted life expectancy, on the other hand, increased from 70.7 years to 72.9 years in the decade to 2003, an annual growth of 0.31% (or 0.22 of a year per year). If, in addition to past mortality trends, trends in non-fatal health conditions that give rise to disability continue into the future as projected, health-adjusted life expectancy will increase to 74.5 years in 2013 and 76.2 years in 2023. This represents an annual growth of 0.22% (or 0.16 of a year per year) over the 20-year period. 116 Complex dynamics in population health will drive the slower gains in health-adjusted life expectancy relative to total life expectancy. The most important of these is the decline in mortality rates between 1993 and 2023 across the life span, but particularly in the elderly (Figure 6.1a). One of the consequences of declining mortality is that, in combination with ongoing declines in fertility, Australias population will continue to age. Of particular relevance is the number of people aged 80 years and older. Over the last decade, the proportion of the total population in this age group increased from 2.4% in 1993 to 3.3% in 2003. Based on recent Australian Bureau of Statistics (ABS) projections (ABS 2003a), this is expected to increase to 4.2% in 2013 and 5.4% in 2023 (Table 6.2). 0 100 200 300 100 200 300 400 60 70 80 90 100 60 70 80 90 100 Mortality Disability 1993 2023 Rate per 1,000 Age Figure 6.1: Age-specific (a) mortality, and (b) total prevalence-based years lived with disability (PYLD) expressed as rates per 1,000 in the elderly for both sexes combined, Australia, 1993 and 2023 The impact on life expectancy of declining mortality rates is straightforwardit will increase. The impact on health-adjusted life expectancy and its corollary, life expectancy lost due to disability, however, is perhaps less intuitive at first. The key point is that, in most populations, even if the prevalence of disability at each age were to remain at constant levels, a decline in mortality would mean an increase in life expectancy lost due to disability in the future (Figure 6.2). This is because reductions in mortality result in more people surviving through to ages when the probability of being disabled is highest. Ultimately, though, this relationship depends on changes in the rate at which mortality increases with age relative to changes in the rate at which disability increases with age. 117 0 5 10 0 50 100 0 50 100 Males Females 1993 2023 LE lost due to (years) Age Figure 6.2: Impact on life expectancy lost due to disability (years) of declining mortality and constant (1993) levels of disability, Australia, 1993 to 2023 In addition to the increase in the proportion of total life expectancy lost due to disability through reductions in mortality, is the impact of temporal trends in diseases and injuries that give rise to the prevalence of disability. By estimating separately the epidemiology of these causes in a fully temporal model, changes in total prevalence of disability by age, sex and cause can be quantified for the first time over the past as well as into the future. While the prevalence of overall disability appears to decrease when the effect of population ageing is removed (by standardising for age), it will consistently increase over the next two decades in crude terms (Figure 6.3). In other words, the proportion of overall time lived with disability will increase from 7.8% in 2003 to 8.9% in 2023, an increase of 14.1%. 8.0 7.8 7.6 7.5 7.4 7.8 8.3 8.9 7 8 9 1993 2003 2013 2023 1993 2003 2013 2023 Standardised Crude Prevalence Year Figure 6.3: Age-standardised and crude total prevalence-based years lived with disability (PYLD) for both sexes combined, Australia, 1993 to 2023 This is for two reasons. First, the number of people aged 80 years and over is set to expand rapidly due to declining mortality (Figure 6.1a). Second, while the prevalence of disability will drop at most ages, it will actually increase in this age group (Figure 6.1b). In the decade disability (%) 118 to 2003, disability in people aged 80 years and over increased by 2.0%; if past trends continue, by 2023 disability will have increased a further 1.7% (Figure 6.4). This lends support to the hypothesis which predicts that as population health improves, disability is increasingly concentrated towards the end of the life span. 0.0 -6.2 -11.0 -13.8 0.0 -0.5 -6.3 -9.4 2.0 3.3 3.7 -15 0 15 1993 2003 2013 2023 1993 2003 2013 2023 1993 2003 2013 2023 60-69 years 70-79 years 80+ years Change from 1993 (%) Year Figure 6.4: Per cent change in the rates of total prevalence-based years lived with disability (PYLD) since 1993 at selected age groups for both sexes combined, Australia, 2003 to 2023 The effect on health-adjusted life expectancy of the increasing concentration of disability towards the end of life, which until now has been largely unexplored using empirical data, can be illustrated by gains in expectation of life in a model in which disability is included as a dynamic force over time, compared with a counterfactual scenario in which the probability of disability is held constant (Figure 6.5 and Figure 6.6). Such a comparison provides insights into the question What impact will concentration of disability towards the end of the lifespan have on health-adjusted life expectancy? Health-adjusted life expectancy at birth is the most commonly cited measure, and summarises mortality and disability risks across the life span. In males, this will increase at a rate faster than would have been observed through reductions in mortality alone, with the net effect of morbidity being increasingly concentrated towards the end of life. From 1993 to 2003 the gain was about 0.2 years. Over the longer term, however, the gain will be larger, at around 0.8 years of healthy life for males born in 2023. At adult ages, the gains are less and, in the elderly, where gains in health expectancy due to declines in mortality are more easily offset by increases in disability, there were losses in healthy life due to this dynamic in the decade to 2003, but these disappear in the subsequent decade (Figure 6.5). 11.0 0.5 6.3 0.0 2.0 6069 years 7079 years 80+ years 0 119 -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 0 50 100 0 50 100 0 50 100 2003 2013 2023 Net gain (years) Age Figure 6.5: Net years gained in health-adjusted life expectancy (years) from changes in prevalence of disability compared with baseline (1993) levels of disability for males, Australia, 2003 to 2023 In females, the impact of morbidity being increasingly concentrated towards the end of life is not readily apparent in the decade to 2003. Over the next two decades this dynamic will start to have an impact, although the gains in health expectancy at birth will be smaller than for males (around 0.3 years in 2023) and the losses in the elderly will be greater and will be experienced earlier in life. -.8 -.6 -.4 -.2 0 .2 .4 .6 .8 0 50 100 0 50 100 0 50 100 2003 2013 2023 Net gain (years) Age Figure 6.6: Net years gained in health-adjusted life expectancy (years) from changes in prevalence of disability compared with baseline (1993) levels of disability for females, Australia, 2003 to 2023 The correct answer to the question What impact will the concentration of disability in the latter part of the lifespan have on health-adjusted life expectancy?, therefore, is: It depends. This is because health expectancy at any particular age is a summary measure based on the 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 Age Age 120 combination of mortality and disability risks at that age and all subsequent ages. While a detailed decomposition of the drivers of this complex dynamic is beyond the scope of this report, growth in prevalent disability in the elderly is likely to come from increases in diabetes and neurological conditions. Disability from diabetes, in particular, grew 10.4% in the decade to 2003, and will grow a further 29.3% over the next two decades if current trends in obesity continue (Figure 6.7). Neurological conditions grew 2.5% in the decade to 2003, and are likely to grow a further 6.6% in the 20 years to 2023. Most other causes of prevalent disability are likely to decline. -3.3 -13.0 -23.1 10.4 24.7 2.5 9.1 -9.4 -18.8 -27.7 -3.5 -9.2 -16.1 6.8 8.6 10.6 -40 0 40 -40 0 40 -40 0 40 -40 0 40 -40 0 40 -40 0 40 1993 2003 2013 2023 1993 2003 2013 2023 1993 2003 2013 2023 1993 2003 2013 2023 1993 2003 2013 2023 1993 2003 2013 2023 Cancer Diabetes Neurological Cardiovascular Injuries Other Change from 1993 (%) Year Figure 6.7: Percentage change in disability (PYLD) prevalence rates since 1993 for selected causes in people aged 80 years and over, Australia, 2003 to 2023 Figure 6.8 shows the number of healthy years lost due to prevalent disability (PYLD) by cause and age for 1993 and 2023. This figure demonstrates the absolute growth in PYLD that is expected to occur over this period due to increases in population size. It also shows the shift in the distribution of PYLD towards older ages that will occur as a result of population ageing. Trends in epidemiology will interact with these demographic factors to influence the composition of causes of prevalent disability at each age. Neurological conditions will grow substantially over the period 1993 to 2023 and will remain the largest contributor to disability prevalence at older ages. Mental disorders, on the other hand, will grow only slightly from 1993 levels but will remain the largest contributor to disability prevalence until age 60. Disability from cardiovascular disease is expected to decline from middle age onwards over this period but this decline will be more than offset by increases from diabetes. 40 0 40 40 0 40 0 40 40 0 40 40 0 40 3.3 0.0 13.0 23.1 9.4 0.0 18.8 27.7 3.5 0.0 9.2 16.1 0.0 39.7 0.0 6.7 0.0 Year 121 0 100 200 Number (thousands) 0 20 40 60 80 100 Age Other Injuries Cancer Musculoskeletal Cardiovascular Diabetes Chronic respiratory Neurological Mental 1993 0 100 200 Number (thousands) 0 20 40 60 80 100 Age 2023 Figure 6.8: Prevalence of disability (PYLD) due to selected broad cause groups for both sexes combined by age, Australia, 1993 and 2023 Changes in the age-specific trends described above reflect changes in the prevalence of disability experienced at all ages (Figure 6.9). Mental disorders decreased from 26% to 25% of total prevalence of disability in the decade to 2003. The effects of population ageing will mean that mental disorders, which are largely experienced in early to middle adulthood, will further decline to 22% of total prevalent disability in 2023, although they will remain the leading cause of overall prevalent disability. Neurological & sense disorders, on the other hand, will increase as a consequence of population ageing because they are experienced later in life. In the decade to 2003 this group increased from 15% to 17% of total prevalent disability in 2003, and over the next two decades, through population ageing alone, will increase to 21% in 2023. 122 Figure 6.9: Proportion of total prevalence of disability (PYLD) due to selected broad cause groups, Australia, 1993 to 2023 Diabetes was the other strong growth area at all ages, increasing from 5% of total prevalent disability in 1993 to 6% in 2003. If current trends in obesity continue, this figure is set to increase by a further 50% to 9% of total prevalent disability in 2023. 6.3 Burden The remainder of this chapter presents past, present and future burden using the standard burden measureDALYs. It is worth reiterating at this point that, unlike prevalent years lived with disability (PYLD), DALYs are incidence-based and include, in addition to nonfatal health outcomes, time lost due to premature mortality. Observed and projected trends in burden (DALYs) by broad cause group are summarised in Table 6.3. The methods underlying these figures are described in detail in Chapter 2. More detailed data on past, Mental Neurological Chronic respiratory Cardiovascular Diabetes Musculoskeletal Cancer Injuries Other 26% 15% 8% 10% 5% 6% 6% 6% 18% Mental Neurological Chronic respiratory Cardiovascular Diabetes Musculoskeletal Cancer Injuries Other 25% 17% 8% 10% 6% 6% 6% 5% 17% Mental Neurological Chronic respiratory Cardiovascular Diabetes Musculoskeletal Cancer Injuries Other 24% 19% 7% 9% 8% 6% 5% 5% 17% Mental Neurological Chronic respiratory Cardiovascular Diabetes Musculoskeletal Cancer Injuries Other 22% 21% 7% 9% 9% 7% 5% 4% 16% 1993 2003 2013 2023 123 present and future burden by age, sex and cause is available on the web at <www.aihw.gov.au/bod>. Table 6.3: Changes in burden rates (DALYs) by broad cause group and sex, Australia, 1993 to 2023 Standardised rate ratio(a) Rate per 1,000 for 2003 Males Females Broad cause group Males Females 1993 2003 2013 2023 1993 2003 2013 2023 Infectious 2.8 1.7 0.93 1.00 1.02 0.99 0.99 1.00 0.93 0.85 Acute respiratory(b) 1.7 1.8 0.67 1.00 1.00 1.00 0.61 1.00 1.00 1.00 Maternal 0.2 1.09 1.00 1.03 1.02 Neonatal 1.9 1.6 1.32 1.00 0.80 0.68 1.00 1.00 0.82 0.71 Nutritional 0.1 0.5 1.12 1.00 1.03 1.02 1.03 1.00 0.99 0.98 Cancer 26.8 23.5 1.20 1.00 0.85 0.70 1.16 1.00 0.88 0.74 Other neoplasms 0.5 0.6 1.03 1.00 0.83 0.68 0.94 1.00 0.89 0.81 Diabetes 7.8 6.6 0.87 1.00 1.15 1.32 0.89 1.00 1.18 1.40 Endocrine 1.5 1.4 1.88 1.00 1.08 1.03 0.89 1.00 1.16 1.31 Mental 16.8 18.5 1.03 1.00 1.01 0.99 0.99 1.00 1.01 1.01 Neurological 14.9 16.6 0.96 1.00 1.02 1.03 0.96 1.00 1.03 1.05 Cardiovascular 25.6 22.1 1.56 1.00 0.69 0.48 1.51 1.00 0.74 0.53 Chronic respiratory 10.0 8.8 1.22 1.00 0.83 0.73 1.04 1.00 0.96 0.93 Digestive 2.9 2.9 1.01 1.00 0.81 0.71 1.03 1.00 0.85 0.75 Genitourinary 2.9 3.7 0.97 1.00 0.97 0.96 0.97 1.00 0.98 0.95 Skin 1.0 1.0 1.00 1.00 1.00 0.99 1.00 1.00 1.00 0.99 Musculoskeletal 4.5 6.1 0.98 1.00 1.03 1.05 0.97 1.00 1.02 1.02 Congenital 1.9 1.4 1.11 1.00 0.84 0.74 1.19 1.00 0.84 0.72 Oral 1.2 1.3 0.99 1.00 1.02 1.03 0.98 1.00 1.01 1.02 Ill defined 0.5 0.7 1.70 1.00 0.83 0.73 1.31 1.00 0.93 0.89 Injuries 13.1 5.5 1.16 1.00 0.91 0.79 1.08 1.00 0.89 0.76 All causes 138.2 126.7 1.18 1.00 0.90 0.81 1.11 1.00 0.93 0.87 (a) Ratio of age-standardised DALY rates for year to DALY rates for 2003. (b) Age-specific rates for pneumonia post-2003 held at 2003 rates due to coding discontinuities between ICD-9 and ICD-10 for this cause. As observed with PYLD, total burden will most likely decrease after the effect of population ageing is removed (that is, age-standardisation) over the next two decades, yet in crude terms it will most likely increase (Figure 6.10). Again, this is due to a larger proportion of the population alive at older ages. 124 151.0 132.4 121.1 111.4 139.8 132.4 132.4 135.1 100 125 150 175 1993 2003 2013 2023 1993 2003 2013 2023 Standardised Crude Rate per 1,000 Year Figure 6.10: Age-standardised and crude burden (DALY) rates for both sexes combined, Australia, 1993 to 2023 Chapter 3 described the decline of cardiovascular disease relative to cancer as a proportion of overall burden, and stated that for the first time, cancer accounted for the largest share of overall burden experienced by the Australian population in 2003. This is primarily because Australia has been relatively successful at curbing the impact of the cardiovascular disease epidemic, but not nearly as successful to date with cancer. If these trends continue, the burden of cardiovascular disease will further decline to about 13% of the total burden in 2023. The age-standardised rates of cancer mortality and disability are expected to fall somewhat in the future but cancer as a whole will retain its share of around 19% of total burden two decades from now and will remain the largest contributor to total burden in 2023 (Figure 6.11). Despite the steady decline in cardiovascular disease burden over the next two decades, there is likely to be a strong increase in burden due to diabetes, primarily as a consequence of the obesity epidemic. If current trends in obesity continue unabated, diabetes will account for around 9% of total burden in 2023, up from around 5% in 2003 (Figure 6.11). A major consequence of population ageing will be the steady growth in burden from neurological & sense disorders, up from 12% in 2003 to around 16% in 2023. The main contributors here will be dementia and adult-onset hearing loss, both causes for which current treatments are largely ineffectual. The economic consequences of the former in terms of the provision of appropriate care services are likely to be significant and will be evident in the home and community sectors before they are felt in the residential aged care sector. 125 Cancer Cardiovascular Neurological Mental Chronic respiratory Injuries Diabetes Musculoskeletal Other 19% 22% 13% 10% 7% 8% 4%3% 14% Cancer Cardiovascular NeuroMental logical Chronic respiratory Injuries Diabetes Musculoskeletal Other 19% 18% 13% 12% 7% 7% 5% 4% 14% Cancer Cardiovascular NeuroMental logical Chronic respiratory Injuries Diabetes Musculoskeletal Other 19% 15% 13% 14% 7% 6% 7% 4% 14% Cancer Cardiovascular Neurological Mental Chronic respiratory Injuries Diabetes Musculoskeletal Other 18% 13% 16% 12% 7% 5% 9% 5% 14% 1993 2003 2013 2023 Figure 6.11: Proportion of total burden (DALYs) due to selected broad cause groups, Australia, 1993 to 2023 The proportion of burden due to major causes experienced at different ages throughout the life span is unlikely to change dramatically over the next two decades (Figure 6.12). The decline of cardiovascular disease as a proportion of total burden will be experienced at all ages, although, in absolute terms, most notably in the elderly. This will be partially offset by the increase in the proportion due to diabetes at all ages. The proportion of total burden due to cancer at different ages is unlikely to change. 126 0 200 400 Number (thousands) 0 20 40 60 80 100 Age Other Musculoskeletal Injuries Diabetes Chronic respiratory Mental Neurological Cardiovascular Cancer 1993 0 200 400 Number (thousands) 0 20 40 60 80 100 Age 2023 Figure 6.12: Burden (DALYs) due to selected broad cause groups for both sexes combined by age, Australia, 1993 and 2023 In terms of specific causes of disease burden, ischaemic heart disease is the leading cause in males across three of the four time periods. Its share of burden declined from 14.7% in 1993 to 11.1% in 2003 (Table 6.4). If this trend continues, ischaemic heart disease will decline a further 36% to 7.1% of the total burden in 2023. Type 2 diabetes, on the other hand, rose from sixth place to second in the decade to 2003, and is likely to increase a further 65% to first place or 8.6% of the total burden in 2023. Anxiety & depression will retain its third place, at around 4.5% of the total burden in 2023, but lung cancer will drop to sixth place, largely because of the dramatic decline in smoking prevalence in males over the last two decades. In its place, dementia will occupy fourth position in 2023, up from 11th place in 2003. 127 Table 6.4 Leading causes of burden (DALYs) in males, Australia, 1993 to 2023 Rank(a) Per cent of total Specific cause 1993 2003 2013 2023 1993 2003 2013 2023 Ischaemic heart disease 1 1 1 2 14.7 11.1 8.9 7.1 Type 2 diabetes 6 2 2 1 3.6 5.2 6.8 8.6 Anxiety & depression 2 3 3 3 4.5 4.8 4.7 4.5 Lung cancer 3 4 4 6 4.4 4.0 3.8 3.4 Stroke 5 5 6 7 4.2 3.9 3.5 3.2 COPD 4 6 9 11 4.4 3.6 2.9 2.2 Adult-onset hearing loss 11 7 5 5 2.5 3.1 3.7 4.2 Suicide & self-inflicted injuries 8 8 10 10 2.9 2.8 2.8 2.4 Prostate cancer 10 9 8 8 2.5 2.7 3.0 3.1 Colorectal cancer 9 10 11 9 2.6 2.5 2.6 2.4 Dementia 14 11 7 4 1.8 2.5 3.3 4.4 Road traffic accidents 7 12 14 18 3.0 2.3 1.8 1.3 Asthma 12 13 12 12 2.2 2.1 2.0 1.9 Alcohol dependence & harmful use 13 14 13 14 2.0 2.0 1.9 1.6 Personality disorders 16 15 17 19 1.1 1.2 1.2 1.2 Schizophrenia 15 16 20 23 1.1 1.1 1.1 1.0 Osteoarthritis 24 17 15 15 0.8 1.1 1.3 1.6 Back pain 23 18 18 17 0.9 1.1 1.2 1.3 Melanoma 20 19 21 20 0.9 1.0 1.1 1.1 Parkinsons disease 25 20 16 13 0.8 1.0 1.3 1.6 (a) Sorted according to the leading specific causes for Australia in the year 2003. Anxiety & depression is ranked first in females across three of the four time periods, although in percentage terms its share of the total burden will decrease from 10.0% in 2003 to 8.7% in 2023 (Table 6.5). Ischaemic heart disease will remain in second place over the next decade, but fall to fourth place by 2023. In its place will be dementia, which increased by 1.1 percentage points to 4.8% of the total burden in the decade to 2003, and, if current projections of population ageing eventuate, will be ranked third at 7.4% of the total burden in 2023. As with males, Type 2 diabetes is set to increase steadily and is likely to occupy second position in 2023, at around 8% of the total burden. 128 Table 6.5 Leading causes of burden (DALYs) in females, Australia, 1993 to 2023 Rank(a) Per cent of total Specific cause 1993 2003 2013 2023 1993 2003 2013 2023 Anxiety & depression 2 1 1 1 9.8 10.0 9.6 8.7 Ischaemic heart disease 1 2 2 4 12.4 8.9 7.5 6.1 Stroke 3 3 5 5 5.9 5.1 4.4 3.8 Type 2 diabetes 6 4 3 2 3.7 4.9 6.4 8.0 Dementia 5 5 4 3 3.7 4.8 5.9 7.4 Breast cancer 4 6 6 6 5.1 4.8 4.3 3.5 COPD 7 7 8 8 3.1 3.0 2.9 2.8 Lung cancer 10 8 7 7 2.3 2.7 3.1 3.5 Asthma 8 9 9 9 2.9 2.7 2.5 2.4 Colorectal cancer 9 10 10 12 2.6 2.3 2.2 1.9 Adult-onset hearing loss 11 11 11 11 1.5 1.8 2.0 2.2 Osteoarthritis 12 12 12 10 1.4 1.6 1.9 2.2 Personality disorders 15 13 14 16 1.2 1.3 1.3 1.3 Migraine 14 14 17 18 1.3 1.3 1.2 1.1 Back pain 16 15 15 15 1.1 1.2 1.3 1.3 Lower respiratory tract infections 38 16 13 13 0.5 1.1 1.3 1.6 Falls 20 17 18 19 0.9 1.0 1.1 1.1 Parkinsons disease 19 18 16 14 0.9 1.0 1.2 1.5 Schizophrenia 17 19 20 25 1.0 1.0 1.0 0.9 Rheumatoid arthritis 21 20 19 20 0.9 1.0 1.0 1.0 (a) Sorted according to the leading specific causes for Australia in the year 2003. 129 7 Discussion and conclusions 7.1 Potential applications A detailed description of the burden of disease and injury in a population is not sufficient for setting priorities in public health. It is, however, an important foundation on which to build assessments and evaluations that underpin health policies. This report contributes most obviously by identifying the magnitude of health problems in a population and by quantifying the contribution to these problems of major modifiable risks to health. The present study greatly extends the scope of the previous study in this respect by presenting burden estimates for a greater range of population subgroups in Australia. It also provides a cogent analysis of past trends in burden in this country, and suggests the likely state of the populations health in 20 years from now if these trends were to continue. Furthermore, it quantifies the contribution to overall burden of an expanded set of risks to health. Equally important, however, is the contribution of burden estimation to down-stream analyses by the creation of a consistent set of epidemiological parameters for a full range of health conditions, and a detailed description of the relationship between these parameters and risks to health. Again, the study upon which this report is based expands the scope of such analyses through the creation of a comprehensive database of all relevant epidemiological and burden parameters both for a number of different subpopulations and through time. The most obvious synergy here is with cost-effectiveness analyses of the potential outcomes of health interventions. Estimates from the previous study have been used extensively in a number of economic evaluation studies to date (for example, Nelson et al. 2005; Stone et al. 2004; Vos et al. 2005). This new set of results has been incorporated into models under development for the project funded by the National Health and Medical Research Council Assessing Cost-Effectiveness (ACE)Prevention, at the University of Queensland and University of Melbourne, the aim of which is to comprehensively model the cost-effectiveness of preventive intervention options for non-communicable disease in Australia. The essential link between burden of disease and injury data and cost-effectiveness results is acknowledged by the Pharmaceutical Benefits Advisory Committee, which now requires companies to present evidence on the likely uptake of a new drug in the population. This requires knowledge about the number of people with a particular condition or risk profile for whom the drug is intended. This report will be invaluable as a common reference point. An important new application of the results of this study will be to further improve estimates of future health expenditure in this country. As life expectancy continues to increase and populations continue to age, this is an area of concern to governments, not only in Australia but around the world. Projections for the 2002 Intergenerational Report relied on models that do not take into account major shifts in epidemiology and expenditure for some diseases. The projections in this report have already been linked to health expenditure data (AIHW 2005c), thus enabling more detailed health expenditure projections for these causes (for example, Vos et al. 2007). 130 7.2 Policy implications The preceding section illustrated that the analyses underpinning this report are not sufficient on their own as a basis for setting future directions in health policy. A number of other inputs are also necessary, not the least of which is evidence on the costs and effectiveness of available interventions. Nevertheless a number of important implications for policy arise from the findings presented in the preceding chapters. This report has presented for the first time a comprehensive overview of the likely effects of population ageing on patterns of disease and injury in Australia over the next 20 years. These projections are based on analyses of past trends in health and are presented as a business as usual scenario (that is, the rate of change in policy responses to emerging problems in the future is consistent with the rate observed in the historical period upon which the projections are based). While all projections regarding the future are uncertain, including those presented in this report, some are more uncertain than others. This is particularly true for the projections for diabetes, where information on trends is limited to one cross-sectional survey and some assumptions regarding changes in case-fatality relative to those observed for ischaemic heart disease (assumptions which are corroborated by the second round of the Australian Diabetes, Obesity and Lifestyle study). Notwithstanding these caveats, the general tenor of these analyses is clear. A key finding of this study is that while life expectancy is likely to continue increasing steadily, growth in health-adjusted life expectancy will not be as rapid. This is because the number of very old people is set to expand rapidly, mostly due to declining mortality. In addition, while the prevalence of disability will drop at most ages, it is expected to increase for people aged 80 years and over. A major consequence of this population dynamic will be the steady growth in the burden from diseases associated with old age such as dementia, Parkinsons disease, hearing and vision loss, and osteoarthritis, all causes for which current prevention and (with the exception of osteoarthritis and cataract) treatment strategies are largely ineffectual. The impact of increasing disability from these diseases is likely to be significant and will be felt in the home and the community care sectors before it is felt in the residential aged care and palliative care sectors. While future research into prevention and treatment may yield unexpected results, relevant stakeholders should be planning for growth in the number of elderly people requiring appropriate services in each of these care settings. The economic consequences of these changes on future health care expenditure have been quantified in a separate report (Vos et al. 2007), which it is hoped might assist the development of appropriate policy responses in this area. In addition, cardiovascular disease are likely to continue to decline relative to cancer as a proportion of the overall burden, primarily because current health care has been relatively successful at curbing the effects of the former, but not nearly as successful with the latter. Successful reductions of cardiovascular disease should not obscure the fact that additional gains could be made by further reductions in levels of cholesterol, blood pressure and smoking, the primary risk factors for these diseases (Taylor et al. 2006). Increasing the coverage and targeting of interventions known to be effective (for example dietary modification, cholesterol and blood pressure lowering drugs, and smoking cessation) is one way of achieving this. There is also likely to be scope for increased efficiency through the adoption of a more cost-effective mix of interventions. 131 Lastly, the projected strong growth in the burden from diabetes over the next 20 years is an area of concern. This is mostly a consequence of increases in body weight. The consequences of increasing obesity will be further magnified by reductions in case-fatality from cardiovascular disease the major cause of mortality in people with diabetes through successful tobacco control and cholesterol and blood pressure lowering strategies. This increased survival will mean an increase in the risk of developing other largely non-fatal but disabling consequences of diabetes such as renal failure, retinopathy, neuropathy and peripheral vascular disease. Efforts to find new approaches to stem rising levels of obesity need to continue. 7.3 Precision of estimates Fatal burden The calculation of fatal burden (YLL) is relatively straightforward and the precision of these estimates is almost entirely dependent on the quality of information on underlying cause of death in official mortality data. While every effort has been made to remedy likely distortions to the overall reported cause of death structure by reallocating deaths with certain codes known to be non-specific to valid and specific underlying causes of death, by world standards the extent of these distortions is small (around 6% to 10%, depending on the codes included in this definition). Of greater concern are the deaths coded to valid and specific causes of death. With the exception of a few studies on sensitivity and specificity for specific conditions, relatively little is known about the accuracy of causal attribution for the majority of cases. It is likely that accuracy varies with the location of the death due to differential access to diagnostic information (for example in an institutional setting versus at home), but the assumption that these inaccuracies will cancel each other out at the population level is largely speculative. Further research in this area would greatly enhance the integrity of the vital registration system in this country. Non-fatal burden The accuracy of estimates of disability is not quantifiable using formal statistical techniques. This is because, in the construction of these estimates, data of widely varying levels of quality, ranging from population level disease registers and high quality research findings at one end to guesstimates and expert opinion at the other, was drawn upon. Precision is likely to vary greatly between different individual estimates, and ultimately depends on the type of model used and the source and nature of the underlying data. Using simulation methods, it is feasible to quantify an uncertainty interval for each estimate that accounts for confidence in the underlying epidemiological data as well as uncertainties associated with the various assumptions and additional information used. Such an analysis has not been possible in the time frame of this report, however, but may constitute the subject of future research. In the absence of such analyses, it is worth noting where major sources of uncertainty are likely to lie in more qualitative terms. Among major causes of burden presented in this report, uncertainty is probably highest for hearing loss, neurological conditions, osteoarthritis and cirrhosis. The reasons for suspecting higher levels of uncertainty in these conditions are discussed below. 