Epidemiologic approaches to the study of drug dependence


1. Distribution of drug dependence in the population
2. Completing the clinical picture
3. Factors affecting the risk of developing drug dependence
4. Identification of new syndromes
Summary and recommendations


Author: Judith B. COHEN, S. Leonard SYME
Pages: 1 to 9
Creation Date: 1973/01/01

Epidemiologic approaches to the study of drug dependence

M.A., M.P.H. Judith B. COHEN
Ph.D. S. Leonard SYME
Professor and Chairman, Programme in Epidemiology School of Public Health, University of California, Berkeley, U.S.A.

Epidemiology may be defined as the study of the distribution of a disease or condition in a population and of the factors that influence this distribution. Information obtained from such study is useful:

  1. In describing the distribution of health and disease in the population;

  2. In completing the clinical picture by the study of cases that do not come to clinical attention;

  3. In estimating the individual's chances and risks of developing the disease or condition; and

  4. In identifying new syndromes based on the observed association of apparently unrelated clinical phenomena in the population [ (1)] . *

These various uses are based on the principle that, in epidemiologic study, whole populations (or their samples) are observed and compared and not particular individuals or patients. Epidemiologic study is therefore helpful in describing the community's burden of disease and disability, in estimating future trends, and in assessing the magnitude of need for community resources and services. Studies of disease distributions may also identify especially susceptible and vulnerable groups in the population leading to a better understanding of causes and means of prevention.

The epidemiologic approach has been useful in the control and prevention of many infectious diseases and is now being vigorously pursued in the study of such non-infectious diseases as coronary heart disease, cerebrovascular disease, and cancer. In contrast, relatively little epidemiologic research has been undertaken in the study of drug usage and dependence. The social problems posed by dependence on drugs are enormous and the challenge to develop effective control programmes is great. Epidemiologic study directed toward identifying particularly susceptible sub-groups in the population as well as possible etiologic factors would therefore be most useful and timely.

1. Distribution of drug dependence in the population

The development of a vigorous programme of epidemiologic research on drug dependence is complicated by several difficulties. One of these difficulties is that of ascertainment. In studying drug dependence, it is obviously necessary to have clear definitions of the phenomenon under investigation and to have the means of detecting the condition when it is present. This is a fundamental requirement of all epidemiologie research. Until the condition of interest can be defined clearly, systematic measurement of the disease in a population is impossible. Clear definition ideally has two aspects.

* Figures between parentheses refer to bibliographical references placed at the end of the article.

First, the condition of interest should be definable in a way that can be clearly communicated across disciplinary and cultural boundaries. Secondly, and equally important, it must be possible to operationally define the condition so that its presence can be determined with satisfactory precision and accuracy, but with sufficient economy to allow measurement on a population-wide basis.

Even where the problem of definition has been resolved, it is extremely difficult to detect cases of drug dependence in the population. Instead, one is likely to see only the "tip of the iceberg". Most known cases involve persons whose dependence has reached a point where it has become a severe liability to themselves or to others within their social settings, and hence their problem has come to the attention of medical or legal authority. Many cases of dependence in the community remain undetected and uncounted.

Problems in ascertainment can result in widely varying estimates of the magnitude of the drug problem. For example, the United States Bureau of Narcotics and Dangerous Drugs estimated that there were 68,088 narcotic addicts in the United States for the year ending 31 December 1969 [ 2] . More than 95 per cent of that number were reported to be heroin users. It is somewhat unsettling to note, however, that the number of new addicts reported to the Bureau doubled from 1968 to 1969. While it is possible that approximately 38,000 new addicts entered the population during one year, it seems equally likely that new methods of detection and reporting contributed to this enormous increase in the estimate. When an estimate can vary by a factor of 100 per cent within one year, it is necessary to regard such figures with caution and to explicitly recognize their imprecise nature.

