Bulletin on Narcotics
Volume LII, Nos. 1 and 2, 2000
Economic and social costs of substance abuse
Introduction: Improving economic data to inform decisions in drug controlD. COLLINS
Adjunct Professor in Economics, Macquarie University, Sydney, Australia
Senior Lecturer in Health Economics, University of New South Wales, Sydney, Australia
Senior Associate, Canadian Centre on Substance Abuse, Ottawa, Canada
Research Associate, Canadian Centre on Substance Abuse, and Professor of Public Health Sciences, University of Toronto, Canada
Key substance abuse issues from a policy-making perspective
A conceptual framework for estimating the costs of substance abuse
Estimating avoidable costs of substance abuse
Estimating the cost-effectiveness of interventions
Estimating performance over time
BackgroundThere is enormous variation between countries and regions of the world with regard to illicit drug use and the problems associated with illicit drugs . In South America and south-east and south-west Asia, where a large share of the world's cocaine and heroin is produced, the illicit drug trade has created substantial underground economies with problems for law enforcement and economic control. In the drug-consuming countries in Europe and North America, the problems are more related to the adverse health, social and economic consequences of illicit drug use.
Although international treaties provide a common framework for drug policy and a certain degree of uniformity in social responses to the problems caused by illicit drugs, there is also wide variation in national drug policies. Responses to illicit drug problems range from strict enforcement of punitive drug laws to benign neglect. In parts of Australia, Europe and North America, harm reduction policies have been implemented to reduce the adverse consequences of illicit drug use for users who cannot be expected to cease their drug use at the present time . Even within countries, there are often cycles of panic over emerging drug problems, followed by periods of indifference when other pressing issues push the problem of illicit drugs to a relatively low place on the national policy agenda .
Thus, there are substantial differences in the nature and magnitude of illicit drug problems as well as the social responses to such problems, over time, both between countries and regions of the world and within countries. Nonetheless, there is a strong consensus among people involved in addressing addiction issues that success requires long-term commitment and investment. As for vaccination programmes, progress can be made only over extended periods of time as consistent efforts are applied to each new generation. There is, therefore, a strong need for more rigorous and comprehensive economic data on substance abuse to promote evidence-based decision-making and a more consistent response to substance abuse. The economic ramifications of illicit drug use are not well understood, either in the producing countries or the consuming countries and regions.
The purpose of the series of articles appearing in the present issue of the Bulletin on Narcotics is to bring together related approaches of several authors who examine various methodological issues and data requirements involved in developing more complete, reliable and comparable data on economic aspects of substance abuse. The articles are based on papers presented at the Third International Symposium on the Economic and Social Costs of Substance Abuse, organized by the Canadian Centre on Substance Abuse in Banff, Canada from 31 May to 3 June 2000. The aim of the symposium was to extend the scope of the cost estimation of substance abuse and to facilitate improvements in cost-estimation methodology.
In the article entitled "Economic evaluation of policies and programmes: further uses of estimates of the social costs of substance abuse", Collins and Lapsley review areas of cost estimation that still need to be addressed or that are in need of further development. It proceeds to consider the extension of the use of the data derived from those studies into the more policy-oriented areas of drug programmes and project evaluation. The other articles deal with particular issues referred to in general terms in the Collins and Lapsley article.
In the article entitled "The cost to employers of employee alcohol abuse: a review of the United States literature", Harwood and Reichman review literature from the United States of America concerning the costs borne by employers as a result of alcohol abuse by employees. The authors' analysis of the types of impact of alcohol abuse is also relevant to the workplace effects of smoking and illicit drug use.
One of the most intractable issues in the cost estimation of substance abuse has been how to identify the extent of crime that is attributable to alcohol and drug use. Crimes may be associated with drug use but such an association does not necessarily imply a causal relationship. In the article entitled "Attributable fractions for alcohol and drugs in relation to crime in Canada: conceptualization, methods and internal consistency of estimates", Pernanen and Brochu examine the problems involved in developing crime-attributable fractions in Canada.
There are difficulties in identifying the size of public expenditure that arises as a result of substance abuse. Drug-related expenditures can be spread across the budgets of many individual government departments and the issue of the development of attributable fractions arises once again, in areas such as health and justice where similar types of expenditures can be attributable to a range of disparate causes. In the article entitled "Estimating the costs of substance abuse to state budgets in the United States of America", Foster and Modi present the analysis underlying a study that attempts to estimate the costs of substance abuse across the complete range of state budgetary units of the United States.
Most of the existing research into the social costs of substance abuse has been undertaken in and for Western developed economies. Those results are likely to be of limited applicability to developing economies that may have radically different institutional structures and may face different types of drug problems. In the article entitled "Estimating the economic costs of drug abuse in Colombia", Pérez and Wilson, drawing on the Colombian case, illustrate the types of issues that may limit the applicability of existing research results to developing countries.
The trade in illicit drugs leads to extensive but largely unquantified underground economic activities. Such activities involve various forms of crime, including drug-dealing, violence, tax evasion and smuggling. The issues involved in estimating the size of the shadow economy are reviewed in the article by Schneider entitled "Illegal activities and the generation of value added: size, causes and measurement of shadow economies".
In the present introduction, the key issues that policy makers must address in dealing with illicit drug problems are first discussed. A conceptual framework for estimating the economic costs of illicit drug use is then presented. Focus is placed on the data requirements, methodological issues and prospects for future development of cost estimation studies in more countries and regions of the world. The problems involved in estimating the avoidable costs of illicit drug abuse and the cost-effectiveness of interventions are then addressed. A summary of the present situation regarding the current understanding of economic aspects of illicit drugs and a discussion of the implications for data systems and research on drug problems conclude the introduction.
Key substance abuse issues from a policy-making perspectiveFour key issues need to be addressed to help policy makers make well-informed decisions on drugs:
(a) The cost of drug abuse to society;
(b) The portion of those costs that are realistically avoidable;
(c) What investments policy makers should make to avoid such costs and where they should make them;
(d) How well such investments are performing over time.
