1 Introduction

1.1 Objectives of Report

This report provides an epidemiological and economic evaluation of five Australian public health programs: namely programs to reduce tobacco consumption, coronary heart disease, HIV/AIDS, measles and Hib-related diseases, and road trauma.

For tobacco consumption, coronary heart diseases, immunisation against measles, and road trauma, our study reports on the effects of public health programs over the last 30 years because this covers the period of modern public health interventions in these cases. For the HIV / AIDS and Hib public health programs, the evaluation starts in the mid-1980s and early 1990s respectively. Outcomes are forecast to year 2010, except for the immunisation programs where outcomes are forecast to 2003.

For each of these public health programs, the study estimates:

The costs of the public health programs,
the reduction in cases of disease that can be attributed to public health interventions since the programs were started and up to 2010,
the benefits of the reductions in disease in terms of increased longevity, improved quality of life, and reduced health care expenditures, and
the overall returns on investment to public health interventions.

As requested by the Department of Health and Aged Care, the study estimates the return on each of the five public health programs as a whole rather than the marginal return on separate parts of each program.

The study estimates both the financial returns of the public health programs to the government, based on the savings in health expenditures, and the total return of the programs to the community. Of course, from the societal perspective, the return to the community is the important outcome rather than the narrower measure of the financial return to government.

It should also be noted that the five case studies are essentially of a survey nature. They bring together a great deal of literature and data that are already in the public domain. The consultancy was not designed to undertake major new research. Notwithstanding this, all five case studies present some material and analysis that has not been previously available.

In this introductory chapter, we discuss the nature of public health and some implications for the study, our general approach to the evaluation of the public health programs, and the basic principles of economic evaluation that are applied to all five case studies.

1.2 The Nature of Public Health

The National Public Health Partnership between the Commonwealth, States and Territories describes public health as ‘the organised response by society to protect and promote health and to prevent illness, injury and disability’.

Health protection measures are designed to mitigate risks to health arising from the environment and activities over which individuals have little or no control. They are usually mandated by legislation or regulation. Examples are legal bans on smoking in certain places and regulations on tobacco advertising, regulations on food, and legislation on vehicle safety belts, vehicle speeds and drinking and driving. Environmental measures to ensure high standards for air, water, soil and waste management are also designed to protect public health.

Health promotion measures are designed to help people make healthy lifestyle decisions. These measures include campaigns to promote physical exercise, weight loss and healthy diet, to reduce consumption of tobacco, alcohol and harmful drugs, to promote safe sex, and to stop driving when tired.

In addition, disease prevention and early detection programs aim to provide information about health risks, to prevent certain, usually common, preventable diseases, and to obtain early warnings of possible disease. Examples include surveillance of communicable diseases, major immunisation programs, and screening programs, such as breast and cervical cancer screening programs.

The boundaries between public health programs and other government actions are often blurred. For example, the taxation of tobacco or other drugs may be regarded as a health protection measure or as a revenue raising measure. Physical education in schools may be regarded as health promotion or as individual self-fulfillment. Road safety programs include accident black spot projects as well as random breath testing. The lines between health research, disease prevention and health care may also be fuzzy.

The size, diversity and imprecise nature of many public health programs create practical issues for the evaluation of public health programs. One set of issues relates to the definition of the programs. It is to some extent a matter of opinion whether tobacco taxes are part of the tobacco reduction program, what road improvements count as road safety measures, and what are the major components of programs to reduce coronary heart disease. In this report we state our assumptions about the nature of the public health programs in each case study, with a focus on programs designed to regulate or influence behaviour but acknowledge that this is not a precise concept.

A related issue is the cost of these public health programs. Because of the diverse nature of the programs, some programs are administered outside health care agencies. Even within health care agencies there may be no clear accounts for public health programs, certainly going back to the 1980s. A further complication for the costing of public health programs is the division of responsibility for many programs between the Commonwealth and the States and Territories. Thus, it has not been easy to determine the true costs of public health programs.

1.3 General Approach to the Evaluation of Public Health Programs

The following are the main steps in the evaluation for each public health program. As allowed by the data, this study:

Describes the relevant public health programs over 20 or so years.
Reviews the effects of the public health programs on behaviour.
Determines whether changes in behaviour caused changes in health outcomes.
Estimates mortality and morbidity with and without public health programs.
Estimates the costs of the public health programs.
Estimates the benefits of the programs in terms of longevity, improved health status, and lower health care expenditures.
Evaluates the benefits and costs of the public health programs from the perspective of (i) the whole Australian community and (ii) government expenditures.

