PUBLIC HEALTH AND ADDICTION: IS PREVENTION BETTER …



PUBLIC HEALTH AND ADDICTION: IS PREVENTION BETTER THAN CURE?

Christine Godfrey

Steve Parrott

Paper prepared for CHE conference, 4th May 2006.

Introduction

One of the major ongoing discussions in public health is why resources are focussed on treatment rather than prevention. Further, even among prevention programmes, the health practitioners, commissioners and to some extent researchers, favour individually delivered interventions over broader population based policies, for example legislation, mass media interventions etc. Traditionally such debates have been evidence free. The growth in systematic reviews in many lifestyle interventions should increase the quality of arguments but there is a history of over-interpretation of review evidence with a limited scope – for example limited to RCT evidence favouring individually based intervention. As health economists we should be able to give some guidance to practitioners and policy makers. The purpose of this paper is to outline some of the issues that are specific to the economic analyses of public health interventions for addictive behaviours and other factors with more general application which existing research on addictive substances has highlighted.

For the conference, three illustrative areas have been chosen:

• Exploring the choice between preventions and treatment

• Exploring issues of evaluating population interventions compared to individually focussed prevention initiatives

• Examining prevention programmes in a more macro economic and policy environment.

The paper is constructed in order to promote discussion and reflect our experience in research rather than being a traditional academic paper. The choice of areas is not meant to be comprehensive as there are addiction slants on a number of evaluation issues likely to be covered by other teams which we are happy to consider in discussion.

Economics of prevention versus treatment

In its simplest form, the question of prevention versus treatment for a specific population group with specified problems should be an empirical issue capable of being addressed by well designed research. Factors influencing the answer include the “numbers needed to prevent” which can be considerably larger than the numbers requiring treatment combined with the relative cost per individual, generally considerably smaller for prevention, between the two approaches. It also clearly depends upon the consequences of the lack of prevention – can the adverse consequences of the lifestyle/risk be easily identified or is there a risk of death or serious morbidity which is irreversible? Are treatments available which can improve quality and quantity of life? Are there any effective prevention policies? While most of these arguments have been outlined previously (Russell, 1986, Godfrey, 1993), the Wanless reviews seem to suggest there has not been a significant increase in economic evaluation studies.

However, the question as to whether treatment is more cost-effective than prevention is far too broad and not a direct question which decision makers would generally address. Unless shown to be relatively cost-ineffective compared to no treatment alternative, decision-makers would not be in a position to withdraw treatment programmes in favour of prevention. Rather they are faced with marginal shifts of finance across different programme areas. However, for any shifts in resources to take place there does need to be some evidence of specific interventions and this is part of a “Catch 22” situation.

Even with limited evidence, it is possible to explore the relative cost-effectiveness of different policy strategies and health economists within the illicit drug field have built such models. In particular, a body of work by RAND demonstrated that individual treatment for problem drug misusers is more cost-effective than international prevention efforts to reduce the supply of drugs. Some of the models produced have been mathematically complex but the data employed within them have been poor and the impact on policy has been very limited (see Reuter, in Godfrey et al, 2001). Our own research for the Home Office used rather better input data (although seriously flawed in parts) but a much simpler structure. The costing of drug related consequences for this project is now built into the performance management tool used to evaluate drug policy (McDonald et al, 2005). This research provided a 2000 estimate of social costs of £35,466 per problem drug user compared to £72 per young recreational user and £3 per older recreational user (Godfrey et al, 2002). An updated version of the model described in Godfrey, Parrott et al (2005) is currently in use by the Home Office to evaluate different schemes to guide more problem users into treatment. How far this work has influenced actual decisions is debatable and it should be noted that it was commissioned at a time when the policy decision had already been made to switch a higher proportion of the drug policy budget into treatment compared to prevention or enforcement strategies.

As a footnote, the RAND group also conducted a study which was published as a whole book and several articles on the potential cost-effectiveness of school education programmes which had no empirically based input parameters (Caulkins et al, 1999). A feat that may be considered even more heroic when systematic reviews of such interventions have struggled to find any positive impacts of such relative expensive prevention programmes.