With regard to those issues, researchers need to pay greater attention to economic aspects of drug abuse and interventions. Policy makers are rarely concerned with the details regarding the data requirements and methodological issues involved in addressing those issues. Nevertheless, those aspects need to be addressed at least in part, if policy makers are to make well-informed, empirically-based decisions on drug policy.
Substantial attention has been paid by researchers to the costs that society bears as a result of the use and abuse of alcohol and tobacco. Very little attention, on the other hand, has been paid to the costs borne by society as a result of the abuse of illicit drugs. There is no doubt that such a deficiency can be attributed largely to the serious data problems inherent in any attempt to quantify the social costs of illicit drug use. It is difficult to quantify the production, consumption, import, export or price of illicit drugs. In addition, although significant information is available on the causal links between drug abuse and health, the causal links in other areas, in particular crime, are extremely difficult to quantify. For those reasons, little quantitative information exists on the social costs of illicit drug abuse.
Although it is inevitable that economic cost data in that area will be deficient, it is possible to provide much more comprehensive information than is currently available. Reasonable estimated values can be placed on some drug abuse costs, narrowing the range of costs on which policy makers will be required to make qualitative judgements. That is not an unusual circumstance. In most social benefit-cost analyses, some benefits or costs cannot be valued and so must be judged qualitatively against those which can be quantified. Assigning values where it is possible to do so narrows the range of uncertainty in the decision-making process.
The composition of the total community costs of drug abuse is illustrated in figure I below.
|Figure I. Community costs associated with drug abuse|
Private costs are those borne by the drug abusers themselves, having made a rational decision to consume in the full knowledge of the effects upon them of drug abuse. If drug abusers do not bear the full costs, for example, if their health costs are subsidized from the public purse, then those costs become the concern of public policy. If the actions of drug abusers are determined by perceived costs that are less than actual costs, the difference between the two can be viewed as the social cost because the abusers have not adjusted their behaviour to reflect those higher costs and so the latter are not accounted for . The three conditions to be satisfied simultaneously if costs are to be classified as private costs are as follows: (a) costs fully borne by abuser; (b) full knowledge of the effects of drug abuse; and (c) rationality in decision to abuse drugs. That is a stringent set of requirements. In practice it means that a high proportion of the total costs of drug abuse is likely to be external rather than internal.
The social (external) costs of drug abuse can be subdivided into tangible and intangible costs. Tangible costs are those that, if reduced, would release resources to the rest of the community for alternative consumption or investment purposes. For example, a reduction in public health-care costs will release resources for government expenditure in other areas. A market exists for the resources used in those areas, therefore, it is possible to attach a price to them.
Intangible costs (for example, loss of life, pain and suffering), when reduced, do not release resources for other uses. The reduction of such costs is very important but does not yield benefits that can be redistributed to other people. No market exists in such benefits, thus it is difficult to assign a value to them.
The difficulty of valuing intangibles should not lead to the avoidance of including those costs in policy analysis. It is possible that a reduction in drug abuse might lead to an increase in certain tangible costs. If the inevitable reduction in intangible costs is not set against the increase in tangible costs, the erroneous conclusion might be drawn that drug abuse was in the public interest.
The example of health-care costs attributable to drug abuse illustrates the point. It is possible that the lifetime health-care costs of abusers are less than those of non-abusers, because the former have lower life expectancies and so draw upon the health-care resources of the community for a much shorter period. There is some evidence that that might be the case for smokers; however, the necessary research has not been undertaken in relation to drug abusers. If it were assumed, for the purposes of argument, that drug abusers imposed lower lifetime health costs than non-abusers, the nonsensical conclusion might be drawn that drug abuse was, in some sense, in the public interest. Such a conclusion could only be drawn if the high intangible costs of abuse (such as loss of life, pain, suffering and bereavement) were ignored, and they clearly should not be ignored.
A summary of the types of external costs of drug abuse is presented in figure II. Production losses can occur in the paid workforce but they can also represent the loss of unpaid work (for example, household work and volunteer and community work). Unpaid work, though productive, is not counted in conventional national accounts statistics. It should, however, be taken into account in the assessment of production losses attributable to drug abuse.
Drug-attributable mortality can reduce the size of the workforce. Its impact on the level of employment may be alleviated by the existence of a pool of unemployed individuals who are ready to fill the employment gap, although the skills available among the unemployed may not match the skills lost as a result of drug-attributable mortality. Drug abuse may also result in reduced production, as a result of increased absenteeism or reduced on-the-job productivity arising from drug-attributable morbidity.
Health-care costs result from a range of medical conditions attributable to drug abuse. As indicated above, health costs in any given period of time will also be affected by the previous premature mortality of drug abusers who otherwise would still have been alive and imposing costs on the health-care system. Given that the health-care costs of drug abuse come earlier in the typical life cycle than the health-care benefits (that arise from the foregone use of health-care services by those who die prematurely), a drug abuse epidemic would lead to a substantial increase in current health-care costs for a considerable period of time, even if the lifetime costs of drug abusers were lower than those of non-abusers. The health-care costs would exceed the health-care benefits.
Crime costs relate to the policing, judicial and penal expenditures directly and indirectly attributable to drug-related criminal activity. That is one of the most difficult areas of quantification of drug-attributable costs. Although it is widely believed that a significant proportion of burglaries and physical assaults is drug-related, crime statistics are usually inadequate for the purposes of identifying such a relationship. Most of the property stolen by drug abusers is redistributed rather than destroyed and thus, from a social point of view, cannot be treated as a cost of illicit drug use. Only the value of property that is destroyed and the insurance administration associated with losses by drug-related theft should be incorporated into that component of drug-attributable costs.
|Figure II. External costs|
The production and consumption of illicit drugs involve the use of resources that would otherwise be available for alternative production or consumption purposes. If the drugs are produced domestically, the production resources can be considered to be a cost of illicit drug use. If the drugs are imported, they are drawing on foreign exchange that would otherwise be available for the purchase of other goods or services.
