VicHealth commissioned Deakin Health Economics (DHE) to undertake this research project into the economic benefits of reducing disease risk factors in mid 2008. DHE and the National Stroke Research Institute formed a partnership to complete this project.

Six potentially modifiable risk factors were nominated for assessment:

1. intimate partner violence (IPV),

2. high risk alcohol consumption,

3. inadequate fruit and vegetable consumption,

4. physical inactivity,

5. tobacco smoking; and

6. high body mass index (BMI).

The research was completed in 2008 and the reports finalised, following review, in early 2009.

The beneficiaries of these ‘health status’, ‘economic’ and ‘financial’ benefits are made up of individuals, businesses and government. Government should benefit through future savings in health care expenditure on treatments for preventable disease, through increased taxation transfers from higher individual incomes and through fewer welfare payments. Businesses should benefit from reduced absenteeism from work and from less recruitment and training costs associated with replacing staff that die or retire prematurely from ill health. Individuals should benefit from increases in income, from reduced absenteeism from work or time spent out of role at home and from increased quality of life from reduced levels of ill health.

The research methods used to estimate the potential benefits were:

1. A detailed literature review on each risk factor to provide the theoretical background and context for assumptions and estimates used in the prevalence scenarios modelled;
2. Regular consultation with external and independent experts through an Advisory Committee convened by the Victorian Health Promotion Foundation (VicHealth);
3. Direct attainment and use of databases from the Australian Bureau of Statistics (ABS) and the 2003 Australian Burden of Disease (BoD) Study to ensure consistent national data inputs for each risk factor;
4. Redevelopment of existing workforce productivity decision analytic models prepared as part of an earlier research project completed for the Victorian Department of Treasury and Finance;
5. Use of both the Human Capital Approach (Human Capital Approach) and the Friction Cost Approach (Friction Cost Approach) to estimate production gains/losses in the general economy;
6. Development of new decision analytic models for the estimation of household production and leisure time costs;
7. Use of probabilistic multivariable uncertainty analyses to improve the precision and reliability of primary outcome variables and provide 95% uncertainty intervals (refer section 2.8);
8. Adjustment for the joint effects of multiple risk factors, since it is more common for people to report two or more risk factors than one risk factor; and
9. The analysis was limited to the prevention of new cases of disease attributable to the six risk factors over the lifetime of the 2008 population.

The risk factor prevalence scenarios were modelled separately for each risk factor using best available evidence to inform decisions on what constituted realistic and feasible reductions. However, the decision on what constituted ‘best available’ evidence, varied between the risk factors. In the estimations for alcohol, tobacco and intimate partner violence it was agreed that the feasible reductions should be modelled against attainment of prevalence levels observed in a comparable country (referred to as an ‘Arcadian’ ideal).

In contrast, for inadequate consumption of fruit and vegetables, high BMI and physical inactivity, a consensus approach informed by best available evidence was preferred. For these risk factors, the Advisory Committee agreed that international comparisons were too problematic, mainly due to country specific socio-economic and cultural variations. Irrespective of which approach was used to provide the basis of the ‘feasible’ reduction in risk factor prevalence targets, systematic methods and data sources were applied to ensure comparability. The systematic approach also facilitated correction for co-morbidities across risk factors.

Overall, large potential opportunity cost savings from the avoidable disease burden are possible if we achieve the ‘feasible’ reductions in the prevalence of the nominated risk factors. Over the lifetime of the 2008 Australian adult population, opportunity cost savings were conservatively estimated to be $2,334 million (using Friction Cost Approach) or $3,057 million (using Human Capital Approach). The total opportunity cost savings are the sum of the health sector offsets and the combined workforce, household and leisure production effects (see column with mean estimates in Table 1).