In these cases the models were developed not to evaluate specific interventions as may be undertaken for the Public Health NICE evaluations discussed by Karl but broader programme comparisons. A number of discussion points arise:

• Do such models have any role to play?

• What level of sophistication should be used both in their structure and evidence synthesis?

• Are there roles for models where data is extremely limited? Or can they be used without the body of micro-evaluations which would ideally underpin them?

• What are the dangers from using such models?

Individual versus population level policies

Systematic reviews of prevention programmes have the danger, if limited by research design, to focus on interventions delivered to individuals (and therefore generally amenable to a randomised controlled trial research evaluation) compared to broader public health interventions. The evidence on effectiveness and cost-effectiveness has occasionally been over interpreted to suggest that all prevention programmes are ineffective when the review has failed to consider all alternatives. The WHO-CHOICE programme has made some attempt to consider the relative cost-effectiveness of broad band of different types of prevention policies in a number of areas, including alcohol and tobacco (see WHO CHOICE website). The policies considered range from tax policies, advertising restrictions, bans of smoking in public places, or server intervention for alcohol through to individual secondary prevention measures such as NRT for smoking cessation or brief interventions for alcohol interventions. Ignoring the issues around the use of DALYs, there are a range of technical issues which could be debated such as how to derive costs of legislative approaches and other non standard interventions.

However, the WHO analyses of alcohol and tobacco policies does not seem to have explicitly taken into account some of the wider consequences of population policies which restrict individual choice or the nature of addictive substances. This is contrast to media led debates in the UK which focus on “nanny state” arguments. How have economists contributed to these issues?

Most of the economic developments in both theoretical and empirical analyses have been directed at explaining individual behaviour. In particular there have been developments of theory to explain addictive behaviour which have included such behaviour within a “rational” framework. In simple terms, consumers could rationally pursue their welfare even if they are aware of the nature of dependence and how this may influence their future choices. There is some, if limited, empirical support for this model and considerable evidence that consumers do respond to both prices and incomes. The model can be extended to account for observed behaviour by including learning, regret and endogenous determination of time preferences. Other models include time-inconsistent preferences generally biasing utility towards the present for example by using hyperbolic discounting. Also, and importantly, evaluation welfare analysis can be extended to suggest that consumers may rationally demand government action to reduce their choices. Interestingly Becker et al (2004) suggest that the most restrictive policies, such as prohibition, could be preferred by the rich and the full welfare costs of such policies’ fall disproportionately on the poor.

Far less research has been undertaken on how different theoretical welfare models may impact on comparative interventions of different policies (Buck et al, 1996). Where consumers are fully informed on the risks of consumption, and are economically rational, there can still be justification for policies that restrict choice if there are external costs. For alcohol, however, this is further complicated in that the external costs are not at an individual level simply correlated with level of consumption. However, there is also the issue of the level of information about the risks of consumption. While individuals will sometimes overestimate the risks of long-term effects such as lung cancer risks, there is less knowledge of impacts on morbidity. Is the role of governments simply to ensure that consumers can make an informed choice or are there further economic arguments that could be used to support more interventionist policies? Empirical evidence suggests that giving information either through mass media campaigns or through face-to-face or group interventions is not straightforward. This can perhaps be best illustrated by exploring those policies which receive the strongest support from industry (school programmes, non threatening, low level mass media campaigns, sometimes disinformation campaigns) compared to those they oppose (clear and simple labelling, for example note the food industry’s reluctance to adopt the traffic light scheme, counter-advertising – schemes that are linked to direct advertising by the industry, more sophisticated social marketing approaches).

In evaluating different prevention approaches, should the wider consequences of the impact in restricting choices be included, for example, by some estimate of the loss of consumption benefits of consumers not currently causing harm to others? If, however, addiction is believed to restrict rational choice are there arguments that suggest that there would be greater gains than traditionally evaluated from legislative or prohibitive approaches. Some have argued in evaluations of addictive substances that current consumption of “addicts” produces no consumer benefits and therefore any reduction in consumption would yield social benefits over an above a reduction in addiction related problems at an individual or social level (see Buck et al, 1996). There is a potential area for further research exploring the different implications of theoretical approaches to addictive behaviour formation and policy evaluation.