It is debatable whether public expenditures on education and research related to drug abuse should be counted as a drug-attributable cost. On the one hand, those expenditures can be considered to be discretionary policy responses to drug abuse rather than directly attributable to that abuse. On the other hand, it could be argued that such programmes should yield benefit-cost ratios of at least unity, so that in the absence of the programmes, social costs would have been higher by at least the cost of those programmes. In the absence of a resolution of that issue, a compromise is to distinguish the social cost estimates from the costs of research and education programmes related to drug abuse.
Intangible costs are additional costs borne by society over and above the tangible resource costs. For example, the premature death of a drug abuser of working age will cause an actual or potential loss of production, but that is not the total loss to society. Such deaths will cause suffering and bereavement to others; the abusers themselves, like most members of society, are likely to value their lives more highly than simply the value of their production. Many societies place a higher value on life as shown by the substantial resources devoted to extending lifespans and improving the quality of life of citizens who are over the age of retirement.
The cost estimates referred to here are aggregate cost estimates of illicit drug abuse. They attempt to compare the actual situation of drug abuse with a hypothetical counterfactual state that would have existed had there been no drug abuse. Such a comparison yields the total costs attributable to drug abuse. There is no suggestion that the counterfactual situation is achievable and, for that reason, no suggestion that there exists a set of feasible policies capable of reducing the drug abuse level to zero. One implication of this is that there is no conceivable set of public policies capable of eliminating the costs of drug abuse. A significant proportion of current costs arises as a result of past abuse. In addition, there can be no prospect of reducing drug abuse to zero. The component of abuse costs that is susceptible to elimination as a result of public policies can be identified as avoidable costs. The extent of avoidable costs indicates the potential benefits available to public expenditures on drug policies. The major public controversies about the most appropriate approaches to drug policies, for example, zero tolerance or harm minimization, make it difficult to estimate what proportion of the total costs of drug abuse is avoidable.
The most efficient drug policies will be those that yield the highest social rate of return. Without the information underlying social cost analysis, it is not possible to undertake policy and programme evaluation. In spite of the high cost of many public policies and programmes, surpassingly little formal evaluation of them is undertaken. In general, there is little information available about how well those investments are performing over time.
A conceptual framework for estimating the costs of substance abuse
Estimates of the economic costs of drug abuse serve several purposes. Firstly, economic cost estimates are frequently used to argue that policies on drugs should be given a high priority on the public policy agenda. Without such a standard for assessing cost estimates, there is a tendency by the advocates for each social problem to overbid, adding in items to make their concern a suitably high (even exaggerated) number. Secondly, cost estimates are helpful in targeting specific problems and policies. It is important to know which aspects of drug abuse involve the greatest economic costs. The specific types of cost may also indicate specific areas where public attention is needed or where specific measures may be effective. Thirdly, economic cost studies are helpful in identifying information gaps, research needs and desirable refinements to national statistical reporting systems. There is probably no better way to set a national research agenda than to conduct a cost estimation study and use it to map out key areas and topics where information is lacking.
The development of improved estimates of the costs of drug abuse also offers the potential to provide baseline measures for determining which policies and programmes are the most effective. International comparisons of reliable cost estimates could provide important indicators of the effectiveness of national policies. Such comparisons could indicate, for example, whether the costs of drug abuse are lower or higher in less restrictive societies, whether the social costs of cannabis are greater in countries where it has been decriminalized or whether there is less drug abuse in countries where a greater proportion of the costs are borne by the individual, other things being equal. Ultimately, cost estimates could be used to construct social cost functions for optimal tax policy and national target-setting. Perhaps most immediately promising is the prospect for cost estimates to be extended to more comprehensive cost-benefit analyses of specific drug policies and programmes.
The present section includes a brief discussion of cost estimation studies to date and a description of a new set of international guidelines for estimating the economic costs of substance abuse. The data requirements for carrying out cost estimation are detailed and areas for further development are discussed.
Cost estimation studies
As noted in a recent review , the majority of studies estimating the economic costs of drug abuse have been conducted in developed countries, in particular in Australia [5, 6], Canada [7, 8], Switzerland , the United Kingdom of Great Britain and Northern Ireland  and the United States [11-13]. Those studies generally use a prevalence-based approach, measuring the costs in a given year associated with the prevalence of substance-related morbidity and mortality attributable to past and present substance use.
The vast majority of cost estimation studies use a cost-of-illness approach, in which the impact of drug abuse on the material welfare of a given society is estimated by examining the direct costs of resources expended for treatment, prevention, research and law enforcement, plus losses of production due to increased morbidity and mortality, relative to a counterfactual scenario in which there is no drug abuse [14, 15]. The focus of cost-of-illness studies is on the tangible social costs of substance abuse, which equal the sum of the private and external costs after adjusting for transfers within society. An alternative model is the "externality" approach, which strictly limits estimates to external costs .
Within the cost-of-illness framework, there are two major variations regarding the valuation of productivity losses due to premature mortality attributable to drug use: the more commonly adopted human capital approach (see, for example, Rice and others ) and the more recent demographic approach pioneered by Collins and Lapsley . In the human capital approach, the lost value of a deceased worker's production is estimated by present earnings plus a discounted rate of future earnings. The demographic approach compares the actual population size and structure to that of an "otherwise healthy" population, that is, an alternative population in which there were no drug-related deaths. Those two approaches are complementary rather than contradictory. The demographic approach is based on the supposition that there had never been any substance abuse or problems associated with the use of psychoactive substances. The human capital approach is based on the supposition that all substance abuse and problems associated with the use of psychoactive substances were to end immediately. The human capital approach generates an estimate of the present and future costs attributable to drug-related mortality in the current year, while the demographic approach estimates the present costs of drug-related mortality in past and present years.