A weak link in any part of this process weakens the final result. It is possible in a few cases to estimate a plausible relationship between a public health program and health improvement with little knowledge of intermediate behaviour changes. For example, there may be an observable relationship between a road safety program such as random breath testing (RBT) and a decline in road accidents, but no detailed data on the impact of RBT on drinking and driving. But these cases are exceptional. We generally want to know both the impact of the public health program on behaviour and the impact of changes in behaviour on health.

Table 1.1 outlines some main relationships for the five public health areas in the study. It may be observed that these public health programs include not only public health promotion, but also public health regulations and related penalties, and the use of market prices. As we have seen, all these actions may be regarded as components of a public health program. However, from an evaluation perspective, there is a difference between public health promotion that encourages voluntary lifestyle changes and regulations that require people to change their behaviour, sometimes against their will.

There are several constraints on estimates of all causal relationships shown in Table 1.1. For a start, data on some public health programs are limited. For example, there is no agreed inventory of the programs that have been undertaken to reduce coronary heart disease. It is also difficult to distinguish between public health programs designed specifically to reduce heart disease and programs designed to promote generally healthy lifestyles. This lack of information reflects the diverse nature of the implementation of public health programs, which are delivered variously by Commonwealth, State and local governments and non-government organisations. Even within one level of government, public health programs may be delivered by different agencies through schools, workplaces, and national advertising campaigns, and so on.

Table 1.1 Examples of links in the evaluation of public health programs

Public health program	Behavioural objectives	Health objectives	Indicative behavioural and health outcomes^a	Economic analysis

Anti-smoking promotions, regulations, prices	Reduced tobacco consumption	Reduced lung cancer, heart disease, strokes, Bronchitis, other cancers	Programs achieved behaviour and health changes	Compare program cost with health benefits
Lifestyle programs: improved diet, more exercise	Reduced body mass, lower cholesterol, lower blood pressure, increased exercise	Reduced heart disease, strokes, diabetes, and other disease	Some behavioural changes and some unchanged behaviours, improved health outcomes	Compare program cost with health benefits
HIV/AIDS educational, needle/syringe exchange programs	Safer sex, use of clean needles	Reduced HIV / AIDS diseases	Programs achieved behaviour and health changes	Compare program cost with health benefits
Various immunisations (measles and Hib)	Increased participation in vaccinations	Reduced diseases	Reduced incidence of measles and Hib-diseases	Compare program cost with health benefits for selected vaccination programs
Programs to reduce road accidents (regulations and penalties)	Wearing seat belts, lower speed, less drinking and driving	Lower road injuries	Programs achieved behavioural and health changes	Compare program cost with health benefits

(a) Indicative outcomes compared with no public health intervention.

Second, in order to estimate the impacts of public health programs on behaviour, data are needed on behaviour changes with, and without, the program. Good data exist for some behavioural changes, for example for changes in tobacco consumption and changes in the sexual practices of men at risk of HIV/AIDS. On the other hand, only limited data are available on risk factors for coronary heart disease. There have been only a few national surveys of blood pressure, cholesterol levels, body weight, and physical exercise over the last 20 years.

However, by themselves, trend data are of little use. Counterfactual estimates of the behaviour that would have occurred without public health programs are required. For example, on the basis of sparse data, the Australian population is putting on weight and doing little to increase physical exercise. It might be inferred that public health programs have been ineffective in influencing diet and exercise. But without the public programs advocating weight loss and exercise the population might have become even more obese and less physically active!

Estimating the impacts of changes in behaviour (or risk factors) on health is also complicated:

The relationships between risk factors and mortality or morbidity may be non-linear.
Relationships between risk factors and health conditions are often interdependent, and may not be independently additive.
In some cases, notably tobacco consumption, there are long lags between changes in behaviour and changes in health.
Behavioural changes may have other by-products (for example, some people who give up smoking may turn to another adverse addictive behaviour).

The social environment is highly complex. Identifying the impacts of specific variables on health outcomes is difficult in a complex, multi-variable world. Even in a relatively straightforward case, like the introduction of immunisation programs, there is generally a concurrent improvement in environmental conditions. Thus reductions in measles or Hib diseases may be due partly to improved living standards as well as to vaccinations.

Issues with causal relationships are general ones that appear in all the case studies. However, drawing on both our own work and on other studies in each area, this report makes conservative estimates of the impacts of public health programs on health outcomes for each case study.

1.4 Economic Evaluation

For each public health program, benefits and costs are estimated and discounted to present value terms, as is standard economic practice. In this section we summarise the study approach to estimating benefits and costs and the overall evaluation, and to the choice of discount rate.

Benefits of public health programs

There are two main sets of benefit to be considered: savings in health care expenditures and improvements in personal health (longer and better quality life).