Further documentation of this imprecision is obtained by noting estimates of narcotic use developed by the United States Bureau of Narcotics and Dangerous Drugs for New York State compared to those developed by the Register of the New York City Department of Health for the same period [ 3] . The City Health Department Register uses a number of different sources including hospitals, addiction service agencies, police and correctional agencies, private physicians, and other health and social agencies. In 1968, the City Register estimated there were 52, 104 heroin users in the city. The U.S. Bureau of Narcotics and Dangerous Drugs, using the broader definition of narcotic use and basing its count on the whole State of New York, estimated there were only 32,240 narcotic users in the entire state. Clearly, our understanding of the magnitude of the drug dependence problem is directly related to the intensity of the search for cases.

Cases of dependence which come to medical or legal attention constitute the "tip of the iceberg ". Estimates of the magnitude of the problem that are restricted to this "tip" will clearly underestimate the extent of the problem.

When an effort is made to compare the magnitude of the drug dependence problem from city to city, or from region to region, or from country to country, the difficulties of developing comparable methods of ascertainment become even more compelling. In their review of similar difficulties in the study of mental illness, the Group for the Advancement of Psychiatry [ 4] cited three major factors making such comparability especially difficult to achieve. These factors seem applicable to the study of drug dependence as well as mental illness and may, in modified form, be summarized as follows:

  1. Social attitudes towards drug dependence differ from group to group and from time to time; these differences affect the number of persons who seek help;

  2. Available resources for treatment and assistance differ from group to group and from time to time; these differences contribute to differences in the number of reported cases; and

  3. Diagnostic skills, fashions and nomenclature differ from group to group and from time to time; these differences affect the number of cases reported in any diagnostic category.

In addition, and unlike mental illness, many forms of drug dependence are illegal: laws regarding drug use vary from group to group and from time to time affecting the likelihood that certain forms of drug behaviour will come to public attention. The extent of these problems varies among countries according to variability in public policy; clearly, such diversity affects comparability of cross-cultural findings.

The development of epidemiologic studies among different social and cultural groups both nationally and internationally is important to enhance understanding of this complex problem. Comparability of research methods for the study of drug dependence is essential to the conduct of such cross-cultural studies. The difficulties described regarding problems of assessment of drug dependence from group to group and from time to time are obviously crucial to the conduct of such international studies. These problems of measurement require solution or, at least, explicit recognition so that their consequences for study results can be estimated.

One method for obtaining information on the submerged portion of the "iceberg" is the cross-sectional study. Cross-sectional studies can provide representative information on population groups or communities and yield useful data on the extent, distribution and character of drug dependence in the entire population. Several recent studies of this kind in the United States among general populations have indicated that the extent of psychotherapeutic drug use, for example, may be of greater magnitude than had previously been anticipated. To illustrate this point, Friedman and his associates [ 5] studied drug dispensing patterns at the Kaiser Permanente Medical Center in San Francisco, California. In this study, drug data were analysed from 220,000 consecutive clinic and pharmacy visits made by 75,000 unselected patients during a three-month period in 1969. During this period, a total of 53,500 prescriptions were dispensed by the pharmacy. Twenty-eight drug products accounted for 26,468, or about half, of this total number of prescriptions. When attention is restricted to these 26,468 common prescriptions, it is interesting to note that 6,552 prescriptions, or 25 per cent, are for drugs classified as tranquillizers. Of the seven most frequently prescribed drugs, four were for tranquillizers, accounting for 53 per cent of these most commonly prescribed drugs.

People may obtain drugs from sources other than from group medical and prescription plans. For example, two national sample surveys [ 6] of adults in the United States conducted in 1967 found that psychotropic drugs were being used rather extensively: approximately one-fourth of each national population sample had used sedatives, tranquillizers or stimulants in the year prior to study. When drugs are in such wide usage, they may of course be used in excess of recommended dosages contributing to the over-all dependence problem.

Such surveys of representative populations at single points in time can provide detailed snapshots of usage patterns prevalent in a population group but they cannot provide data on time trends or on incidence. To detect and study incidence, longitudinal studies are needed which follow populations over time. Such studies permit the description of patterns of disease spread and illuminate factors associated with high susceptibility. From such studies, it becomes possible to pinpoint ways in which prevention and control efforts can be most usefully concentrated.