132 Hearing loss although population data on measured hearing loss thresholds were used toestimate disability for this condition, there was considerable uncertainty associated with the modelling of the average durations associated with progressing from mild through moderate to severe hearing loss and, to a lesser extent, the effect of hearing aids on reducing the severity of disability from hearing loss. Osteoarthritis estimates for this condition were based on the same overseas studies ofincidence and severity used in the previous Australian Burden of Disease and Injury Study. These estimates are lower than would be suggested by the Australian self-reported population data on osteoarthritis. Considerable uncertainty remains about the true incidence of this condition at the population level. Selected neurological conditions information on dementia and Parkinsons disease camefrom meta-analyses of international community-based studies of prevalence. For the estimates presented in this report, these analyses were updated to include all such studies in Western countries. There is uncertainty about the variations in the level of disease among these countries. Cirrhosis there is no easy way to measure the prevalence of cirrhosis at a populationlevel. This report relied on a published modelling effort that projects the progression to cirrhosis in people with hepatitis C. These figures were used to extrapolate cirrhosis from hepatitis B, alcohol dependence and other causes. Considerable uncertainty surrounds these extrapolations. Mental disorders these estimates were based on the mental health survey conducted in1997; more recent data on this large cause of non-fatal burden would have been desirable. Because of high levels of comorbidity between depression and anxiety, as well as the largely similar treatment pathways, these conditions were combined into one entity and modelled over a life course. This departs from the approach taken in the previous study where each condition was treated separately and depression was modelled as an episodic condition. General levels of uncertainty In more general terms, it is likely that uncertainty in the estimates of burden presented in this report may not be excessive. Overall, about half of the total burden in Australia was attributable to mortality, for which estimates are fairly robust. Of the remainder, half was attributable to non-fatal burden from a small number of diseases (including cardiovascular disease, cancers, diabetes, common mental disorders, and injuries) for which reasonably good Australian data are available. This leaves around a quarter of the total burden with varying and probably higher levels of uncertainty. What is clear is that a number of key estimates presented in this report are likely to be much more accurate than those of the previous study. This is due to the availability of considerably better quality data in some cases, one source of which deserves special mention. Access to the linked hospital and mortality databases in Western Australia allowed greater accuracy in the modelling of cardiovascular disease, the second leading cause of burden in Australia. As Western Australia adds more health information data sets to its linkage programincluding health surveys, disease registries, Medicare and Pharmaceutical Benefits Scheme (PBS) datathis will become an even more valuable resource, both for Western Australia and nationally. Various efforts are in train to encourage data linkage in other jurisdictions and at the national level. For example, the Statistical Information Management Committee (a multijurisdictional committee established by health CEOs) has commissioned the development of a framework for national data linkage, which takes careful account of such concerns as 133 privacy, data protection, and custodianship. These efforts are to be encouraged, as they will underpin improvements in future analyses of the burden of disease and injury in Australia. Aside from data inputs, it is also worth mentioning the tools used. The second version of the epidemiological modelling software DisMod allowed much more accurate modelling of consistent epidemiological parameters than was possible in the previous study. In using this software, particular attention was directed towards a consistent application of the concept of excess risk (that is, people with a condition are often more at risk of dying than would be indicated by the mortality actually coded to that condition) by accessing information from international cohort studies and the linked hospital and death databases in Western Australia. DisMod also allows for past trends in incidence and case-fatality to be modelled, which is particularly important for diseases with prevalence as the main observed parameter and for which there have been significant trends in the past. This is because prevalence is a stock variable and is simply a reflection of past trends in incidence and case-fatality. In this study, there was a strong upward trend in the incidence of diabetes, accompanied by an improvement in case-fatality. Ignoring these trends would have lead to an underestimate of the true incidence of this disease. One area where DisMod was unable to help, however, was in replicating final epidemiological models to multiple subpopulations and to various points through time. For this application a custom-built routine was developed within a statistical package environment based on DisMods underlying equations. A final issue that is relevant to precision considerations is the set of disability weights assigned to each health state. To date, a comprehensive set of weights has not been derived in the Australian context. It was therefore necessary to continue to use an assortment of weights derived largely from a Dutch study supplemented with weights from the original Global Burden of Disease study and weights derived from a regression model on the Dutch weights and the six domains of the EQ5D+ (an instrument to quantify quality of life). Funding requested as part of the current study to validate these disability weights in the Australian context has not been forthcoming. Also, internationally there has not been the expected further development of measuring health state preferences to determine disability weights. The large effort of collecting data in the World Health Surveys by the World Health Organization in the first part of this decade has not yet resulted in any publication. While this may raise concerns about the construct validity of the non-fatal estimates of burden presented in this report, it should be noted that the rank order of weights for most conditions has strong face validity and has been documented to be reasonably constant for a set of tracer conditions when replicated in different countries. Of greater concern are the disability weights for common but low-severity conditions such as mild hearing loss, mild vision loss, uncomplicated diabetes, asthma and anaemia. Existing health state evaluation methods do not seem to accurately capture differences in severity between such conditions. The lack of valid disability weights for distinguishing between high-prevalence low-severity conditions is more important than it sounds because a small absolute difference in the disability weight for a highly prevalent condition has a major bearing on the size of the burden attributable to that condition. This is an area in need of further development. Lastly, with respect to disability weights, a major improvement of this study compared with most burden studies has been a comprehensive correction for coexisting health states. In the previous study, an attempt was made to deal with comorbidities within mental disorders, injuries and common causes of burden in the elderly, although each group was treated separately and the latter was not comprehensive. Furthermore, the methods used assumed no dependence between health states (that is, groups of conditions being more likely to 134 coexist due to common causal pathways). The methods developed for the present study drew upon multiple surveys and hospital data to derive probabilities of coexistence for all possible combinations of health states that were modelled using an innovative microsimulation approach. Dependence is implicitly accounted for in this approach. Corrections for comorbidity in burden estimates are necessary because disability weights are typically derived in isolation from each other, meaning that coexisting disability in the same person is unlikely to be simply additive across two or more health states. 7.4 Access to data Previous burden estimation work has been done mostly within a spreadsheet environment. The benefits of this approach are that transparency and portability are maximised and, to date, no other approach has achieved the level of flexibility afforded by working in this way. However, accuracy can be a problem in that, in a spreadsheet, location is critical and incorrect cell references are common. More importantly, spreadsheets rapidly become unwieldy when more than several dimensions are being represented. In this study, spreadsheets were retained for all basic disability modelling work, but a statistical package environment was used for all subsequent analyses. The first practical implication of this decision is that a set of spreadsheets containing all disability models exists and will be made available to those who are interested. Users of this resource should note, however, that the disability estimates in these spreadsheets will usually not correspond exactly with final estimates in this report because comorbidity corrections and some trend corrections occur outside this environment. The other more interesting implication is that having extracted all relevant information from the spreadsheets and derived final burden estimates outside this environment, the results can be reassembled in any requested structure. This is particularly relevant for the estimates for subpopulations and for various points through time. In collaboration with various jurisdictional stakeholders, this information can be grouped into meaningful aggregations for specific health policy and planning purposes. In addition, a web-based interface could be developed whereby users could extract the desired information by cause, age, gender, time and subpopulation. The website developed for disseminating the 1996 and 2001 Victorian Burden of Disease study results is a useful model (see <www.dhs.vic.gov.au/health/healthstatus/bod/bod_reg.htm>). 7.5 Future directions This study comes seven years after the original Australian Burden of Disease and Injury Study. It has taken three years to complete, with the equivalent of 23 full-time staff and considerable intermittent assistance from researchers from several state health departments and masters students. Such a commitment of resources to this type of research is unlikely to occur again in this country in the short term. One of the aims of the study, therefore, was to develop a less resource-intensive way of retaining an up-to-date set of burden estimates going forward in time. To this end, a database of estimates for each year between 2003 and 2023 has been developed. This will provide an invaluable set of base-level results for those wanting to make assessments of burden in the period prior to the next major update. Such assessments may entail varying levels of sophistication, from simple updates of fatal burden using the most recent mortality data to ad hoc changes to specific non-fatal estimates based 135 on advances in knowledge and better data. Through the adoption of strategies such as these, a major revision may not be needed for at least another five years. The final report of the previous Australian Burden of Disease and Injury Study identified seven areas where it was felt priority should be placed in future research. Progress has been made in a number of these areas over the last seven years, four of which are addressed specifically by the present study. These include the development of burden estimates for Indigenous Australians, a more detailed assessment of differentials in burden across Australia, including estimates by socioeconomic status, remoteness and jurisdictional boundaries, and more detailed modelling of National Health Priority Area diseases. A fifth suggestion regarding the value of linking burden research to cost-effectiveness analyses has been picked up in a number of separate studies around the country over this period. Progress on two recommendations, however, has been less rapid. It has not been possible to estimate and validate a set of disability weights in the Australian context and this remains an important area requiring further development. Finally, there has been no formal evaluation of the usefulness of burden of disease and injury analyses for policy makers and health planners. There has been enough informal feedback from planners, researchers and the media to know that there is consistent demand for this type of information. But a more formal analysis of the impact on health assessment, policy and planning of this research over the last seven years would be welcome. 137 Appendix 1: Methods for estimating disability burden In this section we describe our methods for calculating disability for the large number of diseases and injuries and their sequelae for which models were developed, including all those that make significant contributions to the total non-fatal burden. While this list is extensive, it is not exhaustive, and explicit models were not developed for many conditions. Table A1.1 lists the full names of many of the data sources underlying our models and our abbreviations of these names, which we use in this section for ease of reference. Table A1.1: List of full and abbreviated names of commonly used data sources Abbreviated name Full name AusDiab The Australian Diabetes, Obesity and Lifestyle Study, 19992000 (Dunstan et al. 2001) Australian dialysis and transplant data 2002 Australian and New Zealand Dialysis and Transplant Registry (McDonald & Russ 2002). The interpretation of this data is the responsibility of the authors of this report and should not be seen as the interpretation of the Australian and New Zealand Dialysis and Transplant Registry. Australian disability survey Survey of Disability, Ageing and Carers (1993, 1998 or 2003) (ABS 1993, 1998b, 2003b) Australian general practitioner data 200001 and 200203 Bettering the Evaluation and Care of Health (AIHW: Britt et al. 2001) Australian hospital data 200203 National hospital morbidity database (AIHW 2003a) Australian mortality data 2003 Cause of Death dataset (ABS 2005) Australian notification data National Notifiable Infectious Disease Surveillance System (CDA, 2003) except for HIV/AIDS which is from the National Centre for HIV Epidemiology and Research (National Centre in HIV Epidemiology and Clinical Research, 2003) Australian perinatal data 2002 Australias mothers and babies and various state and territory perinatal data collections (AIHW: Laws & Sullivan 2004; Queensland Health 2004; Riley & King 2003). Disability weight regression model Regression model of Dutch disability weights which requires inputs of health state description based on the six domains of the EQ5D+ (p. 158 of AIHW: Mathers et al. 1999) DisMod DisMod version II (Barendregt et al. 2003) GBD study Global burden of disease and risk factors, 2000 (Lopez et al. 2006) Low prevalence study 199798 Low Prevalence (Psychotic) Disorders Study (Jablensky et al. 1999) National Health Survey 2001 National Health Survey (unless otherwise specified as the 1995 National Health Survey) (ABS 1995, 2001c) National mental health survey 1997 National Survey of Mental Health and Wellbeing (ABS 1997) National Trachoma Survey 1980 National Trachoma and Eye Health Program (Royal Australian College of Ophthalmologists 1980) Previous Australian burden study Australian Burden of Disease and Injury Study, 1996 (AIHW: Mathers et al. 1999) Victorian birth defect data 200102 Victorian Birth Defects Register (Riley & Halliday 2004) Victorian linked hospital dataset Analyses of Victorian hospital data 19962002 & 200102 from the 2001 Victorian Burden of Disease and injury study (DHS, 2005) Womens health Australia Australian longitudinal study on womens health (Lee et al. 2005) 138 1A Infectious and parasitic diseases Tuberculosis We estimate the incidence of tuberculosis using Australian notification data on new cases of tuberculosis. We assume that the average duration for tuberculosis is 8 months, reflecting 6 months for the shortest treatment cycle available and another 2 months of symptoms before treatment. Sexually transmitted diseases (excluding HIV/AIDS) We base our incidence estimates for syphilis, chlamydia and gonorrhoea on Australian notification data. Following expert advice, we assume that annual notifications for syphilis and gonorrhoea represent all incident cases. We model syphilis using a staged approach applying proportionate distributions for primary, secondary, and tertiary syphilis from the GBD study. We adjust our estimates for chlamydia to account for under-reporting due to asymptomatic infections and the reluctance of some patients to consult general practitioners about sexually transmitted diseases. We base our incidence estimates of pelvic inflammatory disease, a complication of both chlamydia and gonorrhoea in women, on Australian hospital data. Following expert advice we adjust these estimates to account for under-identification. Common sequelae of pelvic inflammatory disease include ectopic pregnancy, chronic pelvic pain, infertility and tubo-ovarian abscess. We base our rates of complications following pelvic inflammatory disease on GBD assumptions. We adjust our incident estimates of infertility resulting from pelvic inflammatory disease for women who do not wish to have a child and therefore do not experience disability. In the absence of Australian data on child wish, we make this adjustment using findings from a recent German study (Stobel-Richter et al. 2005). The GBD reports urethral stricture and epididymitis as complications following chlamydial and gonorrhoeal urethritis in men. These complications were thought by experts to be rare, and so have not been included in the Australian estimates. We model disability weights and durations for syphilis, chlamydia and gonorrhoea and their sequelae using the assumptions of the GBD study. HIV/AIDS We model HIV as a progressive condition with four stages: (1) asymptomatic HIV; (2) symptomatic HIV; (3) AIDS prior to terminal phase; and (4) terminal AIDS. We assume that the annual number of new HIV diagnoses from Australian notification data represent all incident cases of HIV. We use the Dutch disability weights for each of the stages (stage 10.2, stage 20.31, stage 30.56 and stage 40.95) and adjust the weight for stage 1 to account for the estimated proportion of undiagnosed asymptomatic HIV cases to whom we assign a disability weight of 0 (Aalen et al. 1997). We calculate the mean durations for stages 1 to 3 using Weibull regressions of published data accounting for background mortality (Kaldor & McDonald 2003; Mocroft et al. 1997; Porter et al. 2003). This gives average durations of 30 years for the combined stages 1 and 2 and 5.5 years for stage 3. We adjust our duration estimates for stage 1 and 2 based on the assumption that an equal amount of time is 139 spent in each stage based on work by Aalen and colleagues (1997). In the absence of new evidence, we assume that stage 4 lasts an average 0.5 of a year. Diarrhoeal diseases Diarrhoeal diseases include a number of notifiable diseases as well as non-notifiable diseases. Given that notifications are generally considered a gross underestimate of the incidence for notifiable diarrhoeal diseases, and that there is often even less reliable information on the incidence of non-notifiable diarrhoeal diseases, we do not model diarrhoeal diseases using notification data or by specific cause. Instead, we derive the incidence of diarrhoea not requiring hospitalisation using annualised self-reported data from the 200102 National Gastroenteritis Survey (Hall & OzFoodNet Working Group 2004). We base our duration of 2 days from the findings of this survey and use age-specific weights for uncomplicated diarrhoea (average weight of 0.093) from the GBD study. We use Australian hospital data to estimate the incidence of diarrhoea cases requiring hospitalisation. We use the age-specific GBD weight for diarrhoea (0.093) since the Dutch weight is implausible. We assume 2 weeks duration for complicated diarrhoea and derive an average weight (0.42) based on 1 week of disability equivalent to the regression model of health state (323311) and 1 week of disability for uncomplicated diarrhoea. Childhood immunisable diseases We do not model poliomyelitis and diphtheria for 2003. This is because there were no notifications of poliomyelitis from 1993 to 2003 and only one notification of diphtheria in 2001. Pertussis We estimate the incidence of pertussis using Australian notification data averaged over 20002003, an epidemic cycle. We adjust our incidence estimates for under-reporting based on the literature (Andrews et al. 1997; Torvaldsen et al. 2002). We apply the age-specific GBD disability weights for untreated cases for pertussis (04 years: 0.178; 514 years: 0.166; 15 years or over: 0.156), since the weight for treated cases is implausible, along with the GBD duration of 1 month. Following expert advice we estimate the incidence of intellectual disability attributable to pertussis as the proportion of intellectual disability cases from the total episodes of infection for 04 year olds in the GBD study (that is, 0.3% of pertussis cases). We derive a disability weight (0.58) for pertussis-related intellectual disability by weighting the number of cases of intellectual disability due to infectious diseases by the level of severity (using the Dutch weights for intellectual disability). Tetanus We estimate the incidence of tetanus using Australian notification data and apply the GBD disability weight for 60 years or over of 0.612, and duration of 2 weeks. 140 Measles We derive the incidence of measles using Australian notification data. We assume annual notifications in 2003 represent all cases of measles due to enhanced surveillance (Brotherton et al. 2004). For acute measles episodes we apply the GBD duration and disability weights (2 weeks, 0.152). We use Australian hospital data to estimate the incidence of measles sequelae. In 2003 there were no hospitalisations for measles encephalitis, and only one for sub-acute sclerosing panencephalitis. For the latter sequelae we apply the Dutch disability weight for end stage disease with a duration of 9 months. Rubella We derive the incidence of rubella using Australian notification data which we adjust for over-reporting. Enhanced surveillance of rubella notifications in Victoria found that 27% were laboratory confirmed (Guy et al. 2004). As there is no GBD or Dutch weight for rubella we use the measles disability weight (0.152) with a duration of one week. We use Australian notification data to derive incidence estimates of congenital defects due to rubella; there were only three such cases in 2003. The classic triad of complications associated with congenital rubella infection are cataract, heart disease, and deafness. In the absence of more specific information, we derive an average disability weight and durations to reflect each of these complications. Haemophilus influenzae type bWe derive the incidence of Haemophilus influenzae type b from Australian notification data.We only model the disability associated with the following sequelaemeningitis, epiglottitis, septicaemia, pneumonia and other using data from an Australian study (Herceg 1997). Following expert advice we assume that all cases of epiglottitis and meningitis are confined to the 014 year age group and pneumonia and septicaemia to the 15 years or older age group. We assume that meningitis from Haemophilus influenzae type b is included in thehospitalisation-based estimates of total meningitis and subtract these cases from the total incidence estimates of meningitis to avoid double-counting. We use the same disability weights and durations for these sequelae as per the previous Australian burden study. Meningitis We estimate the incidence of meningitis from Australian hospital data which we adjust to avoid double-counting of meningitis from Haemophilus influenzae type b. We modelmeningitis as a progressive condition with acute episodes of one month, after effects lasting up to six months and subsequent lifelong effects, in some, for a range of conditions (including hearing loss, ventriculoperitoneal shunt, seizure disorder, less severe developmental problems, mental retardation and motor deficit and physical deformities). We make minor modifications to the assumptions in the Dutch study regarding proportions of meningitis cases progressing to sequelae and their associated disability using the results of a seven-year follow-up study of meningitis in Melbourne children (Grimwood et al. 1995) and expert opinion. 141 Septicaemia We estimate incident cases of septicaemia from Australian hospital data. We do not adjust our estimates to account for meningitis-related septicaemia as Victorian data suggests that less than 2% of cases are due to meningitis. In the absence of a weight for this condition in its uncomplicated state, we use the Dutch weight for meningitis for an average duration of 1 month (Stouthard et al. 1997). Arbovirus infections We estimate the incidence of arbovirus infections using Australian notification data. Because there are no specific disability weights for arboviruses we use comparable weights from the Dutch study. For Ross River and Barmah Forest viruses we adjust estimates by 100% to account for under-reporting in endemic areas. We model Ross River and Barmah Forest viruses as a febrile illness in children aged up to 14 years and as an illness with acute and chronic stages for incident cases aged 15 years or over. Based on Australian literature we use the Dutch weight for influenza for children (1 month duration) and the Dutch weights for moderate rheumatoid arthritis (1 month duration) and mild arthritis (3.5 months in Ross River fever and half of this duration for Barmah Forest virus) for acute and chronic stages respectively in adults (Mylonas et al. 2002; Russell 2002). In general, arthralgia persists longer in Ross River virus infection than in Barmah Forest virus infection (Mackenzie et al. 1998; Russell & Dwyer 2000), therefore we halve the duration of the chronic phase in the latter. We adjust notifications for dengue fever by 10% to account for under-reporting. Based on the literature we use the Dutch weight for malaria with a duration of 6 days (Russell & Doggett 1998; Solomon & Mallewa 2001). We use Australian hospital data to estimate the incidence of the rare and disabling sequelae dengue haemorrhagic fever. There were only two cases in 2003. The GBD weight for this condition appears too low and so we apply the Dutch weight for meningitis for just over 1 week. We model the following flavivirus infections as other arbovirus infections: Murray Valley encephalitis, Kunjin virus infection, Japanese encephalitis, and flavivirus not elsewhere classified. In 2003 there were no notifications for Murray Valley encephalitis and only one case of Japanese encephalitis notified. We apply GDB estimates of the incidence of sequelae (episodes, cognitive impairment and neurologic sequelae), average disability weights, and duration for Japanese encephalitis to all other arbovirus infections. Hepatitis Hepatitis A We estimate the total incidence of hepatitis A using Australian notification data, which we adjust for under-reporting (Amin et al. 1999). We assume that the 10% of incident cases represent prolonged hepatitis A. We assume that Australian hospital data on hepatitis A represent all cases of complicated hepatitis A. We calculate the number of incident cases of uncomplicated hepatitis A by deducting the prolonged and complicated cases from our total estimate. Due to the implausibility of the Dutch weight for uncomplicated hepatitis A we use the average GBD weight of 0.093 with a duration of 3 weeks (Amin et al. 1999). We assume that prolonged hepatitis A cases experience depression or fatigue for 6 months at disability 142 weight equivalent to the Dutch weight for mild depression (0.14) (McIntyre 1990; Willner et al. 1998). We assume durations of 4 weeks for children and 6 weeks for adults (Melnick 1995). We apply a severe disability weight for half of this time (DW 0.747), and the remaining time at the same weight as uncomplicated cases. This gives an average weight of 0.42. Hepatitis B We estimate the incidence of acute hepatitis B using Australian notification data and assume that all infections reported as incident are symptomatic. We derive incidence estimates for acute symptomatic hepatitis B infection in infants from birth data and probabilities of perinatal transmission for at risk mothers as reported by Kaldor and colleagues (1996). Based on the literature we assume a 40% probability of transmission if exposed. Using this estimate we can calculate the number of infants who would be infected in the absence of vaccination (Kaldor et al. 1996). As current vaccination coverage in children born to mothers at risk is 95% (Menzies et al. 2004), we reduce the number of carriers from perinatal transmission accordingly. Similarly we adjust the number of perinatal infections for the probability of symptomatic infection which is 5% (Kaldor et al. 1996). Based on expert opinion, we assume a similar number of infections by casual contact in childhood and for males and females. We base our estimates of chronic hepatitis B on a series of DisMod models. First we estimate the prevalence of adult carriers using an overall prevalence of 0.47% (OSullivan et al. 2004), a remission of 0.5%, and an overall relative risk of mortality of 1.5. Next we estimate the prevalence of adult carriers using incidence estimates of carriers from perinatal and casual childhood transmission assuming no vaccination had occurred. We then subtract the prevalence of carriers from childhood infections from the first model so we can use DisMod to derive the incidence of chronic hepatitis B infection in adults. This model assumes a steady state of hepatitis B infection in the population, with vaccination only recently affecting perinatal and childhood transmission rates. This is unlikely to reflect the pattern of disease over time, but in the absence of data on the trends over time, this was considered the most plausible method of modelling the disease following expert consultation. We assume the average duration for an acute symptomatic episode to be 4 weeks (Lee 1997). We use the Dutch disability weight for acute hepatitis infection (0.21). We adjust the Dutch weight for chronic hepatitis B infection with active viral replication (0.36) following expert advice that only 15% of chronic cases have a symptomatic episode for 2 weeks each year (giving an average weight of 0.002). The methods we use to derive YLD for hepatitis B-related cirrhosis and liver cancer are described in the following section on hepatitis C. test Hepatitis C Due to the asymptomatic nature of hepatitis C infection we assume that all YLD are a result of hepatitis C sequelae, that is, cirrhosis and liver cancer. There is a paucity of information on the occurrence of cirrhosis at a population level. Instead, we make use of estimates of hepatitis C-related cirrhosis occurrence from an Australian study which modelled the progression rates to various sequelae from hepatitis C incidence (Law et al. 2003). The major problem in estimating the occurrence of hepatitis C-related cirrhosis is the dramatic change in hepatitis C incidence over the last 5 decades, the relevant time period for the development of current cirrhosis cases. The best available approximation 143 of the pattern of hepatitis C epidemiology over the last 40 years is based on the pattern of injecting drug use over time (Law et al. 2003). We make largely the same assumptions in the modelling of hepatitis C-related cirrhosis as in Law and colleagues (2003): 75% of people exposed to hepatitis C develop chronic infection an annual progression rate of 2% to cirrhosis a hepatitis C-related mortality rate of 1.5% following cirrhosis mean age of hepatitis C seroconversion among injecting drug users of 25 years a male to female ratio of 2:1 for persons who inject drugs and are hepatitis C-infected unlike in Law et al. (2003), we assume that 80% of those exposed to the hepatitis C (the estimated proportion of hepatitis C carriers exposed through injecting drug use) have a relative risk of mortality of 13 (Darke & Ross 2002) for an average of 14 years from the moment of exposure (as estimated for heroin dependence) background mortality is calculated from life tables constructed from Australian mortality and population data from 1950 to 2003. Based on this model, we estimate 447 new cases of hepatitis C-related cirrhosis and 5,804 people living with cirrhosis due to hepatitis C in 2003. Next, we examine the Australian hospital data for cirrhosis. In all cases in 2003 with a stated underlying cause, 49.4% are alcohol related and 50.6% non-alcohol related. Based on expert opinion we attribute 5% of non-alcohol related cirrhosis to other causes and the remainder to hepatitis. We estimate the occurrence of cirrhosis due to other causes (that is, hepatitis B, alcohol abuse, and other) by adjusting Australian hospital data by the admissions-to-prevalence ratio observed in hepatitis C-related cirrhosis cases. We only give a disability weight for the last 3 years lived with cirrhosis at 0.31 (minus 2 months) and 0.84 for the last 2 months (effectively interpreting the Dutch weight for compensated cirrhosis as relevant for the time spent in decompensated cirrhosis and the decompensated cirrhosis weight of the Dutch as the weight for terminal liver failure). In the previous study we assumed liver cancer occurred in around 19% of people with hepatitis C and hepatitis B. More recent data, including from a large multi-centre study of liver cancer patients in Europe, indicates that hepatitis B and C are responsible for around 19% and 40% of liver cancer respectively (Brechot et al. 1998; CDC 2001). Malaria We derive incidence estimates for malaria from Australian notification data. We model two aspects of malaria, episodes and neurologic sequelae and adopt GBD assumptions for disability weights and durations. Trachoma We model the disability associated with mild, moderate and severe vision impairment resulting from trachoma infection. We assume that trachoma related visual impairment is a problem only in remote Australia. We estimate the prevalence of trachoma-related visual impairment using the 1980 National Trachoma and Eye Health Program. Based on expert advice we adjust the prevalence downwards by one-third to account for observed decreases 144 in the prevalence of scarring stages that follow infectious trachoma since the national survey was conducted (Landers et al. 2005; Mak & Plant 2001). In the absence of more specific information we assume that mild and moderate vision loss have the same cause distribution by age as severe vision loss. We make minor adjustments to the prevalence of each stage by age to ensure plausibility and to reflect published estimates. We estimate the incidence and duration of trachoma in DisMod using our derived prevalence estimates. We initially model the prevalence of severe vision loss in DisMod assuming no remission and a relative risk of mortality of 1. We then use the incidence of severe vision loss from the DisMod output as mortality in the moderate vision loss DisMod model. This takes the cases of severe vision loss out of the pool of susceptible cases for moderate vision loss and therefore gives more accurate average durations than if we were to use remission as remitted cases in the DisMod model, as the cases continue to be subject to the hazard of incidence. Similarly we use the incidence of moderate vision loss as mortality in mild vision loss. 1B Acute respiratory infections Lower respiratory tract infections We base our incidence estimates for lower respiratory tract infections, including episodes of influenza, acute bronchitis and pneumonia, on Australian general practitioner data. For pneumonia, general practitioner data was thought to be more representative than hospital data as it should include those who do and do not go to hospital. We use the same assumptions for disability and durations as in the previous Australian burden study. GBD duration estimates were halved to 3.5 days for acute bronchitis, and left at 1 and 2 weeks respectively for influenza and pneumonia. Disability weights were derived using the regression model (influenza 0.047; acute bronchitis 0.132; pneumonia 0.373). Upper respiratory tract infections We base our incidence estimates for episodes of acute nasopharyngitis and acute sinusitis on annualised self-report data from the National Health Survey, while we model tonsillitis and pharyngitis using Australian general practitioner data. We use the data from the 1995 National Health Survey because the 2001 survey did not include questions on acute conditions. We adjust the tonsillitis and pharyngitis incidence estimate upwards by twofold to reflect the much higher rate (13 times) for the broader condition of sore throat that was reported in the survey. We use derived weights and assume GBD durations, with minor adjustments where we consider this to be appropriate. For employed adults, the average number of days off work due to upper respiratory tract infections was around 0.5 of a day. The GBD assumed an average duration of 3.5 days. The self-report prevalence data probably includes a considerable number of minor infections with minimal disability. Hence we use days off work plus half a day on either side to give an average duration of 1.5 days for acute nasopharyngitis. For tonsillitis and pharyngitis and sinusitis we use the GDB durations of 3.5 days. 145 Otitis media We model the following stages of otitis media; acute infection, bilateral chronic infection, and life-long deafness. We estimate the incidence of acute episodes using Australian general practitioner data. We assume that those who have relatively low disability do not seek treatment and base YLD estimates on treated numbers. We adjust our incidence estimates to allow for a higher rate of acute otitis media in Indigenous Australians in remote areas based on findings from the 1980 National Trachoma and Eye Health Program. We use the disability weight regression model to derive an appropriate weight (0.090) and assume a duration of 1 week. We estimate the prevalence of chronic otitis media in non-Indigenous and Indigenous Australians from the National Health Survey for those people reporting otitis media as a long-term health problem. We assume that these estimates represent non-Indigenous Australians in all areas and Indigenous Australians in major city or regional areas. We adjust these prevalence estimates downwards to account only for bilateral cases using a ratio of bilateral to unilateral cases from the 1980 National Trachoma survey. We estimate the prevalence of bilateral chronic otitis media in Indigenous Australians in remote areas from the 1980 National Trachoma survey and assume that the epidemiology of bilateral chronic otitis media has not changed since the survey was undertaken. We derive the incidence and duration of bilateral chronic otitis media in DisMod using prevalence, a relative risk of 1 and remissions equivalent to durations of 3 months and 3 years for non-Indigenous and Indigenous Australians, respectively, based on Australian data (McGilchrist & Hills 1986). We base our estimates for permanent hearing loss resulting from acute infections on the GBD study. For chronic infection we apply the Dutch weight for early acquired mild to moderate hearing loss (0.110). For the small number of cases that experience lifelong deafness, we use the Dutch weight for early acquired severe hearing loss (0.233). 