The results of cost estimation studies show sizeable costs attributable to drugs (see table). There is enormous variation in the results. Cost estimates in the 1980s ranged from 349 United States dollars per capita in Canada in 1984  to US$ 51 in Switzerland in 1988 . It is difficult to compare the results of those studies, however, because of differences in methodology and study design. For example, the estimate for Canada included the costs of licit drugs while the estimate for Switzerland was limited to illicit drugs. In all of the studies, the largest economic cost is the productivity losses from drug-related morbidity and premature mortality .
Comparison of selected estimates of the economic costs of illicit drug abuse in various countries
|Author of the study (and year of publication)||Country||Year in which the data were gathered||Original total cost a estimate (millions of local currency units)||Cost (millions of US dollars)||Cost per capita (US dollars)|
|Studies undertaken prior to the development of the international guidelines for estimating the costs of substance abuse|
|Adrian and others (1989) b, c||Canada||1984||Can$ 11 840||8 960||349|
|Rice and others (1990) b||United States||1985||US$ 44 050||44 050||185|
|Collins and Lapsley (1991)||Australia||1988||$A 1 441||1 233||75|
|Faze and Stevenson (1990) c||United Kingdom||1988||£1 820||3 293||58|
|Institut suisse de prophylaxie de l'alcoolisme (1990)||Switzerland||1988||SwF 514||342||51|
|Studies utilizing the international guidelines|
|Single and others (1996 d and 1998)||Canada||1992||Can$ 1 371||1 079||38|
|Collins and Lapsley (1996)||Australia||1992||$A 1 684||1 160||66|
|Harwood and others (1998)||United States||1992||US$ 98 000||98 000||384|
Source: Exchange rates are based on International Monetary Fund,
International Financial Statistics Yearbook 1997, pp. 14-15.
a Including all indirect and direct costs, unless otherwise indicated.
b Including costs of licit as well as illicit drug abuse.
c Including estimates of external costs only.
d E. Single and others, The Costs of Substance Abuse in Canada (Ottawa, Canadian Centre on Substance Abuse, 1996).
International guidelines for estimating the economic costs of drug use
Because of the difficulties in comparing the results of cost estimation studies in different countries, two international symposia were held in Canada under the auspices of the Canadian Centre on Substance Abuse with funding provided by the United Nations International Drug Control Programme and a number of provincial, national and international agencies specializing in addictions. The symposia resulted in the development of a set of guidelines for estimating the costs of substance abuse .The guidelines begin with a discussion of the purposes of estimation studies. Instead of giving details on the exact procedures to be followed in every setting, the guidelines provide a general framework for the development of cost estimates and include a matrix of the types of costs to be considered and a detailed discussion of the following theoretical issues:
(a) Definition of substance abuse;
(b) Determination of causality;
(c) Comparison of the demographic and human capital approaches to cost estimation;
(d) Treatment and measurement of addictive consumption;
(e) Treatment of private costs;
(f) Measurement of intangible costs;
(g) Treatment of non-workforce mortality and morbidity;
(h) Treatment of research, education and law enforcement costs;
(i) Estimation of avoidable costs;
(j) Budgetary impact of substance abuse.
The guidelines conclude with a discussion of future directions that places emphasis on the need to include developing countries in economic cost studies and the implications of the guidelines for research agenda and data collection systems.
The development of the international guidelines has resulted in fewer differences in the methodological approach followed in different cost estimation studies, as the more recent studies have used the same basic cost-of-illness methodology. Nonetheless, there is still considerable variation in results (see table). The estimated costs of illicit drug abuse in 1992 range from US$ 38 in Canada , to US$ 66 in Australia  and to US$ 384 in the United States . The higher estimates in the study conducted in the United States may reflect the fact that there is greater illicit drug use, as well as greater problems and costs arising from such use, in that country, but part of the variation in findings may have resulted from differences in how estimates were made of drug-attributed mortality and morbidity.
Types of costs and data requirements in cost estimation studies
Although studies vary with respect to how the economic costs of drug use are combined into different categories, the costs described below have often been included in prior cost estimation studies [4, 16] or have been recommended for inclusion . The data requirements for each of those major types of costs are also noted. The general cost categories are presented in approximate order of their magnitude in prior cost estimation studies.
Indirect productivity costs
Indirect productivity costs refer to productivity losses due to premature mortality, lower productivity resulting from drug use (such as absenteeism) and the removal of some individuals from the legitimate market economy due to crime and crime careers. Estimating indirect productivity losses (and health-care costs) firstly requires estimates of mortality and morbidity attributable to drug use. Some deaths and hospitalizations are attributable simply because, by definition, they are caused by drug use (for example, drug-overdose deaths). In other cases, drug use is a contri-butory cause that accounts for a certain proportion of deaths and hospitalizations. To estimate the portion of specific chronic diseases attributable to drug use, information on the relative risk of drug use is combined with data on the number of persons using drugs at levels associated with a particular relative risk, in order to generate an estimate of the "aetiologic fraction" or attributable proportion of the cause of disease or death that can be reasonably attributed to drug use. For acute conditions where drug use is a contributory but not a necessary cause (for example, assaults, homicide, suicide and motor vehicle trauma attributable to drug use), the attributable fractions must be determined from special studies based on local information. Once the attributable proportion is estimated, it is then applied to the number of recorded deaths and hospitalizations due to the particular cause in order to generate an estimate of drug-attributable mortality and morbidity.
Thus, in order to estimate morbidity and mortality, prevalence data are required on drug use and on drug use by injection. Also required are the recorded number of deaths and hospitalizations, ideally by cause, age and gender; the list of conditions that epidemiological research has shown to be attributable to drug use; and the associated relative risks. Meta-analyses are available that review the epidemiological literature and estimate the relative risk of drug use for various causes of disease and death (see, for example, English and others  and Fox and others ). Estimates of the attributable fractions for certain causes of death (such as assaults, homicide and suicide) and disease, however, should be based on local information. To standardize the results in terms of per capita rates, data on population structure by age and gender and on life expectancy by age and gender are also required. In order to then estimate the value of lost productivity due to morbidity or premature mortality, data on mean income by age and gender are required (for estimating morbidity costs) and data on present value of lifetime earnings by age and gender are required (for estimating costs of premature mortality).