2. Completing the clinical picture

The preceding comments have suggested that cases coming to official attention constitute only the tip of the iceberg. Beyond this, it should be emphasized that those cases coming to attention are themselves composed of sub-groups with differing characteristics; cases of dependence which come to official attention cannot be considered a homogeneous category. An illustration of this issue is provided by the work of Turner and his associates [ 7] . In their study of schizophrenia, these investigators have provided impressive documentation of the fact that different cases of this disease are differentially perceived by record-keeping agencies. In the conduct of this study, the investigators used the Monroe County (New York) Psychiatric Case Register which records virtually all psychiatric contacts-whether diagnostic or treatment, inpatient or outpatient, public or private-that occurred within Monroe County. The research questions posed by the investigators were as follows: how many of these known register cases would have been detected by the usual ascertainment techniques? Do different ascertainment techniques yield different pictures of the magnitude and nature of the problem?

To answer these questions, the investigators drew from this register a random sample of white, male, schizophrenics, aged 20-50 years. All of these men had been reported to the register for the first time between 1 January 1960 and 30 June 1963, had no prior history of psychiatric hospitalization, and had received a diagnosis of schizophrenia during the 3½ year period. The clinical status of the 212 men chosen in the sample ranged from those who were nearly asymptomatic to those who were severely impaired, and their total psychiatric experience varied from a single outpatient visit to many lengthy periods of hospitalization. Four different but overlapping ascertainment techniques were used:

  1. First admissions to the State Hospital during the 3½ year period;

  2. First admissions to one or more of the public and private psychiatric inpatient facilities in the county;

  3. Any subject seen at some time by a private psychiatric practitioner during the 3½ year period; and

  4. Any subject who had sufficient contact with any public or private outpatient clinic to be placed on their records whether or not treatment was actually received.

Obviously, no one of these ascertainment methods accurately reflected the total magnitude of the problem in the county; each method, however, revealed a portion of the picture. More importantly, the various ascertainment methods yielded quite different prescriptions of the patient population. The patient populations detected using these four methods differed in terms of age, social class, marital status, living situation, occupation, and education. Had a research project depended upon only one or another of these approaches, a highly selective and limited perspective on the total problem would have been obtained. While this study was concerned with schizophrenia, it would seem that its conclusions have implications for the study of drug dependence as well.

Even when all possible ascertainment sources are utilized in detecting cases, those cases that do come to official attention cannot be considered a representative sample of all those affected by the problem. In almost every disease or condition which has come to the attention of the epidemiologist, those who seek help or in other ways come to official attention are inevitably different from those who do not. This suggests the limitation of using observed drug cases to develop an understanding of drug dependence. Thus, while ethnographic descriptions of drug use in particular settings, groups, or times have provided useful details on the nature of dependence under particular sets of circumstances, and among various selected groups, they have a very limited generality for understanding problems of drug dependence in the general population.

This limitation is analogous to the study of characteristics of those who die from the late effects of syphilis as descriptive of all those who have contracted this disease. Indeed, information obtained from this latter type of study would lead to grossly inappropriate inferences and programme planning regarding the current venereal disease epidemic. As in the problem of drug dependence, officially recognized cases of syphilis do not permit the description of the nature, extent, or characteristics of the problem in the population.

It has been argued by some, however, that we need be concerned only with those cases of drug dependence which come to medical or legal attention since these are obviously the most severe and troublesome problems. In fact, we have no idea of the consequences that hidden cases of dependence have for morbidity and mortality from many other causes. Thus, although one consequence of being dependent on drugs is that the problem is brought to the attention of officials concerned with drugs, other consequences of drug dependence may be less obvious and may involve such outcomes as automobile accidents, suicide, or criminal behaviour where the connexion to drugs may not be evident. It would be of interest to assess drug use patterns among those involved in crimes of violence or among suicides. This would be especially interesting to study among members of the middle and upper social classes where information on drug dependence is not typically sought or apparent.