1C Maternal conditions We base our incidence estimates for maternal haemorrhage, maternal sepsis, hypertension in pregnancy, obstructed labour, abortion and other maternal conditions on Australian hospital data. We adopt GBD methods except in the following instances. On expert advice we assume hypertension in pregnancy results in restricted activity (due to advised bed rest or hospitalisation) for 2 months at a derived weight of 0.117 (health state 122111), with 1 in 2,500 cases developing neurological sequelae. We model the sequela caesarean section with 2 weeks of disability at a derived weight of 0.349 (health state 222111). We base our incidence estimates for abortions using South Australian data on terminations of pregnancy as a proportion of total births (Chan et al. 2003). For abortion we model the disability of infertility resulting from the sequela pelvic inflammatory disease. We assume that 20% of hospitalised cases of pelvic inflammatory disease following abortion experience infertility from age at infection to post-reproductive age which we assume to be 45 years. We adjust our incident estimates of infertility, in the abortion and maternal sepsis models, for women who do not wish to have a child and who therefore do not experience disability. In the absence of Australian data on child wish, we make this adjustment using findings from a recent German study (Stobel-Richter et al. 2005). Although stress incontinence was considered a sequela of obstructed labour in the GBD study, most stress incontinence occurs in the absence of such a history. We therefore treat this condition as a category in its own right, classified under genitourinary conditions. 146 1D Neonatal causes Birth trauma and asphyxia We estimate the incidence of mild, moderate and severe birth asphyxia using Australian hospital data. We separate the mild and moderate incident cases using data from the GBD study. We base our sequela estimates of neurological disability by severity of birth asphyxia (0% of mild, 25% of moderate and 100% of severe) on the GBD study. We use the estimates of intellectual disability due to birth trauma from the overall calculations for intellectual disability by all underlying causes (see Section 2K). Stanley and colleagues (1995) estimated that 8% of cerebral palsy is associated with birth trauma. The balance of the incident cases of permanent disability is divided equally between deafness and seizures. We assume that the duration of cerebral palsy without intellectual disability and severe hearing loss is the same as those with mild intellectual disability. We base the duration of seizure on life expectancy at birth assuming a twofold risk of dying to indicate a greater likelihood of premature mortality. Low birth weight We estimate the incidence of low birth weight (>=1500g and <2500g) and very low birth weight (<1500g) in neonatal survivors using Australian perinatal data. We apply the sex distribution of low birth weight from the 2002 Victorian perinatal data to the Australian combined proportion for both sexes which we then apply to the total number of live births in Australia in 2003 (ABS 2004). We adjust our estimates of total neonatal deaths in 2003 (ABS 2005) using a proportion for those due to low birth weight. This was derived using an average of 2002 Victorian, Queensland, and South Australian data. We assume the probability of disability among low birth weight survivors is 25% for very low birth weight (<1500g) and 5% for low birth weight (>=1500g and <2500g) as per the GBD study. This corresponds to a total of 1,230 incident cases (596 males, 634 females) of disability in low birth weight survivors in 2003. For hearing loss, vision loss, epilepsy, and other disability we distribute the incident cases of disability in low birth weight survivors to disability type from the GBD study. We use the estimates of intellectual disability due to low birth weight from the overall estimates of intellectual disability (see Section 2K). In addition we attribute 60% of total incident cerebral palsy cases (at 2.25 per 1,000 live births) to low birth weight. Just over one half of the low birth weight survivors with permanent disability do not have severe neuro-developmental disability. In the absence of a defined disability weight for this health state we assume that these cases have a level of disability similar to the Dutch weight for permanent early childhood acquired moderate hearing loss. For all other sequelae we apply the relevant Dutch disability weight. We assume the duration of cerebral palsy without intellectual disability, severe hearing loss, moderate vision loss, and mild permanent disability to be the same as those with mild intellectual disability. We base the duration of epilepsy on life expectancy at birth assuming a twofold risk of dying as compared to the mortality rates of the general population. 147 Neonatal infections We estimate the incidence of neonatal infections from Australian hospital data. We assume 1 month of acute disability using the Dutch weight of 0.894 (same as for meningitis) for acute episodes. The main long-term sequelae are deafness, motor deficit disability and intellectual disability. We estimate intellectual disability attributable to neonatal infections as part of overall estimates for all causes of intellectual disability (see Section 2K). Other conditions arising in the perinatal period Here we include YLD for intellectual disability due to other conditions arising in the perinatal period. 1E Nutritional deficiencies Iron deficiency anaemia We model the following levels of severity for iron deficiency: non-anaemic, mild anaemia, moderate anaemia and severe anaemia. We define anaemia in terms of blood haemoglobin levels as per the GBD study. We derive our incidence estimates for iron deficiency anaemia using DisMod. We base our prevalence estimates for mild, moderate and severe anaemia for men and women aged 25 years and above from AusDiab. For the younger ages we use a variety of Australian studies, assuming 60% of cases are mild and the remaining 40% moderate (English & Bennett 1990; Karr et al. 1996; Nguyen et al. 2004; Oti-Boateng et al. 1998; Sadler 1996;). Iron deficiency causes anaemia but people can be iron-deficient and not anaemic and vice versa. To calculate iron deficiency without anaemia, we first have to estimate the prevalence of total iron deficiency which includes those with and without anaemia. We assume prevalence estimates of 10% and 1% in children aged 04 years and 514 years respectively (English & Bennett 1990; Mira et al. 1996; Oti-Boateng et al. 1998; Rangan et al. 1998; Ranmuthugala et al. 1998; Sadler 1996), with figures for other ages taken directly from 1989 National Risk Factor Prevalence SurveyIron status study. In the absence of population data on the overlap between iron-deficiency and anaemia, we assume half the cases with mild anaemia and all cases with moderate anaemia are also iron-deficient. For adults aged 15 years or over, we subtract the prevalence of iron deficiency combined with anaemia from the total prevalence of iron deficiency to avoid double-counting the disability. We use the same assumptions for disability and duration as in the previous Australian burden study. 2F Malignant neoplasms As in the previous study, the basis of YLD estimation for malignant neoplasms is a series of models of disease progression developed by the Dutch burden of disease study team for 26 cancers for which they determined disability weights (Stouthard et al. 1997). The disease model commences with an initial phase of diagnosis and primary therapy, with a duration of up to 12 months. After this, cases are classified as those who will and will not 148 be cured. Those who will be cured enter a phase of up to 5 years after which they are considered cured and have (with some exceptions as discussed below) no further cancerrelated disability. Those who will not be cured enter a phase (of variable length) of remission followed by a phase of disseminated carcinoma (lasting 12 months or less), then a terminal phase (lasting 1 month) and death. We allocate a Dutch weight to each of these phases. Where no Dutch weights were available for a specific cancer site, we extrapolate weights based on the cancer that it most resembles. The Dutch study did not derive a weight for the terminal phase of any of the cancers, so we use instead the Dutch weight for general end-stage disease. We modify this general model to each cancer site with results from studies in the peerreviewed literature and input from local clinicians to reflect local treatment practices. Disease incidence data The primary source of cancer incidence data is the AIHW & AACR National Cancer Statistics Clearing House database (AIHW & AACR 2001). This database records all cancer cases (except non-melanoma skin cancer) notified in Australia from 1982 to 2001. Cancer incidence rates in the Australian population change very slowly (AIHW et al. 2005). We apply the 2001 age- and sex-specific cancer incidence rates to the 2003 Australian population counts to estimate the 2003 cancer incidence. Two exceptions to this approach are breast cancer and non melanoma skin cancer. The breast cancer disease model requires details of size of tumour at diagnosis which are not available from the Clearing House database. Instead, we extrapolate the proportion of new cases in each size category from 2001 BreastScreen Australia data (AIHW 2005a) and all incident cases from the AIHW breast cancer size and nodal status report for 1997 (AIHW et al. 2001). We then apply these proportions to the 2003 incident cases projected from the Clearing House database. Non-melanoma skin cancer is not a notifiable disease in Australia and so is not within the scope of the Clearing House database. Instead, we extrapolate the incidence from the results of a 2002 Australian population survey of the incidence of non-melanoma skin cancer (NCCI 2003). Cure rate and mean survival time To estimate the cure rate and mean time to death for those not cured for each cancer we assume a Weibull distribution for the time from diagnosis to death and apply a non-linear model to the survival curves for each cancer (Verdecchia et al. 1998). We base the survival curves on all cases recorded in the Clearing House database with a diagnosis date between 1982 and 1997 which we follow-up for death until the end of 1999 (AIHW & AACR 2001). We base the durations of the initial treatment, disseminated and terminal stages separately for each cancer, using Dutch study assumptions, peer-reviewed literature and input from local clinicians. For those not cured, we base duration of the remission stage as the total average time to death (estimated from the Weibull model) less the sum of the other stages. For those cured, we base the duration of the stage following initial treatment as 5 years less the duration of the initial treatment stage. Again, breast cancer and non melanoma skin cancer are the two exceptions to this approach. Since the Clearing House database does not record tumour size, we base the survival times and cure rates on an analysis of breast cancer cases by tumour size published by the South 149 Australian Cancer Registry (South Australian Cancer Registry 2000). Because there are no national data on non melanoma skin cancer we estimate survival times and cure rates using assumptions modelled from published studies. Long-term sequelae of cancer The model for cancer in the previous Australian burden study assumed, with the exception of bone cancer, that cancer sufferers have no further burden following cancer cure. However, there are some cancers that are likely to have major sequelae causing long-term burden following successful treatment. The GBD study included long-term sequelae for colorectal cancer, breast cancer, female reproductive cancers and male genitourinary cancers. In addition, we include removal of one eye for eye cancer, removal of the larynx for larynx cancer, amputation for bone cancer and long-term brain injury for brain cancer. These sequelae and their associated severity weights are listed in the table below (Table A1.2). We estimate cancer-related rates of amputation, stoma, mastectomy, larynx, eye removal and infertility from Australian hospital data. We estimate infertility rates from cancer-related hysterectomies and assume these only apply to survivors under 40 years of age. We derive impotence and incontinence rates from a review of the literature. Results published in the literature note the similarity between the effects of treatment for brain cancer and other forms of traumatic head injury, so we assume that the rates of long-term brain injury from brain cancers are the same as the equivalent rates for head injury. We use the GBD disability weights for stoma, mastectomy, infertility, impotence and incontinence. For the disability associated with removal of an eye, amputation, and longterm brain injury we use comparable weights from the Australian study for long-term weight for an injury to an eye, major amputation and long-term effects of a brain injury in a non-fatal accident or injury, respectively. For removal of the larynx we assume that the Dutch weight for mild hearing loss, which is defined as some difficulty in actively participating in a conversation with one or more persons, is appropriate. 150 Table A1.2: Extra sequelae for cancer model Site/sequelae Proportion of survivors with sequelae (%) Severity weight Colorectal cancerstoma 0.09 0.21 Bone & connective tissueamputation 0.08 0.30 Breast cancermastectomy 0.51 0.09 Female reproductive cancerinfertility Cervix: 0.46 Uterus: 1.00 Ovary: 0.64 0.18 (ages under 40) Male genitourinary cancerimpotence and incontinence Prostate: 0.53 Bladder: 0.12 0.20 Brain cancerlong-term brain injury 0.05 0.35 Eye cancerremoval of an eye 0.45 0.30 Larynx cancerremoval of the larynx 0.35 0.04 2G Other neoplasms Benign neoplasms are not notifiable in Australia. As a result we base our incidence estimates for uterine myoma and benign brain tumour on Australian hospital data. Specifically, for uterine myoma we use the numbers of myomectomies and hysterectomies for fibroids. We assume that surgical treatment is undertaken for all cases of rapidly growing or large tumours and myoma-related symptoms. We assume a six month pre-operative state equivalent to the GBD weight for chronic pelvic pain and an additional three-week postoperative state equivalent to laparotomy (derived weight of 0.349 for health state 222211). Based on expert advice, we assume reproductive disability occurs in 3% of hysterectomy cases to whom we apply the GBD weight for infertility. We assume the additional burden associated with menorrhagia in undiagnosed women is included in our YLD estimates for this condition under the other genitourinary category. Our model for benign brain tumour is based on the model for malignant brain tumours where we model the disease in stages for survivors (diagnosis and initial treatment, and post-curative treatment) and non-survivors (diagnosis and initial treatment, pre-terminal and terminal). We adjust our incidence estimates on the assumption that 20% of hospitalisations are readmissions (Jaaskelainen 1986; Simoca et al. 1994). We base our survival estimates on Australian mortality data and assume successfully treated cases recover normal efficiency (Steiner et al. 1998) with a period of worry after treatment of 2 years. In the absence of specific disability weights, we use those for malignant brain tumours. 2H Diabetes Diabetes cases We estimate the incidence of insulin dependent diabetes mellitus (Type 1) from the National Diabetes Register (AIHW 2003b). We use DisMod to estimate prevalence and duration, 151 assuming no remission and age-specific risks of dying for all diabetes from the Asia Pacific Cohort Studies Collaboration a meta-analysis of 24 cohort studies from Asia, Australia, and New Zealand that assessed the effects of diabetes on the risks of major cardiovascular disease and death (Woodward et al. 2003). We estimate the incidence of non-insulin dependent diabetes mellitus (Type 2) for ages less than 25 years from the National Diabetes Register. We estimate the incidence of Type 2 diabetes for ages 25 years and above by subtracting the prevalence of Type 1 diabetes from the total prevalence of diabetes from AusDiab and then deriving incidence and duration in DisMod including an annual trend for the period 19801999 for incidence of 2.5% for males and 1.5% for females (Dunstan et al. 2002). There is no direct measurement of the trend in incidence/prevalence of Type 2 diabetes in Australia. Instead, we analyse the historical trend in diabetes mortality (which is relatively flat) and assume that this reflects the net effect of an increase in incidence and a decrease in case-fatality which in turn we assume to be equivalent to the trend in cardiovascular disease case-fatality (as the main causes of death in people with diabetes are of cardiovascular origin). Thus, we also incorporate a 20 year trend for the case-fatality rate (2% annual for males and 1% for females). We then project incidence and case-fatality forward to the year 2003 using the same trends as above and enter these into a DisMod model for total diabetes for 2003. We subtract out those with diabetic nephropathy to avoid double-counting as the Dutch disability weight for diabetic nephropathy includes the disability associated with diabetes per se. We use the Dutch disability weight for an uncomplicated diabetes case (0.070). Complications from diabetes for which we calculate YLD include retinopathy, cataract, glaucoma, renal failure, neuropathy, peripheral vascular disease, diabetic foot, amputations, ischaemic heart disease and stroke. Retinopathy We estimate the prevalence of mild and moderate vision loss from proliferative diabetic retinopathy in the Melbourne Visual Impairment Project (Weih et al. 2000). Experts confirmed that most retinopathy is treated before it leads to more serious vision loss. Therefore we estimate the incidence and duration of diabetic retinopathy in DisMod from the prevalence estimates from the Melbourne project, assuming no remission and twice the excess risk of mortality as for all diabetes. We base the proportion of cases due to Type 1 and Type 2 diabetes on the ratio of expected cases derived from modelling data on the progression of proliferative diabetic retinopathy from time of diagnosis (NHMRC 1997b; Tapp et al. 2003b). The Dutch disability weights for mild and moderate vision loss apply. Cataract and glaucoma We estimate the proportion of YLD from cataract and glaucoma attributable to Type 1 and Type 2 diabetes using population attributable fractions. We base the risks of cataract and glaucoma in diabetics from the Blue Mountain Eye Study (Mitchell et al. 1997) and use severity distributions from the Melbourne Visual Impairment Project (Weih et al. 2000). 152 Renal failure We estimate the incidence of diabetes-related renal failure using 2002 data from the Australian dialysis and transplant data. We use DisMod to estimate the average duration for people on dialysis, assuming a case-fatality rate reflecting observed deaths from the register. We base our annual remission estimates on observed transplant data: 85% in Type 1 diabetes cases aged 085 years or over for males and females combined; 6% in Type 2 cases under 65 years for males and females combined; and 0% in Type 2 cases aged 65 years or over for males and females combined. We use the Dutch disability weight for diabetic nephropathy (0.29). We estimate YLD for transplant patients assuming a case-fatality ratio reflecting observed deaths from the register and 3% remission due to graft failure (as these patients return back to the pool of dialysis cases). We assume a high disability weight (0.29) for the first 6 months following the transplant and a GBD weight of 0.11 thereafter. Neuropathy Tapp and colleagues provide estimates of diabetic neuropathy prevalence by time since diagnosis (2003a). We estimate by linear regression an annual increment in prevalence, which we then apply to survivors of incident cases of Type 1 and Type 2 diabetes by age as they progress to other age groups. Based on the Rochester Diabetic Neuropathy Study only 15% of Type 1 and 13% of Type 2 cases with diabetic neuropathy are symptomatic, of which 6% of Type 1 and 1% of Type 2 are severely affected (Dyck et al. 1993). The disability weight for Type 1 is 0.099 using the disability weight regression model (health state: 11111185%; 2222219%; and 2223316%) and for Type 2 is 0.074 (using the corresponding percentages of 87%, 12% and 1%). Peripheral vascular disease Tapp and colleagues provide estimates of peripheral vascular disease incidence and prevalence (Tapp et al. 2003a). We assume that only those with claudication are symptomatic. We estimate by linear regression an annual increment in the prevalence of diabetes-related peripheral vascular disease in order to derive incidence, similar to the approach for diabetic retinopathy. In the absence of Dutch or GBD disability weights for this condition we derive a weight of 0.19 using the disability weight regression model. Remission from surgery by vascular grafts is assumed to be 20%. Amputation and diabetic foot We estimate the incidence of diabetes-related amputations from Australian hospital data. We use GBD disability weights for these conditions and base our durations and proportions treated on expert opinion. We use amputation rate data for diabetics with foot ulcers from the Diabetes Research Foundation (Yue & Molyneaux 2005). From 19942005 the amputation rate for diabetics with foot ulcers was 5.3%. We calculate an average duration of 8.9 months after fitting a log normal function to follow-up data on the duration of foot ulcers. As there is no Dutch disability weight, we apply the GBD weight of 0.113. 153 Ischaemic heart disease and stroke We estimate the proportion of ischaemic heart disease YLD attributable to Type 1 and Type 2 diabetes using a population attributable fraction based on prevalence and the relative risk (2.0 and 2.5 for males and females respectively) of dying from ischaemic heart disease and stroke (2.0 for both males and females) amongst diabetics from the Asia Pacific Cohort Studies Collaboration (Woodward et al. 2003). 2I Endocrine and metabolic disorders Haemolytic anaemia We use Australian hospital data to estimate the incidence of hereditary haemolytic anaemia, assuming that annual admissions at age 0 years represent incidence. We model beta thalassaemia and other haemolytic anaemia separately to account for different durations. We assume that the average duration for beta thalassaemia is 35 years based on a USA review (US Preventive Services Taskforce 1996) and we assume that the life expectancy of persons with other haemolytic anaemia is the same for sickle cell anaemia, that is, around 25 years lower than the population average. In the absence of specific weights we use the GBD weight for very severe anaemia (0.25) and severe anaemia (0.09) for beta thalassaemia and other haemolytic anaemias respectively. Other non-deficiency anaemia We model the disability associated with aplastic anaemia and autoimmune anaemia. We base the prevalence of aplastic anaemia on Australian hospital data. We derive incidence and duration in DisMod using prevalence data, Australian mortality data where aplastic anaemia was an underlying condition and a remission of zero. We estimate the incidence of autoimmune anaemia using hospital data and assume that the average duration is 3 months. In the absence of a specific weight for other non-deficiency anaemias we use the GBD weight for very severe anaemia (0.25). Cystic fibrosis Massie and colleagues (2000) found the incidence of cystic fibrosis in Victoria over a 9-year period to be 3.5 per 10,000. This estimate is very similar to information from Queensland and Western Australia (Bower et al. 2004; Queensland Health 2004). We apply the Victorian estimate to the whole of Australia. We estimate the duration of cystic fibrosis in DisMod using the above incidence, no remission, and an age- and sex- specific risk of mortality from a patient-based USA study (Kulich et al. 2003). There is no disability weight for cystic fibrosis available. As obstructive lung disease is a major sequela, and the disease is progressive and fatal, we use the disability weight for severe chronic obstructive pulmonary disease (0.53). 154 Haemophilia We base our estimate of the incidence of moderate and severe haemophilia on Australian data (Street & Ekert 1996). We do not model mild cases of haemophilia since we assume they have zero disability, as bleeding only occurs as a result of injury. We use the same assumptions about severity distribution, duration and disability weights as the previous Australian burden study. 2J Mental disorders The 1997 National Survey of Mental Health and Wellbeing, including the child mental health and low prevalence disorder components, remains the only population-based data source for our estimates of most mental disorders (ABS 1998a; Jablensky et al. 1999; Sawyer et al. 2000). Table A1.3 summarises the mental disorders for which we calculated YLD, along with the sources of data on which our incidence estimates are based. Table A1.3: Sources of data for mental disorders Data source Mental disorder National Survey of Mental Health and Wellbeing 1997 Depression & anxiety; bipolar disorder; most substance abuse (alcohol, sedative and cannabis drug dependence or abuse); and borderline personality disorder Low Prevalence (Psychotic) Disorders Study Psychotic disorders Child and Adolescent Component of the National Survey of Mental Health and Wellbeing 1997 (Sawyer et al. 2000) Childhood disorders (separation anxiety disorder, attentiondeficit hyperactivity disorder) Epidemiological StudyNational Drug and Alcohol Research Centre Technical Report No. 198 (Degenhardt et al. 2004) Heroin dependence Alcohol and Other Drug Treatment Services National Minimum Data Set (AODTS-NMDS) collection <www.aihw.gov.au/drugs/datacubes/index.cfm> (accessed 15 December 2005) Stimulant dependence Reviews of epidemiological studies Eating disorders (anorexia nervosa and bulimia nervosa), autism, and Aspergers syndrome Depression & anxiety, substance abuse (excluding heroin and stimulant dependence), borderline personality disorder and bipolar disorder While the data sources have remained mostly the same as were used for the previous Australian burden study, there are a number of key methodological changes. First, we have grouped all anxiety disorders (panic, agoraphobia, social phobia, generalised anxiety disorder, obsessive-compulsive disorder, post-traumatic stress disorder and separation anxiety disorder) and the unipolar depressive disorders (major depression and dysthymia) that were previously modelled separately into a single disease category. This is based on the argument that the high degree of comorbidity and the similarity in psychological and drug treatment means that all these disorders can be considered as part of the same entity, with a continuum between mostly depressed to mostly anxious (for example (Andrews et al. 1990; 155 Andrews & Slade 2002). The advantage of this approach is that it takes away some of the difficulties of dealing with the frequent comorbidity among these disorders. Second, disability weights for all conditions derived from the national mental health survey continue to be based on the mental component score of the SF-12 but for this update we calculate a per unit change in disability weight for each unit change in the mental component score and apply this to all disorders. Dutch disability weights exist for mild, moderate and severe depression as well as for six different anxiety disorders for a combined mild moderate state and a severe state. Assuming that mild, moderate and severe are 1, 2 and 3 standard deviations, respectively, below the population mean of the mental component score we sought a mathematical function that best describes the range of disability weights. A second-order polynomial function gave the best fit. This transformation of categorical weights into a continuous scale allows us to calculate a disability weight for each respondent in the survey. Any mental component score value greater than the population mean of 52 is set to 0 and the weight for a mental component score of 20 is taken as the highest disability weight even if the mental component score is lower (this is done because otherwise the lowest mental component scores would correspond with a disability weight of greater than 1). Third, to deal with comorbidity, we apportion the disability weights calculated in the National mental health survey equally between the comorbid mental health diagnoses for each individual. In the previous Australian burden study we did the correction for comorbidity at the level of the number of people affected and hence reported lower than actual numbers of incident and prevalent cases. Our general model for these conditions derives incidence figures from the National mental health survey prevalence figures, using DisMod and assuming appropriate remission rates and relative risks of mortality from a meta-analysis (Harris & Barraclough 1998). We use the proportion of one-year prevalent cases reporting symptoms in the previous two weeks as an approximation of the proportion of time with symptoms and thus assume that all these conditions have a chronic nature with periods of remission in between. For children aged 517 years, we use prevalence estimates for depression and anxiety from the Child and Adolescent Component of the national mental health survey (Sawyer et al. 2000). In DisMod we use a remission rate of 0.043, a pooled estimate from follow-up studies of people with various anxiety disorders (Steketee et al. 1999; Wewetzer et al. 2001; Yonkers et al. 2003) and an increased relative risk of mortality of 1.5, a value in between the range of meta-analysis estimates reported for anxiety and depressive disorders (Harris & Barraclough 1998). The prevalence estimates for bipolar disorder in the previous Australian burden study were based on the international literature. This was because the prevalence figures from the National mental health survey were considered inaccurate due to a technical problem during the conduct of the survey. Subsequently Mitchell and colleagues (2004) have re-analysed the data and defined the prevalence of euphoric hypomanic/manic syndrome. They argue that with this definition around 95% of cases of bipolar disorder are captured. For the current estimates we use the same definition and adjust the 12-month prevalence by 100/95. In DisMod we use a remission rate of 0.035 calculated from a follow-up study (Angst & Preisig 1995) and an increased relative risk of mortality of 1.96 in men and 1.76 in women (Harris & Barraclough 1998). In this study we include all personality disordersrather than borderline personality disorder onlybut limit our estimates to those without any comorbid mental disorders. The 156 proportion of comorbidity between personality disorders and other mental disorders is so high that we argue that in most cases it ought to be seen as a risk factor rather than a separate condition. However, in order to capture all disability from mental disorders we include a category isolated personality disorder. The remission estimate of 17% is consistent between two follow-up studies (Grilo et al. 2004; Zanarini et al. 2003). The relative risk of mortality is 1.84 (Harris & Barraclough 1998). In the previous Australian burden study, estimates for alcohol use disorder were made separately for alcohol dependence and harmful alcohol use, and then presented as one disease category. In the current update we combine the two categories and create one DisMod model based on 12 month prevalence of any alcohol use disorder in the National mental health survey. The two other parameters in DisMod are a remission rate of 23.7% calculated from a two-year follow-up study (Booth et al. 2001), and an elevated mortality risk of 1.8 in males and 3.84 in females (Harris & Barraclough 1998). For cannabis dependence, we assume a remission of 8% (Swift et al. 2000) and no excess risk of mortality. There are no follow-up studies of people with sedative dependence. We use the same remission as in the cannabis model and apply an excess mortality risk of 2.1 reported for legal drug use (Harris & Barraclough 1998). Heroin dependence and harmful use Household surveys are likely to underestimate the true prevalence of heroin use (differential response between users and non-users and a greater proportion of users not living in households). Instead, we use higher estimates of regular heroin users based on triangulation between five data sources: ABS opioid deaths, ambulance attendances for drug overdose in New South Wales, New South Wales Health heroin pharmacotherapy client database, New South Wales data on arrests for drug offences, and data from the Alcohol and Drug Information Service on calls related to heroin use (Degenhardt et al. 2004). While the detailed comparison of databases was done for New South Wales, extrapolations were made for all jurisdictions by extrapolation of relationship between numbers under treatment or in contact with police and opioid mortality figures from New South Wales and the opioid deaths in each jurisdiction. In the previous Australian burden study, we assumed very high remission after age 45 years to reflect the low prevalence of heroin use. However, expert advice that this is a cohort effect rather than a high remission effect explains the drop in prevalence at older ages. In current estimates we allow DisMod to build up prevalence figures at older ages. Back projection methods by the National Drug and Alcohol Research Centre assumes a risk of dying from overdose of 0.8% per year (Law et al. 2001). We assume a case-fatality rate of 1% to account for raised mortality from other causes. The overall relative risk calculated in DisMod is of the same order of magnitude as reported elsewhere (AIHW: Ridolfo & Stevenson 2001; Darke & Ross 2002). The disability weight for heroin dependence of 0.27 was derived by Victorian mental health experts for the previous Australian burden study and is close to the GBD disability weight estimate of 0.252. Stimulant dependence We decided to use treatment figures rather than the estimates of prevalence of stimulant dependence from the National mental health survey as there has been a marked increase in 157 the use of stimulants since 1997 and the survey results show an erratic age pattern as only few cases were identified. Instead we estimate the prevalence of stimulant dependence from the number of closed treatment episodes in 20022003 where the principal drug of concern was listed as amphetamines (Alcohol and Other Drug Treatment Services National Minimum Data Set) collection. We inflate these figures by 5.5 as described by McKetin and colleagues (2005). We estimate remission by first entering prevalence, a relative risk of 0 and a case-fatality rate of 0, into DisMod. We thus get DisMod to produce an estimate of remission that best replicates the age pattern of prevalence. The average remission across all ages was 12%. We then run the DisMod model again with same prevalence, this remission rate and a relative risk of 2.1 for excess mortality as reported for legal drug use (Harris & Barraclough 1998). We derive a disability weight for stimulant dependence as we have done for all other conditions in the National mental health survey and thus assume that the same average severity found among the lower number of cases with stimulant dependence in the survey reflects that of all cases in the population. Psychotic disorders Estimates for psychotic disorders are based on prevalence from the Low Prevalence (Psychotic) Disorders Study conducted in Australia in 1997 as part of the National Survey of Mental Health and Wellbeing. This survey measured an overall estimate of 4.7 per 1,000 population. The low prevalence study suffered from a low response rate by general practitioners contacted in the study areas and therefore under-represented people with psychotic disorders who are solely managed by their general practitioner (Lewin & Carr 1998). Before conducting further analysis, we adjust upwards to one in three the number of people in the survey who are wholly treated by a general practitioner and adjust downwards by a factor of 0.841 to reflect only those with schizophrenia and related diagnoses and not those with a diagnosis of bipolar or affective psychosis. Annual remission is based on a number of longer term studies and is set at the median of the reported rates (1.5%) (Ciompi 1980; Harding et al. 1987; Harrison et al. 2001; Helgason 1990; Huber et al. 1980). We derive incidence and duration figures from DisMod using a 54% higher risk of mortality overall for people with schizophrenia (Harris & Barraclough 1998), with an age pattern imposed by the relative frequency by age that schizophrenia is mentioned in death records. The DisMod incidence output indicates that almost all psychotic disorders have their beginning in late adolescence or early adulthood, with a small second peak in post-menopausal women. We assume that the average time spent in psychosis is 30% (Leff et al 1992). We use a composite weight based on 30% of the GBD weight for psychosis corresponding to the estimated time spent in this state and 70% of the treated weight (0.3 x 0.627 + 0.7 x 0.351 = 0.434). The low prevalence study reported a higher proportion (61%) of people with a psychotic disorder having current delusions or hallucinations. It also stated that 86% are taking prescribed medication and that 83% of the total reported that their psychotic symptoms respond to pharmacological treatment. The first finding would indicate that our composite disability weight is too low but the second finding would support a lower weight. For the Assessing Cost-Effectiveness (ACE)Mental Health study, disability weights for each individual in the low prevalence study were estimated using a sliding scale between the highest and lowest of Dutch disability weights for schizophrenia and anchoring individuals on this scale based on their score on the diagnostic interview for psychosis disability module that was included in the survey (Haby et al. 2004). The mean disability weight across the sample using this 158 method is 0.39. We decided to continue to use the 0.434 disability weight as in the previous Australian burden study. Eating disorders Estimates for bulimia are based on a prevalence rate of 0.7% among Swiss 1417 year old females (Steinhausen et al. 1997). This is the mid-point in the range of prevalence between 0.5% and 1% reported from more rigorous epidemiological studies (Gilchrist et al. 1998). We calculate a remission rate of 0.21 from figures reported in a review of follow-up studies (Keel et al. 1999). We derive incidence and duration estimates for women from these figures using DisMod, assuming the age at onset is between 14 and 29 years with no increased risk of mortality. Estimates for anorexia are based on a 0.5% prevalence among females older than 15 years (Gilchrist et al. 1998; Keel et al. 1999) and a remission rate of 0.11 calculated from a follow-up study (Strober et al. 1997). We use DisMod to derive incidence and duration estimates for women from these figures, assuming the age at onset is between 14 and 29 years with an increased annual risk of mortality of 0.59% (Sullivan 1995). We assume the incidence in males is 10% of the rate in females. We use the Dutch weight of 0.28 for both types of eating disorder. Childhood disorders Australian prevalence data for childhood attention deficit with hyperactivity disorder come from the Child and Adolescent Component of the 1997 National Survey of Mental Health and Wellbeing (Sawyer et al. 2000). We define attention deficit with hyperactivity disorder to include children with a diagnosis on the survey and whose parents report the child having more emotional or behavioural problems than have other children of the same age. The estimates of burden of attention deficit with hyperactivity disorder were derived from prevalence rates of 6% in male children, 3% in female children, 2% in male adolescents and 1% in female adolescents. Our incidence figures were derived from DisMod, assuming an age at onset of 36 years and a remission rate of 0.15 (Hill & Schoener 1996). To reproduce the prevalence pattern we use a higher remission rate of 0.25 in adolescents aged 1019 years and 0.3 thereafter. We assume no increased risk of mortality. We use the Dutch weights for both mild and moderate-to-severe attention deficit with hyperactivity disorder (0.02 and 0.15), and weight these by the severity distribution found in the 1997 survey to derive a composite disability weight. Autism is part of pervasive developmental disorders; the other important condition in that category is Aspergers syndrome, which was described at about the same time as autism. Autism is characterised by the triad of language or communication impairment, social impairment and behavioural impairment (obsessions, rituals). However, Aspergers syndrome has only the latter two components and is not associated with intellectual disability, as is the case with 80% of autistic children. Behavioural problems are a predominant feature in children with Aspergers syndrome. We derive the incidence of autism and Aspergers syndrome from an Australian study with data from treatment and educational support services in Western Australia and New South Wales. We assume no remission and an elevated risk of mortality as reported by Shavelle and colleagues (2001). We use the average duration of mild intellectual disability and the Dutch disability weight of 0.55 for autism, and for Aspergers syndrome an estimated weight 159 of 0.25 based on expert advice that the condition is worse than moderate to severe attention deficit with hyperactivity disorder but much less severe than autism. 