Economic models are very individualistic by nature. Public health approaches in contrast focus on the population impacts. The individualistic nature of economics has been criticised with authors suggesting that evaluation methods are both ill-suited to evaluate community level interventions (Shiell and Hawe, 1996) and ignore some of the social diffusion impacts of public health interventions (Rosen and Lindholm, 1992). Again there is evidence on the importance of peer effects on the consumption of tobacco, alcohol and illicit drugs (e.g. Molyneux et al, 2002). Also drug use patterns can be analysed in terms of dynamic collective behaviour, (Cave and Godfrey, 2005). Furthermore, “contagion” effects can be negative, users “infecting” non users, or positive as changes in social acceptability of behaviours may mean non-hazardous users may influence problem or high risk individuals. This has been empirically illustrated in terms of binge drinking and different cultures in US colleges. However, much more sophisticated theoretical models can be built combining some of the individual theories of economics with social theories from sociology and criminology. A model developed in Cave and Godfrey (2005) explored different levels of interactions between

• Individuals’ intrinsic preferences

• Peer pressure

• Anticipated price and risk relationships

• Social networks

• Drug use among the network neighbours.

It was shown that strong peer effects (pressure and learning) produce a model with multiple equilibria, catastrophic jumps and the sort of S shaped paths often seen in epidemiological dynamic disease models. An attempt was also made to explore some of the unintended consequences of different policies – for example policies that could induce clustering. Again there is some empirical simulation of the changing effectiveness of policies in different stages of a drug use cycle. This is a fruitful area for further theoretical and empirical research in itself and also how these dynamic models interact with policy cost-effectiveness.

Prevention policies in their wider context

Interactions between individuals have been shown to influence addictive and lifestyle behaviours. However, prevention policies do not occur in isolation, nor can many population based policies be fully evaluated without exploring some of the wider consequences – good or bad. This is particularly important in a public health context where there is some attempt to assess the overall population level of problems, as well as looking at effectiveness or cost-effectiveness. In recent work for NICE, Parrott et al (2006) it was suggested that brief interventions for smoking cessation delivered across a broad population group yields health gains at a comparative cost per health gain of more intensive smoking cessation interventions. Indeed there is some evidence that intensive interventions may be more cost-effective than brief interventions and well below the NICE threshold. However, as West (2005) and Parrott et al (1998) demonstrate such intensive interventions are only likely to reach a small proportion of smokers. The ban on smoking in public places is predicted to have a much larger overall population impact (and indeed is likely to be cost saving through reductions in smoking related disease costs).

However, econometric demand work from the NBER has demonstrated the importance of “policy sentiment”. That is those states with a favourable attitude towards more restrictive policies are more likely to be early adopters of new restrictive policies. Holder (1998) summarising a lifetime work in alcohol prevention evaluation has taken these ideas of interaction between demand, supply, populations and policies a stage further as illustrated in Figure 1. This model and the computer simulations model constructed indicate that a relationship clearly exists between economic prosperity and levels of consumption which in turn impacts on levels of problems. However, the level of problems also impacts on social norms which change the policy sentiment and the ability to adopt regulations or taxes. Similarly in analysing drug markets, particularly looking at enforcement policies it is necessary to examine impacts on supply as well as demand. Naive or restrictive evaluations which do not take account of the full round of impacts could seriously mislead decision makers.

Conclusions

The evaluation of addictive behaviours and interventions to change them can be used to illustrate many of the general issues raised in “public health economics” but they also raise some interesting new ones. In particular, they highlight the need to at least identify some of the wider consequences of policy choice comparisons.

A great potential exists to use simulation modelling to illustrate the question initially posed in this paper – prevention or treatment – and the answer is likely to vary between different drugs of abuse. There are also varying levels of information available about different substances. Cigarette use has the richest epidemiological and effectiveness evidence base but perhaps surprisingly very few primary economic evaluations. As in other areas cost and other consequence data has been neglected in comparison to effectiveness synthesis. In the case of illicit drug use, data is difficult to find but this area contains some of the most innovative uses of economic techniques. Alcohol as well as being our favourite drugs poses some of the most challenges. The level of physical dependence is lower than many others; many drink hazardously but at relatively low levels of risk, but external harms are high (and spread across different levels of drinking).