Epidemiologic research depends upon the reliability and validity of ascertainment techniques. While these techniques will never be perfect, it is important that we have a better understanding of the consequences that follow from the use of various observational approaches. Thus, it would be useful to know which surveillance procedures and measurements are most appropriate for which classes of drugs and for which groups of people; similarly, it would be useful to have better estimates of the comparability that exists between various alternative ascertainment approaches. Only in this way can we obtain the essential information necessary to complete the clinical picture of drug dependence.

3. Factors affecting the risk of developing drug dependence

The foregoing considerations are primarily directed towards the dual challenges of assessing the magnitude of the drug dependence problem in the population and in assuring that a representative picture of that problem is obtained. An additional and major concern of epidemiologic research is the investigation of high-risk or susceptible sub-groups in the population so that etiologic factors can be described. Some major issues here include the following questions: Are the personality or social characteristics of those dependent on different types of drugs in a population similar or dissimilar? [ 8] Is the particular drug used the result of a specific set of causal influences or simply a result of the relative availability or acceptability of that drug in a particular social milieu?

Unfortunately, most studies directed towards these questions have been retrospective and have focused on the description of the psychological or social histories of persons already dependent in an effort to discover the causes of the dependence. Reviewing such studies of those who are already dependent, Abrams et al. [ 9] noted that the "dependent, immature" personality type often described is tautologically based on information about those who are by definition already dependent on drugs. There is no way of determining whether the so-called etiologic factors preceded or followed the condition of drug dependence. This is a limitation of many retrospective studies and seriously affects their usefulness.

In addition, studies which have attempted to describe etiologic factors associated with drug dependence have rarely utilized appropriate control groups. Factors found to have strong associations with drug behaviour in such studies may or may not be of etiologic importance; without appropriate control groups, it is difficult to determine whether or not an observed association is in fact unique to excess drug use. One example of this difficulty is illustrated by the oft-heard statement that marihuana use leads to heroin addiction. Most of the research evidence cited in support of this belief is based on the observation that the vast majority of heroin addicts have had previous experience with marihuana. Before concluding that marihuana use does in fact lead to heroin addiction, however, it would be important to observe the extent of marihuana use in other appropriate comparison groups. It is possible that marihuana use is quite prevalent in such other groups as well. The question at issue is the risk of heroin use which follows upon the use of marihuana. Research is needed to determine the proportion of marihuana users who subsequently become addicted to heroin. Studies of heroin addicts without such control groups cannot retrospectively illuminate this issue.

The prospective study is specifically designed to provide data on time sequences among affected and non-affected persons. Thus, in the prospective design, potential etiologic factors can often be measured prior to the onset of dependence. Not only can questions of time order be clarified in this way but assessments can be made regarding the predictive usefulness of these factors. It would be of great interest and usefulness to observe the beginnings of drug use and abuse over time in a group initially free of drug use. It would also be most interesting to observe the evolutionary consequences of drug dependence over time. Thus, some patterns of dependence may lead to organic pathology, others to social dysfunction, and others to mental illness. Studies of the natural history of drug use would be most helpful in enhancing understanding of etiology, management, prevention, and control. Of particular interest in this regard would be the initiation of follow-up studies of young persons dependent on drugs. What happens to such persons over time? Can factors be identified which predict favourable outcomes for some and unfortunate outcomes for others? Interesting and provocative evidence regarding the natural history of alcoholism has recently been provided by Cahalan [ 10] . In this research, Cahalan found that drinking problems disappeared among many persons as they became older. Winick [ 11] observed a similar phenomenon regarding narcotic addiction, noting a sharp decline in addiction as patients reached their mid-thirties. Winick referred to this as a" maturing-out" process. O'Donnell [ 12] , Vaillant [ 13] , and Maurer [ 14] have also pointed to this phenomenon and have reached similar conclusions. The implications of a "maturing-out" process for treatment and control programmes is obvious and this phenomenon certainly deserves further careful and detailed consideration. Such longitudinal or follow-up studies are very difficult of execution; however, the ultimate goal of such investigation is worthy of the effort involved: the identification of causative factors and their elimination or modification.