2K Nervous system and sense organ disorders Dementia A door-to-door population-based two-phase investigation method (screening followed by detailed neurological examination by a psychiatrist) is the most accurate epidemiologic approach to estimate the epidemiology of dementia and Parkinsons disease (Benito-Leon et al. 2004). We base our estimates of the prevalence of dementia for people aged 65 years or over on a recent European meta-analysis of population-based door-to-door studies conducted by the Neurologic Diseases in the Elderly Research Group (Lobo et al. 2000). We proportionately redistribute the one-third of cases that constitute other or mixed type to Alzheimers and vascular dementia. We estimate the prevalence of dementia below the age of 65 years from a recent UK study of patients aged 3064 years (Harvey et al. 2003). We use relative risks of mortality for Alzheimers disease and vascular dementia from a survival study of incident cases that controlled for comorbidity (Aguero-Torres et al. 1999). The estimated mortality risk for all dementia from this is comparable to the results of the meta-analyses of dementia prevalent cases and survival (Dewey & Saz 2001; Jagger et al. 2000). We prefer using the former because it provides type-specific survival data. We derive incidence and duration using DisMod, based on the aforementioned representative population-based studies of prevalence, assuming no remission and relative risks from the incident-based survival study. This model gives average durations across all ages for both sexes of around 4 years which was in keeping with the literature on the survival of prevalent cases (Aguero-Torres et al. 1999; Helmer et al. 2001). We model dementia as a progressive illness and discount the latter stages back to incidence of disease. We use the disability weights derived by the previous Australian burden study (which combined the Dutch weights with a severity distribution from a European population-based cohort study). Epilepsy We base our incidence estimates for primary epilepsy on the 198084 Rochester Epidemiology Project medical record linkage system (Zarrelli et al. 1999). We use these incidence estimates, assuming no differentials by sex, with age-specific remissions (Annegers et al. 1979) and an overall standardised mortality ratio of 1.3 (Tomson 2000) to derive estimates of incidence and duration using DisMod. We use the Dutch disability weight for epilepsy (0.110). Parkinsons diseaseWe only explicitly model primary Parkinsons disease (ICD-10 code G20). We assume that secondary Parkinsonism is accounted for under other relevant disease categories. 160 We base our estimates of the prevalence of Parkinsons disease from a recent European metaanalysis of population-based door-to-door studies conducted by the Neurologic Diseases in the Elderly Research Group (de Rijk et al. 2000). There are no sex differences in the prevalence of Parkinsons disease. We do not use the Australian studies on the prevalence of Parkinsons disease since we consider them to be outliers; they give prevalence estimates two to three times higher than most of the literature (Chan et al. 2001, 2005). We base our relative risk of mortality on the meta-analysis of prevalent cases of Parkinsons disease and survival undertaken by the Neurologic Diseases in the Elderly Research Group (Berger et al. 2000). We plot and fit the risk of mortality by age using an exponential trendline to smooth the irregular pattern by age. We derive incidence and duration using DisMod assuming no remission and a relative risk of mortality for males and females of 3.1 and 1.8 respectively, resulting in average durations of 4.5 and 9.8 years. These durations are broadly consistent with durations reported in the literature (Elbaz et al. 2003; Fall et al. 2003; Herlofson et al. 2004; Hughes et al. 2004; Morgante et al. 2000). We derive disability weights from an Australian patient-based cohort study (Hely et al. 1999) reporting on the distribution of Hoehn and Yahr stages (which corresponds with the descriptions of the severity states of Parkinsons disease for which Dutch disability weights are available) and survival at each 2-year interval. We model Parkinsons disease assuming that all cases start with mild symptoms and progress over time to moderate and then severe symptoms over time. From simple linear regression lines we derive an annual increase in those with moderate and severe symptoms. The proportion of cases over time who are in the moderate category is the balance between those moving from mild to moderate and those exiting moderate by shifting to the severe category. For each age group we calculate the average disability weight during the estimated average duration. As severity progresses with time since incidence and younger age groups have longer durations, disability weights are highest in the younger age groups. Motor neurone disease We base our incidence estimates for motor neurone disease on Australian mortality data. We assume that incident cases equal annual deaths due to motor neurone disease. Our estimates for males and females are consistent with international literature for males (Chancellor et al. 1993). We assume average durations of 2.9 years for those aged 064 years and 1.9 years for people aged 65 years or over. We base our duration assumptions on Australian and international literature (Forbes et al. 2004; Sach 1995). In the absence of a specific disability weight we use the Dutch weight for progressive multiple sclerosis (0.67). Multiple sclerosis We estimate the prevalence of multiple sclerosis using 1981 and 1996 estimates of multiple sclerosis for some Australian states and territories (Barnett et al. 2003; Simmons et al. 2001) with extrapolations based on latitudinal differences for jurisdictions with no estimates. We assume that changes over time represent improvements in identification rather than changes in epidemiology. We derive incidence and duration using DisMod assuming no remission and age and sex specific case-fatality rates based on a 25-year New Zealand cohort study (Miller et al. 1992). 161 In 10.8% of patients the disease has a progressive course from the onset (Roxburgh et al. 2005). The median time it takes to reach an Expanded Disability Status Scale score of 6 (equivalent to having to use a cane) in those with a relapsing-remitting course is 30 years (Tremlett et al. 2006). We use the Dutch weights for relapsing-remitting (0.33) and progressive (0.67) phases and assume that those with relapsing-remitting disease have 30 years at the lower disability weight and the remainder at the higher disability weight level. Huntingtons choreaHuntingtons chorea is modelled in DisMod using prevalence from the literature (McCusker et al. 2000), assuming no remission and mortality data. We assume a duration of 20 years for the younger age groups and apply the durations from DisMod for ages 65 years or over. Assuming similar progression of disease as in Parkinsons, we adopt the weights for the three stages of this disease. Muscular dystrophy For muscular dystrophy in males, we use the average incidence rates from New South Wales, Victoria, Queensland, Western Australia and the Australian Capital Territory (Cowan et al. 1980; Emery 1991). The incidence for females is calculated by applying the sex ratio from mortality data. In the absence of specific weights for this condition, we assume the initial symptomatic phase is similar to the initial stage of Parkinsons disease, the phase in which walking becomes impossible is similar to that of paraplegia, and the final stage is equivalent to quadriplegia. Vision loss Our incidence estimates for vision loss are based on the results of the Melbourne Visual Impairment Project, which assessed visual acuity and the prevalence by cause of mild, moderate and severe visual impairment in a sample representative of Victorians (Weih et al. 2000). For glaucoma, refraction errors, macular degeneration and the category other vision loss, we derive incidence and duration of related visual impairment using DisMod, assuming no remission and a relative mortality risk of 1. For glaucoma we use Dutch disability weights for mild, moderate and severe vision loss to derive a composite disability weight from the severity pattern across all ages (as the age-specific data are based on small numbers). For macular degeneration, refraction errors and other vision loss we derive agespecific disability weights. We estimate the incidence of mild and moderate cataract-related vision impairment using Australian hospital data assuming that 50% of surgically corrected cases had vision loss in both eyes prior to operation for 1 year on average and that 90% of cases are mild and 10% are moderate. We estimate the prevalence of un-operated cataracts as the difference between the prevalence of cataract-related visual impairment estimated by the Melbourne study and the number of surgical corrections. This leads to a small estimate of un-operated cataracts in the elderly over 80 years of age. We use this to estimate the incidence of un-operated cases of cataract-related severe vision loss in DisMod, assuming no remission and a relative risk of 1.5. For cataract-related vision loss at ages 014 years we assume duration of 2 years and for ages 15 years or over we assume a 1-year duration. Incident cases of un-operated cataract were assumed to be prevalent cases waiting on average 1 year for cataract surgery. We use 162 Dutch disability weights for mild and moderate cataract-related vision loss. For severe cataract-related vision loss we estimate a combined disability weight using the Dutch weights for each of the stages along with prevalence data from the Melbourne study to derive combined stages age-specific disability weights. The proportion of glaucoma and cataract-related vision loss attributable to diabetes is then determined from relative risks from the Blue Mountain Eye Study (Mitchell et al. 1997) and only non-diabetes-related vision loss is included in the YLD estimates for these categories. Hearing loss We model hearing loss as a progressive condition with mild (2534 dB and 3544 dB), moderate and severe stages so that prevalent cases with moderate or severe impairment are regarded as incident cases of mild impairment at an earlier age. We use survey prevalence data from South Australia (Wilson et al. 1999), initially modelling the prevalence of severe hearing loss, no remission and a relative risk of 1 in DisMod. We use incidence of severe hearing loss from the DisMod output as mortality in the moderate hearing loss DisMod model; this takes the cases of severe hearing loss out of the pool of susceptible cases for moderate hearing loss and hence gives more accurate average durations than if remission were used as remitted cases in the DisMod model, as the cases continue to be subject to the hazard of incidence. Similarly, we use incidence of moderate hearing loss as mortality in mild hearing loss (3544 dB) and incidence of mild hearing loss (3544 dB) as mortality in mild hearing loss (2534 dB). From examination of the prevalence data by level of severity and age, and assuming that all cases progress from the mildest to most severe category, it seems reasonable to assume that on average progression to the next severity level occurs at 5 year intervals between mild (2534 dB) and mild (3544 dB), and at 10 year intervals from mild (3544 dB) to moderate and moderate to severe. From the cross-sectional data on prevalence it is not possible to estimate these progression times exactly. However, to be consistent with other disease models where subsequent severity levels for the same health state are discounted back to first incidence, we apply a 25-year lag for severe hearing loss, 15 years for moderate and 5 years for the mild (3544 dB) categories. Dutch weights of 0.04, 0.12 and 0.37 apply for mild (3544 dB), moderate and severe hearing loss, respectively. For the mild (2534 dB) category we assume a disability weight of 0.02, half that of the mild (3544dB) category. Intellectual disability Intellectual disability is categorised into the following levels: mild, moderate, severe and profound, with intelligence quotient (IQ) ranges of 5069, 3549, 2034, <20 respectively. This categorisation is based on the Dutch disability weight criteria. We estimate the incidence of mild-moderate and severe intellectual disability using the Intellectual Disability Exploring Answers Database, a Western Australian population-based dataset of children with intellectual disability identified through disability and educational services between 19831996. We adjust the severity distribution of incidence data to account for unspecified cases and redistribute cases so that the severity level as defined by IQ is comparable to the Dutch disability weight criteria. Then we extrapolate incident cases by the two severity levels (mild-moderate and severe) to four levels of severity (mild, moderate, severe and profound) using the average severity distribution from two Australian studies (Einfeld & Tonge 1996; Wellesley et al. 1992). We assume that because neither study 163 recruited cases from school services, mild cases were underestimated and base our estimation of mild cases on the balance of the mild-moderate category. This gives the following proportionate distribution of incident cases by severity: mild (76%), moderate (14%), severe (7%) and profound (3%). In order to derive plausible durations of intellectual disability by the four stages of severity we calculate the proportional difference in life expectancy by level of severity of intellectual disability in comparison to the life expectancy of the general population from a 35-year Finnish follow-up study (Patja et al. 2000). We model the incidence and duration of intellectual disability in DisMod assuming that 90% of intellectual disability, based on Australian population data, occurs in the first year of life and the remaining 10% occurs in the 14 age group, no remission, and a relative risk of mortality that gives an average duration by severity level based on the extrapolation of Finnish data to the 2003 Australian life table. We do not include the YLD for intellectual disability as a discrete category in the main listings of this burden study. Instead, incident cases of intellectual disability are attributed to underlying causes (such as congenital disorders, epilepsy, autism, perinatal conditions, meningitis, brain tumours and cerebral palsy) using findings from the Australian Child to Adult Development Study, a longitudinal study of behavioural and emotional problems in 429 young people with intellectual disabilities. We use data from two publications of this study to produce the proportionate distribution of the underlying cause of intellectual disability by severity level and sex (Mowat et al. unpublished; Partington et al. 2000). We calculate YLD for each underlying cause using incidence and duration derived from DisMod and the Dutch disability weights for mild, moderate, severe, and profound intellectual disability. Migraine We base our prevalence estimates for migraine on the National Health Survey data and our incidence estimates for migraine on international data (Stewart et al. 1991). We estimate the incidence and duration of migraine in DisMod using prevalence, incidence, and a casefatality rate of zero. Within DisMod, we use manual smoothing to extrapolate incidence to older ages. We assume that 20% of cases receive treatment in developed countries. We assume that the average duration for untreated and treated episodes is 24 hours and 6 hours respectively. We derive average disability weights for untreated and treated models using frequency, severity, and disability weight data from Global burden of migraine in the year 2000(Leonardi & Mathers 2003). 2L Cardiovascular disease Ischaemic heart disease Three health states are modelled separately for ischaemic heart disease: angina pectoris, acute myocardial infarction and heart failure. We model the incidence of angina pectoris as the number of admissions to hospital without any mention of angina in any previous admission in 15 years of linked hospital records in Western Australia (Department of Health of Western Australia et al. 2005; Holman et al. 1999) and adjust by the ratio of admissions for 164 angina pectoris between Western Australia and the whole country. We model angina pectoris pre- and post-myocardial infarct together. The duration is determined in DisMod, assuming remission estimated from the number of revascularisation procedures from Australian hospital data and age- and sex-specific case-fatality rates calculated over the period 19982003 in prevalent cases of angina pectoris (that is, anyone with an admission for angina pectoris since 1988 and still alive over the follow-up period). Assuming that about half of the declining ischaemic heart disease mortality reflects change in the case-fatality rate rather than incidence (Unal et al. 2004), we apply half of the ischaemic heart disease mortality trend observed over the period 19792003 in DisMod to incidence and the other half to case-fatality. We assume 95% of angina is experienced at the mild-moderate level with the corresponding Dutch disability weight of 0.08, and the remaining 5% with a weight of 0.57. For people discharged alive following acute myocardial infarction in 2003, we calculate a period of 3 months of disability at the GBD treated disability weight of 0.395. Heart diseases resulting in heart failure Population-level prevalence or incidence information on heart failure is absent in Australia and scarce elsewhere. In 2001, by extrapolation from US studies a rough estimate was made of about 300,000 prevalent cases of heart failure in Australia (Krum 2001). Complicating factors in the estimation of heart failure prevalence are that estimates from other countries and different time periods may not apply to the current Australian situation. Ischaemic heart disease is the underlying cause of heart failure in the majority of cases, and there has been a steady decline in the risk of ischaemic heart disease since the early 1970s combined with improved survival due to improvements in therapeutic options. The first would cause a reduction in prevalence, while the latter would lead to higher prevalence. It is not clear what the net effect of these two influences would be on the prevalence of heart failure. Using hospital data is also not straightforward as the current wisdom is that there has been a change in the case load of people presenting to tertiary health facilities with this condition, following the wider use of improved pharmacological treatment combinations since the 1990s, resulting in a greater proportion of cases being successfully treated in primary care. Nevertheless, our model for heart failure starts with a description of the epidemiology of hospitalised heart failure, for which we have extensive information from Western Australia. From the linked data set of all hospitalisations and deaths in this state, we identify people who presented to hospital with heart failure (either as a primary diagnosis or as an associated condition) at any time in the period 19902003. To derive case-fatality, we calculate the number of years lived between 1998 and 2003 by anyone who had ever been admitted with a diagnosis of heart failure since 1990. The case-fatality rate was then taken as the number of deaths over person-years of follow-up in 5-year age groups after subtracting out the background mortality. The complete descriptive epidemiology in this group is derived in DisMod from incidence and case-fatality, the third parameter being zero remission (that is, people do not recover from heart failure). We include in this model a declining trend in case-fatality over the last 10 years of 3% per year for males and 1% per year for females (derived from our survival model), and a 2% decline in incidence per year for both males and females over the last 35 years. This latter figure is half the annual decline we observe for ischaemic heart disease mortality over this period, ischaemic heart disease being the major driver of heart failure 165 risk. The other half of the decline in ischaemic heart disease mortality we assume to be due to improvements in case-fatality (see above) (Unal et al. 2004). There is little information on the incidence of heart failure in the community (that is, not yet diagnosed cases and those diagnosed but treated in the primary case setting without requiring hospitalisation). We assume that this group has less severe disease with better survival compared to their hospitalised counterparts. We also assume that when they die, it is less likely that heart failure will be mentioned as the underlying cause of death. We have data on the number of hospitalised cases of heart failure who died with heart failure as the underlying cause of death and we know the overall number of deaths coded to heart failure. Assuming that the linkage of hospital and death records in Western Australia is complete, we then assume that the balance of heart failure coded deaths occur in never-hospitalised cases of heart failure. In the absence of data to characterise the never-hospitalised cases of heart failure we make two assumptions. First, to account for lower severity we assume that their case-fatality rate is lower by 25%. Second, we assume that deaths in non-hospitalised heart failure cases are 25% less likely to be coded to heart failure. Among hospitalised cases that die, the probability of receiving an underlying cause of death code of heart failure (428 in ICD-9 and I58 in ICD-10) is 3.6% in males and 5.3% in females. If non-hospitalised cases are 25% less likely to be assigned a code of heart failure the percentage of total excess deaths coded to heart failure would be 2.7% in males and 3.9% in females. From this we can derive the total number of deaths due to heart failure in nonhospitalised cases (3,199 in males and 3,958 in females over the period 20012003). By adding in the 3,833 deaths from ever-hospitalised cases of heart failure in males and 4,186 in females, we can calculate the average population mortality rate of heart failure over the period. These rates (calculated by age and sex) are the inputs to a second iteration of DisMod, together with zero remission and the case-fatality rate of the first DisMod model of hospitalised heart failure cases adjusted downwards to reflect the proportion of never-hospitalised cases having 25% lower case-fatality. We continue to use the same assumptions on trends in casefatality and incidence as in the first model. The output of the second DisMod iteration then gives us the incidence, prevalence and average durations for all heart failure, which feed into our YLD calculations. The total prevalence of heart failure in Australia in 2003 is thus estimated to be 220,000 cases. We then identify the underlying causes for all heart failure casesrheumatic heart disease, hypertensive heart disease, ischaemic heart disease, pulmonary heart disease, inflammatory heart disease, non-rheumatic valvular heart diseasein the Victorian linked hospital admission dataset between 1996 and 2002, if any of these were mentioned as a cause in the six years of hospital admission data. We then adjust the proportions, by age and sex, of all underlying causes so they add up to 100% to account for cases with none or more than one underlying cause identified. We use the duration, together with the incidence and prevalence estimates initially obtained from the heart failure model described above, multiplied by the proportion of heart failure cases for each of the above six underlying causes, to calculate the YLD for each of these conditions (including ischaemic heart disease). Stroke We model stroke in terms of the following health states: a short period of disability for those who die in the first 28 days, survival beyond 28 days with no permanent impairment at one year after onset, and survival beyond 28 days with permanent impairment. Admissions for stroke in the year 2003 are the starting point for our estimate of incidence. To get an 166 approximation of first-ever stroke incidence we take the ratio of hospital admission figures from the North-East Melbourne Stroke Incidence Study (NEMESIS) area during the time of the study to the reported NEMESIS first-ever incidence figures (Thrift et al. 2000) and apply this ratio to 2003 Australian hospital admissions for stroke. Next, we subtract a proportion of cases that die, using a 28-day case-fatality rate by stroke subtype as reported by Thrift and colleagues. The case-fatality rate of stroke comes from the Western Australian linked database using a similar approach to that described above for heart failure. The case-fatality rate for DisMod is the excess mortality in prevalent cases, defined in our analyses as anyone still alive at the beginning of the follow-up period (mid 1998mid 2004) with a mention of stroke during any admission between 1989 and 1998 as well as any new cases of admitted stroke during followup. Follow-up time and numbers of deaths were analysed for each 5-year age group and the overall case-fatality rate reduced by the relevant background mortality. Analyses by Judy Katzenellenbogen in Western Australia for her PhD indicate that after the first 28 days the case-fatality rate does not vary significantly by type of stroke and that the DisMod assumption of a case-fatality hazard that varies with age but not with time since stroke is plausible. Disability weights are derived from one-year follow-up data of stroke survivors in the Perth Community Stroke Study analysed by Judy Katzenellenbogen to compare health status information before and at 4 months and 1 year after the stroke event. Other cardiovascular disease Heart failure is the main disability from rheumatic heart disease, non-rheumatic valvular disease, hypertensive heart disease and the group of inflammatory heart diseases (including myocarditis, cardiomyopathy, endocarditis and pericarditis). The proportions of heart failure cases for each of these causes are derived as described above for all heart failure. For rheumatic heart disease and non-rheumatic heart disease we do a separate DisMod model based on heart failure prevalence for these causes and taking into account the remission through surgical interventions using Australian hospital data. For aortic aneurysm, we assume the hospitalisation rate reflects incidence. For peripheral vascular disease, we assume the hospitalisation rate reflects prevalence at all ages. We derive the incidence from DisMod, assuming a relative risk of 2 and a remission rate of 0.1, which approximates the number of surgical interventions as a proportion of total prevalent cases. For aortic aneurysm, we assume a one-month period of disability during treatment and no residual disability for those who survive treatment. Without a disability weight for this health state, we use the derived weight for laparotomy (0.349). For peripheral vascular disease, we use derived weights of 0.243 and 0.257 for men and women respectively, based on severity distributions from the 1993 Australian disability survey. Weights for amputations are from the GBD study. 167 2M Chronic respiratory diseases Chronic obstructive pulmonary disease We estimate the prevalence of chronic obstructive pulmonary disease for cases with a forced expiratory volume in one second of less than 70% of predicted (excluding those with a doctor defined diagnosis of asthma) using the 199495 Busselton Study (Knuiman et al. 1999). While this study sample comprises a selected rural population in Western Australia, we assume the data are representative of prevalence in all areas of Australia. We use DisMod to estimate the incidence of chronic obstructive pulmonary disease in 1994, assuming no remission and a relative risk equivalent to that calculated from death rates attributed to smoking (see section on risk factors). We include a trend of 2% per year for males and 3% per year for females based on trends in chronic obstructive pulmonary disease mortality since 1979. We calculate 2003 incidence estimates by applying age- and sex-specific trends (based on mortality) to the 1994 incidence estimates. We then use DisMod with the same assumptions about remission and relative risk of mortality to model prevalence, age of onset and duration for 2003. We derive a composite average disability weight for males (0.168) and females (0.159) using the Dutch weights for mild, moderate and severe chronic obstructive pulmonary disease and the proportionate distribution by level of severity of dyspnoea from the Busselton Study. We add the proportion of heart failure cases attributed to pulmonary heart disease on the basis that chronic lung disease is the underlying cause. Asthma We estimate the prevalence of asthma for cases that have a positive airway hyperresponsiveness test and wheezing in the last 12 months from the literature (Bauman et al. 1992a; Peat et al. 1992, 1994, 1995; Toelle et al. 2004). Although these two criteria may underestimate the true prevalence of asthma, a reliance on self-reported wheeze alone overestimates figures by up to a third (Toelle et al. 1992; Van Asperen 1995). We estimate the prevalence of asthma in children aged 12 years to be 5.75%, using a report of wheeze from the US (Martinez et al. 1995) which we adjust by 42% to obtain an estimate that reflects those with wheeze having asthma (Peat et al. 1994, 1995; Toelle et al. 2004). We estimate the prevalence of asthma to be 12.3% in boys and 8.8% in girls aged 318 years, using an average of 3 studies from 1992 to 2002 (Peat et al. 1994, 1995; Toelle et al. 2004) and a male-to-female ratio of 1.4:1 (Gergen et al. 1988). For adults, we average the prevalence data from the early 1990 studies (Bauman et al. 1992a; Peat et al. 1992, 1994, 1995) since these were the last studies in adults to have used a positive airway hyper-responsiveness test and assume no change in the prevalence of asthma over time based on the literature and the observed trend in children. We use a male-to-female ratio of 1:1.5 (DHS 2002) to give an estimated 2003 asthma prevalence of 5% and 7.5% in male and female adults. We derive incidence estimates from DisMod assuming age-specific remission rates from a follow-up study in the US (Bronnimann & Burrows 1986), which are consistent with overall remissions reported by Australia studies (Xuan et al. 2002). From findings reported by Bauman and colleagues, we calculate that asthmatics are symptomatic 12% of the time (Bauman et al. 1992b). Rather than use the Dutch weight for this health state (0.36), which we consider to be for a more severe health state than the average for symptomatic asthmatics in the population, we use a derived weight of 0.229 based on the severity distributions found in the 1998 Australian disability survey (ABS 1998b) and the disability weight regression model. The remainder of the time 168 we assume is spent in a state equivalent to the Dutch weight for asthma controlled by treatment (0.03). This results in a combined weight of 0.054. 