Health-care costs comprise treatment in general and psychiatric hospitals, co-morbidity costs, ambulance services, residential care, treatment agencies, ambulatory care (physicians' fees and other professional services), prescription drugs and other health-care costs (such as household help and rehabilitation equipment). The required data include hospitalization costs; physicians' fees, costs of other professional services and number of cases seen by physicians and other professional service providers, by age and gender; ambulance costs (total costs, total number of trips, number of trips for drug-related causes); and costs of pharmaceuticals used to treat drug-related conditions (total number of prescriptions and number of prescriptions by cause).
Law enforcement costs
The costs of law enforcement consist of the portion of police, court, corrections and customs costs attributable to drug use. The costs for enforcement of drug laws per se are relatively uncontroversial and all of the associated costs may be included. An argument can be made to include some of the costs for enforcement of property crime (such as burglary or theft) and violent crime (such as assault and homicide); however, estimates of an appropriate attributable fraction of those crimes that can be causally attributed to drug use are generally lacking.
Prevention, research, training and averting behaviour costs
The costs of prevention, research, training and promotion of averting behaviour include drug prevention campaigns, training for physicians and other health professionals, specialized training on drug issues for law enforcement officials, and programmes for the promotion of averting behaviour (such as community crime prevention). The inclusion of such policy costs is debatable, as they concern discretionary expenditures in response to drug abuse rather than costs directly attributable to drug abuse. In most cost estimation studies, those costs are included but identified separately as policy costs. The inclusion of such costs requires data on the costs for prevention and specialized drug training for the health professionals and law enforcement officials.
Administrative costs for transfer payments
The cost of welfare payments, such as social welfare assistance or workers' compensation for persons disabled due to drug abuse, is limited to administrative costs. That is, the actual payouts to recipients are not included, as the productivity lost due to drug-related morbidity is also included. To count both the productivity lost due to drug-related illness and the welfare payments to persons unable to work because of the effects of drug abuse would represent double counting.
Other direct costs
Other costs include fire and accident damage attributable to substance abuse and direct workplace costs, such as the costs of drug-testing in the workplace or the attributable portion of such schemes as the Employee Assistance Programs and other health programmes.
Selected areas for further development
The choice of which costs to include is not a simple issue and it depends in part on the availability of data. If there are no data, it is perilous to exclude a particular cost item, as ignoring it would effectively treat the cost as zero. Indirect estimates, therefore, may be required for some cost items. For example, estimating productivity lost due to absenteeism, high job-turnover and accidents, in most studies, relies on estimates of the lower earnings of drug users or drug-dependent persons. Such estimates do not provide good control of confounding factors that may account for both drug use and lower productivity. Clearly, more exact estimation procedures would be desirable.
Another issue that requires further development is the attribution of crime to drug use. As Brochu and Pernanen note in their article in the present issue of the Bulletin on Narcotics, there is little doubt that drug abuse is a contributory cause for some property crimes and violent crimes. In most countries, high rates of illicit drug use have been found among criminal offenders. In Canada, for example, as many as 80 per cent of convicted criminal offenders reported having used illicit drugs during their lifetime, 50-75 per cent showed traces of drugs in their urine at the time of arrest and close to 30 per cent were under the influence of drugs when they committed the crime for which they were accused. Similarly, disproportionate numbers of drug addicts admitted for treatment have criminal records. Chronic or dependent use of heroin, cocaine or crack, amphetamines or hallucinogens is often implicated as a contributory cause of property crime, particularly burglary and theft. Assault, homicide and other crimes of violence are a result of disputes between sellers and buyers or other sellers in the illicit drug market. Illicit drug users are disproportionately involved in incidents of spouse and child abuse.
While drug abuse is related to crime, the relationship is not always causal, as Brochu and Pernanen point out. The fact that a crime is committed by someone using illicit drugs does not necessarily mean that the use of drugs caused the crime to be committed. There are several plausible causal connections. The pharmacological effects of drugs such as cocaine, other stimulants and phencyclidine (PCP) might induce violence; however, the pharmacological effects of most other illicit drugs would not lead to violent behaviour and some drugs may even have the opposite effect. Most addicts do not commit violent crimes, and those who do commit assault or other forms of violence began doing so before becoming drug-dependent. Thus, the pharmacological effects of drugs are at best only a partial explanation for violent behaviour.
As noted by Brochu and Pernanen, another potential causal connection between drug use and crime is the need for addicts to commit property crimes to support drug use. There are heroin and cocaine addicts who commit property crimes to purchase drugs, but the majority of illicit drug users are not dependent and even most dependent drug users do not commit property crimes. In the majority of cases, addicts who committed property crimes had been engaging in criminal behaviour prior to drug use, and many former addicts continue to commit property crimes even when they no longer use drugs. As with the pharmaceutical explanation, the connection between drugs and crime undoubtedly plays a role in many cases, but it is only a partial explanation.
Brochu and Pernanen also note that some crimes result from territorial disputes between rival distributors, and arguments and robberies involving buyers and sellers on the illicit market. That is most common in areas that are disadvantaged economically and socially, and that have traditionally high rates of violence.
Perhaps the most plausible causal connection is that both criminality and drug use, in particular drug-use dependency, are related to a similar set of socio-demographic and personality variables such as poverty, poor future career or income prospects and low investment in social values. Those factors may represent common underlying causes of both criminality and illicit drug use. Illicit drug use and crime may be mutually reinforcing, but the real cause of both drug use and criminal behaviour may be a complex set of underlying personality and social determinants.
In summary, there is little doubt that drugs are a contributing causal factor in many crimes, but the fact that a crime is committed by a drug addict or by an individual under the influence of drugs does not necessarily mean that the crime can be causally attributed to drug use. Some crimes, in particular crimes of violence, are undoubtedly caused by the use or marketing of illicit drugs. Credible estimates of the proportion of those crimes that are causally attributable to drug use are generally lacking; thus, the article by Brochu and Pernanen fills an important gap, at least with regard to the progress of research in Canada. Similar studies are required in other countries.