Estimated savings in health care expenditures are based on detailed estimates of health care expenditure per disease in 1993-94, the only year for which such estimates are available.¹ Drawing on these data, savings in health care expenditures due to reduced morbidity are then estimated for a specific year in the 1990s. These estimated savings in health care expenditures are projected back to earlier years and forwards to 2010 in proportion to estimated morbidity changes. The implicit assumption is that the unit costs of treatment are constant over time.

Of course, these estimated savings in health care expenditures ignore any additional health care expenditure incurred because people live longer. But this is reasonable. In valuing longevity, the estimated benefit of extra years of life implicitly allows for, or should allow for, any public or private resources that individuals consume in the health care sector or elsewhere.

In order to value improved health, this study draws mainly on the concept of a disability-adjusted life year (DALY) lost. A DALY is one year lost of healthy life. For any disease, the total health cost to an individual is the sum of years of life lost due to premature death (YLL) and equivalent years of healthy life lost due to disability (YLD). Thus estimated total DALYs are:

DALYs = YLLs + YLDs (1.1)

If a health state has a DALY of say 0.33, being in that state for three years is equivalent to losing one year of healthy life (i.e. one DALY).

Years lived with disability are calculated by multiplying a disability weight representing the quality of life associated with a particular health state by the number of years lived in that health state. With a DALY, a disability weight of 0 represents perfect health and a weight of 1 represents death.² In this study, years lost due to disability account overall for about a quarter of total DALYs. However, the proportion varies with the disease or risk factor, with the proportion due to disability being relatively high for obesity and low for high blood cholesterol.

Mathers et al. (1999) estimate DALYs in 1996 for a large number of diseases and health states in Australia, including most diseases relevant to this study. Drawing on our review of the impacts of public health programs on health, we can then estimate the reduction in DALYs associated with each public health program.

In order to convert DALYs into economic benefits, a dollar value per DALY is required. In this study, we follow the standard approach in the economics literature and derive the value of a healthy year from the value of life. For example, if the estimated value of life is $2.0 million, the average loss of healthy life is 40 years, and the discount rate is 5 per cent per annum, the value of a healthy year would be $118,000.³ Tolley et al (1994) review the literature on valuing life and life years and conclude that a range of US$70,000 to US$175,000 per life year is reasonable. In a major study of the value of health of the U.S. population, Cutler and Richardson (1997) adopt an average value of US$100,000 in 1990 dollars for a healthy year.

Although there is an extensive international literature on the value of life (Viscusi, 1993), there is little Australian research on this subject. As the Bureau of Transport Economics (BTE, 2000) notes, international research using willingness to pay values usually places the value of life at somewhere between A$1.8 and A$4.3 million. On the other hand, values of life that reflect the present value of output lost (the human capital approach) are usually under $1.0 million.

The BTE (2000) adopts estimates of $1.0 million to $1.4 million per fatality, reflecting a 7 per cent and 4 per cent discount rate respectively. The higher figure of $1.4 million is made up of loss of workforce productivity of $540 000, loss of household productivity of $500 000, and loss of quality of life of $319 000. This is an unusual approach that combines human capital and willingness to pay concepts and adds household output to workforce output.

For this study, a value of life of $1.0 million and an equivalent value of $60 000 for a healthy year are assumed.⁴ In other words, the cost of a DALY is $60 000. This represents a conservative valuation of the estimated willingness to pay values for human life that are used most often in similar studies.⁵

Costs of public health programs

Ideally, an economic evaluation should include all costs of public health programs: that is, expenditures on resources and other welfare costs. Expenditures on resources include government expenditures on public health programs and required private expenditures (such as expenditures on car seat belts and out-of-pocket expenses for visits to a doctor for a vaccination). Other welfare costs are the welfare losses associated with changes in otherwise preferred private behaviour. These losses may occur for example with regulations (such as lower driving speeds and restrictions on drinking), and even with voluntary changes in otherwise preferred behaviour (for example eating less red meat). Only if the benefits of a public health program exceed total resource expenditure and welfare costs, can it be truly said that a public health program is socially beneficial in the standard economic calculus of social welfare.

In practice, few economic evaluations of public health programs are so comprehensive. The measurement of welfare losses as behaviour changes is a complex task. Indeed, such a comprehensive calculus is outside the scope of this report. This study focuses on government expenditures on public health programs. Other welfare costs are considered only if they appear to be very significant, as with controls of traffic speeds.

As noted, expenditure data on public health programs are also limited or difficult to obtain, especially for coronary heart programs. Deeble (1999) estimated that total community and public health expenditure in Australia rose from $194 million in 1975-76 to $2157 million in 1995-96. The public health component accounted for an estimated $130 million and $515 million in the two years respectively. The latter figure was about 2 per cent of the total health care expenditure budget.