2N Diseases of the digestive system Peptic ulcer disease In the absence of Australian population data on the frequency of peptic ulcer disease, we assume that all incident cases of peptic ulcer disease visit a general practitioner and base our estimates on the Australian general practitioner data. We assume that 83% of cases are treated by Helicobacter pylori eradication therapy, which has a cure rate of 90% (Mollison et al.1999). We model those who are cured using eradication therapy as being symptomatic for one month, with no residual disability. We assume that the remainder of those who are treated but not cured (including those receiving alternative treatments) receive relief from their treatment but remain with the condition for the GBD duration. Untreated cases we assume to be symptomatic for the same period. Because the annualised Dutch weight for peptic ulcer disease is implausible, we use derived weights from the Dutch study for both symptomatic and treated states. Cirrhosis of the liver The methods of deriving estimates of alcohol-related cirrhosis and the category of other cirrhosis have been described in the section on hepatitis. Inflammatory bowel disease We model two manifestations of inflammatory bowel disease: Crohns disease and ulcerative colitis. We estimate the incidence of inflammatory bowel disease in adults from a European study (Shivananda et al. 1996) and for children we pool estimates from a number of international studies based on a recent review (Griffiths 2004). The relative risks of mortality due to the two types of inflammatory bowel disease were based on the findings of a recent large UK study which showed that inflammatory bowel disease was associated with a small overall increase in mortality after controlling for smoking and sex (Card et al. 2003). We assume no remission and derive a composite disability weight (0.224), assuming that 20% of time is spent with active exacerbation and the remainder is in remission (Griffiths 1995; Hendriksen et al. 1985; Stonnington et al. 1987). For inflammatory bowel disease (and vascular insufficiency of the intestine, diverticulitis and intestinal obstruction), we assume that a proportion of cases have more complicated surgery involving the creation of a stoma (a surgical opening in the skin of the abdomen for excretion of faeces) that can be either permanent or temporary. We estimate the incidence of inflammatory bowel disease cases that receive a temporary or permanent stoma from Australian hospital data. We apply the ratio of stoma for Crohns disease to stoma for ulcerative colitis from an analysis of Victorian linked hospital data. Similarly the average duration of temporary stoma was estimated from Victorian hospital data from 199899 to 200102 to determine if they were closed and, if closed, the time to closure. The duration of permanent stoma was taken to be the same as the duration of the respective condition. We 169 assume stomas not yet closed within this period remain open indefinitely. In the absence of a specific weight for this condition, we derive a weight (0.204) from the disability weight regression model. Other diseases of the digestive system We base the incidence estimates for appendicitis, intestinal obstruction, diverticulitis, gall bladder and bile duct disease, pancreatitis and vascular insufficiency of the intestine on the numbers of people with a relevant hospital procedure or diagnosis from Australian hospital data. With the exception of appendicitis, these conditions were not considered in either the GBD or Dutch studies. We adopt a 2-week duration for appendicitis, and a 3-week duration for gall bladder and bile duct disease, intestinal obstruction, vascular insufficiency and pancreatitis. For each of these conditions, we assume the GBD weight for appendicitis. For gall bladder and bile duct disease, we use cholecystectomies or bile duct incisions but ignore people admitted with un-operated cholelithiasis on the assumption that these people are largely asymptomatic. 2O Genitourinary diseases Nephritis & nephrosis We base the incidence of dialysis and transplant patients on the Australian dialysis and transplant data from which we derive durations for both categories of patients using DisMod. For dialysis patients, we use case-fatality rates to match observed deaths and remission through transplant, and apply the Dutch weight for diabetic nephropathy (0.290). In the first 6 months after transplant, we assume a health state equivalent to the Dutch weight for diabetic nephropathy (0.290). For the remaining period with the transplant, we use a weight of 0.11, which is equivalent to both the GBD weight for treated renal failure and the Dutch weight for uncertain prognosis. We derive untreated end stage renal failure from the difference between dialysis or transplant deaths and total renal deaths, to which we apply an average duration of 1 year prior to death at the GBD weight for untreated renal failure (0.104). We use Australian dialysis and transplant data on underlying renal disease distribution to attribute YLD from diabetic nephropathy to diabetes, analgesic nephropathy to the injury category of medical misadventure, and congenital dysplasia and polycystic kidney disease to congenital urogenital disease, and retain only those for primary renal disease in the nephritis & nephrosis category. Benign prostatic hypertrophy We base the incidence of benign prostatic hypertrophy on Australian hospital data. Based on expert advice we adjust the number of benign prostatic hypertrophy cases upwards to account for the proportion of cases that receive medical instead of surgical treatment. We also assume, based on expert opinion, that half of all benign prostatic hypertrophy cases receive surgical treatment, a proportion of whom experience complications or continuing symptoms following surgery (1% with lifelong incontinence at a derived weight of 0.204, 15% with lifelong impotence at the GBD weight of 0.195, and 5% with urethral stricture for 4 weeks at the GBD weight of 0.151). Of those opting for medical treatment, we assume 70% 170 use alpha-blocker drugs, of which half are cured. The other half may then try surgery. We assume none of those receiving drugs other than alpha-blockers are cured. We apply the GBD weight for symptomatic benign prostatic hypertrophy to each of these intervention pathways assuming the following durations: 1.5 years for surgery, 1 year for successful medical treatment, 2 years for unsuccessful medical treatment then surgery, and lifelong for unsuccessful medical treatment but no surgery. Urinary incontinence We derive incidence rates of incontinence from DisMod using prevalence figures reported in a review of Australian and international literature (AIHW: Lea 1993) and from Womens Health Australia. We assume that a number of diseases and injuries are associated with this condition, most of which are more prevalent at older ages, and that the underlying causes are multi-factorial and interrelated. Based on a multivariate analysis (Chiarelli et al. 1999), we assume that, while all disability from incontinence among younger men and younger and middle-aged women belongs under this category, half that experienced by middle-aged and older men and older women is already captured under other conditions either explicitly (for example, as a sequela for benign prostatic hypertrophy among men) or implicitly as part of the overall weightings for these conditions (for example, severe stroke). For unaccounted incontinence, we apply an average of the GBD weight for moderate incontinence and the derived weight for benign prostatic hypertrophy-related severe incontinence using severity distributions from the 1998 Australian disability survey. Infertility We estimate the prevalence of infertility from a 1988 population survey of infertility, surgical sterility and associated reproductive disability in Perth, Western Australia (Webb & Holman 1992). This survey indicates that of the 3.5% of couples with non-surgical infertility, 68% have an associated reproductive disability defined in terms of the couple being unable to achieve a desired level of reproductive function. From a review of patients at an Adelaide infertility clinic indicating that 83% of couples with reproductive disability seek assisted reproductive technologies, 30% of whom achieve a pregnancy within 2 years (Weiss et al. 1992), we derive a net prevalence of 1.02% and 0.73% for short-term reproductive disability and 0.67% and 0.48% for long-term reproductive disability in females and males respectively. The causes of infertility are derived from recent national data on assisted conception and reproduction (AIHW: Dean & Sullivan 2003; AIHW: Ford et al. 2003). For short-term cases, we assume incident cases equal prevalent cases divided by the duration, which we assume is 2 years. For long-term cases, we derive incidence and durations from DisMod assuming nonzero remission rates from ages 45 years or over to account for declining prevalence of reproductive disability reflecting adoptions and changes in reproductive goals. For women, we subtract from the total number of long-term incident cases the estimated incidence of infertility as a sequela to maternal sepsis, abortion and pelvic inflammatory disease, the disability of which is calculated under chlamydia and gonorrhoea. We determine the duration of long-term infertility by subtracting the age at onset estimated in DisMod from 45 years. GBD weights are used for both short- and long-term reproductive disability. 171 Other genitourinary diseases For this residual category, we assume the application of a simple YLD to YLL ratio of one across the age groups is sufficient to capture the morbidity from other genitourinary diseases in men. This method, however, does not capture the significant burden experienced by women, particularly at younger ages. We therefore calculate separate models for menstrual disorders and hysterectomies for menorrhagia, genital prolapse and endometriosis. We base our estimates for menstrual disorders on women who report they often have severe period pain or premenstrual tension in the last 12 months from Womens health Australia. For severe period pain we assume a duration of 1 day per month and a disability weight similar to that for caesarean section. For menstrual tension we assume a duration of 2 days each month and we use the disability weight for mild depression. We use DisMod to model the conditions, assuming no excess mortality and remission of 0.1 for ages less than 50 years. We model disability from hysterectomies associated with menorrhagia, genital prolapse and endometriosis in terms of disability from both the procedure and the resulting infertility. We derive the number of procedures from hospital data and we assume a 2-week duration at the derived weight for laparotomy of 0.349 (compare with estimates for caesarean section). Following the findings of a survey of surgical sterility in Perth (Webb & Holman 1992), we assume the majority of women who undergo a hysterectomy have completed their reproductive objectives, and that infertility leads to disability in 3.3% of cases with endometriosis. We apply the GBD weight for infertility. 2P Skin diseases Eczema, acne and psoriasis We model the incidence of severe eczema (that is, an episode in the past 12 months that disrupts sleep on average one or more nights per week) using self-reported prevalence data from a study of Melbourne school children (Robertson et al. 2004) and from the National Health Survey for adults (ABS 2001c). For other skin conditions we limit our estimates to severe acne and moderate and severe psoriasis. Prevalence figures for acne are based on a study of Australian school children and a study of adults in Central Victoria (Kilkenny et al. 1998; Marks et al. 1999). Prevalence figures for psoriasis were derived from the National Health Survey and from the central Victorian study (Marks et al. 1999; Plunkett et al. 1999). We derive incidence and duration estimates from DisMod assuming no excess mortality and a remission rate of 0.1 for eczema (Thestrup-Pedersen 2003), 0.27 for acne (assuming 70% spontaneous remission after 4 to 5 years) and 0.3 for psoriasis. For eczema we derive a disability weight (0.019) from the disability weight regression model which we adjust for 3 symptomatic episodes per year lasting 6 weeks in total. For acne we use the unadjusted disability weight for eczema from the disability weight regression model (0.056) and for psoriasis we apply the GBD weight for vitiligo. Other skin diseases We model the disability associated with chronic leg, skin and varicose ulcers, excluding decubitus and cellulitis which we assume are captured elsewhere. We use the weighted 172 incident cases of skin ulcers from Australian general practitioner data to estimate the incidence of other chronic skin ulcers. YLD for diabetic foot is included within the diabetes mellitus model. To avoid double-counting diabetic foot we adjust our incident estimates for skin ulcers using Western Australian aetiological data on the proportion of leg ulceration cases that had diabetes (Baker et al. 1992). In the absence of more specific information we use the same assumptions for duration (8.9 months) and disability (0.131) as the diabetic foot model. 2Q Musculoskeletal diseases Musculoskeletal diseases are highly prevalent in the population. The fair to good testretest reliability of self-reported musculoskeletal diseases and the consistent correlation with pain make health survey self-reports of some use to measure musculoskeletal conditions. Although the prevalence of most musculoskeletal diseases differs substantially depending on the measurement method, with self-report showing the highest prevalence, the pattern of prevalence in men and women is often similar. A higher prevalence of herniated disc of the back and gout is found in men, whereas for most other musculoskeletal diseases the prevalence is higher among women than among men (Picavet & Hazes 2003). Rheumatoid arthritis Given the small numbers in Australian studies on rheumatoid arthritis and problems with proper incidence and remission measurement, we base our incidence estimates for this condition on the international literature. For juvenile chronic arthritis, we use findings from a population study during 19841988 in Sweden (Gare & Fasth 1992). For adults, we use results from a 40-year follow-up study of a population-based cohort in Rochester, Minnesota, USA (Doran et al. 2002). We derived durations from DisMod assuming a relative risk of mortality of 1.6 at ages 15 years or over (Pincus et al. 1994), with no increased risk for children, and a remission rate of 0.04 (Prevoo et al. 1996) indicating that, while drug treatment may slow the disease process and remission is the ultimate endpoint of treatment, most therapeutic options have fallen short of achieving this (Sesin & Bingham 2005). Because progression through the three stages of rheumatoid arthritis described by the Dutch weights is relatively rapid, we do not model this condition as progressive. Rather we apply an average of the Dutch weights using severity distributions for American adults (Hakala et al. 1994) and those relating to Swedish children (Gare & Fasth 1992). Osteoarthritis While there are a few Australian population-based studies on self-reported osteoarthritis (Jones et al. 1995; March et al. 1998), we prefer to base our estimates for this condition on reported findings of radiographic osteoarthritis (grade 2 and above) by affected joint, age and sex from a large-scale study in Massachusetts, USA (Jones et al. 1995; March et al. 1998). We model hip and knee osteoarthritis only, given the high correlation between osteoarthritis of the hip, hand and fingers (Spector et al. 1997). We used DisMod to derive average durations, assuming a slightly increased risk of mortality (1.1) and the observed remission rate from joint replacement surgery. Because osteoarthritis is a relatively slow progressive disease, with few patients showing symptomatic progression over an 11-year period (Ahern 173 & Smith 1997), we apply an average of the relevant Dutch weights, assuming a severity distribution based on the Framingham study (Guccione et al. 1990). Back pain Back pain is a very common condition, with about 7090% of people suffering from it in some form at some point in their lives (Hicks et al. 2002). Back pain may be viewed as running either an acute or chronic course. Acute back pain is usually considered to have a short duration and tends to resolve within days to weeks. However, recurrence of acute episodes is common and there is some contention as to the difference between recurring acute back pain and long-term chronic back pain. A duration of back pain lasting at least 3 months commonly underlies the definition of chronic back pain (NINDS 2006), and is often likely to continue indefinitely (Von Korff & Saunders 1996). Our estimates for back pain are based on self-reported prevalence of recent episodes, and long-term back pain from the 2003 Australian disability survey and the 1995 National health survey. We model recent episodes of (acute) back pain and long-term (chronic) back pain separately. Prevalence of long-term back pain resulting in at least mild disability is obtained from the Australian disability survey. Of these, the cases that were due to recent episodes of back pain were not identified separately. We therefore estimated the proportion due to recent episodes by applying the percentage of recent cases of long-term back pain from the 1995 National Health Survey. We use the Dutch weight for low back pain (0.06) as the disability weight for recent episodes of back pain, which applies to an average health state involving some problems in walking about and in usual activities, as well as moderate pain or discomfort. We assume an average duration of 4 days for painful and limiting episodes of back pain. To model chronic back pain, we use the prevalence of long-term back pain (not identified as recent episodes as described above) from the 2003 Australian disability survey. We use DisMod to derive the incidence and duration of chronic back pain, assuming a remission rate of 10% and no increased risk of mortality. For many people, there are few treatment alternatives and complete relief is rare (Atkinson 2004). We assume that 14% of long-term cases experience constant or persistent pain (Quittan 2002), and 86% experience pain 1 day per week. We use the GBD disability weight for chronic intervertebral disc pain of 0.103. Slipped disc Our estimates for slipped disc are based on numbers of intervertebral disc procedures from Australian hospital data. We assume only 7.5% of incident cases of disc displacement receive surgery (Deyo et al. 1990), and derive total annual episodes from this proportion. We assume on average an episode of discomfort lasts 4 weeks. For those who receive surgery, we take the median time of 224 days from onset of symptoms to recovery reported in the literature (Rasmussen 1996). In the absence of weights for both these health states, we use the Dutch weight for low back pain (0.06). Based on a 5-year follow-up study (Kurth et al. 1996), we model 14% of operated cases as going on to experience long-term chronic pain with a lifelong duration at the GBD disability weight for chronic intervertebral disc of 0.103. Occupational overuse syndrome Occupational overuse syndrome (formerly known as repetition strain injury) is a contentious condition with considerable disagreement within the literature about its aetiology and 174 pathophysiology (Byrne 1992; Cohen et al. 1992; Helme et al. 1992). Our model uses selfreport prevalence data on repetition strain injury from the 2003 Australian disability survey from which we derive incidence figures using DisMod assuming an average duration of 3 years and no mortality. In the absence of Dutch or GBD weights for this condition, we use sex-specific derived weights to account for the fact that all males in the 1993 Australian disability survey had mild or no handicap, whereas 26% of females had moderate handicap and 17% had severe or profound handicap. Gout Our estimates for gout are based on self-reported prevalence from the National Health Survey which has the same overall result as found in a general practitioner study in the UK (Mikuls et al. 2005). We assume a slight increased risk of mortality associated with gout (relative risk=1.1) and no remission, based on information that at 1 year 62%, at 2 years 78% and at 10 years 93% has had at least one repeat attack (Alamo Family Foot and Ankle Care 2005). Fitting a Weibull function to these figures gives an average time to the next episode of 2.2 years, but this is rather high because of the skewness of the function. The median time to next episode is 0.44 years. We assume that 10% has chronic symptoms and the remaining 90% has an attack of 1 week every 0.44 years. Given that people may suffer gout at varying levels, from acute attacks of a short duration to chronic gout, we assume on average one attack per 2 months lasting 1 week in 90% of people and the remaining 10% suffer chronic ongoing disease at the GBD disability weight of 0.061. Other musculoskeletal disorders Because mortality for musculoskeletal conditions is low and because 49% of deaths from musculoskeletal disorders do not fall within the above categories, a derivation of disability for this rest category by applying a ratio of YLD to YLL for the explicitly modelled musculoskeletal conditions is not plausible. Therefore we try to model disability from all other conditions explicitly. In the absence of detailed information, we define an other category comprising both prevalent minor conditions and more serious diseases (for example joint derangement and disorders; osteopathies; chondropathies and other bone disorders; connective tissue diseases; and soft tissue problems such as rheumatism, ganglions, bunions, bursitis, cramps, tenosynovitis and tennis elbow). We base our estimates for these conditions on the prevalence of other musculoskeletal disorders that have not been accounted for in each of the musculoskeletal models described above from the 2001 National Health Survey. Based on figures from the 2003 Australian disability survey, we assume a proportion of prevalent cases report on refer to musculoskeletal sequelae of other diseases or injuries, which we account for by adjusting overall prevalence figures downwards by 50%. For recent non-chronic cases, we assume the same duration and weight as for recent episodes of back pain. For chronic cases, we derive incidence rates and durations from DisMod assuming no excess mortality and a remission rate of 0.1. We take the proportion reporting symptoms in the 2 weeks before interview as an approximation of the proportion of time spent symptomatic and assume symptomatic chronic cases experience a health state equivalent to the weight for low back pain. 175 2R Congenital anomalies Congenital heart disease We model the disability associated with four types of congenital heart disease for live-born infants: surgically treated atrial or ventricular septal defect, surgically treated Fallots tetralogy or transposition of great vessels, surgically treated pulmonary stenosis, and complex but not curatively operable congenital heart disease. We derive the incidence of the first three conditions from Australian hospital data by assuming that all curative procedures represent an incident case with disability. We assume a duration of 1 year before operation with disability equivalent to the Dutch weight for moderate heart failure (0.35) and postsurgery we use relevant Dutch weights and assume reduced life expectancy, except for those with septal defects (Miyamura et al. 1993; Nollert et al. 1997a, 1997b). We assume disability starts at birth and we discount YLD back to birth to account for this. We derive the incidence of other congenital heart malformations from Victorian birth defects data. Following expert advice, we assume that 50% of these cases are complex but not curatively operable. We assume that duration is half of those with surgically treatable conditions and use the relevant Dutch weight (0.72). Digestive system malformations We model the disability for anorectal and oesophageal atresia and other digestive system malformations. We estimate the incidence of digestive system atresia for cases surviving 28 days using Victorian birth defects data. We assume 26 weeks of disability from birth at the GBD weight for anorectal atresia (0.85). After this period, we assume that a proportion of both types of atresia cases have lifelong problems (15% and 20% respectively) and decreased life expectancy (by 10 and 5 years respectively) and disability equivalent to health state 111211 for two-thirds of the time (0.037) (Ludman & Spitz 2003). We estimate the incidence of other digestive system malformations using data from the Australian congenital malformations dataset (AIHW: Hurst et al. 2001). We assume no long-term disability, and a 1-month period of disability from birth equivalent to the GBD weight for anorectal atresia. Renal agenesis We estimate the incidence of unilateral and bilateral renal agenesis for cases surviving 28 days using Victorian birth defects data. For unilateral cases we assume that 20% of survivors have ongoing problems, with a life expectancy of 70 years and a disability of 0.067. For bilateral cases we assume an average duration of 3.5 days and use the GBD weight for renal agenesis (0.85). We also calculate YLD for renal failure due to renal dysplasia based on attributions from Australian dialysis and transplant data. Other urogenital tract malformations We model the disability associated with the following urogenital tract malformations: cystic kidney disease, obstructive defects of renal pelvis and ureter, and other urinary tract malformations. We estimate the incidence of cases of other urogenital tract malformations surviving beyond 28 days from the Victorian, Western Australian and Australian birth 176 defects data. We assume 30% of cases have chronic lifelong problems, with a life expectancy of 50 years and a disability weight of 0.067. YLD were also calculated for end-stage renal failure due to cystic kidney disease. Other congenital anomalies We estimate the incidence of anencephaly using Australian mortality data for newborns, assuming deaths are equivalent to incident cases. We assume a duration of 1 week with a disability weight of 1. For spina bifida, we estimate the average annual number of live births that survive the first 28 days from Victorian birth defect data (Riley & Halliday 2004). We derive an average disability weight (0.52) based on the Dutch weights for each level of severity combined with severity distributions from expert advice. We estimate the incidence of surgically treated cleft lip and cleft palate from Australian hospital data, assuming that all curative procedures represent a case and that all cases are treated within the first year. We assume disability equivalent to the treated GBD weights (0.016, 0.015 respectively). YLD estimates for Down syndrome and other chromosomal anomalies are calculated as described in the section on intellectual disability (see Section 2K). We estimate the incidence of abdominal wall defects (exomphalos and gastroschisis) in infants surviving >28 days using 2001 Australian birth defects data (AIHW NPSU 2004) and survival data from the Victorian birth defects data. We assume a duration of 4 weeks based on Australian and international literature (Dimitriou et al. 2000; Sharp et al. 2000) and apply the GBD weight for abdominal wall defect. Based on expert advice we assume that 20% of cases have lifelong problems, a shortened life expectancy by 20 years, and disability weight of 0.200 (the Dutch weight for young adult in permanent stage after surgical repair to Fallots tetralogy). 2S Oral conditions Caries The incidence of caries is measured by one or more new dental cavities (caries increment). The occurrence of dental caries in an individual is measured using the DMFT or DMFS index: the number of decayed (D), missing (M) and filled (F) primary or permanent teeth (T) or surfaces (S). A review of the relationship between DMFT and DMFS suggests that DMFS data should be adjusted by a factor of 1/3.5 to be consistent with DMFT data (Carvalho et al. 2004; Hopcraft & Morgan 2005; Rosen et al. 2004). For children and adults we estimate the incidence from representative Australian caries prevalence data: the 2000 Child Dental Health Survey (AIHW: Armfield et al. 2004) and the 198788 National Oral Health Survey of Australia (Barnard 1993). Fitting linear regression lines to the prevalence data gives slopes in children (114 years) of 0.25 (AIHW: Armfield et al. 2004; Davies et al. 1997) and in adults (1559 years) of 0.27 (Barnard 1993). For older adults (60 years or over) and nursing home residents (60 years or over) we estimate the incidence of caries from the South Australia Dental Longitudinal Study (AIHW DSRU 2002) and the 1998 Adelaide Dental Study of Nursing Homes (AIHW: Chalmers et al. 2001), respectively. Based on the 5-year increment of all new carious surfaces, the 1-year increment (assuming that the incidence of carious surfaces over the 5-year period was evenly 177 distributed) is 0.98 (AIHW DSRU 2002). The 1-year increment of new carious surfaces in nursing home residents is 3.5 (AIHW: Chalmers et al. 2001). We use our DMFT/DMFS adjustment factor to give annual caries increments of 0.28 and 1.0 respectively for older adults in the general population and nursing homes. The previous Australian burden study assumed a symptomatic duration of 10 weeks based on advice from the Australian Research Centre for Population Oral Health. More recent work, based on patient self-report, by this group suggests durations in the order of 81 weeks (Brennan & Spencer 2004, 2005). However, both of these estimates refer to time spent with and without symptoms. A review of the literature shows that there is a paucity of information on symptomatic caries, specifically mean duration of symptoms and proportion of people who are symptomatic. A patient-based study in children in the UK reported that 78% of the children sampled presented within 1 month of pain onset (Mason et al. 1997) whereas a patient-based study in New Zealand observed that 67% of adults presented within 1 month of pain onset (Whyman et al. 1996). Patient-based samples are biased as they do not reflect all cases of caries in the community. Neither of these studies provided data on the mean durations for those people experiencing symptoms for greater than 1 month. We estimate the average time symptomatic for those people presenting with caries problems by fitting a lognormal distribution to the midpoint of the observed durations. This gives mean durations of symptomatic caries of 28 days in children and 55 days in adults. We base our estimate of people with symptomatic caries (32.4%) on the findings of the 1998 Australian Longitudinal Study of Dentists Practice Activity (Brennan & Spencer 2002). Following the first Australian burden of disease study the Australian Research Centre for Population Oral Health developed disability weights for oral disease using a patient-based sample in South Australia (Brennan & Spencer 2004, 2005). Disability weights for caries (0.044), periodontal disease (0.023) and denture problems (0.026) in this study were higher than comparable Dutch weights used in the previous Australian burden study (0.005 for caries involving a filling and 0.014 for caries involving an extraction, 0.007 for periodontal disease, and 0.004 for edentulism). We did not use these Australian-derived disability weights because patient-based samples are likely to under-represent asymptomatic people, and questions with limited response categories are likely to bias results. For instance, the duration-related question was During the period that you have had this dental problem, what percentage of the time (0% = none of the time, 50% = half of the time, 100% = all of the time) have you experienced the limitations listed above in relation to: mobility, self-care, usual activities, pain/discomfort, anxiety/depression, cognition? (Brennan & Spencer 2004, 2005). Both of these limitations are likely to over-estimate the percentage of people reporting problems for each of the health dimensions as well as the duration of their symptoms. We follow expert advice and derive a disability weight for symptomatic caries (0.057) using the disability weight regression model (health states: 20%111211 and 80%111111). Edentulism We estimate the prevalence of edentulism (loss of all natural teeth) for the general population and nursing home residents using the 2002 National Dental Telephone Interview Survey (AIHW: Carter & Stewart 2003) and the 1998 Adelaide Dental Study of Nursing Homes (AIHW: Chalmers et al. 2001), respectively. We derive incidence and duration using DisMod, based on these studies, assuming no remission and no excess case-fatality. We model a 2% declining time trend to reflect the observed decline of the prevalence of 178 edentulism from 20.5% in 1979 to 8.0% in 2002 (Sanders et al. 2004). We use the same disability weight as in the previous Australian burden study. Periodontal disease We estimate the prevalence of periodontal pockets larger than 6 mm using data from the 198788 National Oral Health Survey of Australia (Barnard 1993). Expert advice suggested that periodontal disease is a largely asymptomatic risk factor for tooth loss; pain occurs in around 1% of time when an abscess forms in a periodontal pocket; and the typical duration of periodontal disease is around 15 years. We derive incidence and duration using DisMod based on the Australian prevalence data, remission rates that reflect 15 years average duration and no case-fatality. A new disability weight for periodontal abscess was estimated (0.056) based on the disability weight regression model (health state 111211). Pulpitis We estimate the incidence of pulpitis using the proportion of patients sampled in the 1998 Longitudinal Study of Dentists Practice Activity (AIHW: Spencer & Brennan 2002) who had a main diagnosis of pulpal infection and the total number of dental consultations in Australia in 2003. We assume that most people with pulpal infection will visit a dentist. We estimate the total number of dental consultations by multiplying the proportion of people who visited a dentist in the last 12 months by the mean number of dental visits per person (from the 2002 National Dental Telephone Information Survey (AIHW: Carter & Stewart 2003)) and 2003 population data (excluding the edentulous population). Expert consultation suggests that a symptomatic duration of 1 month is plausible for pulpitis with the first few weeks consisting of intermittent pain and the last week being of more severe and consistent pain. We assume that 71.3% of people with pulpitis presenting in pain, a figure which we derive from the 1998 Longitudinal Study of Dentists Practice Activity. We use the disability weight regression model to estimate a disability weight for pulpitis assuming that 1 week is spent in moderate pain and 3 weeks are spent at a level of the disability for moderate pain for 10% of the time. 2Z Chronic fatigue syndrome We base our model for chronic fatigue syndrome on the internationally accepted US Centers for Disease Control and Prevention criteria, which state that for a patient to receive a diagnosis of chronic fatigue syndrome, they must have severe chronic fatigue of 6 months or longer duration with other known medical conditions excluded by clinical diagnosis, and concurrently have four or more of the following symptoms: substantial impairment in shortterm memory or concentration; sore throat; tender lymph nodes; muscle pain; multi-joint pain without swelling or redness; headaches of a new type, pattern or severity; unrefreshing sleep; and post-exertional malaise lasting more than 24 hours. The symptoms must have persisted or recurred during six or more consecutive months of illness and must not have predated the fatigue (Fukuda et al. 1994). Following expert consultation we conceptualise two manifestations of chronic fatigue syndrome: (a) post-infective fatigue syndrome which constitutes between 3040% of cases and is characterised as an acute outcome of viral and non-viral infections, has a disability starting point of moderate severity, a median duration of 12 months, and around 99% 179 recovery at 2 years (Hickie et al. submitted 2005; Wilson et al. 2001); and (b) protracted chronic fatigue syndrome, which constitutes the remaining 6070% of chronic fatigue syndrome cases, where cases have an insidious onset with initially severe disability followed by cases fluctuating around 5080% of their previous healthy state, and a median duration of around 7 years. We assume that the disability associated with post-infective fatigue syndrome is included within the disability weights and durations in the relevant infectious disease models (explicitly in the arbovirus estimates but not for other viral infections such as Q fever and Epstein-Barr virus which are subsumed in the rest of infectious disease category). We base our estimates of prevalence for protracted chronic fatigue syndrome on the population-based study of chronic fatigue syndrome conducted in Wichita, Kansas, USA in 1997 (Reyes et al. 2003). In the previous Australian burden study we used prevalence estimates based on an Australian prevalence study of chronic fatigue syndrome (Lloyd et al. 1990). This studys applicability in the current context is limited due to the different diagnostic criteria used and the physician referral sample. The population-based study by Reyes and colleagues (2003) showed that only 16% of people identified with chronic fatigue syndrome had previously been diagnosed as such by a medical practitioner. Although it is not clear how similar the epidemiology of chronic fatigue syndrome is between the US and Australia, the findings from an international multi-centre study of the prevalence of chronic fatigue syndrome in patients lend support to the notion that the epidemiology of chronic fatigue syndrome is similar in the two countries (Wilson et al. 2001). We model incidence and duration using DisMod, assuming no excess mortality and remission rates which gave an average duration of 7.3 years (Reyes et al. 2003). We assume that 90% of the time people with chronic fatigue syndrome are symptomatic, using findings from the 1993 Australian disability survey. In the absence of an established disability weight for chronic fatigue syndrome we use the disability weight estimated for the previous Australian burden study. 