Notes: The total opportunity cost savings are the sum of the health sector offsets and the combined workforce, household and leisure production effects. The mean estimates can be added together in this way, but not the uncertainty intervals as both the components and the total are run as independent distributions. These financial outcomes are opportunity cost estimates and not immediately realisable cash savings. No probabilistic uncertainty analysis was conducted for health sector offsets. Taxation is treated as a transfer payment and should not be added to production effects or health sector offsets. Human Capital Approach: Human Capital Approach; Friction Cost Approach Friction Cost Approach (preferred conservative estimate). Leisure and home based production estimates are based on persons 15+ years. Production gains/(losses) and taxation effects are based on persons 15-64 years. LL: lower limit; UL: upper limit. Recruitment and training costs are included in production gains/losses using the Friction Cost Approach but not counted using the Human Capital Approach. Values are net present value using a 3% discount rate. Numbers in brackets ( ) indicate the possibility of losses resulting from achieving the target, rather than gains.

The upper and lower limits of the uncertainty interval indicate the range of possible values that these estimates might be. The wider the interval the larger the degree of uncertainty there is around the estimates. The presence of negative numbers at the lower limit indicates there is some chance (albeit small) of a financial loss occurring.

For individual risk factors, the relative value of potential opportunity cost savings varied. This reflected the difference in prevalence of the risk factors; the age and gender distributions of the population with the risk factor; the size of the realistic targets selected; the impact of the risk factor on health status and the quality of the data to inform differences between individuals with and without the risk factor of interest, such as number of days off work from ill health. The largest potential opportunity cost savings estimated using Friction Cost Approach, could be gained from reductions in alcohol consumption, followed by reductions in tobacco smoking, IPV, physical inactivity, BMI and lastly from increases in the consumption of fruit and vegetables (Table 2).

Notes: Friction Cost Approach: Friction Cost Approach for valuing workforce production gains/ (losses); Human Capital Approach Human Capital Approach to valuing workforce production gains/(losses). IPV: Intimate Partner Violence; F&V: Fruit and Vegetables; BMI: Body Mass Index; JE: Joint Effects. Values are net present value using a 3% discount rate.

While the opportunity cost savings for each individual risk factor can be compared to each other in size, they cannot be simply added together to determine the overall potential savings. This would lead to a serious overestimation of benefits. If several analyses were added together it could appear as if more than 100% of the burden for any one disease or injury was being accounted for by the risk factors in combination. This is illustrated in Table 2 where the potential opportunity costs savings results for the combined risk factors are provided.

The total economic and health status benefits of achieving the ideal targets (with adjustment for joint effects) are presented in Table 3. The largest potential savings in terms of lost days prevented are estimated to occur in the workforce, followed by home based production and leisure time. Where uncertainty was able to be estimated, it clearly indicates the precision of the estimates as well as the possibility of losses occurring. The chance of loss associated with a number of particular factors, including data quality, and risk factor by risk factor is detailed in individual risk factor chapters.

Notes: Disability Adjusted Life Years (DALYs), incidence of disease and mortality were calculated for all age groups. Leisure and home based production were calculated for persons aged 15+ years. Absenteeism and early retirement were calculated for persons aged 15-64 years. LL: lower limit; UL: upper limit. Values are net present value using a 3% discount rate.

The spread of the potential health benefits in DALYs from avoidable disease related to each of the risk factors was consistent with the potential total production gains in Table 2. That is, the greatest health gains in DALYs could be achieved from reducing high risk alcohol consumption and tobacco smoking. This is because alcohol consumption and tobacco smoking are associated with a larger number of fatal and non-fatal diseases compared to the other risk factors. Individual risk factor chapters in this report provide further detail on the absolute health benefits.

The proportionate gains in household production, leisure time, workforce participation, health expenditure and fewer incident cases of disease and deaths did vary amongst each of the risk factors (Table 4). For example, achieving the ideal feasible reduction in physical inactivity prevalence would create more household productivity and leisure time than reductions in alcohol consumption which, in turn, had a greater influence on workforce productivity. These findings reflect the differences in workforce status, age and gender distributions in each of the populations at risk.