4. Identification of new syndromes

Earlier in this paper it was suggested that epidemiologic approaches to the study of drug dependence served four purposes. To this point, attention has been directed towards fundamental problems affecting [ 1] ascertainment of the extent of dependence in the population, [ 2] description of the characteristics of the affected population and [ 3] identification of causal factors. A fourth use of epidemiologic study has not yet been discussed in any detail but has an important bearing on the clarification of these problems. This fourth use concerns the identification of new syndromes based on the association of apparently unrelated behavioural phenomena in the population. This point is based on the observation that there are susceptible, vulnerable, or high-risk individuals and groups in the population and that their drug use behaviour is merely one reflection of this vulnerability. In epidemiologic terms, the host is vulnerable to some sort of dependence and drugs can be seen as one agent to satisfy this vulnerability due to a fortuitous set of circumstances in the environment. In other environmental settings, other means of coping with vulnerability might be utilized and might include a wide range of self-destructive actions ranging from passive neglect of health to avoidance of medical services to excessive intakes of food, tobacco and alcohol to accident-proneness and suicide.

The implications of this view are that research on drug dependence should perhaps not focus narrowly on drug use behaviour but should address itself to the broader issue of generalized vulnerability. Thus, research might profitably be directed towards the study of factors which affect a person's vulnerability not only to drug dependence but to a wider range of deviant behaviours. One such factor that has recently been suggested in this connexion is the concept of "anomie" or alienation [ 15] which has been found useful in describing a person's vulnerability not only to the abuse of drugs but also to a broad range of other deviant behaviours such as suicide and juvenile delinquency [ 16] [ 17] [ 18] .

This broader view of drug use has obvious implications for the development of appropriate methods of measurement and ascertainment; epidemiologic studies of this kind would hopefully generate data on the association of different behaviours in the community so that new coping or adaptive behaviour patterns could be identified. In this way, population studies which focus on a wide range of coping behaviours may extend our understanding of the fundamental nature and causes of drug dependence and thereby yield important insights for control and prevention. Thus, epidemiologic approaches to the study of drug dependence may be useful not only in describing that phenomenon but in viewing this behaviour in its wider social and cultural context. To the extent that drug dependence is intimately bound up with societal functioning, this broader epidemiologic perspective may be essential to the development of control and prevention programmes.

Summary and recommendations

Epidemiologic research can be of value in the understanding and control of drug dependence by describing the distribution of this problem in population groups and in helping to identify factors that influence this distribution. Current epidemiologic research on this problem is complicated by several difficulties which have been discussed above.

Specific recommendations for future work on these research issues may be summarized as follows:

  1. Problems of ascertainment call for additional work in two related areas. Clarification of nosologic and measurement issues on the concept of drug dependence is needed. Resolution of these problems will require the design and execution of comparable descriptive studies in a variety of cultural settings. Such studies must provide for the comparison and evaluation of different ascertainment techniques.

  2. Research efforts to obtain a more representative picture of drug dependence in whole populations are needed to complete the clinical picture on this problem. Since the behaviour under study is defined as illegal in many populations, unobtrusive or non-threatening measurement techniques will have to be developed to provide a less biased picture of the nature and extent of drug dependence.

  3. Retrospective case-control studies are needed in the search for etiologic factors. While this method of study has its limitations, it is an approach which economically generates important etiologic hypotheses. Only after such appropriate retrospective research has been done should one be willing to undertake the expensive and difficult task of developing prospective, longitudinal studies of drug dependence in the population. Such prospective follow-up studies among healthy populations are necessary to test etiologic hypotheses and to accurately assess relative risks but they should be initiated only after sufficient preliminary research has been completed.

    Another type of follow-up study begins not with "healthy populations" but with those already dependent on drugs. Natural-history studies are needed in which cohorts of identified users are followed over time. Among the several uses of this type of study, this approach can provide valuable information on factors associated with the "maturing-out" of dependence.