More recently, the AIHW estimates that the total public health expenditure budget in 1998-99 was $869 million (see Table 1.2). At the time of writing, more detailed figures are not available.

Table 1.2 Public health expenditures in 1998-99

Public health category	$m	% of total
Communicable disease control	149.4	16.8
Selected health promotion	187.7	21.1
Immunisation	183.0	20.6
Environmental health	71.7	8.1
Food standards and hygiene	22.4	2.5
Breast cancer screening	90.8	10.2
Cervical cancer screening	80.9	9.1
Research (Commonwealth only)	17.0	1.9
Other core public health expenditure	85.3	9.6
Total expenditure	888.2	100.0

Source: AIHW and National Public Health Partnership, forthcoming,
National Public Health Expenditure Report 1998-99.

Overall economic evaluation

Finally, in an economic evaluation, the estimated benefits and costs of the public health programs are converted into present values and a summary net present value (NPV) is estimated. Formally, this can be represented as:

NPV = (b-c)₁ / (1+r) + (b-c)₂ / (1+r)²… + (b-c)_n / (1+r)ⁿ (1.2)

where b and c represent the estimated benefits and costs respectively, r is the rate of discount, and the benefits or costs are experienced over n years. If the NPV is positive, estimated benefits exceed estimated costs.⁶

Equation (1.2) is usually estimated in constant prices, so that the discount rate is also estimated free of inflation. The choice of discount rate may have an important bearing on the outcome, especially when benefits occur a long time after costs, as may happen with public health programs.

The choice of discount rate has been much debated. The debate centres on the choice between a marginal return on capital measure and the marginal rate at which individuals discount future consumption (an individual rate of time preference). The estimated risk-free marginal real return on capital for major projects is usually about 7-8 per cent. However, because tax creates a wedge between the return on investment to society (including tax) and the return that private individuals receive, the real rate at which individuals discount future marginal consumption is generally in the order of 3-4 per cent per annum. This is lower than the gross return on capital⁷

Actually, it can be shown that whenever resources used to develop a project can be invested in an alternative project, the appropriate rate of discount is the opportunity cost of capital (Pearce, 1983; Abelson 1996). Accordingly, most public financial agencies recommend that economic evaluations adopt an opportunity cost of capital of 7-8 per cent as the discount rate (for example, Commonwealth Department of Finance, 1991). However, if we are solely concerned with comparing an individual’s welfare over time, and there are no opportunities for investment, the private consumption rate of discount is the appropriate rate of discount.

In this study, a compromise 5 per cent rate of discount is adopted as the benchmark rate for the case studies, but the implications of a 7 per cent and a 3 per cent rate of discount are also considered. In addition, at the request of the Commonwealth Department of Health and Aged care, undiscounted results are reported as part of sensitivity testing.

1.5 Summary

This report provides an epidemiological and economic evaluation of five public health programs: namely programs to reduce tobacco consumption, coronary heart disease, HIV/AIDS, measles and Hib-related diseases, and road trauma.

In order to make these evaluations, the study:

Describes the relevant public health programs over 20 or so years.
Reviews the effects of the public health programs on behaviour.
Determines whether changes in behaviour caused changes in health outcomes.
Estimates mortality and morbidity with and without public health programs.
Estimates the costs of the public health programs.
Estimates the benefits of the programs in terms of longevity, improved health status, and lower health care expenditures.
Evaluates the benefits and costs of the public health programs from the perspective of (i) the whole Australian community and (ii) government expenditures.

The study is based on existing literature and data and is essentially a survey of the main findings of existing knowledge on each topic. However, the study provides new epidemiological findings in several areas along with new economic results for each case study.

¹ See Mathers et al., (1998a), Mathers and Penm (1999) and other sources in relevant chapters.

² A disability-adjusted life year may be viewed as the converse of a quality adjusted life year (QALY). With a QALY, 1.0 usually represents perfect health and 0 represents being dead.

³ In round numbers, $2,000,000 = $118,000/1.05 + $118,000/(1.05)²…+ $118,000/(1.05)⁴⁰.

⁴ The equivalent value of $60 000 assumes, in broad terms, 40 years of lost life and a discount rate of 5 per cent.

⁵ In addition to the cited references in the text, see for example Murphy and Topel’s study (1999) on the economic value of medical research.

⁶ Other summary evaluation criteria are the internal rate of return and the benefit-cost ratio (see Commonwealth Department of Finance, 1991)

⁷ The gap between the real total return on capital and the return to private individuals also depends on the interaction of taxation and inflation.

Applied Economics