Notes: Friction Cost Approach: Friction Cost Approach for valuing workforce production gains/(losses); IPV: Intimate Partner Violence; F&V: Fruit and Vegetables; BMI: Body Mass Index; JE: Joint Effects. Values are net present value using a 3% discount rate.

This research provides a wealth of new information to inform policy decisions on public health strategies to prevent chronic disease in Australia.

The key messages are:

i)         the potential benefits of reducing risk factor prevalence are substantial;

ii)        the gains vary by risk factor and reduction target considered; and

iii)       further research is required to reduce the uncertainty surrounding the estimates.

The main strengths of the research reported here are the consistent methods and data sources applied, including comprehensive assessment by age, gender and workforce status to account for variations within each risk factor population.

There are a number of major assumptions to note.

The first was our current reliance on the accuracy of self-reported cross sectional data to identify the association between the presence of risk factors and the amount of time and its use away from work due to ill health. Assuming causality between risk factors, illness, absenteeism and workforce participation in the absence of rigorous longitudinal data, means that our results must be regarded as broadly indicative, rather than authoritative, until further testing and validation of causal relationships can be completed.

A second major assumption was the adoption of an incidence based approach to the measurement of health benefits (i.e. looking at new cases avoided, but not health benefits among people who were already ill). This approach was necessary to complete the research in the time available for the project, but biases the estimates in a conservative direction. Offsetting this bias, was the omission of any time lag effects in the modelling between the reduction in the prevalence of a risk factor, the assumed reversal of the elevated risk of disease and consequential reductions in the incidence of diseases associated with that risk factor. The omission of time lags biases the results in an optimistic direction and means we are unable to specify exactly when the benefits will be realised.

A fourth major assumption was our preference for the adoption of the Friction Cost Approach, rather than Human Capital Approach, as the preferred method for measuring workforce production gains and losses. We present estimates, however, using both the Human Capital Approach and Friction Cost Approach, as the adoption of one approach rather than the other has a dramatic effect upon the results and there are valid arguments for adopting both methods. The essence of our position is that the Friction Cost Approach is more suitable for answering the research question we were charged with – that is, for estimating production gains/losses in the general economy.  For this question it was important to us to take into account the fact that businesses will adjust to short term and long term absences. Further, we argue that the Human Capital Approach is more suited to answering a different research question – that is, placing a monetary value on human life, where the total forgone income stream due to premature death provides a sensible floor estimate (refer section 2.4.5).

In addition to these assumptions, it should be noted that there is a paucity of effectiveness evidence for specific interventions to adequately inform judgements about feasible reductions in prevalence of many risk factors. Future research in this area is an important way forward if we are to have better evidence for prioritising specific interventions to achieve risk factor prevalence reductions and to support further modelling for health promotion.

Finally, we recommend that caution in the interpretation of the presented ‘opportunity cost savings’ is necessary for three main reasons. Firstly, these benefits will only be achieved by the adoption of effective interventions that will certainly have implementation costs attached to them. Inclusion of intervention costs was not a part of our brief for this analysis and we have assumed that effective interventions exist to achieve the target reductions in prevalence of the risk factors. One might argue that the large potential opportunity cost savings we found could enable the upfront investment in suitable prevention interventions.

Secondly, we have assessed the benefits as occurring over the future lifetime of the 2008 population. The estimates we present are particularly conservative in that they represent the benefits for a single population group. Similar benefits would be expected in subsequent cohorts if the same magnitudes of risk factor reductions could be achieved in future years. However, the prevalence of each risk factor would fall in each future year. Thirdly, the opportunity cost savings are not estimates of immediately realisable financial savings, but rather estimates of resources reflecting current practice that could be available for other purposes. In the health context, they are estimates of resources devoted to the treatment of preventable disease that could be released for other activities. A number of steps would be required, for example, to close/restructure nursing wards if hospital services for cardiovascular disease or cancer were reduced. Therefore, the results presented are broadly indicative of potential financial savings.