3 Injuries We model the disability from non-fatal injuries where a person has an injury severe enough to warrant emergency department or inpatient hospital treatment but that does not lead to death. This method assumes that injuries treated outside the hospital system do not result in significant disability. We derive non-fatal incident injuries from Australian hospital data. We classify incident cases according to a matrix of 14 external cause of injury categories (12 unintentional and two intentional) and 32 nature of injury categories (for example fractures, burns, wounds, brain injury, spinal cord injury). We exclude admissions for the same ICD-10 code within 90 days, on the assumption that these are re-admissions, as well as, those resulting in death. Given that it is not uncommon for multiple sites of the body to be damaged from a single accident, we estimate disability for only the most disabling ICD-10 code associated with each incident, on the assumption that the disability for the other ICD-10 codes is captured in the weight for the more severe injury. We redistribute ill defined injuries and adjust estimates for amputated finger as in the previous Australian burden of disease study. We use disability weights, durations and the risk of mortality as per the GBD study. 180 Appendix 2: Methods for attributing risk In this section we describe our methods for assessing the contribution of 14 health risks to the total burden of disease and injury in Australia. For most risks, our analyses are based on methods developed by the WHO CRA project and described in detail elsewhere (Ezzati et al. 2004a). Briefly, the main inputs are the prevalence of exposure to a health risk in a population and information on the risk of disease, injury or death (referred to here as relative risk or hazard) from this exposure, which is typically derive from systematic reviews of the international literature. Our analyses are not comprehensive since choices had to be made about which risks to include on the basis of certain criteria, as outlined at the beginning of Chapter 4. We begin by describing the methodological basis of our analyses, the population attributable fraction. Estimating population attributable fractions The population attributable fraction (PAF) is a subtype of a more general measurethe potential impact fraction (PIF). The PIF measures the proportional reduction in disease or injury burden experienced by a population that would occur if the population were subjected to an alternative or counterfactual distribution of exposure to a particular health risk. If the alternative exposure scenario is set to a level such that it represents the lowest possible risk in a population (no exposure, for example), the PIF represents the total amount of burden that is attributable to that risk; in this instance it is called the population attributable fraction (Eide & Heuch 2001; Miettinen 1974). For health risks that are measured on a continuous scale, the PIF can be defined thus: 0 0 0 ( ) ( ) ( ) ( ) ( ) ( ) m m x x m x RR x P x dx RR x P x dx PIF RR x P x dx = = = − ′ = Where RR(x) = relative risk at exposure level, P(x) = population distribution of exposure, P(x) = counterfactual distribution of exposure, and m = maximum exposure level (Equation 1) When a risk is measured on a categorical scale, the discrete version of the PIF formula is (Eide & Heuch 2001; Walter 1980): − = c c c c c c c c c P RR P RR P RR PIF * Where c = an index for category, P = prevalence, and P* = prevalence after a change, and RR = relative risk (Equation 2) 181 The difference between Equation 1 and Equation 2 in practical terms is that the latter can easily be resolved in a spreadsheet environment, whereas the former requires more advanced mathematical techniques. Equation 2 is mathematically the same as the PAF formula for risk factors with multiple categories given by English and colleagues (Equation 3), if the counterfactual is set as the hypothetical minimum distribution (English et al. 1995). ( 1) 1 ( 1) − + − = c c c c c c P RR P RR PAF (Equation 3) Choice of theoretical minimum Calculating a PAF requires the explicit characterisation of an exposure distribution that represents the lowest possible level of risk in a population. This has been termed the theoretical minimum exposure distribution and corresponds to zero exposure for some risks (for example smoking). For other risks, however, zero exposure is inappropriate because it is physiologically impossible (for example systolic blood pressure, BMI and cholesterol). In this case the lowest levels observed in specific populations and epidemiological studies described in the literature are used instead. For example, a theoretical minima of 115 mmHg for systolic blood pressure and 3.8 mmol/L for total cholesterol (each with a small standard deviation) are the lowest levels at which the doseresponse relationships have been characterised (Chen et al. 1991; Eastern Stroke and Coronary Heart Disease Collaborative Research Group 1998; Law et al. 1994). For factors with protective effects (fruit and vegetable consumption and physical activity), the theoretical minimum exposure distribution is based on information from high exposure populations about the level to which the benefits continue to accrue given current scientific evidence. Estimating attributable burden Age- and sex-specific PAFs are calculated for each health risk and heath outcome pair using the relationships in Equations 1 and 2. Where a relative risk of disease or injury is different to the relative risk for death, two PAFs are calculated, one for non-fatal burden and the other for fatal burden. PAFs are then multiplied with the relevant burden estimates for that health outcome and the sum of the burden across all outcomes affected by a health risk constitutes the total attributable burden for that risk. For example, if there are 1,000 deaths from ischaemic heart disease and 500 from stroke in a particular age and sex category, and the PAFs for cholesterol leading to ischaemic heart disease and stroke are 0.5 and 0.3 respectively, the mortality attributable to high cholesterol equals 1,000 Χ 0.5 + 500 Χ 0.3 = 650. In other words, if the population had been exposed to the hypothetical minimum cholesterol distribution instead of the current distribution, 650 fewer deaths would have occurred. Table A2.3 summarises the exposure levels, theoretical minima, health outcomes and sources of relative risks for each of the 14 health risks analysed in this report. Table A2.4 summarises 182 our estimates of exposure in the Australian population to each of these risks. A brief description the specific methods we used for each risk is provided below. Tobacco Given the long lag time between exposure to tobacco smoke and the occurrence of cancers and COPD, the attributable burden cannot be estimated from the current prevalence of smoking. Even with good historical information on smoking prevalence, it is not straightforward to determine the current amount of illness that is due to smoking because the lag time between the relevant exposure and disease is variable. Therefore, we used the method of Peto and colleagues, who proposed an artificial compound prevalence measure of the relevant past exposure to tobacco (Peto et al. 1992). This smoking impact ratio is derived from a comparison of lung cancer mortality rates in the population of interest and lung cancer mortality rates among non-smokers and smokers observed in a large long-term follow-up study in the United States. We used this smoking impact ratio instead of the current prevalence in the standard calculation of attributable fractions for the other cancers and COPD. Compared with cancers and COPD, the mean time between exposure to tobacco and all other adverse health outcomes is considerably shorter. We therefore used prevalence estimates of smoking for adults aged 18 years and over in 2001, two years before our baseline year of 2003 (ABS 2001c). Our previous calculations of attributable mortality burden included only those diseases for which English and colleagues report strong evidence of an association (English et al. 1995). For this report, we added other conditions for which reasonable evidence of an association with tobacco exists (AIHW: Ridolfo & Stevenson 2001): cancer of the stomach, endometrial cancer, peripheral vascular disease, pneumonia, inflammatory bowel disease, injuries from fires and Parkinsons disease. Tobacco has a small protective effect against Parkinsons disease and endometrial cancer. We omitted peptic ulcer disease, given evidence of its largely infectious aetiology. We also added the burden attributable to smoking from macular degeneration (Mitchell et al. 1999; Tomany et al. 2004). In addition, we calculated the burden from passive smoking using attributable fractions for lung cancer, ischaemic heart disease, and asthma in children (NHMRC 1997a). For lower respiratory tract infection, sudden infant death syndrome and otitis media in children due to passive smoking, we used the prevalence of maternal smoking (Turrell et al. 2002) and relative risks from the US Surgeon Generals Report (US Department of Health and Human Services 2006) and NHMRC (NHMRC 1997a). We also estimated the burden of low birth weight due to smoking during pregnancy using the relative risk from Ridolfo and Stevenson (2001) and estimates of smoking during pregnancy from Laws and Sullivan (2005). High blood pressure We used the AusDiab study (Dunstan et al. 2002) to estimate distributions of high blood pressure by age and sex in the Australian population. Despite a low response rate AusDiab is the only recent and representative study that has measured this risk in Australia. Relative risks came from Lawes and colleagues (2004a). We used the CRA theoretical minimum distribution for blood pressure (mean 115, SD 6 mmHg) as the counterfactual in this analysis. 183 High body mass We used the AusDiab study (Dunstan et al. 2002) to estimate distributions of body mass index (BMI) by age and sex in the Australian population. Relative risk of type 2 diabetes came from the Asia Pacific Cohort Collabortation (2006); the relative risk of all remaining conditions associated with high body mass came from James and colleagues (2004). We used the CRA theoretical minimum distribution for BMI (mean 21, SD 1 kg/m2) as the counterfactual in this analysis. Physical inactivity Recent developments have led to the treatment of physical inactivity as a four-level categorical variable by subdividing the exposure group labelled as sufficiently active in the CRA project into those meeting current recommendations and highly active. While physical activity levels equivalent to 2.5 hours per week of moderate-intensity activity (approximately 4000kJ/week) are considered an important target for population health benefits, the protective effects are expected to continue to higher levels. Therefore, the theoretical minimum exposure distribution was chosen to be the whole population in the high active category to increase consistency with the counterfactual exposure distribution of other risk factors (Bull 2003; Murray et al. 2003; Powles & Day 2002) (Table A2.1). The required prevalence data were derived from the NHS 2001 (ABS 2001c). The exercise related questions in this survey relate to physical exercise undertaken for recreation, sport, health or fitness purposes, conceptually excluding physical activity undertaken as a part of work or for other purposes. This may underestimate the amount of physical activity undertaken, and therefore our analyses may overestimate the burden of disease attributable to physical inactivity. The associated hazards were modified to correspond to the new referent category of highly active. Given no available quantitative meta-analysis with comparable categories, risk estimates were derived from a synthesis of recent reviews (Kelley & Goodpaster 2001; Kesaniemi et al. 2001; Kohl 2001; Oguma et al. 2002; Thune & Furberg 2001; Williams 2001) and findings from several recent studies in which the results were reported separately by intensity of activity as well as total volume of activity (Manson et al. 2002; Sesso et al. 2000). The relative risk of ischaemic heart disease for the inactive group compared to high active was set at 2.0, based on reviews of studies with both physical activity and fitness measures as well as a recent studys differential results for moderate versus vigorous activity. The likely linear dose-response relationship (Kesaniemi et al. 2001) was represented by the arithmetic midpoints for those classified as meeting current recommendations and insufficiently active. For stroke, the mean of nine studies summarised in the systematic review and metaanalysis by Blair and colleagues (2001) was used (relative risk of 2.0). The findings from the review by Thune and Furberg (2001) were used to derive the risk estimate for colon and breast cancer. There has been no quantitative review of diabetes and physical activity; therefore the relative risks from the CRA project were adjusted by the same magnitude as for ischaemic heart disease. It is recognised that these estimates of risk are derived from a synthesis of the available scientific evidence and alternative interpretations are possible. 184 Table A2.1 Physical activity exposure categories Physical activity level Definition High 3 sessions x at least average 40 minutes vigorous AND total of at least 1500 METmins/week(a) Recommended 3 sessions x at least average 20 minutes vigorous OR 5 x 30 minutes moderate OR 600 METmins/week Insufficient Some activity but not meeting recommendation Inactive No activity (a) The standard metabolic equivalent, or MET, level. This unit is used to estimate the amount of oxygen used by the body during physical activity. One MET = the energy (oxygen) used by the body sitting quietly, perhaps while talking on the phone or reading a book. The harder the body works during the activity, the higher the MET. High blood cholesterol We used the AusDiab study (Dunstan et al. 2002) to estimate distributions of high blood cholesterol by age and sex in the Australian population. Relative risks came from Lawes and colleagues (2004b). We used the CRA theoretical minimum distribution for serum cholesterol (mean 3.8, SD 0.5 mmol/L) as the counterfactual in this analysis. Alcohol There are a number of recent data sources on the prevalence of alcohol consumption in the Australian population, including the 2004 National Drug Strategy Household Survey (NDSHS 2004) (AIHW & DoHA 2005) and the 2001 National Health Survey (NHS 2001) (ABS 2001c). The NHS 2001 focuses on the quantity of alcohol consumption on the three most recent days on which alcohol was consumed in the week prior to interview, while the NDSHS 2004 explicitly quantifies the amount of alcohol drunk on the day prior to interview. Of these, only the NHS 2001 collected information on the type and brand of alcoholic drinks consumed as well as the number. Also, the NHS 2001 gives average daily alcohol consumption over the previous week in millilitres. For this reason, we used the NHS 2001 to estimate the prevalence of alcohol consumption for adults aged 18 years or over. We categorised the prevalence of alcohol consumption into the four levels used in English and colleagues analysis of the risks of alcohol consumption (English et al. 1995), and with the NHMRCs recommendations on alcohol consumption (NHMRC 1992) (Table A2.2). The prevalence of each level of alcohol intake was estimated by age and sex from the average weekly consumption of alcohol after conversion to standard drinks per day. Data for people interviewed on each day of the week were reweighted to obtain prevalence of alcohol consumption based on equal samples for each day of the week. Those that last drank alcohol more than 1 week ago were classified as abstainers. 185 Table A2.2 Classification and prevalence of alcohol intake levels used in this report Average number of standard drinks (= 10 g alcohol) per day Alcohol intake Male Female Abstinence 00.25 00.25 Low 0.264.00 0.262.00 Hazardous 4.016.00 2.014.00 Harmful >6 >4 Source: English et al. 1995 We used relative risks and population attributable fractions from Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001) for conditions for which there is evidence of causation by alcohol consumption. English and colleagues (1995) estimated that 44% of fire injuries are attributable to alcohol; this was not updated by Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001). We revised these estimates with the addition of more recent studies, and produced a separate PAF for fire injuries and scalds or other burns for both YLD and YLL. We also updated the drowning PAF of 0.34 from Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001) with age-specific estimates from Driscoll and colleagues (AIHW: Driscoll et al. 2004) who found that 17% of unintentional drownings were attributed to alcohol (blood alcohol content of at least 0.10 g/100 ml). English and colleagues (1995) derived a PAF of 0.07 for alcohol, and occupational and machine injuries. We applied this to all machinery accidents, and to the occupational YLD PAFs for injury codes not already covered elsewhere in alcohol. For YLL we applied a PAF of 0.051 (Driscoll et al. 2001) to the occupational YLL PAFs for injury codes not already covered elsewhere in alcohol. Low fruit and vegetable consumption We used the National Health Survey (ABS 2001c) to estimate distributions of fruit and vegetable consumption by age and sex in the Australian population. Relative risks came from Lock and colleagues (2004). We used the CRA theoretical minimum risk distribution for fruit and vegetable consumption (mean 600, SD 50g/day) as the counterfactual in this analysis. Illicit drugs In addition to being a direct cause of death, illicit drugs are also risk factors for conditions such as HIV/AIDS, hepatitis, low birth weight, inflammatory heart disease, poisoning, and suicide & self-inflicted injuries. By definition, heroin, benzodiazepine, cannabis and other drug dependence and harmful use are due to illicit drug use; therefore the entire burden due to these conditions was attributed to this risk factor category. For infective endocarditis and suicide we used the attributable fractions for illicit drugs developed by English and colleagues (1995). The proportion of inflammatory heart disease that was due to infective endocarditis was derived from hospital data. The infective endocarditis PAF was then applied to this proportion only. The proportion of HIV due to injecting drug use was based on diagnosed HIV from the Australian HIV Public Access Dataset (National Centre in HIV Epidemiology and Clinical Research 2005b).We use diagnosed rather than newly acquired HIV, which is in keeping 186 with YLD estimates, and due to the apparent stabilisation of HIV incidence over recent years. AIDS cases and deaths attributable to injecting drug use were from the Australian AIDS Public Access Dataset (National Centre in HIV Epidemiology and Clinical Research 2005a). Time to death (year of death minus year of AIDS diagnosis) was added to the midpoint of the age at diagnosis range to approximate age range at death. For those age-at-death ranges available in the AIDS Public Access Dataset, we used the age-specific proportion attributable to injecting drug use. For all other ages we applied the all-age proportion. For cases of HIV and AIDS, and AIDS deaths, we assumed that all cases with exposure category male homosexual contact and injecting drug use were attributable to male homosexual contact. The proportion of newly acquired hepatitis B and C cases due to injecting drug use was from the HIV/AIDS, viral hepatitis and sexually transmissible infections in Australia Annual Surveillance Report 2004 (National Centre in HIV Epidemiology and Clinical Research 2004). The proportion of road traffic accidents due to illicit drug use was derived from Drummer and colleagues (2004). We applied the methodology used by Ridolfo and Stevenson on earlier data from Drummer (1994). For low birth weight we used prevalence of cannabis and opioid diagnosis during pregnancy in New South Wales and relative risks from Burns and colleagues (2006). The relative risk for antepartum haemorrhage attributable to illicit drug use was from English and colleagues (1995), with the prevalence being heroin or cocaine use in past 12 months for females aged 1549 years from the NDSHS 2004. The odds ratio for schizophrenia attributable to cannabis use was from Semple and colleagues (2005). This odds ratio is the result of a meta-analysis of seven studies with different classifications of psychosis and cannabis use. Despite these differences, there was consistency in the unadjusted odds ratios. We used prevalence of daily cannabis use over the last 12 months from the 2004 NDSHS (AIHW & DoHA 2005) to calculate the PAF to be applied to schizophrenia. Occupational exposures and hazards The attributable burden of occupational exposures and hazards was based on the following methods. Work-related fatal injuries were derived from the National Workers Compensation Statistics Database accessed online via National Occupational Health and Safety Commission (NOHSC) Online Statistics Interactive (NOSI) <www.nosi2.nohsc.gov.au/>. Since compensation statistics do not cover all occurrences of occupational injury deaths, we inflated these figures according to a study carried out by Driscoll and colleagues (AIHW: Driscoll et al. 2004), that investigated the coverage of workrelated traumatic deaths by official occupational health and safety and compensation agencies in Australia. Work-related deaths by mechanism, nature and industry from the NOSI database were inflated according to the proportion of all work-related deaths in 1989 92 covered by compensation agencies by industry (Table 3 in AIHW: Driscoll et al. 2004). In the absence of more reliable information, the attributable fractions for non-fatal injuries were derived from an analysis of the National Hospital Morbidity Database 200203. For each agesexinjury group, the attributable fraction for occupational injuries was estimated as the ratio of hospital episodes where workplace was specified as the place where the injury occurred to the total hospital episodes where a place of occurrence was specified. Where possible we derived non-injury attributable fractions by following the CRA methods (Concha-Barrientos et al. 2004). This produced age- and sex-specific attributable fractions for 187 lung cancer, leukaemia, COPD, asthma, adult onset hearing loss, and chronic back pain (which we also applied to slipped disc). For each of the remaining cancer categories, we derived attributable fractions from a study carried out for the National Institute of Occupational Health and Safety (Kerr et al. 1996). This study also provided attributable fractions for a number of other chronic diseases, including neurological disorders, cardiovascular diseases, chronic respiratory diseases and renal disease. Attributable fractions for osteoarthritis were derived separately, based on relative risks of self-reported arthritis for blue collar workers compared to managers, administrators and professionals (AIHW: Turrell et al. 2006). Child sexual abuse and intimate partner violence Girls that experience child sexual abuse are more likely to experience intimate partner violence than non-abused girls (Mouzos & Makkai 2004). Women that experience multiple types of abuse, including child sexual abuse and intimate partner violence, have a higher risk of depression than those subject to only one form of abuse (Arias 2004; Messman-Moore et al. 2000; Nicolaidis et al. 2004). The 2001 Victorian Burden of Disease Study produced estimates of the burden attributable to intimate partner violence but did not calculate the burden attributable to child sexual abuse (DHS 2005; Vos et al. in press). Conversely the CRA project (Andrews et al. 2004) produced estimates of the burden attributable to child sexual abuse but not intimate partner violence. In this study we estimated the burden attributable to child sexual abuse and intimate partner violence. Further, to avoid over-estimating the burden when both of these risk factors are present we estimated an adjusted relative risk to account for the combined exposure state of having experienced both child sexual abuse and intimate partner violence. We estimated the prevalence of intimate partner violence without child sexual abuse and child sexual abuse and intimate partner violence combined from the Womens Safety Survey (ABS 1996). We used two categories of exposure to intimate partner violence, namely physical or sexual violence by a partner in the last 12 months, and physical or sexual violence by a partner more than 12 months ago. Given that the Womens Safety Survey asks only one question regarding child sexual abuse (Whether experienced sexual abuse when a child) we used the CRA project priors for Australia for the prevalence of child sexual abuse (based upon epidemiological studies) and assumed no trend in prevalence of child sexual abuse. We subtracted the prevalence of child sexual abuse and intimate partner violence combined from the child sexual abuse priors to estimate the prevalence of child sexual abuse without intimate partner violence. Messman-Moore and colleagues (2000) looked at the mean psychological functioning indices for women who had experienced (a) both contact child sexual abuse and adult victimisation (revictimisation); (b) adult victimisation only (multiple or once only); (c) contact child sexual abuse only; and (d) no abuse history. From these group means and standard errors we calculated an effect size using Hedges adjusted g for standardised mean difference (Egger etal. 2001). We then converted the effect sizes into odds ratios for risk of depression, anxiety and post-traumatic stress disorder by exposure group using the methods described by Hasselblad and Hedges (1995). These odds ratios, along with relative risks for contact child sexual abuse from the CRA project (Andrews et al. 2004), and relative risks for intimate partner violence from the Womens health Australia study (see DHS 2005: page 29) were then used to derive relative 188 risks for contact child sexual abuse only, intimate partner violence only, and child sexual abuse and intimate partner violence combined. Since Messman-Moore and colleagues (2000) define child sexual abuse as contact only, for non-contact child sexual abuse we used the CRA project relative risks and prevalence for that category unadjusted. In our main results we combined anxiety and depression together into one category. We therefore found the mean relative risk from the derived relative risks for depression, anxiety, and post-traumatic stress disorder symptoms. We applied the same relativities from the anxiety and depression relative risks for child sexual abuse only, intimate partner violence only, and combined child sexual abuse and intimate partner violence, to the intimate partner violence and child sexual abuse relative risks for other conditions (alcohol use disorders, other drug use disorders, and self-inflicted injuries). For ease of reporting, the population attributable fraction calculated for the combined child sexual abuse and intimate partner violence category was proportionately redistributed to either child sexual abuse or intimate partner violence. To calculate the population attributable fractions for those disease categories that only apply to intimate partner violence (smoking, cervical cancer, sexually transmitted diseases, eating disorders and physical injuries) we used the relative risks for intimate partner violence from the Womens health Australia study, and the prevalence of intimate partner violence (including those women who may have also experience child sexual abuse) from the Womens Safety Survey. The proportion of homicide due to intimate partner violence (52%) was from the 20032004 National Homicide Monitoring Program Annual Report (Mouzos 2005). Violence YLD was based on the proportion of hospitalisations for assaults where the relationship of the victim of assault to the perpetrator was recorded as spouse or domestic partner (including exspouse and ex-partner). The assaults where this relationship was unspecified were proportionately redistributed. Due to a lack of data on the prevalence of intimate partner violence among males, and on the related health outcomes, for males we only estimated the burden due to child sexual abuse. Analyses were based on methods developed for the CRA project described elsewhere (Andrews et al. 2004). We used the CRA priors for Australia for the prevalence of male child sexual abuse (based upon epidemiological studies). Urban air pollution Numerous studies have documented that urban air pollution has a range of effects on health, from irritated eyes to death. The effects of short-term exposure are generally demonstrated through time-series studies on daily events (for example mortality, hospitalisations, emergency department attendance) (Cohen et al. 2004; Simpson et al. 2005a, 2005b). The effects of long-term exposure have been demonstrated in large cohort and cross-sectional studies, mainly in the US and Europe (Cohen et al. 2004; Pope et al. 2002). We estimated the burden due to both long- and short term exposure to urban air pollution, and present the results for long-term exposure only as a minimum estimate, and the combination of longand short-term exposure as a more inclusive but less certain higher estimate. Long-term exposure For chronic exposure to urban air pollution, our analyses are based on methods developed for the CRA project (Cohen et al. 2004). The main data inputs were: (a) annual 24-hour average particulate matter concentrations (particulate matter with an aerodynamic diameter 189 10 and PM2.5) as an indicator of exposure to pollutionfrom combustion sources; and (b) information on the relative risk of mortality. In the CRA method, the population attributable fraction was calculated from these inputs as the difference in disease experience in a population and the hypothetical disease experience if the population were exposed to the hypothetical minimum of particulate matter (PM 2.57.5g/m 3; PM10 15g/m3). However, there is evidence that there may be no safe level ofexposure to particulate matter (WHO Europe 2004). We therefore set the theoretical minimum to zero in our analyses. Our estimates for long-term exposure are based on the contributions of two health outcomes: cardiopulmonary disease and lung cancer in adults aged 30 years and older. Attributable burden was estimated using risk coefficients from a large cohort study of adults in the United States (Pope et al. 2002). We did not use the CRA method of attributing acute respiratory infection in children aged 04 years as this method applies a relative risk based on daily exposure to annual exposure levels. Given the availability of daily urban air pollution data in Australia, and more appropriate relative risk estimates from Australian pollution concentration and mortality data, we used the estimates generated using the shortterm effects methods described below. We based exposure on annual mean levels for 2002 in the following urban areas: Sydney, Newcastle, Wollongong, Melbourne, Geelong, Brisbane, Perth, Adelaide, Canberra (including Queanbeyan), and Hobart. Annual concentrations were derived from data supplied by the state and territory environmental protection authorities, except for Adelaide and Hobart where we used published estimates (DPIWE 2004; Gooding & Riordan 2004). PM 2.5 concentration was not available for Geelong, Hobart or Canberra. For Geelong, weestimated the concentration from Melbournes PM 10:PM2.5 ratio. For Hobart and Canberra,we based our estimates on the average PM 10:PM2.5 ratio for those cities with original data(that is, Brisbane, Melbourne, Perth, Sydney, Adelaide, Newcastle and Wollongong). Due to temporal trends in particulate matter concentration, the linking of current exposure to chronic outcomes may underestimate the attributable burden if exposure levels were higher in the past. However, the use of recent exposure data is in keeping with the CRA methods. Short-term exposure Short-term exposure to urban air pollution has been associated with day-to-day variations in hospital admissions and mortality (Simpson et al. 2005a, 2005b). However, translating these findings into burden of disease estimates is not straightforward. The difficulty with estimating attributable morbidity is that published risks are established for the impact on hospitalisations only. An increase in hospitalisations for causes related to urban air pollution is likely to largely reflect exacerbation of existing disease rather than new disease events. Our YLD estimates are based on incident cases and their average duration at a particular level of severity. Thus the impact of urban air pollution on morbidity needs to be estimated as either a proportion of new cases of disease or a worsening of the condition for an undefined period of time. Until these methodological issues can be resolved we consider only a mortality component of the short-term health consequences of urban air pollution. The problem with attributing mortality to the short-term impact of urban air pollution is that there is equivocal evidence regarding the extent of harvesting, that is, imminent deaths brought forward by only a short period of time (less than a month) that were imminent anyway, or new deaths that would not have occurred in the absence of urban air pollution. This has a major bearing on our estimates of YLL: if harvesting occurs, YLL will be only a of less than 10 and 2.5 micrometres, P M 190 fraction of that normally calculated for each death. There is much debate in the literature on this topic. There are some arguments that harvesting does not play a role in the effects of urban air pollution. For instance there is an increase in deaths when longer lags between exposure and outcomes (up to 4 months) of urban air pollution are estimated, rather than a decrease (Schwartz 2001; Zeger et al. 1999). (The need to control for seasonal variation in these analyses makes it difficult to extend these analyses over the longer term as longer lags become strongly correlated with seasonal changes). This finding has been interpreted to indicate that harvesting is not an important issue. However, it could also be the case that urban air pollution exposure leads to chronic rather than acute effects on mortality. A further argument put forward by the same authors is that the largest increase in deaths was seen in people dying outside a hospital, while one would have expected a greater increase in hospital deaths if harvesting were bringing deaths forward in people who were already ill. The authors do not comment, however, on whether this may be due to the protective effect of the hospital environment. We concluded that there is no consensus on the relative contribution of deaths brought forward by urban air pollution nor on the size of the true acute impact on mortality. We therefore present the chronic impact as a lower estimate of the burden due to urban air pollution and add an alternative estimate of the combined shortterm and long-term effects, ignoring any harvesting, as an upper bound. Recent Australian research has provided the most applicable risk coefficients describing the effect of short-term exposure to urban air pollution on mortality (Simpson et al. 2005b). We applied these to daily urban air pollution data to estimate the attributable mortality burden of this risk. Following expert advice, our estimates were based on an averaged 01 day lag (that is, exposure to urban air pollution on the day of death and the day before death) of the contributions of two pollutants to two causes of death: all cause mortality (excluding accidental and other external causes of death) due to particle exposure (in units of light scattering by nephlometry, bsp), and respiratory deaths due to exposure to ozone. The choice of including these two pollutants and excluding others was made after discussion with the researchers (Simpson, Williams and Barnett) and justified by the finding that the impact on mortality of NO 2, CO and particles largely overlaps and hence including all three would leadto overestimation. The impact of SO 2 is considered small in Australia but ozone has asignificant impact on respiratory mortality independent of that of other pollutants. Estimates were calculated with a theoretical minimum exposure level of zero. This is based on evidence that at the population level there appears to be no safe level of exposure to particles or ozone (WHO Europe 2004). A decision was made to work with exposure data from 2002 rather than 2003 (the reference year for our study) because 2003 is considered an outlier year by the environmental protection authorities for pollutant readings. Daily urban air pollution data were supplied by the Victorian, New South Wales, Australian Capital Territory, Queensland and Western Australian environmental protection authorities. We calculated a PAF for each day by urban area, pollutant, and underlying cause of death, with the assumption that the entire population of that urban area was exposed. This was applied to daily 2002 mortality data, and aggregated to age- and sex-specific annual PAFs. We aggregated the number of deaths and YLL attributable to urban air pollution in specific areas (Sydney, Newcastle, Wollongong, Melbourne, Geelong, Brisbane, Perth, Adelaide, Canberra including Queanbeyan, and Hobart), calculated this as a proportion of all deaths or YLL in Australia and, finally, applied this proportion to 2003 mortality estimates. 