Estimation of the economic costs of pharmaceutical abuse is another area in which further research is needed. Most studies are limited to consideration of illicit drug abuse, largely because of difficulties in determining when medication use becomes abuse. A further difficulty in estimating the costs of pharmaceutical abuse is that health-care recording systems, such as the International Classification of Disease, sometimes fail to make a clear distinction between a disorder caused by licit drug use and a disorder caused by illicit drug use.
Further research is also required in order to take advantage of recent advances in willingness-to-pay methods that are used to value the cost of mortality. In the willingness-to-pay approach, information from surveys, insurance data sets or other sources is examined to determine how much people are willing to pay for relatively small changes in the risk of death. From those figures, an estimate of the value of life is produced. While that technique appears to have a reasonably sound theoretical basis, there continue to be considerable problems in terms of the accuracy and consistency of estimates obtained using such an approach. Another difficulty is that the cost estimates from studies utilizing willingness-to-pay methods cannot be compared with the gross domestic product. The results can only be meaningfully compared with the total value of life in a society, a figure that is invariably much higher than the gross domestic product and that generally lacks intuitive meaning. Nonetheless, a growing number of economic studies use willingness-to-pay methods, and future cost estimation studies may be able to utilize that emerging methodology to produce reasonable estimates of the largely intangible costs of premature mortality.
Prospects for expansion of cost studies to more countries
Economic cost studies utilizing the international guidelines are currently under way in several countries in Europe. The data required to conduct such estimates, however, are extremely difficult to obtain in most developing countries. Several developing countries, for example, Chile and Colombia, are planning to estimate the economic costs of drug abuse. A host of economic impacts arising from the production and distribution of illicit drugs will need to be taken into account in those studies, and the international guidelines will need to be revised to take those variables into account.
Prospects for a satellite account for substance abuse in the System of National Accounts
It has been suggested that the development of improved, internationally comparable methods for estimating the costs of substance abuse should be attempted, as far as possible, within the framework of the existing System of National Accounts . The development of estimates of the costs of substance abuse in the framework of the System of National Accounts would be a further step towards the improvement and refinement of national accounting systems, increasing their relevance and usefulness. It took decades for the development of standardized measures such as the gross domestic product under the System of National Accounts, but the ultimate utility of such measures has been worth the time and effort.
Estimating avoidable costs of substance abuse
Economic cost estimates include both avoidable and unavoidable costs. The estimates do not represent the amount of money that could realistically be saved through effective government and social policy and programming. The counterfactual situation in economic cost studies, in which there are no problems associated with drug use, is hypothetical and generally not realizable under any circumstances. Even if completely effective policies could be found with no appreciable costs for enforcement, treatment and prevention programming, implementation would not be instantaneous and there would still be lingering adverse consequences from past use of the psychoactive substances.
Economic cost studies in Australia [5, 6] estimate the percentages of mortality and morbidity, and associated economic costs, that are avoidable. They utilize an "Arcadian normal" , which is the lowest age-standardized mortality rate for the relevant mortality or morbidity category among 20 comparable Western countries. The "Arcadian normal" is used to estimate the lowest percentage of preventable morbidity and mortality yet achieved in any of the chosen countries. While that is an extremely conservative assumption, such a method is nevertheless a useful tool for quantifying the percentage of preventable morbidity and mortality and their associated costs, which can be reduced and ultimately avoided.
Many of the estimated avoidable costs of drug abuse may be reduced or eliminated only over long lead times. There are several reasons for the slow reduction in avoidable costs. Firstly, lead times for policy implementation will not be effective instantaneously. Secondly, even after the implementation of effective policies, there will be long lead times before the health effects of policy changes are achieved. It may take years before the health status of former drug users is equivalent to persons who never used drugs. Finally, as some costs apply to premature mortality, it may take years before there is an appreciable decline in deaths attributable to drug use.
Estimating the cost-effectiveness of interventions
Estimation of the cost-effectiveness of interventions to eliminate, reduce or minimize the harm associated with drug abuse requires, firstly, comprehensive data reflecting the costs of abuse. Without such data, it is not possible to determine the total amount of resources that should be allocated to interventions. Some of those costs may be crude estimates that are necessary for policy purposes, to justify and support resources for interventions and their evaluations. Researchers cite burden-of-illness studies in support of research and of interventions; the same estimates should be required for the costs of illicit drugs. Prior to making any decisions on methodological issues, it should be determined whether the cost-effectiveness of only one particular intervention is being considered or whether the initial policy objective is to determine the relative cost of alternative interventions or programmes.
Cost-effectiveness, effectiveness and cost-benefit analysis
For cost-effectiveness, effectiveness and cost-benefit analysis, essential requirements include determining the costs that are to be included, estimating those costs which are considered essential and for which there are no reliable data and specifying how to measure whether the desired outcome or effectiveness has been achieved. In measuring effectiveness or outcome, it is especially important to specify the time period to be considered. If, for example, the intended outcome of a programme is for an addict to be drug-free, the time period by which effectiveness is to be measured must be determined. That specification can be compared with measures such as years of survival in therapeutic drug trials.
The choice of methodology depends on the question to be answered. If there has already been a decision to undertake an intervention, then the appropriate analysis is a cost-effectiveness, or value-for-money, analysis. In a cost-effectiveness analysis, either the required outcome or the amount of money available is specified, that is, one parameter is fixed. The objective is therefore to determine the least-cost way of achieving a specified objective, for example, providing a predetermined number of addicts with detoxification treatment or providing the largest number of addicts possible with detoxification treatment for a predetermined amount of money. In each case, one parameter is fixed and the relative merits of the intervention are not being evaluated.