  4. While epidemiologic research directed specifically to drug behaviour is to be encouraged, it is also possible that a broader understanding of this problem will be obtained by the more general study of vulnerabilities and deviant behaviours. Research attention should be focused on studies of factors affecting vulnerability to many types of deviant behaviour. Varying definitions and operational measurements of vulnerability should be compared to identify common and disparate elements. Aspects of vulnerability associated with varying types of deviant behaviour should be tested as predictors of alternative behavioural outcomes.

This recommended programme of research can extend our understanding of the problem of drug dependence. By studying drug behaviour in community and population settings, the special importance of social and cultural influences is emphasized. Enhanced understanding of the role of these factors is essential to the development of community and environmental intervention programmes of primary prevention. Comparative international research provides a unique opportunity to study these social and cultural factors. It is only within the framework of cross-cultural study that a full range of both etiologic and environmental parameters is available (19). The advantage of an international perspective is that the potential variation of dependence and etiologic characteristics can be maximized. All of the recommendations summarized above, therefore, would benefit from such collaborative international research.



J. N. Morris, Uses of Epidemiology. Baltimore, Williams and Wilkins, 1964.


President's Commission on Law Enforcement and Administration of Justice. Task Force Report on Narcotic and Drug Abuse. Washington, D.C., U.S. Government Printing Office, 1967.


L. C. Richards, and E. E. Caroll, "Illicit drug use and addiction in the United States: review of available statistics ". Pub. Health Rep. 85:1035-1041, 1970.


Group for the Advancement of Psychiatry. "Problems of estimating changes in frequency of mental disorders". Report No. 50:469-517, New York, August 1961.


G. D. Friedman, M. F. Collen, L. E. Harris, E. E. Van Brunt and L. S. Davis, "Experience in monitoring drug reactions in outpatients: the Kaiser-Permanente drug monitoring system ". J.A.M.A. 217:567-572, 1971.


H. J. Parry, "Use of psychotropic drugs by U.S. adults ". Pub. Health Rep. 83:799-810, 1968.


R. J. Turner, L. J. Zabo, J. Raymond and J. Diamond," Field survey methods in psychiatry: the effects of sampling strategy upon findings in research on schizophrenia ". J. Health & Soc. Beh. 10:289-296, 1969.


S. L. Syme, "Personality characteristics of the alcoholic ". Quart. J. of Studies on Alcohol. 18:288-302, 1957.


A. Abrams, J. H. Gagnon and J. J. Levin, "Psychological aspects of addiction ". Am. J. Pub. Health 58:2142-2155, 1968.


D. Cahalan, Problem Drinkers. San Francisco, Jossey-Bass, 1970.


C. Winick, "Maturing out of narcotic addiction ". Bulletin on Narcotics, XVI, 1.


J. A. O'Donnell, "The relapse rate in narcotic addiction: a critique of follow-up studies ". In Wilner, D. and Kassenbaum, G. (eds.). Narcotics. New York, McGraw-Hill, 1965.


G. Vaillant, "Twelve-year follow-up of New York narcotic addicts ". New Engl. J. Med. 275:1282-1287, 1966.


D. W. Maurer and V. H. Vogel, Narcotics and Narcotic Addiction. 3rd ed. Springfield, III., Chas. C. Thomas, 1967, pp. 225-229.


L. Srole, "Social integration and certain corollaries ". Am. Sociol. Rev. 21:709-716, 1956.


M. Zaks, "Anomia, psychological dysfunction, and self-injurious behavior: a study in suicide potential ". Abstract VIII, Interamerican Congress of Psychology, April 1963.


I. Chein, The Road to H: Narcotics, Delinquency and Social Policy. New York, Basic Books, 1964, pp. 82-104.


M. Dolkart, et al. "Suicide preoccupations in young affluent American drug users ". Bull. Suicidology 4:70-73, 1972.


S. L. Syme, "Contributions of social epidemiology to the study of medical care systems: the need for cross-cultural research ". Medical Care 9:203-213, 1971.