191 We did not gain access to daily Tasmanian or South Australian urban air pollution data. Particle levels for Adelaide and Hobart were therefore extrapolated from published annual mean PM 10 levels (Air Monitoring Unit, EPA SA 2003; DPIWE 2004), and the average ratio ofbsp:PM 10 for Brisbane, Sydney and Melbourne from Simpson and colleagues (2005b). Theratio of the extrapolated mean bsp for Adelaide and Hobart to the annual mean bsp level for the cities for which we had detailed exposure data was then applied to the annualised PAF for these cities to extrapolate the PAFs for the two cities with missing exposure data. Ozone levels for Adelaide were based on the published average for 2002 (Gooding & Riordan 2004). Ozone is not routinely monitored in Hobart (DPIWE 2006); we therefore did not include this region in our analysis of respiratory deaths due to ozone exposure. Unsafe sex All sexually transmitted diseases were attributed to unsafe sex. The PAFs for HIV/AIDS and hepatitis B and C due to unsafe sex were derived as described in the section on illicit drugs. Previous Australian and Victorian burden of disease studies have used a PAF of 0.90 for cervical cancer. In this study we attributed all cervical cancer to sexual transmission of the human papilloma virus. Munoz and colleagues (2003) found that 90.7% cases had HPV DNA detected. Similarly, in a meta-analysis Clifford and colleagues (2003) found that HPV DNA was present in 8089% of cases. However, research by Walboomers and colleagues (1999), in which they revisited a previous study, suggests that nearly all cases that were negative for HPV DNA were false negatives. They revised up the estimates of cases testing positive for HPV DNA from 93% to 99.7%. Bosch and Munoz (2002) suggest that in most studies where 515% of cases are negative for HPV these are false negatives. Osteoporosis Osteoporosis causes no disability or death per se; it does, however, increase the risk of fracture. Therefore we treated osteoporosis as a risk factor in this study rather than as a disease in its own right, as was done in the previous Australian burden study. The WHO Task-Force for Osteoporosis recommends that the condition be defined by level of bone mineral density (BMD). We therefore based our PAF calculations on the population distribution of BMD, and relative risks associated with decreasing BMD. In Australia there are two large studies that have measured population BMD, one based in Geelong and the other in Dubbo. Both the Geelong Osteoporosis Study and the Dubbo Osteoporosis Epidemiology Study state that the population they cover is representative of the Australian population (Nguyen et al. 2001; Sanders et al. 1998). Mean BMD and standard deviations (SDs) for the Geelong and Dubbo studies were supplied by the study custodians. We used Geelong data for ages <60 and combined Geelong and Dubbo data for 60 years or over by fitting a Weibull distribution. From this distribution we plotted BMD by age for ages 25 years or over and fitted a polynomial distribution (R 2=0.998). We then predicted meanBMD from this equation for 5-year age groups from 60 years. For males, we assumed that the difference between the Dubbo and Geelong BMD means for women would also apply to males if Geelong data were available. We therefore increased Dubbo means by the ratio of female Dubbo sampled mean to the combined mean. We assumed deviations from the line were sampling error, and predicted mean BMD by age group from the fitted quadratic equation (R 2=0.925). We assumed the SD for Dubbo applied.192 The WHO Task-Force for Osteoporosis recommends that the condition be defined in Caucasian women as a BMD 2.5 SDs or more below the young female reference mean (Genant et al. 1999). The Australia and New Zealand Bone and Mineral Society and Osteoporosis Australia recommended that data from the Geelong Osteoporosis Study be used to establish a standardised reference range for Australia (Henry et al. 2004). We therefore used the mean BMD and SD for young women aged 2029 from this study as the theoretical minimum, and also used this population for the osteoporosis cut-off (Henry et al. 2004). There is currently no Australian reference mean BMD and SD for young adult men. We estimated these values by multiplying the Australian young female mean and SD (Henry et al. 2004) by the ratio of male to female mean and SD from the USAs National Health and Nutrition Examination Survey (NHANES) (Looker et al. 1998). The NHANES used Hologic densitometers while both the Dubbo and Geelong studies used Lunar densitometers. These machines do not give standardised results. We therefore converted the Hologic estimates to Lunar by applying the formula available at <www.courses.washington.edu/bonephys/ opBMDs.html>. Relative risks and odds ratios from a number of studies were pooled to estimate the relative risk of low impact fracture per 0.1g/cm 2 decrease in BMD measured at the femoral neck(EPOS Group 2002; Fujiwara et al. 2003; Kroger et al. 1995; Nguyen et al. 2005a, 2005b; Papaioannou et al. 2005; Schott et al. 2005; Schuit et al. 2004; Stone et al. 2003). Where a study used Hologic or Norland densitometers, and the relative risk was per SD change in BMD, we converted the studys SD estimates to Lunar. We derived PAFs for a number of fracture sites. Where possible these sites were linked directly to a single nature of injury category. In some cases (for example hip) we applied the PAF to a proportion of a category based on the distribution of fracture sites in the National Hospital Morbidity Database 200203. Since most studies that we included in the calculation of relative risks excluded fractures resulting from high impact causes, we applied the PAFs to fractures resulting from falls, striking and crushing accidents, and other unintentional injuries. For attributable YLL, we applied the site-specific fracture YLD PAFs to the site-specific mortality distribution for vertebral, pelvis and femur fracture to derive a site-specific YLL PAF. This was applied to deaths with an underlying cause of falls, striking and crushing accidents, other unintentional injuries, ill-defined falls or osteoporosis, where a fractured spine, pelvis or femur was mentioned. If more than one fracture was mentioned we applied the larger PAF, that is, for fractured pelvis and femur we applied the PAF for femur. We assumed that all deaths with an underlying cause of osteoporosis but no mention of vertebral, pelvis, or femur fracture, were attributable to osteoporosis. To determine the burden of disease code-specific YLL PAF for osteoporosis we calculated the proportion of burden of disease code-specific deaths attributable to osteoporosis. Osteoporosis and ill defined fall deaths were redistributed to falls. If we were to limit the deaths attributable to osteoporosis to only those that were coded to osteoporosis, the overall number of deaths would have been considerably smaller. 193 Table A2.3: Definitions, theoretical minima, health outcomes and data sources for 14 selected health risks Health risk Exposure variable Theoretical minimum Outcomes Sources for exposure estimates Sources for hazard estimates High blood pressure Level of usual systolic blood pressure 115 (SD 6) mmHg Ischemic heart disease, stroke, hypertensive heart disease AusDiab study (Dunstan et al. 2002) Meta-analysis of 61 cohort studies with 1,000,000 North American and European participants (Prospective Studies Collaboration (Lawes et al. 2003)) High blood cholesterol Level of usual total blood cholesterol 3.8 (SD 0.6) mmol/L (147 (SD 23) mg/dL) Ischemic heart disease, ischemic stroke AusDiab study (Dunstan et al. 2002) Meta-analysis of 10 cohorts with 490,000 North American and European participants, and 29 cohorts with 350,000 participants from the Asia-Pacific region High body mass index (BMI) Body mass index, BMI (weight over height squared) 21 (SD 1) kg/m2 Ischemic heart disease, stroke, hypertensive heart disease, diabetes, osteoarthritis, endometrial cancer, kidney cancer, colon cancer, post-menopausal breast cancer AusDiab study (Dunstan et al. 2002) Meta-analysis of 33 cohorts with 310,000 participants for cardiovascular disease risks, 27 cohorts for cancer risks, and systematic review of cohort studies for diabetes risk Low fruit and vegetable consumption Fruit and vegetable intake per day 600 (SD 50) g intake per day for adults Ischemic heart disease, stroke, colorectal cancer, gastric cancer, lung cancer, oesophageal cancer National Health Survey 2001 (ABS 2001c) Systematic review and new meta-analysis of published cohort studies Osteoporosis Bone mineral density of the femoral neck Males 1.107 (SD 0.140) g/cm2; Females 1.018 (SD 0.127) g/cm2 With osteoporosis defined 2.5 or more SD below this mean Fractured hip, femur, humerus, clavicle, forearm/wrist, elbow, spine, rib, pelvis, lower leg, patella, foot, heel, toe, hand, finger from falls, striking and crushing accidents, other unintentional injuries Dubbo Osteoporosis Epidemiology Study (Nguyen 2005) and Geelong Osteoporosis Study (Kotowicz 2005) Pooled analysis of 10 studies (EPOS Group 2002; Fujiwara et al. 2003; Kroger et al. 1995; Nguyen et al. 2005a, 2005b; Papaioannou et al. 2005; Schott et al. 2005; Schuit et al. 2004; Stone et al. 2003) (continued) 194 Table A2.3 (continued): Definitions, theoretical minima, health outcomes and data sources for 14 selected health risks Health risk Exposure variable Theoretical minimum Outcomes Sources for exposure estimates Sources for hazard estimates Physical inactivity Four categories: inactive, insufficient, recommended level and highly active All in highly active group Ischemic heart disease, stroke, breast cancer, colon cancer, diabetes National Health Survey 2001 (ABS 2001c) Systematic review of published literature and new meta-analysis of cohort studies Past smoking No smoking COPD, cancers of mouth, oesophagus, lung, pancreas, larynx, bladder, kidney, stomach and uterus PetoLopez method Systematic reviews by English and colleagues (1995) and Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001) Current daily smokers No smoking Ischemic heart disease, stroke, peripheral vascular disease, Parkinsons disease, pneumonia (adults), fire injuries, macular degeneration National Health Survey 2001 (ABS 2001c) Systematic reviews by English and colleagues (1995) and Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001); Tomany and colleagues (2004) for age related macular degeneration Passive smoking No smoking Ischemic heart disease, stroke National Health Survey 1995 (ABS 1995) Systematic reviews by English and colleagues (1995) and Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001) Tobacco Maternal smoking; smoking while pregnant No smoking Asthma, pneumonia (children), sudden infant death syndrome, otitis media, low birth weight National Health Survey 2001 (ABS 2001c), Australias Mothers and Babies 2003 (AIHW: Laws & Sullivan 2005) Systematic reviews by English and colleagues (1995) and Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001); US Surgeon Generals Report on Involuntary exposure to tobacco smoke (US Department of Health and Human Services 2006); systematic review by Anderson and Cook (1997). (continued) 195 Table A2.3 (continued): Definitions, theoretical minima, health outcomes and data sources for 14 selected health risks Health risk Exposure variable Theoretical minimum Outcomes Sources for exposure estimates Sources for hazard estimates Alcohol Average number of standard drinks per day Low level of drinking Cancers of the mouth and oropharynx, oesophagus, liver, larynx and breast; inflammatory heart disease, hypertensive heart disease, ischemic heart disease, stroke, alcohol dependence and harmful use, gallbladder and bile duct disease, pancreatitis, road traffic accidents, falls, fires/burns/scalds, drowning, machinery accidents, suffocation and foreign bodies, suicide and self-inflicted injuries, homicide and violence, occupational injuries National Health Survey 2001 (ABS 2001c) Systematic reviews by English and colleagues (1995) and Ridolfo & Stevenson (AIHW: Ridolfo & Stevenson 2001); the National Coroners Information System (Driscoll et al. 2001, 2004) for alcohol-related drownings and occupational YLL; fire injuries and fatalities pooled results from published studies; scalds and burns from Levy and colleagues (2004) Illicit drug use Use of illicit drugs Abstinence Heroin or polydrug, benzodiazepine, cannabis, and other drug dependence and harmful use AusBoD drug use and dependence models PAF = 1 by definition Use of illicit drugs Abstinence HIV/AIDS, hepatitis B, hepatitis C, inflammatory heart disease, suicide and self-inflicted injuries, road traffic accidents Population attributable fraction direct from the literature Incorporated findings from a multi-centre case-control study on 3,398 fatally injured drivers over Victoria, NSW and Queensland (examining psychoactive drugs); viral hepatitis and sexually transmissible infections from Australia Annual SurveillanceReport 2004; and systematic literature reviews by English and colleagues (1995) and Ridolfo and Stevenson (AIHW: Ridolfo & Stevenson 2001) Daily cannabis use No cannabis use, or use less often than daily Schizophrenia National Drug Strategy Household Survey 2004 Meta-analysis of 7 published case-control or cohort studies (examining link between psychosis and cannabis use). (continued) 196 Table A2.3 (continued): Definitions, theoretical minima, health outcomes and data sources for 14 selected health risks Health risk Exposure variable Theoretical minimum Outcomes Sources for exposure estimates Sources for hazard estimates Unsafe sex Unprotected sex Abstinence/prote cted sex Sexually transmissible diseases, abortion, cervical cancer, HIV/AIDS, hepatitis B & C AusBoD sexually transmissible diseases, abortion, and cervical cancer models; PAF direct from the literature PAF=1 (sexually transmissible diseases, abortion, cervical cancer); HIV/AIDS proportion from the Australian HIV and AIDS Public Access Datasets (National Centre in HIV Epidemiology and Clinical Research 2005a, 2005b); hepatitis B & C fraction from the National Centre in HIV Epidemiology and Clinical Research (National Centre in HIV Epidemiology and Clinical Research 2004) Child sexual abuse Non-contact only, contact only, intercourse No abuse Anxiety & depression, alcohol dependence & harmful use, heroin or polydrug use & dependence, benzodiazepine dependence & harmful use, cannabis dependence & harmful use, other drug dependence & harmful use, suicide and self-inflicted injuries CRA priors for Australia Systematic review and new meta-analysis of published studies (Andrews et al. 2004) Intimate partner violence Physical or sexual violence by current or previous partner No history of sexual or physical violence by an intimate partner Anxiety & depression, alcohol dependence & harmful use, heroin or polydrug use & dependence, benzodiazepine dependence & harmful use, cannabis dependence & harmful use, other drug dependence & harmful use, suicide and self-inflicted injuries, tobacco smoking, cervical cancer, syphilis, chlamydia, gonorrhoea, other sexually transmissible diseases, anorexia nervosa, bulimia nervosa, other eating disorders, falls, other unintentional injuries, homicide & violence Womens Safety Survey 1996 (ABS 1996) Australian Longitudinal Study on Womens Health (Brown et al. 1999); 20032004National Homicide Monitoring Program (NHMP) Annual Report (Mouzos 2005);National Hospital Morbidity Database 2002 03 (AIHW 2003a) (continued) 197 Table A2.3 (continued): Definitions, theoretical minima, health outcomes and data sources for 14 selected health risks Health risk Exposure variable Theoretical minimum Outcomes Sources for exposure estimates Sources for hazard estimates Occupational exposures and hazards Exposure in the workplace to diseasecausing agents such as carbon monoxide, dyes, inorganic and organic dusts, pesticides, metals, metal fumes, petrochemica ls, plastics, solvents, isocyanate and nitroglycerine or nitroglycerol No exposure All accidents, intentional and unintentional injuries, cancers, heart disease, neurological disorders, chronic respiratory disorders, renal disease, osteoarthritis, slipped disc, occupational overuse syndrome National Workers Compensation Statistics Database 2003, National Coroners Information system 2003, and Bestestimates of the magnitude of health effects of occupational exposure to hazardous substances (Kerr et al. 1996) Systematic review of published literature, hospital inpatient data, mortality datasets, National Health Survey results, workers compensation data, notified industrial accident reports and special disease registry datasets Urban air pollution Exposure to particulate matter and/or oxygen (i.e. total population of cities of interest) No exposure Short-term exposure: cardiovascular, respiratory, and other deaths Long-term exposure: lung cancer, ischemic heart disease, stroke, inflammatory heart disease, hypertensive heart disease, COPD Assume all residing in relevant geographical areas exposed; particulate and ozone levels from state environmental protection agencies Time series analysis of short-term effects of urban air pollution in four Australians cities (Simpson et al. 2005a); long-term exposure effects from Pope and colleagues (2002) 198 Table A2.4: Prevalence of health risks by age and sex Males Females Health risk Category 04 514 1529 3044 4559 6069 7079 80+ 04 514 1529 3044 4559 6069 7079 80+mean . . Blood pressure (mmHg) . . . . 124 131 140 148 154 . . . . . . 115 126 138 146 150 SD . . . . . . 11 16 17 19 19 . . . . . . 12 17 19 22 21 Blood cholesterol (mmol/L) mean . . . . . . 5.5 5.8 5.6 5.6 5.3 . . . . . . 5.2 5.8 6.0 6.1 5.9 SD . . . . . . 1.0 1.1 0.9 0.9 1.0 . . . . . . 1.0 1.1 0.9 1.0 1.0 BMI (kg/m2) mean . . . . . . 26.8 27.5 27.2 27.1 25.8 . . . . . . 25.4 27.2 28.5 27.0 24.9 SD . . . . . . 4.1 4.0 3.7 3.8 3.5 . . . . . . 5.4 5.7 5.8 5.2 4.5 Fruit and vegetable mean . . . . 445 452 496 538 538 538 . . . . 484 506 569 602 577 577 consumption (g/day) SD . . . . 241 235 245 230 219 219 . . . . 237 228 240 234 217 217 Bone mineral density (BMD) (g/cm3) Mean . . . . . . . . . . 0.93 0.87 0.77 . . . . . . . . . . 0.85 0.78 0.66 SD . . . . . . . . . . 0.15 0.14 0.16 . . . . . . . . . . 0.13 0.12 0.12 High . . . . 10% 3% 3% 1% 1% 0% . . . . 4% 2% 1% 1% 0% 0% Recommended . . . . 47% 37% 37% 41% 44% 30% . . . . 37% 32% 35% 38% 27% 17% Insufficient . . . . 23% 29% 29% 26% 22% 21% . . . . 35% 38% 33% 28% 28% 24% Physical activity (% population in categories) Inactive . . . . 20% 31% 32% 33% 33% 49% . . . . 25% 28% 30% 33% 45% 59% Current smoker . . . . 30% 31% 23% 16% 9% 7% . . . . 25% 25% 18% 12% 9% 2% Prenatal exposure 16% . . . . . . . . . . . . . . 16% . . . . . . . . . . . . . . Tobacco (% population in categories) Maternal smoking 27% . . . . . . . . . . . . . . 27% . . . . . . . . . . . . . . Abstainer . . . . 37% 35% 33% 43% 49% 56% . . . . 57% 59% 58% 66% 73% 76% Low . . . . 48% 51% 52% 43% 45% 40% . . . . 35% 32% 32% 25% 20% 22% Hazardous . . . . 7% 7% 8% 8% 4% 2% . . . . 7% 7% 7% 7% 6% 2% Alcohol (% population in categories) Harmful . . . . 7% 7% 7% 6% 2% 2% . . . . 1% 2% 3% 2% 1% 0% (continued) 199 Table A2.4 (continued): Prevalence of health risks by age and sex Males Females Health risk Category 04 514 1529 3044 4559 6069 7079 80+ 04 514 1529 3044 4559 6069 7079 80+Daily cannabis use . . . . 4% 4% 1% 0% 0% 0% . . . . 2% 2% 0% 0% 0% 0% Prenatal exposure opioids 0% . . . . . . . . . . . . . . 0% . . . . . . . . . . . . . . Prenatal exposure cannabis 1% . . . . . . . . . . . . . . 1% . . . . . . . . . . . . . . Maternal use heroin . . . . . . . . . . . . . . . . . . . . 0% 0% 0% 0% 0% 0% Illicit drugs (% population in categories) Maternal use cocaine . . . . . . . . . . . . . . . . . . . . 2% 1% 0% 0% 0% 0% No abuse 100% 96% 96% 94% 94% 94% 94% 94% 98% 79% 79% 71% 71% 71% 71% 71% Non-contact only CSA 0% 1% 1% 2% 2% 2% 2% 2% 1% 6% 6% 9% 9% 9% 9% 9% Contact only CSA 0% 2% 2% 3% 3% 3% 3% 3% 1% 11% 12% 16% 16% 16% 16% 16% Child sexual abuse (% population in categories) Intercourse CSA 0% 1% 1% 1% 1% 1% 1% 1% 0% 3% 3% 5% 5% 5% 5% 5% Intimate partner violence (% population in categories) Sexual or physical violence . . . . . . . . . . . . . . . . . . . . 15% 22% 21% 10% 10% 10% Low . . . . 14% 9% 8% 3% 1% 1% . . . . 34% 25% 22% 5% 1% 1% Moderate . . . . 44% 41% 35% 13% 3% 3% . . . . 18% 17% 18% 4% 1% 1% Occupational exposure to ergonomic stressors (% population in categories) High . . . . 2% 3% 4% 4% 2% 2% . . . . 0% 0% 0% 0% 0% 0% Occupational exposure to ergonomic stressors (increases risk of osteoarthritis) (% population in categories) Blue collar workers . . . . 40% 40% 34% 12% 3% 3% . . . . 8% 9% 10% 3% 1% 1% (continued) 200 Table A2.4 (continued): Prevalence of health risks by age and sex Males Females Health risk Category 04 514 1529 3044 4559 6069 7079 80+ 04 514 1529 3044 4559 6069 7079 80+8590 dBA Occupational exposure to . . . . 5% 5% 5% 2% 1% 1% . . . . 4% 3% 3% 1% 0% 0% noise (% population in categories) >90 dBA . . . . 4% 4% 3% 1% 0% 0% . . . . 1% 1% 1% 0% 0% 0% Occupational exposure to Low . . . . 3% 4% 4% 1% 0% 0% . . . . 3% 4% 4% 1% 0% 0% leukaemogens (% population in categories) High . . . . 0% 0% 0% 0% 0% 0% . . . . 0% 0% 0% 0% 0% 0% Occupational exposure to Low . . . . 21% 28% 23% 8% 2% 2% . . . . 6% 9% 7% 2% 0% 0% lung carcinogens (% population in categories) High . . . . 2% 3% 3% 1% 0% 0% . . . . 1% 1% 1% 0% 0% 0% Occupational exposure to Low . . . . 18% 27% 25% 11% 4% 4% . . . . 7% 11% 10% 3% 1% 1% agents causing COPD (% population in categories) High . . . . 12% 14% 11% 3% 1% 1% . . . . 1% 2% 2% 0% 0% 0% Background . . . . 23% 10% 17% 67% 90% 90% . . . . 31% 28% 33% 84% 97% 97% Administration . . . . 6% 12% 13% 4% 1% 1% . . . . 16% 21% 18% 4% 1% 1% Technical . . . . 16% 30% 28% 10% 3% 3% . . . . 16% 27% 24% 5% 1% 1% Sales . . . . 10% 4% 3% 2% 0% 0% . . . . 19% 6% 5% 1% 0% 0% Agriculture . . . . 2% 3% 4% 4% 2% 2% . . . . 0% 1% 1% 1% 0% 0% Mining . . . . 1% 1% 1% 0% 0% 0% . . . . 0% 0% 0% 0% 0% 0% Transport . . . . 6% 8% 9% 3% 1% 1% . . . . 1% 1% 1% 0% 0% 0% Manufacturing . . . . 31% 28% 21% 7% 1% 1% . . . . 6% 6% 6% 1% 0% 0% Occupational exposure to agents causing asthma (% population in categories) Services . . . . 6% 4% 4% 2% 1% 1% . . . . 11% 10% 11% 3% 1% 1% 201 Annex tables 202 Annex Table 1: Disease and injury categories and ICD-10 codes Cause ICD10 codesI. Communicable diseases, maternal and neonatal conditions A. Infectious and parasitic diseases 1. Tuberculosis A1519;B90;K230,673,930;M011,490,900;N330,7401;O980;P370 2. Sexually transmitted diseases(a) a. Syphilis A5053;I980;K672;M031,731;N290,742 b. Chlamydia A56;K670;N744 c. Gonorrhoea A54;K671;M730;N743;O982 d. Other sexually transmitted diseases A55,5764 3. HIV/AIDS B2024;F024 4. Diarrhoeal diseases A0009 5. Childhood immunisable diseases a. Diphtheria A36 b. Whooping cough A37 c. Tetanus A3335 d. Poliomyelitis A80;B91 e. Measles B05 f. Rubella B06;M014;P350 g. Haemophilus influenzae type b (Hib) A413,492;G000;J051,14,2016. Meningitis A39;G0019,03 7. Septicaemia A40,4102,4148 8. Arbovirus infection a. Ross River virus B331 b. Barmah Forest virus A92.8 c. Dengue A9091 d. Other arbovirus infection A8384,852,9299 9. Hepatitis a. Hepatitis A B15 b. Hepatitis B(b) B16,170,1801 c. Hepatitis C(c) B171,182 d. Other hepatitis B1728,1889,19;P353 10. Malaria B5054 11. Trachoma A71;B940 12. Other infectious and parasitic diseases A2032,38,4248,4901,4939,6570,7479,8182,8501, 858, 8689;B0004,0709,2530,330,3328,3449,5589,92 (excluding 92.8),941, 9489,9599;G0102,04 07;K231;M00,010,0123,0158, 030 B. Acute respiratory infections 1. Lower respiratory tract infections J1013,1518,200,2029,2122 2. Upper respiratory tract infections J0004,050,06 3. Otitis media H6566 C. Maternal conditions 1. Maternal haemorrhage O441,4546,67,72 2. Maternal sepsis O411,8586 3. Hypertensive disorders of pregnancy O1016 4. Obstructed labour O6466,711,713 5. Abortion O0008 (continued) 203 Annex Table 1 (continued): Disease and injury categories and ICD-10 codes Cause ICD10 codes6. Other maternal conditions O09,2024,2640,410,4189,4243,440,4763,6870, 710,712, 7149,7382,8797,981,9839,99 D. Neonatal causes 1. Birth trauma and asphyxia P03,1021,2428 2. Low birthweight P0507,22 3. Neonatal infections P23,3512,3589,36,3719,3839 4. Other conditions arising in the perinatal period P04,08,29,5096 E. Nutritional deficiencies 1. Protein-energy malnutrition E4045,640;M833;O25 2. Deficiency anaemia D5053 3. Other nutritional deficiencies E0002,031,50,511,5189,5261,6308,6419 II. Non-communicable diseases F. Malignant neoplasms 1. Mouth and oropharynx cancers C0014 2. Oesophagus cancer C15 3. Stomach cancer C16 4. Colorectal cancer C1821 5. Liver cancer(d) C22 6. Gallbladder cancer C2324 7. Pancreas cancer C25 8. Lung cancer C3334 9. Bone and connective tissue cancer C4041,4909 10. Melanoma C43 11. Non-melanoma skin cancers C44 12. Breast cancer C50 13. Cervix cancer C53 14. Corpus uteri cancer C54 15. Ovary cancer C56,5704 16. Prostate cancer C61 17. Testicular cancer C62 18. Bladder cancer C67 19. Kidney cancer C6466,68 20. Brain cancer C71 21. Thyroid cancer C73 22. Lymphoma C8185,96 23. Multiple myeloma C8890 24. Leukaemia C9195 25. Larynx cancer C32 26. Eye cancer C69 27. Other malignant neoplasms C17,2631,3739,4548,5152,5779,5860,63,70,72,7475 G. Other neoplasms 1. Uterine myomas D25 2. Benign neoplasms of meninges and brain D3233 3. Other benign neoplasms D0024,2631,3448 H. Diabetes mellitus 1. Type 1 diabetes E10 2. Type 2 diabetes E1113 (continued) 204 Annex Table 1 (continued): Disease and injury categories and ICD-10 codes Cause ICD10 codesI. Endocrine and metabolic disorders 1. Non-deficiency anaemia a. Haemolytic anaemia D5558 b. Other non-deficiency anaemia D5963,6408 2. Cystic fibrosis E84 3. Haemophilia D6667,681 4. Other endocrine and metabolic disorders D680,6829,6972,7304,7389,7489;E030,0329,0407,1535, 65,6602,6777,7814,7869,7983,85,8734,878,8890; D65,735;E6689,86,8702,8757 J. Mental disorders 1. Substance use disorders a. Alcohol dependence and harmful use(e) E512;F10;G312;X45 b. Heroin or polydrug dependence and harmful use F11;X42 c. Benzodiazepine dependence and harmful use F13 d. Cannabis dependence and harmful use F12 e. Other drug dependence and harmful use F1416,1819 2. Schizophrenia F2029 3. Anxiety and depression F30,3239,4001,4102,42,431,930 4. Bipolar disorder F31 5. Personality disorders(f) F603 6. Eating disorders a. Anorexia nervosa F5001 b. Bulimia nervosa F5023 c. Other eating disorders F5049 7. Childhood conditions a. Attention-deficit hyperactivity disorder F90 b. Autism spectrum disorders F84 8. Other mental disorders F0509, 4029,4139,430,4329,4448,5159,6002,6049, 61-69,8083,8889,9192,9319,9499 K. Nervous system and sense organ disorders 1. Dementia F0001,0201,023,03;G30,3101,3189 2. Epilepsy G4041 3. Parkinsons disease G20 4. Multiple sclerosis G35 5. Motor neurone disease G122 6. Huntingtons chorea F022;G10 7. Muscular dystrophy G710 8. Sense organ disorders a. Glaucoma-related blindness H40 b. Cataract-related blindness H2527 c. Macular degeneration H353 d. Adult-onset hearing loss H9091 e. Refractive errors H5207 f. Other vision loss H54 9. Migraine G43 10. Other nervous system and sense organ disorders F028,04,7079;G0809,11,1201,1289,13,2126,32,3637,44, 4670,711932,7292,9349,94H22;H2834,3502,354-9,36, 4251,53,5562,6783,9295 (continued) 205 Annex Table 1 (continued): Disease and injury categories and ICD-10 codes Cause ICD10 codesL. Cardiovascular disease 1. Rheumatic heart disease I0009 2. Ischemic heart disease I2025 3. Stroke G45;I6069 4. Inflammatory heart disease I3033,4042 5. Hypertensive heart disease I11,130,15 6. Non-rheumatic valvular disease I3439 7. Aortic aneurysm I71 8. Peripheral vascular disease I7008,7209,7374 9. Other cardiovascular disease I26,271,28,4345,4701,479,48,4919,5104,52,7784,8697, 9818,99 M. Chronic respiratory disease 1. Chronic obstructive pulmonary disease (COPD) I270,2789;J4044 2. Asthma J4546 3. Other chronic respiratory diseases J3039,4799 N. Diseases of the digestive system 1. Peptic ulcer disease K2527 2. Cirrhosis of the liver(g) I85;K70,717,7219,7374,7667 3. Appendicitis K3537 4. Intestinal obstruction K4001,4034,4101,4134,4201,4301,4401,4508,4601,56 5. Diverticulitis K57 6. Gallbladder and bile duct disease K8083 7. Pancreatitis K85,8601 8. Inflammatory bowel disease K5051 9. Vascular insufficiency bowel K55 10. Other digestive system diseases K2022,238,2831,38,402,409,412,419,429,439,449,469,52, 58-66, 678,7106,7189,720,75,7605,7689,77,8629, 87-91,9289, 9318 O. Genitourinary diseases 1. Nephritis and nephrosis(h) I12,131;N0001,0316,1719 2. Benign prostatic hypertrophy N40 3. Urinary incontinence N3934 4. Infertility N46,97 5. Other genitourinary diseases N02,2028,2918,3032,338392,3437,3989,4145,4764, 75-96, 9899 P. Skin diseases 1. Eczema L2027 2. Acne L70 3. Psoriasis L40 4. Ulcers L03,0889,89,97,984 5. Other skin diseases L0002,0405,0801,1014,2830,4168,7188,9095,9803, 9859,99 Q. Musculoskeletal diseases 1. Rheumatoid arthritis M0506,080,120,4658 2. Osteoarthritis M1519 3. Back pain(i) M469,47,4803,4889,5389,5459 4. Slipped disc M464,5051,5434,961 5. Occupational overuse syndrome 6. Systemic lupus erythematosus (SLE) M32 (continued) 206 Annex Table 1 (continued): Disease and injury categories and ICD-10 codes Cause ICD10 codes7. Gout M10 8. Other musculoskeletal diseases M02,0326,07,0819,09,11,1218,1314,2031,3345,4603, 4845,4918,5303,5402,6072,738,7579,8302,8349, 84-89,901960,9195,9629,99 R. Congenital anomalies 1. Anencephaly Q00 2. Spina bifida Q05 3. Congenital heart disease Q2028 4. Cleft lip and/or palate Q3537 5. Digestive system malformations a. Anorectal atresia Q42 b. Oesophageal atresia Q3901 c. Other digestive system malformations Q38,3929,4041,4345 6. Urogenital tract malformations a. Renal agenesis(j) Q60 b. Other urogenital tract malformations(k) Q5056,6164 7. Abdominal wall defect Q7925 8. Down syndrome Q90 9. Other chromosomal disorders Q9199 10. Other congenital anomalies Q0104,0618,3034,6578,7901,7969,8089 S. Oral conditions 1. Dental caries K02 2. Periodontal disease K05 3. Edentulism 4. Pulpitis K04 5. Other oral conditions K0001,03,0614 Z. Ill-defined conditions 1. Sudden infant death syndrome R95 2. Chronic fatigue syndrome G933;R53 III. Injuries T. Unintentional injuries 1. Road traffic accidents V0119,0219,0319,0419,0619,0923,1049,1149,1249, 1349,1449,1549,1649,1749,1849,1949,2049,2149, 2249,2349,2449,2549,2649,2749,2849,2949,3059, 3159,3259,3359,3459,3559,3659,3759,3859,3949, 4059,4159,4259,4359,4459,4559,4659,4759,4859, 4949,5059,5159,5259,5359,5459,5559,5659,5759, 5859,5949,6059,6159,6259,6359,6459,6559,6659, 6759,6859,6949,7059,7159,7259,7359,7459,7559, 7659,7759,7859,7949,8035,809,811,8219,8303,8403, 8503,8604,8708,892,899;Y85 2. Other transport accidents V010,020,030,040,05,060,0901,099,1003,1103,1203,1303, 1403,1503,1603,1703,1803,1903,2003,2103,2203, 2303,2403,2503,2603,2703,2803,2903,3004,3104, 3204,3304,3404,3504,3604,3704,3804,3903,4004, 4104,4204,4304,4404,4504,4604,4704,4804,4903, 5004,5104,5204,5304,5404,5504,5604,5704,5804, 590-3,6004,6104,6204,6304,6404,6504,6604,6704, 6804,6903,7004,7104,7204,7304,7404,7504,7604, 7704,7804,7903,8002,8068,810,8129,820,8349, 8449, 8549,8659,879,88,8901,893,9099 3. Poisoning X4041,4344,4649 4. Falls W0019; M8082 5. Fires, burns and scalds X0019 (continued) 207 Annex Table 1 (continued): Disease and injury categories and ICD-10 codes Cause ICD10 codes6. Drowning W6574 7. Sports injuries W21;X50 8. Natural and environmental factors W5359,64,8599;X2039,5157 9. Machinery accidents W24,2731 10. Other unintentional injuries Suffocation and foreign bodies W44, W75W84 Adverse effects of medical treatment Y40Y59, Y60Y69, Y70Y84, Y88 Other unintentional injuries n.e.c. W20, W22W23, W25W26, W32W44, W45, W49, W51, W50, W52, W60, W7584; X58; Y40Y59, Y60Y84, Y86, Y880Y883 U. Intentional injuries 1. Suicide and self-inflicted injuries X6084;Y870 2. Homicide and violence X85Y09;Y871 3. Legal intervention and war Y3536,8901 Redistribution categories 1. Pelvic inflammatory disease N7073,748 2. Unspecified septicaemia A419 3. Hepatitis sequelae B942 4. Neonatal causes coded based on maternal condition P0002 5. Ill-defined nutritional E46,639 6. Ill-defined malignant neoplasms C7680,97 7. Uterus cancerunspecified C55 8. Unspecified diabetes mellitus E14 9. Other anaemia D649 10. Smoking listed as cause F17 11. Hypertensive heart and renal disease I1329 12. Heart failure I50 13. Essential hypertension I10 14. Ill-defined cardiovascular conditions E780,785;I46,472,490,5159,709 15. Gastric haemorrhage K9202 16. Ill-defined unintentional accidents (fall if also fracture) X59;Y9098 17. Other accidentsintent undetermined Y20,2225,2829,33,34,872,899 18. Road traffic accidentsintent undetermined Y32 19. Poisoningintent undetermined Y1019 20. Fallsintent undetermined Y3031 21. Burnsintent undetermined Y2627 22. Drowningintent undetermined Y21 23. Ill-defined non-injuries R0052,5494,9699 Notes (a) Excluding HIV/AIDS. (b) Including hepatitis B-related liver cancer and cirrhosis. (c) Including hepatitis C-related liver cancer and cirrhosis. (d) Excluding hepatitis B and C related liver cancer. (e) Including alcoholic cirrhosis. (f) Excludes those with any other comorbid mental disorders. (g) Excluding alcoholic and hepatic cirrhosis. (h) Excluding diabetic-, congenital- and poisoning-related renal failure. (i) Includes both acute and chronic back pain. (j) Including renal failure due to dysplasia. (k) Including polycystic renal failure. 