In the addiction field, the economic literature has both rich and poor streams. Economic evaluations of prevention programmes are very rare and while there is a more comprehensive literature in treatment, many are of poor quality. However, there is a rich theoretical and empirical literature, mainly published in mainstream economic or health economic journals, focussing mainly on demand models. Much of the evidence and models from these journals is slow or is never disseminated into the addiction specialist literature or is seen to influence policy making. One of the historic strengths of health economists has been to bridge the gaps between health policy, clinical and economic ideas and there is considerable scope in this area.

This paper identifies three areas where some work has already been undertaken, whilst also highlighting ideas for future research. We look forward to gathering the thoughts of our colleagues throughout the conference.

References

Becker, G., Murphy, K. and Grossman, M. The economic theory of illegal goods: the case of drugs.NBER Working Paper 10976. Cambridge, Mass: NBER.

Buck, D., Godfrey, C. and Sutton, M. Economic and other views of addiction: implications for the choice of alcohol, tobacco and drug policies, Drug and Alcohol Review, 1996, 15: 357-368.

Caulkins, J., Rydell, P., Everingham, S., Chiesa, J. and Bushway, S. An Ounce of Prevention: A Pound of Uncertainty. The cost-effectiveness of school-based drug prevention prgrams. Drug Policy Research Center, RAND: Santa Monica, 1998.

Cave, J. and Godfrey, C. Economics of Addiction and Drugs. Foresight Brain Science Addiction and Drugs Project. Office of Science and Technology. 2005.

Godfrey, C. Is prevention better than cure?. Drummond, M F & Maynard, A(eds.), Purchasing and providing cost-effective health care. Edinburgh: Churchill Livingstone, 1993: 183-197.

Godfrey, C., Parrott, S., Eaton, G., Culyer, C and McDougall. Can we model the impact of increased treatment expenditure on the UK drug market. In Lindgren, B. and Grossman, M. eds. Substance Use: Individual Behaviour, Social Interactions, Markets and Politics. Advances in Health Economics and Health Services Research, Volume 16. Elsevier: New York. 2005: 257-275.

Godfrey, C., Parrott, S., Coleman, T. and Pound, E. Cost effectiveness of English smoking cessation services: evidence from practice, Addiction, 2005, 100 (Suppl. 2):70-83.

Godfrey, C., Wiessing, L. and Hartnoll, R. ( scientific eds.) Modelling drug use: methods to quantify and understand hidden processes. EMCDDA Monograph No 6. Lisbon: EMCDDA, 2001

MacDonald, Z, Tinsley, L., Collingwood, J., Jamieson, P. and Pudney, S. Measuring the harm from illegal drugs using the Drug Harm Index, Home Office Online Report 24/05. 2005.

Molyneux, A., Hubbard, R., McNeill, A., Godfrey, C., Madeley, R. and Britton, J. Is Smoking a communicable disease? The effect of exposure to ever-smokers in school year groups on the risk on incident smoking in the first year of secondary school, Tobacco Control 2002, 11: 241-245.

Parrott, S., Godfrey, C. and Kind, P. Cost-effectiveness of brief intervention and referral for smoking cessation. Report for NICE.

Rosen, M. and Lindholm, L. The neglected effects of lifestyle interventions in cost-effectiveness analysis, Health Promotion International, 7: 163-169.

Russell, L.B. Is Prevention better than Cure? Washington, D.C: The Brookings Institute 1986.

Shiell A, Hawe P. Health promotion, community development and the tyranny of individualism. Health Economics 1996; 5: 241-247

West, R (2005) Projected national smoking prevalence and smoking related death rates following actual and potential policy initiatives in the UK. rjwest.co.uk

[pic]

Figure 1: Systems approach to alcohol prevention policies (Holder, 1998)

-----------------------

Pop growth or

decline

Economic trends

Legal sanctions

Social, economic

and health

consequences

Retail

sales

Consumption

Social norms

Formal regulation

and control

................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download