Cost-benefit analysis should be undertaken only when the decision on whether to proceed with the programme, treatment or intervention has not yet been made. It is important to remember that cost-benefit analysis, valuing all the identifiable costs and benefits with a common unit of value, reaches a conclusion from a societal point of view, regardless of which groups or persons bear the costs and which groups or persons obtain the benefits. For that reason, the results may prove to be unacceptable to politicians and policy makers. Much of the criticism levelled at cost-benefit analysis fails to acknowledge that. It is worth noting that when the term cost-benefit analysis is used by non-economists, what is frequently meant is cost-effectiveness analysis.
Cost-effectiveness of prevention
Cost-effectiveness of prevention requires special consideration. For example, a significant proportion of the costs of illicit drugs borne by governments is associated with law enforcement, both preventive and punitive. Disaggregation of those two aspects of law enforcement costs could require comprehensive research, would be difficult and would inevitably involve some arbitrary judgements, and the results would not necessarily be transferable or generalizable.
Outcome measurement has special difficulties when assessing the cost-effectiveness of prevention. The question is whether outcomes, such as abstinence from illicit drugs, should be applied only to high-risk groups, to target groups by age and geographical location at greatest risk, or to those with an already high participation rate in consumption of illicit drugs. An alternative would be for outcome measures to be applied to the entire population. Another point for consideration is whether preventive strategies, to be most cost-effective, should focus on preventing initial consumption of illicit drugs or an early intervention for those who are already consuming illicit drugs. Relative cost-effectiveness studies may provide answers in terms of likely outcomes, but those outcomes may not be politically acceptable. The illegal status of many drugs of addiction in most societies may make prevention programmes more expensive than if those drugs were decriminalized. If costs of a preventive programme are calculated to be large in return for a very small gain, such programmes may still be funded so that governments can be "seen to be doing something". That is, a decision to fund a programme may be socially and politically acceptable even when there is a very low cost-effectiveness ratio. While optimization according to cost-effectiveness ratios may not be acceptable, knowledge of ratios can nevertheless be used to inform present and future policy decisions.
Cost-effectiveness of treatment
Alternative treatment modalities are rarely subjected to economic evaluation, though such evaluation presents fewer problems and fewer ambiguities than other areas of drug abuse control. Firstly, the population is much more readily defined than the population for whom prevention strategies may be applicable. Secondly, a number of treatment modalities and therapeutic interventions are currently being provided in a range of countries, so that comparative cost-effectiveness studies within and between countries would be possible. Those treatment programmes include drug-assisted and non-drug-assisted detoxification, psychotherapy, addict support organizations and prescribed methadone. A portion of the costs of some of those programmes may be collected but there is a dearth of economic evaluations of treatment alternatives, making it difficult for policy makers to make informed decisions on such matters.
Law enforcement considerations
Difficulties associated with the disaggregation of law enforcement costs have already been mentioned. The argument concerning costs that are incurred only because a drug is illicit should also be acknowledged, although there is no agreement as to whether such costs should be calculated as part of the costs of illicit drugs. Punitive costs of law enforcement, in particular costs of incarceration, are usually only crude estimates, partly because there are very few data available and because the decisions relating to the inclusion and exclusion of costs are necessarily arbitrary. Although many crimes are associated with the consumption of illicit drugs, it is not appropriate to assign causality to all of those crimes.
Data requirements have been referred to throughout the present discussion. Both gross cost estimates and micro-costing data relating to drug abuse are often non-existent or very crude. In order to improve the use of economic data to enable informed policy decisions to be made, more comprehensive and detailed cost studies are required. There needs to be agreement of objective criteria for measuring outcomes, so that the cost-effectiveness of programmes can be compared over time. Because of the absence of cost data, it is not possible to estimate marginal costs; thus, informed decisions relating to economic size and expansion or contraction of programmes cannot be made.
It should be emphasized that the costs required for estimates include not only money, but also the intangible costs associated with crime, violence, suicide and premature deaths. Economic analysis is needed to provide information on the cost-effectiveness of interventions on both current costs and future costs. When comparisons are being made between the relative effectiveness of programmes designed to reduce drug abuse, some non-paid inputs must also be costed. For example, the costs of services provided by volunteers in non-governmental organizations should be estimated; otherwise real costs will not be identified and transferability will be limited.
The likelihood of compliance and the costs associated with non-compliance require estimation. Thus, there is a need for persons involved in economics and other disciplines, such as behavioural science and epidemiology, to collaborate in the design and evaluation of effective interventions. The estimation of outcome probabilities can be combined with economic estimates to inform policy decisions about programmes for prevention and intervention. While evidence such as that obtained from randomized controlled trials is not usually available to estimate outcome probabilities, the relative rigorousness of the available data should be acknowledged and addressed.
Determining the right policy mix
Drug policy inevitably involves a mixture of intervention strategies. Illicit drug use is discouraged through the application of criminal laws against the production, distribution and use of illicit drugs. Treatment is provided in various ways and preventive education may be conducted through school programmes and the mass media. Long-term investments include basic research on biomedical aspects of drug use, socio-behavioural risk factors and specialized training for health-care and law enforcement personnel. A key task for policy makers is determining the most appropriate mix of the various strategies, all of which are directed at reducing drug-related problems and associated costs. Enhanced research on the cost-effectiveness of specific policies and programmes not only helps in deciding which interventions should continue to receive support and funding, but it also helps policy makers determine the most appropriate mix of strategies to achieve the overall goals of a national drug strategy.
Estimating performance over time
Knowledge of the costs attributable to drug abuse at one point in time is of limited value. While the magnitude of the estimated costs might help in setting general priorities, indicating the importance of drug abuse on the political agenda in comparison with competing concerns, the true value of cost estimates can only be realized when estimates are available from a series of cost estimation studies indicating trends in total costs and the various cost components.
The utility of cost estimates over time is enhanced when the specific outcome indicators and performance measures are set out. That focuses the assessment of drug programming and policy on the specific impacts on cost components. For example, the initiation of enhanced outreach and treatment programming for intravenous drug users might be targeted to the reduction of drug-related crime and crime costs.