208 Annex Table 2: Principal data sources for epidemiological modelling Primary data source Prevalence/ Incidence Reference period Disease and injury categories A. Disease registers, surveillance and notification systems Incidence 2003 A1 Tuberculosis Incidence 2003 A2a Syphilis Incidence 2003 A2b Chlamydia Incidence 2003 A2c Gonorrhoea Incidence 2003 A5a Diphtheria Incidence 200003 A5b Pertussis Incidence 2003 A5c Tetanus Incidence 2003 A5d Poliomyelitis Incidence 2003 A5e Measles Incidence 2003 A5f Rubella Incidence 2003 A5g Haemophilus influenzae type BIncidence 199396 A5g Hib B sequela Incidence 2003 A8 Arbovirus infections Incidence 2003 A9a Hepatitis A Incidence 2003 A9b Hepatitis B National Notifiable Diseases Surveillance System: includes: notifications; and reports: annual report C ommunicable diseases intelligenceIncidence 2003 A10 Malaria HIV/AIDS National Registry Incidence 2003 A3 HIV/AIDS National Perinatal Data Collection Incidence 2003 D2 Low birthweight Victorian Perinatal Data Collection Unit Incidence 200102 D2 Low birthweight Queensland Perinatal Data Collection Incidence 2002 D2 Low birthweight National Cancer Statistics Clearing House Incidence 2001 F Malignant neoplasms State and territory cancer registries Incidence 1997 F12 Breast cancer BreastScreen Australia Incidence 200102, 1997 F12 Breast cancer National Diabetes Register Incidence 2001 H Diabetes mellitus Incidence 2002 H Diabetes Australia and New Zealand Dialysis and mellitus sequela Transplant Registry Incidence 2002 O1 Nephritis and nephrosis Victorian Cystic Fibrosis Screening program Incidence 19891998 I2 Cystic fibrosis ABS Causes of death data set Incidence 2003 K5 Motor neurone disease Incidence 2003 R1 Anencephaly Western Australian Intellectual Disability Exploring Answers database Incidence 19831996 K9 Intellectual disability Incidence 200102 R2 Spina bifida Incidence 200102 R5 Digestive system malformation Victorian Perinatal Data Collection Unit Birth Defects Register Incidence 200102 R6a Renal agenesis Incidence 200102 K9 Intellectual disability Incidence 1997 R3 Congenital heart disease Incidence 1997 R5 Digestive system malformation Incidence 1997 R6b Other urogenital tract malformations Congenital malformations, Australia Incidence 2001 R7 Abdominal wall defect Western Australian Birth Defects Registry Incidence 2003 R6b Other urogenital tract malformations (continued) 208 Annex Table 2: Principal data sources for epidemiological modelling Primary data source Prevalence/ Incidence Reference period Disease and injury categories A. Disease registers, surveillance and notification systems Incidence 2003 A1 Tuberculosis Incidence 2003 A2a Syphilis Incidence 2003 A2b Chlamydia Incidence 2003 A2c Gonorrhoea Incidence 2003 A5a Diphtheria Incidence 200003 A5b Pertussis Incidence 2003 A5c Tetanus Incidence 2003 A5d Poliomyelitis Incidence 2003 A5e Measles Incidence 2003 A5f Rubella Incidence 2003 A5g Haemophilus influenzae type BIncidence 199396 A5g Hib B sequela Incidence 2003 A8 Arbovirus infections Incidence 2003 A9a Hepatitis A Incidence 2003 A9b Hepatitis B National Notifiable Diseases Surveillance System: includes: notifications; and reports: annual report C ommunicable diseases intelligenceIncidence 2003 A10 Malaria HIV/AIDS National Registry Incidence 2003 A3 HIV/AIDS National Perinatal Data Collection Incidence 2003 D2 Low birthweight Victorian Perinatal Data Collection Unit Incidence 200102 D2 Low birthweight Queensland Perinatal Data Collection Incidence 2002 D2 Low birthweight National Cancer Statistics Clearing House Incidence 2001 F Malignant neoplasms State and territory cancer registries Incidence 1997 F12 Breast cancer BreastScreen Australia Incidence 200102, 1997 F12 Breast cancer National Diabetes Register Incidence 2001 H Diabetes mellitus Australia and New Zealand Dialysis and Incidence 2002 H Diabetes mellitus sequela Transplant Registry Incidence 2002 O1 Nephritis and nephrosis Victorian Cystic Fibrosis Screening program Incidence 19891998 I2 Cystic fibrosis ABS Causes of death data set Incidence 2003 K5 Motor neurone disease Incidence 2003 R1 Anencephaly Western Australian Intellectual Disability Exploring Answers database Incidence 19831996 K9 Intellectual disability Incidence 200102 R2 Spina bifida Incidence 200102 R5 Digestive system malformation Victorian Perinatal Data Collection Unit Birth Defects Register Incidence 200102 R6a Renal agenesis Incidence 200102 K9 Intellectual disability Incidence 1997 R3 Congenital heart disease Incidence 1997 R5 Digestive system malformation Incidence 1997 R6b Other urogenital tract malformations Congenital malformations, Australia Incidence 2001 R7 Abdominal wall defect Western Australian Birth Defects Registry Incidence 2003 R6b Other urogenital tract malformations (continued) 209 Annex Table 2 (continued): Principal data sources for epidemiological modelling Primary data source Prevalence/ Incidence Reference period Disease and injury categories B. Health service utilisation data Incidence 200203 A2b Chlamydia sequela Incidence A2c Gonorrhoea sequela Incidence A4 Diarrhoea Incidence A5e Measles sequela Incidence A6 Meningitis Incidence A7 Septicaemia Incidence A8c Dengue fever sequela Incidence A9a Hepatitis A Incidence A9c Hepatitis B sequela (D)(a) Incidence A9c Hepatitis C sequela (D) Incidence C1 Maternal haemorrhage (P)(b) Incidence C3 Hypertension in pregnancy (P) Incidence C4 Obstructed labour (P) Incidence C5 Abortion (P) Incidence C6 Other maternal conditions (P) Incidence D1 Birth trauma & asphyxia Incidence D3 Neonatal infections Incidence G Benign neoplasms (P) Incidence H Diabetes sequela (P) Incidence I1a Haemolytic anaemia Prevalence I1b Other non-deficiency anaemia Incidence K8b Cataract-related blindness (P) Incidence L2 Ischemic heart diseaseAMI Incidence L3 Stroke Incidence L7 Aortic aneurysm Prevalence L8 Peripheral vascular disease (P) Incidence L8 Peripheral vascular disease sequela Prevalence N2 Cirrhosis of the liver (D) Incidence N3 Appendicitis (P) Incidence N4 Intestinal obstruction (P) Incidence N5 Diverticulitis (P) Incidence N6 Gall bladder and bile duct disease (P) Incidence N7 Pancreatitis Incidence N8 Inflammatory bowel disease (P) Incidence N9 Vascular insufficiency of intestine (P) Incidence O2 Benign prostatic hypertrophy (P) Incidence Oot Other genitourinary diseases (P) Incidence Q4 Slipped disc (P) Incidence R3 Congenital heart disease (P) Incidence R4 Cleft lip and or palate (P) Incidence T Unintentional injuries National Hospital Morbidity Database (diagnoses or procedures) Incidence U Intentional injuries Bettering the Evaluation and Care of Health Incidence 200001 B1 Lower respiratory tract infections Incidence 200001 B2 Upper respiratory tract infections (continued) 210 Annex Table 2 (continued): Principal data sources for epidemiological modelling Primary data source Prevalence/ Incidence Reference period Disease and injury categories Incidence 200001 B3 Otitis media Incidence 200304 N1 Peptic ulcer disease Incidence 200304 P4 Skin ulcers Alcohol and Other Drug Treatment Services National Minimum Data Set Prevalence 200203 J1c Stimulant dependence Incidence 19902003 L1 Heart failure Incidence 19902003 L2 Ischemic heart disease Western Australian Data Linkage System Incidence 19902003 L3 Stroke Victorian Linked Admitted Episodes Database Incidence 19962002 L Heart failure Incidence 19962002 N9 Vascular insufficiency of intestine C. Population health surveys 200102 National Gastroenteritis Survey Incidence 200102 A4 Diarrhoea 1980 National Trachoma and Eye Health Program Prevalence 197678 A11 Trachoma sequela Incidence 197678 B3 Otitis media Incidence 1995 B2 Upper respiratory tract infections Incidence 2001 B3 Otitis media Prevalence 2001 K10 Migraine Prevalence 2001 P1 Eczema Prevalence 2001 Poth Other skin diseases Prevalence 1995 Q3 Chronic back pain (U)(c) Incidence 2001 Q7 Gout Prevalence & incidence 2001 Qot Other musculoskeletal disorders National Health Survey Prevalence & incidence 1995 Qot Other musculoskeletal disorders Australian Diabetes, Obesity and Lifestyle Study Incidence 19992000 E2 Deficiency anaemia (AusDiab) Prevalence 19992000 H Diabetes mellitus Risk Factor Prevalence Study, 1989 Incidence 1989 E2 Deficiency anaemia 2002 National non-melanoma skin cancer survey Incidence 2002 F11 Non-melanoma skin cancer Prevalence 1997 J1a Alcohol dependence Prevalence 1997 J1c Benzodiazepine dependence Prevalence 1997 J1d Cannabis dependence Prevalence 1997 J2 Psychotic disorders Prevalence 1997 J3 Anxiety and depression Prevalence 1997 J4 Bipolar disorder Prevalence 1997 J5 Personality disorders (isolated) National Mental Health and Wellbeing Survey, 1997adult component, Low prevalence (psychotic) disorders component, and child & adolescent component Prevalence 1997 J7a ADHD Australian Child to Adult Development Study Incidence 199096 K9 Intellectual disability Australian Longitudinal Study on Womens Prevalence 19962002 O3 Urinary incontinence Health(d) Prevalence 19962002 Oot Menstrual problems Prevalence 1998 O3 Urinary incontinence Prevalence 2003 Q3 Chronic back pain Prevalence 2003 Q5 Occupational overuse syndrome Survey of Disability, Ageing and Carers Prevalence 1993 Qot Other musculoskeletal disorders Child Dental Health Survey, Australia Incidence 2000 S1 Dental caries National Oral Health Survey of Australia Prevalence 198788 S1 Dental caries Prevalence 198788 S2 Periodontal disease South Australian Dental Longitudinal Study Incidence 19911996 S1 Dental caries (continued) 211 Annex Table 2 (continued): Principal data sources for epidemiological modelling Primary data source Prevalence/ Incidence Reference period Disease and injury categories The Adelaide Dental Study of Nursing Homes, one year follow up 1999 Incidence 1999 S1 Dental caries The Adelaide Dental Study of Nursing Homes 1998 Prevalence 1998 S3 Edentulism The Longitudinal Study of Dentists Practice Activity Incidence 199899 S4 Pulpal infection Prevalence National Dental Telephone Interview Survey 2002 S3 Edentulism Incidence 2002 S4 Pulpal infection D. Epidemiological studies Incidence A2 STIs (apart from HIV/AIDS) Incidence A5b Pertussis sequela Incidence A10 Malariasequela Incidence B3 Otitis mediasequela Incidence C2 Maternal sepsissequela GBD study Incidence C3 Hypertensive disorders in pregnancysequela Incidence A6 Meningitis sequela Prevalence A9b Hepatitis B Prevalence A9c Hepatitis C sequela Incidence D4 Other neonatal causes Prevalence E2 Deficiency anaemia Incidence H Diabetes mellitus sequela Prevalence I3 Haemophilia Prevalence J1b Heroin dependence Prevalence J6b Anorexia Incidence J7b Autism spectrum disorders Prevalence K4 Multiple sclerosis Incidence K6 Huntingtons chorea Incidence K7 Muscular dystrophy Prevalence K8 Sense organ disorders Incidence K9 Intellectual disability Prevalence L3 Stroke Prevalence M1 Chronic obstructive pulmonary disease Prevalence M2 Asthma Prevalence N2 Cirrhosis of the liver Prevalence O4 Infertility Prevalence P1 Eczema Australian epidemiological studies Prevalence Poth Other skin diseases Incidence A2b Chlamydia sequela (i.e. childwish) Prevalence A9b Hepatitis B sequela Incidence A9c Hepatitis C sequela Incidence D1 Birth trauma & asphyxiasequela Incidence D2 Low birthweightsequela Incidence J6a Bulimia Incidence K2 Epilepsy Incidence K10 Migraine Prevalence M2 Asthma International epidemiological studies Incidence N8 Inflammatory bowel disease (continued) 212 Annex Table 2 (continued): Principal data sources for epidemiological modelling Primary data source Prevalence/ Incidence Reference period Disease and injury categories Prevalence O3 Urinary incontinence Incidence Q1 Rheumatoid arthritis Incidence Q2 Osteoarthritis Incidence Q4 Slipped disc Prevalence Z2 Chronic fatigue syndrome Meta-analyses of epidemiological studies Prevalence K1 Dementia Prevalence K3 Parkinsons disease E. Estimates that are distributed to other models Incidence D1 Birth trauma & asphyxia D2 Low birthweight D3 Neonatal infections D4 Other perinatal conditions R8 Down syndrome K9 Intellectual disability R9 Other chromosomal anomalies Prevalence 19962002 L1 Rheumatic heart disease Incidence 19962002 L2 Ischemic heart disease Incidence 19962002 L4 Inflammatory heart disease Prevalence 19962002 L5 Hypertensive heart disease Incidence 19962002 L6 Non-rheumatic valvular disease L Heart failure Incidence 19962002 M1 Chronic obstructive pulmonary disease F. Indirect estimation A12 Other infectious and parasitic diseases D4 Other perinatal conditions I4 Other endocrine and metabolic diseases L9 Other cardiovascular disease M3 Other chronic respiratory diseases N10 Other digestive system diseases R10 Other congenital anomalies YLL to YLD ratio from rest of category Oot Other genitourinary diseases Notes (a) (D) refers to distributions which are used to estimate incidence to underlying causes. (b) (P) refers to hospital data on proceduresmay or may not be in addition to information on principal diagnosis. (c) (U) proportion by underlying cause or type of problem (recent versus long-term). (d) The research on which this report is based was conducted as part of the Australian Longitudinal Study on Womens Health, The University of Newcastle and The University of Queensland. We are grateful to the Australian Government Department of Health and Ageing for funding and to the women who provided the survey data. 213 Annex Table 3: Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+All causes 2,632,770 1,364,614 1,268,156 124,809 102,480 603,937 244,198 289,190 96,727 94,077 514,332 184,705 378,314I. Communicable diseases,maternal and neonatal conditions 123,094 64,993 58,101 24,836 1,807 22,017 6,363 9,970 20,387 3,131 17,027 4,240 13,316 A. Infectious and parasitic diseases 44,685 27,301 17,385 2,004 901 17,093 4,034 3,269 1,644 1,099 8,662 2,318 3,662 1. Tuberculosis 646 330 316 3 14 150 63 100 4 11 92 44 166 2. Sexually transmitted diseases(a) 2,048 83 1,966 5 26 40 12 54 437 1,412 21 41 a. Syphilis 102 26 77 4 1 8 12 36 2 29 9 b. Chlamydia 1,188 49 1,139 22 26 14 264 830 13 19 c. Gonorrhoea 28 9 19 3 6 5 13 d. Other sexually transmitted diseases 730 730 4 166 539 8 13 3. HIV/AIDS 6,660 5,960 700 7 346 5,417 179 12 6 62 610 22 4. Diarrhoeal diseases 1,858 872 986 334 101 309 51 77 348 90 310 63 175 5. Childhood immunisable diseases 557 315 243 99 7 119 66 23 121 8 47 39 27 a. Diphtheria b. Whooping cough 150 70 80 42 7 18 2 1 42 8 26 2 1 c. Tetanus d. Poliomyelitis 197 119 78 32 65 22 21 37 21 e. Measles 1 f. Rubella 25 17 8 17 8 g. Haemophilus influenzaetype b (Hib) 184 108 76 40 68 71 5 6. Meningitis 2,722 1,405 1,317 937 154 212 74 29 631 234 389 36 27 7. Septicaemia 3,987 2,244 1,743 224 49 719 546 704 144 25 405 231 938 8. Arbovirus infection 1,272 658 614 2 58 544 42 13 8 61 506 25 15 (continued) 214 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+a. Ross River virus 649 307 342 1 26 256 18 5 1 27 292 13 8 b. Barmah Forest virus 253 126 128 8 106 8 4 9 111 5 2 c. Dengue 5 3 2 1 2 1 d. Other arbovirus infection 364 222 142 23 180 15 4 6 24 101 6 4 9. Hepatitis 19,889 13,072 6,817 74 36 8,848 2,456 1,659 30 30 3,970 1,305 1,482 a. Hepatitis A 51 26 25 5 3 11 1 7 3 3 8 1 10 b. Hepatitis B(b) 6,961 4,429 2,532 45 31 2,430 967 956 13 16 1,074 530 899 c. Hepatitis C(c) 12,723 8,509 4,214 15 2 6,308 1,488 696 13 12 2,887 775 528 d. Other hepatitis 154 108 46 9 98 46 10. Malaria 89 60 29 30 29 1 28 11. Trachoma 121 55 66 1 42 11 1 1 49 14 2 12. Other infectious and parasitic diseases 4,835 2,247 2,588 288 80 694 545 640 297 140 843 518 790 B. Acute respiratory infections 35,502 17,217 18,285 3,388 833 4,461 2,078 6,456 2,790 851 3,791 1,635 9,219 1. Lower respiratory tract infections 27,354 13,121 14,233 1,067 298 3,360 2,001 6,395 798 268 2,519 1,511 9,137 2. Upper respiratory tract infections 3,451 1,614 1,837 615 282 618 56 43 615 359 731 82 50 3. Otitis media 4,697 2,482 2,215 1,706 254 484 22 17 1,377 223 541 42 33 C. Maternal conditions 2,152 2,152 1 434 1,716 1. Maternal haemorrhage 126 126 19 108 2. Maternal sepsis 332 332 1 95 236 3. Hypertensive disorders of pregnancy 887 887 1 204 683 4. Obstructed labour 147 147 24 123 5. Abortion 25 25 12 14 6. Other maternal conditions 634 634 82 552 D. Neonatal causes 34,558 19,027 15,531 19,027 15,530 1. Birth trauma and asphyxia 9,308 5,086 4,221 5,086 4,221 (continued) 215 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+2. Low birthweight 15,423 8,281 7,142 8,281 7,142 3. Neonatal infections 3,404 2,156 1,248 2,156 1,248 4. Other conditions arising in the perinatal period 6,424 3,505 2,919 3,505 2,919 E. Nutritional deficiencies 6,197 1,449 4,748 417 73 462 251 245 421 746 2,858 287 4351. Protein-energy malnutrition 97 33 64 1 1 31 1 14 48 2. Deficiency anaemia 6,011 1,368 4,643 387 73 461 238 208 421 746 2,842 259 376 3. Other nutritional deficiencies 89 48 42 30 12 6 1 17 14 10 II. Non-communicable diseases 2,324,625 1,170,116 1,154,509 90,683 72,481 501,687 232,155 273,110 69,322 83,085 472,356 175,985 353,761 F. Malignant neoplasms 499,416 264,382 235,034 2,512 2,530 115,797 77,316 66,226 1,577 1,926 117,559 53,828 60,1441. Mouth and oropharynx cancers 13,464 9,483 3,981 36 122 5,902 2,226 1,198 2 53 1,984 910 1,032 2. Oesophagus cancer 14,163 9,983 4,180 29 5,044 2,933 1,977 1,292 1,190 1,698 3. Stomach cancer 15,218 9,073 6,145 1 3 4,120 2,788 2,162 31 2,661 1,388 2,064 4. Colorectal cancer 63,605 34,643 28,962 2 46 15,622 10,531 8,442 2 52 11,693 7,513 9,703 5. Liver cancer(d) 4,716 3,241 1,474 15 2 1,633 948 643 14 12 648 333 468 6. Gallbladder cancer 3,549 1,429 2,121 601 500 327 752 633 735 7. Pancreas cancer 22,680 11,434 11,246 5,415 3,413 2,606 1 4,172 3,023 4,050 8. Lung cancer 88,904 55,028 33,876 62 63 22,112 19,258 13,533 1 30 14,848 9,937 9,059 9. Bone and connective tissue cancer 5,879 3,317 2,562 315 666 1,536 419 380 212 357 1,388 276 329 10. Melanoma 20,236 13,734 6,501 5 238 8,342 2,836 2,313 2 53 3,450 1,519 1,478 11. Non-melanoma skin cancers 4,734 3,233 1,502 2 1,208 933 1,090 391 246 864 12. Breast cancer 60,654 134 60,520 87 23 24 25 41,056 10,445 8,995 13. Cervix cancer 5,231 5,231 24 3,738 741 727 14. Corpus uteri cancer 4,663 4,663 2,448 1,174 1,041 15. Ovary cancer 11,994 11,994 11 164 6,429 2,631 2,758 (continued) 216 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+16. Prostate cancer 36,547 36,547 9,112 11,950 15,484 17. Testicular cancer 862 862 6 123 713 10 11 18. Bladder cancer 10,077 7,010 3,068 1 4 2,046 2,133 2,827 28 598 770 1,671 19. Kidney cancer 12,487 7,794 4,694 47 6 4,128 2,092 1,521 53 2 1,618 1,429 1,592 20. Brain cancer 19,792 11,515 8,276 721 515 7,617 1,693 970 543 194 4,959 1,538 1,043 21. Thyroid cancer 1,762 640 1,122 14 305 201 120 4 45 675 159 237 22. Lymphoma 22,263 12,375 9,888 173 232 6,212 3,161 2,597 27 318 4,029 2,382 3,132 23. Multiple myeloma 8,925 4,778 4,147 30 1,824 1,437 1,487 1 1,343 1,216 1,587 24. Leukaemia 19,956 11,393 8,563 785 444 4,841 2,753 2,570 542 342 3,293 1,909 2,477 25. Larynx cancer 3,751 3,263 488 1,644 1,059 560 237 134 117 26. Eye cancer 952 530 422 39 12 279 125 76 38 12 197 67 108 27. Other malignant neoplasms 22,354 12,945 9,409 276 11 5,455 3,895 3,309 125 181 3,659 2,262 3,183 G. Other neoplasms 10,903 4,615 6,288 155 237 1,377 1,180 1,666 237 45 2,998 1,057 1,9511. Uterine myomas 1,545 1,544 4 1,447 70 23 2. Benign neoplasms of meninges and brain 1,451 518 934 42 7 218 98 153 21 6 495 203 209 3. Other benign neoplasms 7,907 4,097 3,810 113 230 1,159 1,082 1,513 215 35 1,056 784 1,719 H. Diabetes mellitus 143,831 77,437 66,394 975 681 48,711 15,183 11,887 911 802 36,213 12,109 16,3591. Type 1 diabetes 10,891 6,260 4,631 872 548 3,236 980 625 781 404 1,825 592 1,028 2. Type 2 diabetes 132,940 71,176 61,763 103 133 45,476 14,203 11,262 130 398 34,388 11,517 15,330 I. Endocrine and metabolic disorders 28,565 14,556 14,010 3,395 791 5,470 1,972 2,928 2,162 800 4,600 1,815 4,633 1. Non-deficiency anaemia 5,109 2,739 2,370 917 42 787 399 594 636 36 594 332 773 a. Haemolytic anaemia 1,313 774 539 689 2 14 13 56 476 1 2 14 46 b. Other non-deficiency anaemia 3,797 1,965 1,832 228 41 773 385 538 160 35 592 318 727 2. Cystic fibrosis 1,863 926 937 520 140 263 3 492 244 201 3. Haemophilia 205 169 37 59 56 11 43 2 18 16 (continued) 217 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+4. Other endocrine and metabolic disorders 21,387 10,722 10,665 1,899 608 4,364 1,563 2,288 1,033 520 3,803 1,465 3,844 J. Mental disorders 350,545 165,676 184,869 28,633 48,387 82,282 4,711 1,662 21,492 47,683 112,469 1,878 1,3471. Substance use disorders 60,782 46,094 14,687 3 14,711 28,148 2,397 836 79 4,180 9,680 417 331 a. Alcohol dependence and harmful use(e) 34,116 27,225 6,891 4,848 19,181 2,378 817 416 5,749 395 331 b. Heroin or polydrug dependence and harmful use 16,839 12,455 4,383 3 5,657 6,776 14 6 78 2,052 2,233 20 c. Benzodiazepine dependence and harmful use 2,656 1,102 1,554 207 892 3 362 1,189 2 d. Cannabis dependence and harmful use 5,206 4,075 1,131 3,520 554 1 983 148 e. Other drug dependence and harmful use 1,966 1,237 729 478 745 14 367 361 2. Schizophrenia 27,502 14,785 12,717 186 9,795 4,719 25 60 181 3,754 8,639 53 90 3. Anxiety and depression 191,786 65,321 126,464 9,554 17,868 36,126 1,430 343 15,507 29,946 80,515 321 175 4. Bipolar disorder 7,770 3,920 3,849 2,672 1,246 2 2,450 1,347 30 23 5. Personality disorders(f) 32,587 16,248 16,339 3,130 11,955 816 347 2,622 12,044 1,032 642 6. Eating disorders 6,062 375 5,687 103 211 52 9 828 4,639 200 19 a. Anorexia nervosa 2,933 367 2,567 103 211 52 407 2,063 91 5 b. Bulimia nervosa 3,087 3,087 421 2,576 90 c. Other eating disorders 41 9 33 9 19 14 7. Childhood conditions 23,794 18,804 4,990 18,785 19 4,896 93 a. Attention-deficit hyperactivity disorder 9,928 7,082 2,846 7,082 2,840 6 b. Autism spectrum disorders 13,866 11,722 2,144 11,703 19 2,056 88 8. Other mental disorders 262 127 135 1 17 42 66 43 26 67 (continued) 218 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+K. Nervous system and sense organ disorders 312,766 146,645 166,121 8,850 7,613 51,886 32,573 45,723 6,744 9,296 45,104 30,004 74,973 1. Dementia 94,399 33,653 60,747 45 42 4,599 7,872 21,095 155 32 3,340 10,236 46,984 2. Epilepsy 14,821 8,479 6,342 3,249 1,209 3,430 353 237 2,446 848 2,248 336 464 3. Parkinsons disease 26,852 13,664 13,189 3,459 3,958 6,247 2,759 4,979 5,451 4. Multiple sclerosis 5,252 1,609 3,642 18 101 1,345 119 27 79 201 3,112 110 140 5. Motor neurone disease 7,088 3,696 3,392 1 1 1,896 1,124 674 33 1,394 1,114 851 6. Huntingtons chorea 1,779 937 842 35 758 103 41 7 631 124 81 7. Muscular dystrophy 1,046 801 244 221 318 227 35 99 29 64 34 18 8. Sense organ disorders 112,728 63,316 49,412 383 1,073 29,229 17,192 15,439 237 857 18,678 11,675 17,964 a. Glaucoma-related blindness 3,671 1,698 1,974 868 586 244 866 694 414 b. Cataract-related blindness 2,343 883 1,460 5 2 139 228 510 3 1 153 337 966 c. Macular degeneration 11,642 4,383 7,259 13 1,132 3,238 14 1,338 5,906 d. Adult-onset hearing loss 64,853 42,653 22,200 699 22,983 11,920 7,052 432 12,315 5,834 3,618 e. Refractive errors 18,761 8,241 10,520 224 286 2,697 1,941 3,094 90 343 2,861 2,107 5,119 f. Other vision loss 11,457 5,457 5,999 154 87 2,529 1,386 1,301 143 81 2,470 1,364 1,941 9. Migraine 21,848 5,972 15,875 1,523 3,539 910 1 955 6,217 8,671 15 17 10. Other nervous system and sense organ disorders 26,953 14,518 12,435 3,411 1,294 6,034 1,815 1,964 2,741 1,104 4,206 1,381 3,004 L. Cardiovascular disease 473,794 252,405 221,389 2,112 2,414 94,217 59,839 93,822 1,632 1,324 45,697 38,727 134,0091. Rheumatic heart disease 4,091 1,371 2,720 5 65 585 284 432 33 63 809 702 1,112 2. Ischaemic heart disease 263,497 151,107 112,390 35 322 57,210 37,860 55,680 13 120 20,352 21,052 70,853 3. Stroke 118,462 53,296 65,166 1,436 1,128 17,961 10,938 21,834 984 480 14,237 9,635 39,830 4. Inflammatory heart disease 15,904 10,134 5,771 419 305 5,207 2,078 2,125 428 156 2,066 1,181 1,939 5. Hypertensive heart disease 8,982 3,768 5,213 6 5 1,018 885 1,855 7 5 637 708 3,856 (continued) 219 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+6. Non-rheumatic valvular disease 8,951 4,367 4,584 29 141 1,390 912 1,896 26 80 887 772 2,820 7. Aortic aneurysm 11,338 7,189 4,149 59 1,871 2,187 3,071 31 29 580 916 2,594 8. Peripheral vascular disease 18,606 10,604 8,002 50 74 4,816 2,639 3,026 21 93 2,592 1,519 3,777 9. Other cardiovascular disease 23,962 10,569 13,394 133 315 4,161 2,058 3,903 90 297 3,537 2,242 7,228 M. Chronic respiratory disease 186,737 98,925 87,813 23,093 1,936 31,452 17,475 24,968 16,944 6,925 26,552 13,854 23,537 1. Chronic obstructive pulmonary disease (COPD) 86,751 49,201 37,550 378 294 21,936 11,693 14,900 174 278 14,923 8,855 13,318 2. Asthma 63,100 29,271 33,828 21,953 1,314 4,802 738 465 16,490 6,641 8,069 1,412 1,216 3. Other chronic respiratory diseases 36,887 20,453 16,435 762 329 4,715 5,044 9,603 280 5 3,560 3,587 9,003 N. Diseases of the digestive system 57,957 28,613 29,344 1,204 1,281 14,092 5,083 6,953 799 1,134 11,792 4,535 11,084 1. Peptic ulcer disease 6,358 3,292 3,065 32 39 1,622 662 937 6 1,148 334 1,577 2. Cirrhosis of the liver(g) 1,524 687 838 31 17 277 112 249 1 3 178 87 569 3. Appendicitis 648 324 323 53 59 135 13 64 41 60 154 30 38 4. Intestinal obstruction 5,019 2,227 2,792 61 18 665 582 902 10 18 927 381 1,455 5. Diverticulitis 6,118 2,829 3,289 6 1,373 701 749 1 1,072 921 1,296 6. Gallbladder and bile duct disease 3,202 1,212 1,990 2 6 395 359 450 3 55 852 295 785 7. Pancreatitis 2,501 1,464 1,037 2 38 921 232 273 2 37 498 151 348 8. Inflammatory bowel disease 12,176 6,334 5,843 553 1,001 4,369 264 148 523 854 4,044 227 195 9. Vascular insufficiency of bowel 3,982 1,647 2,335 125 29 463 365 664 32 35 557 592 1,119 10. Other digestive system diseases 16,430 8,597 7,832 346 69 3,872 1,792 2,518 186 64 2,362 1,518 3,702 (continued) 220 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+O. Genitourinary diseases 65,249 28,163 37,086 127 1,637 11,461 5,791 9,148 1,068 7,860 14,951 3,273 9,9351. Nephritis and nephrosis(h) 21,133 10,688 10,444 106 107 2,808 1,933 5,734 46 146 1,944 1,632 6,677 2. Benign prostatic hypertrophy 7,622 7,622 2,723 2,950 1,949 3. Urinary incontinence 8,263 1,823 6,440 898 542 383 1 217 4,271 1,053 898 4. Infertility 14,344 6,268 8,076 21 1,502 4,746 19 1,822 6,236 5. Other genitourinary diseases 13,888 1,762 12,126 28 286 365 1,082 1,002 5,676 2,500 589 2,360 P. Skin diseases 20,302 9,852 10,451 1,446 1,679 4,555 1,126 1,045 1,593 1,778 2,672 1,408 3,0001. Eczema 2,730 1,031 1,699 371 47 555 31 27 1,210 42 413 31 2 2. Acne 3,899 1,988 1,910 646 1,013 329 242 1,198 470 3. Psoriasis 4,021 3,122 899 206 578 2,059 174 105 58 192 524 76 49 4. Ulcers 9,324 3,620 5,704 222 41 1,575 886 895 82 346 1,177 1,235 2,864 5. Other skin diseases 329 90 238 1 37 35 18 88 65 84 Q. Musculoskeletal diseases 105,508 44,210 61,298 856 1,289 27,639 8,375 6,052 1,305 1,639 35,570 11,574 11,2111. Rheumatoid arthritis 16,841 4,780 12,062 343 214 2,833 888 502 958 513 7,658 1,710 1,222 2. Osteoarthritis 34,578 14,495 20,083 1 58 7,772 3,863 2,802 7,356 6,088 6,638 3. Back pain(i) 29,658 14,470 15,188 275 541 9,776 2,227 1,650 206 610 10,704 2,012 1,657 4. Slipped disc 6,120 3,439 2,681 13 144 2,711 386 184 29 84 1,956 401 211 5. Occupational overuse syndrome 4,953 697 4,256 9 663 24 65 4,177 13 1 6. Systemic lupus erythematosus (SLE) 1,609 168 1,441 1 43 56 68 1 76 984 186 193 7. Gout 1,988 1,636 352 2 85 1,330 100 119 1 59 131 97 64 8. Other musculoskeletal diseases 9,759 4,525 5,235 222 236 2,511 829 726 109 232 2,605 1,066 1,223 R. Congenital anomalies 33,228 18,770 14,458 14,738 624 2,688 345 374 10,838 528 2,172 439 4811. Anencephaly 387 102 285 102 285 2. Spina bifida 812 408 404 307 31 57 12 270 30 105 3. Congenital heart disease 8,394 4,975 3,419 3,434 282 1,091 97 71 2,202 268 723 135 90 (continued) 221 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+4. Cleft lip and/or palate 221 112 109 112 109 5. Digestive system malformations 493 244 248 207 16 12 8 222 1 12 14 a. Anorectal atresia 31 17 14 17 14 b. Oesophageal atresia 31 19 12 19 12 c. Other digestive system malformations 431 208 223 171 16 12 8 196 1 12 14 6. Urogenital tract malformations 2,575 1,560 1,016 568 1 547 168 276 144 393 191 288 a. Renal agenesis(j) 279 153 126 145 1 7 82 29 11 3 b. Other urogenital tract malformations(k) 2,296 1,407 890 422 540 168 276 62 364 180 284 7. Abdominal wall defect 312 210 102 210 102 8. Down syndrome 3,808 2,181 1,627 1,668 61 429 22 1,059 2 470 79 18 9. Other chromosomal disorders 8,493 4,685 3,807 4,682 1 2 3,754 1 52 10. Other congenital anomalies 7,733 4,293 3,440 3,448 248 546 32 19 2,692 226 429 21 71 S. Oral conditions 24,507 11,402 13,105 1,114 1,098 7,359 1,186 645 1,062 1,065 8,490 1,470 1,0171. Dental caries 12,088 6,026 6,061 665 789 3,860 427 285 631 760 3,819 375 476 2. Periodontal disease 581 280 301 5 20 230 19 7 5 19 237 30 9 3. Edentulism 5,264 1,880 3,384 2 7 1,166 526 179 3 12 2,281 836 252 4. Pulpitis 6,497 3,197 3,300 443 283 2,103 214 155 424 274 2,127 228 248 5. Other oral conditions 77 18 59 18 26 32 Z. Ill-defined conditions 11,317 4,467 6,850 1,470 283 2,701 13 958 280 5,517 14 811. Sudden infant death syndrome 2,428 1,470 958 1,470 958 2. Chronic fatigue syndrome 8,890 2,997 5,893 283 2,701 13 280 5,517 14 81 (continued) 222 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+III. Injuries 185,050 129,504 55,546 9,290 28,191 80,233 5,681 6,109 7,018 7,861 24,949 4,480 11,238T. Unintentional injuries 125,862 84,201 41,661 8,695 19,148 46,795 4,154 5,408 6,293 5,662 14,745 3,963 10,9971. Road traffic accidents 42,425 31,028 11,397 1,991 10,380 17,215 838 605 1,336 3,572 5,253 621 616 2. Other transport accidents 8,601 6,782 1,819 779 1,756 3,996 177 74 556 316 815 62 70 3. Poisoning 12,046 6,922 5,124 54 927 5,501 230 210 55 463 2,722 691 1,194 4. Falls 26,386 13,118 13,269 1,552 1,717 5,171 1,490 3,188 1,086 379 2,119 1,870 7,814 5. Fires, burns and scalds 4,399 2,822 1,577 786 279 1,499 154 103 564 63 775 67 108 6. Drowning 4,812 3,366 1,447 646 672 1,854 114 79 706 94 532 81 34 7. Sports injuries 579 344 234 71 112 147 8 6 44 44 96 16 35 8. Natural and environmental factors 1,927 1,330 597 163 277 780 53 57 190 98 182 54 72 9. Machinery accidents 5,095 4,725 370 214 957 3,255 227 71 37 47 270 11 4 10. Other unintentional injuries(l) 19,591 13,765 5,827 2,440 2,070 7,378 862 1,015 1,718 586 1,981 491 1,050 Suffocation and foreign bodies 5,727 3,930 1,797 736 606 2,133 172 283 734 181 529 55 298 Adverse effects of medical treatment 3,695 2,016 1,678 96 124 812 405 581 55 127 580 347 570 Other unintentional injuries n.e.c. 10,169 7,818 2,351 1,608 1,340 4,433 285 152 929 278 872 89 182 U. Intentional injuries 59,189 45,303 13,886 594 9,043 33,438 1,527 701 726 2,199 10,204 517 2401. Suicide and self-inflicted injuries 49,916 38,717 11,199 176 7,320 29,099 1,437 685 208 1,479 8,854 467 191 2. Homicide and violence 9,221 6,535 2,686 418 1,722 4,289 90 16 518 721 1,349 50 49 3. Legal intervention and war 51 51 1 50 Australian population (000) 19,881 9,872 10,010 2,041 1,404 5,292 656 478 1,938 1,349 5,311 694 718 DALYs per 1,000 population 132.4 138.2 126.7 61.2 73.0 114.1 372.3 605.0 49.9 69.7 96.8 266.1 526.9 (continued) 223 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+Risk factors Alcohol 61,091 52,180 8,911 737 11,648 38,139 2,867 1,211 237 1,518 10,704 296 3,252 Illicit drugs 51,463 36,515 14,948 78 11,892 21,161 2,109 1,276 67 4,246 8,579 1,067 989 Tobacco 204,788 131,616 73,172 2,238 88 62,414 35,879 30,997 1,757 22 28,699 18,915 23,780 Unsafe sex 14,897 6,217 8,679 22 303 5,022 514 357 60 493 5,990 1,004 1,132 Child sexual abuse 23,513 4,166 19,348 704 3,204 173 84 4,309 14,597 209 232 Intimate partner violence 29,360 29,360 5,455 22,325 901 678 Occupational exposures & hazards 51,362 35,492 15,870 2,853 25,291 4,245 3,104 1,387 12,458 1,097 928 Physical inactivity 174,431 87,742 86,689 147 42,424 21,262 23,909 166 32,596 17,172 36,756 High blood pressure 199,315 107,098 92,218 33,296 28,717 45,085 12,100 18,236 61,882 High body mass 197,632 105,616 92,017 65,684 22,683 17,248 46,520 20,534 24,963 Low fruit and vegetable consumption 55,259 36,429 18,830 103 18,804 9,111 8,411 38 6,272 4,150 8,370 High blood cholesterol 163,591 89,669 73,922 45,316 19,718 24,635 17,979 14,247 41,695 Osteoporosis 4,386 1,019 3,368 18 128 873 23 209 3,135 Air pollution - short term 7,781 4,032 3,750 70 37 981 1,010 1,935 54 8 629 702 2,356 Particulates 3,807 1,976 1,831 19 11 590 470 885 12 4 274 323 1,219 Ozone 3,974 2,056 1,918 50 25 391 539 1,050 42 4 355 379 1,138 Air pollution - long term 19,738 10,422 9,316 4,097 2,768 3,557 2,280 1,740 5,296 Joint effect of all risk factors 847,307 478,511 368,796 3,075 27,569 242,892 99,049 105,926 2,122 15,726 155,198 63,319 132,432 Alternative burden of disease categories Diabetes mellitus (attributable) 218,518 112,615 105,904 977 688 56,122 24,190 30,637 913 805 39,282 17,720 47,183 Anxiety and depression (attributable) 215,783 80,770 135,013 9,560 19,056 48,235 2,620 1,298 15,526 30,441 86,556 1,134 1,356 All intellectual disability 44,187 22,822 21,365 18,743 666 3,043 228 141 18,591 270 1,935 244 326 All vision loss 55,539 26,828 28,711 5,354 404 6,746 5,548 8,775 384 425 6,834 6,109 14,959 All nephritis and nephrosis 68,721 37,691 31,030 519 1,029 15,723 8,514 11,905 215 682 8,480 6,281 15,371 (continued) 224 Annex Table 3 (continued): Disability-adjusted life years (DALYs) by age, sex and cause, Australia, 2003 Notes (a) Excludes HIV/AIDS. (b) Includes hepatitis B-related liver cancer and cirrhosis. (c) Includes hepatitis C-related liver cancer and cirrhosis. (d) Excludes liver cancer related to hepatitis B and C. (e) Includes alcoholic cirrhosis. (f) Excludes those with any other comorbid mental disorders. (g) Excludes alcoholic and hepatic cirrhosis. (h) Excludes diabetic-, congenital- and poisoning-related renal failure. (i) Includes both acute and chronic back pain. (j) Includes renal failure due to dysplasia. (k) Includes polycystic renal failure. (l) Includes suffocation and foreign bodies, adverse effects of medical treatment, other mechanical force injuries and other unintentional injuries. 225 Annex Table 4: Deaths by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+All causes 132,287 68,325 63,962 1,004 1,071 15,483 14,039 36,728 776 399 8,895 8,295 45,597I. Communicable diseases,maternal and neonatal conditions 6,847 3,464 3,383 405 15 734 496 1,813 314 14 340 267 2,449 A. Infectious and parasitic diseases 2,416 1,465 952 28 10 606 319 501 23 10 254 150 5141. Tuberculosis 52 26 26 5 5 16 2 3 21 2. Sexually transmitted diseases(a) 12 2 10 2 1 4 5 a. Syphilis 5 2 3 2 1 1 1 b. Chlamydia 4 4 2 2 c. Gonorrhoea d. Other sexually transmitted diseases 3 3 1 2 3. HIV/AIDS 119 108 11 94 12 1 10 1 4. Diarrhoeal diseases 48 18 29 1 3 3 11 1 3 25 5. Childhood immunisable diseases 27 16 11 1 5 6 4 2 1 3 5 a. Diphtheria b. Whooping cough c. Tetanus d. Poliomyelitis 20 12 8 2 6 4 1 3 4 e. Measles f. Rubella g. Haemophilus influenzaetype b (Hib) 7 4 3 1 3 2 1 6. Meningitis 61 29 32 11 4 5 5 4 9 6 12 2 3 7. Septicaemia 304 156 149 5 1 23 35 91 3 9 11 125 8. Arbovirus infection a. Ross River virus (continued) 226 Annex Table 4 (continued): Deaths by age, sex and cause, Australia, 2003 Males Females Cause Persons Males Females 014 1524 2564 6574 75+ 014 1524 2564 6574 75+b. Barmah Forest virus c. Dengue d. Other arbovirus infection 9. Hepatitis 1,455 933 521 2 1 439 212 280 1 1 187 100 233 a. Hepatitis A 2 1 1 1 1 b. Hepatitis B(b) 625 381 244 1 1 121 83 175 49 41 154 c. Hepatitis C(c) 824 550 274 317 129 104 138 59 76 d. Other hepatitis 3 1 2 1 2 10. Malaria 3 2 1 1 1 1 11. Trachoma 12. Other infectious and parasitic diseases 335 174 161 8 2 30 41 92 6 3 28 27 97 B. Acute respiratory infections 3,724 1,630 2,095 23 5 128 176 1,297 19 1 78 110 1,8861. Lower respiratory tract infections 3,709 1,624 2,085 21 5 126 176 1,295 18 1 76 110 1,880 2. Upper respiratory tract infections 9 3 6 1 2 1 1 4 3. Otitis media 6 3 3 2 1 1 2 C. Maternal conditions 9 9 2 7 1. Maternal haemorrhage 1 1 1 |
|
|
|