While the attribution of cost reductions to specific programmes and policies is at best a difficult undertaking, changes in specific costs attributable to drug abuse can help assess the effectiveness of programming and serve as a general barometer of the effectiveness of a drug strategy. The process is dialectic in nature: the policy and programme mix must be periodically evaluated and adjusted, based on changes in the economic costs and other considerations.
Efforts should be made to promote comparability, across countries and world regions, over time. The development of the international guidelines for estimating the costs of substance abuse is only a first step in that direction. The guidelines should be periodically reassessed and revised, based on the experience in applying them in different settings.
In the present article, the authors examined methodological issues involved in developing more rigorous and comparable data on economic aspects of substance abuse. After discussing the conceptual framework for estimating the economic costs of illicit drug use, it was noted that there were numerous contentious methodological issues and data requirements. Nonetheless, the prospects for future development of cost estimation studies have been improved by the recent development of international guidelines and the growing experience in cost estimation in a number of countries.
There is a great need for improved data on economic aspects of drug issues. Economic cost studies can identify the economic impact of drug abuse on the total value of goods and services in the economy and specify which sectors of the economy bear the greatest costs. Although not all of the economic costs are avoidable, reasonable estimates of avoidable costs are possible. The continuing development of economic analyses, such as cost-effectiveness studies and cost-benefit analyses of policies and programmes, should direct policy makers to make better-informed choices on where to invest and help them to assess performance.
Improving data on economic aspects of drug abuse will necessarily entail improving data collection systems. Economic cost studies to date have identified many knowledge gaps. Particular care must be taken in estimating productivity losses due to drug abuse and in developing a reasonable estimate of the proportion of crime that can be causally attributed to drug use.
There are many potential benefits to be gained by filling such knowledge gaps and improving the general level of understanding of the economic ramifications of substance abuse. Policy makers want to know the answers to four key questions:
(a) The costs of drug abuse to society;
(b) The proportion of those costs that are avoidable;
(c) How best to invest in order to avoid or minimize such costs;
(d) How well those investments are doing.
More complete and rigorous data on economic aspects of drug abuse would provide policy makers with a more comprehensive understanding of the ramifications of drug abuse. In addition, such data would provide synergy for greater multilateral and multisectoral cooperation in prevention.
With leadership from international drug agencies such as the United Nations International Drug Control Programme (UNDCP), a process has already begun that will enhance the development and use of economic data and, ultimately, improve the quality of decisions made on drug issues at the national, regional and international levels.
- United Nations International Drug Control Programme, World Drug Report (London, Oxford University Press, 1997).
- P. Giffin, S. Endicott and S. Lambert, Panic and Indifference: the Politics of Canada's Drug Laws (Ottawa, Canadian Centre on Substance Abuse, 1991).
- A. Markandya and D. Perce, "The social costs of tobacco smoking", British Journal of Addiction, vol. 84, 1989.
- L. Robson and E. Single, Literature Review of Studies on the Economic Costs of Substance Abuse (Ottawa, Canadian Centre on Substance Abuse, 1995).
- D. Collins and H. Lapsley, Estimating the Economic Costs of Drug Abuse in Australia, National Campaign against Drug Abuse Monograph Series, No. 15 (Canberra, Australian Government Publishing Service, 1991).
- D. Collins and H. Lapsley, The Social Costs of Drug Abuse in Australia in 1988 and 1992, National Drug Strategy Monograph Series, No. 30 (Canberra, Australian Government Publishing Service, 1996).
- M. Adrian, P. Jull and R.Williams, eds., Statistics on Alcohol and Drug Use in Canada and Other Countries (Toronto, Addiction Research Foundation, 1989).
- E. Single and others, "The economic costs of alcohol, tobacco and illicit drugs in Canada, 1992", Addiction, vol. 93, 1998, pp. 983-998.
- Institut suisse de prophylaxie de l'alcoolisme, Le Problème de la drogue-en particulier en Suisse-considéré sous son aspect social et préventif (Lausanne, Office fédéral de la santé publique, 1990).
- C. Fazey and R. Stevenson, The Social and Economic Costs of Drug Abuse in the U.K. and the Netherlands (London, Commission of the European Communities, 1990).
- D. Rice and others, The Economic Cost of Alcohol and Drug Abuse and Mental Illness 1985, report submitted to the Office of Financing and Coverage Policy of the Alcohol, Drug Abuse and Mental Health Administration (San Francisco, Institute for Health and Aging, University of California, 1990).
- W. Manning and others, The Costs of Poor Health Habits, RAND Study (Cambridge, Harvard University, 1991).
- H. Harwood, D. Fountain and G. Livermore , The Economic Costs of Alcohol and Drug Abuse in the United States 1992 (Rockville, Maryland, United States Department of Health and Human Services, 1998).
- D. Rice, Estimating the Cost of Illness, Health Economics Series, No. 6 (Rockville, Maryland, Department of Health, Education and Welfare, 1966).
- E. Single and others, International Guidelines on Estimating the Costs of Substance Abuse (Ottawa, Canadian Centre on Substance Abuse, 1995).
- B. Choi, L. Robson and E. Single, "Estimating the economic costs of the abuse of tobacco, alcohol and illicit drugs: a review of methodologies and Canadian data sources", Chronic Diseases in Canada, vol. 18, 1997, pp. 149-165.
- M. French, J. Rachal and R. Hubbard, "Conceptual framework for estimating the social cost of drug abuse", Journal of Health and Social Policy, vol. 23, 1991, pp. 1-22.
- D. English and others, The Quantification of Drug Caused Morbidity and Mortality in Australia, 1992 (Canberra, Commonwealth Department of Human Services and Health, 1995).
- K. Fox and others, "Estimating the costs of substance abuse to the Medicaid Hospital Care Program", American Journal of Public Health, vol. 85, 1995, pp. 48-54.
- B. Armstrong, "Morbidity and mortality in Australia: how much is preventable?", J. McNeill and others, eds., A Handbook of Preventive Medicine (Sydney, Edward Arnold, 1990).