The Epidemiology of Opiate Addiction in the United States
The Incidence and Prevalence of Opiate Addiction in the United States
JOHN C. BALL, DAVID M. ENGLANDER, AND CARL D. CHAMBERS
In 1928 Terry and Pellens in their classic work The Opium Problem estimated the extent of chronic opiate use in the United States.' They reviewed local clinic records, state studies, and one national survey of opium use. From four studies of opiate use during the period from 1915 to 1920, they estimated that there were from 140,554 to 269,000 addicts in the continental United States. These estimates were limited to ad hoc extrapolations from the available data, as they did not derive a method for calculating the prevalence of opium use in the nation. Furthermore, as these figures were based upon reports from physicians, clinics, and public health authorities and hence dealt only with known addicts, Terry and Pellens held them to be conservative estimates of the extent of opium use.
From 1928 to the present, numerous and varying opinions have been expressed as to the extent of opiate use in the United States.' For the most part, however, the published figures have been mere guesses inasmuch as they are not extrapolations from either local, state, or national data (as was the case with Terryand Pellen's figures), and no method of estimation is employed.
In the present paper we present a method of calculating the incidence and prevalence of opiate addiction for the United States. In deriving our method, two principles were adopted. First, we accepted only enumerations of known addicts as a source of data. Second, our estimates had to be derived from local, state, or national figures according to a method which could be verified by others.
Our method of determining the extent of opiate addiction in the United States is based upon three sources of data: (a) The File of Active Addicts of the Bureau of Narcotics, (b) the New York City Department of Health Register, and (c) addict admissions to the NIMH hospitals at Lexington, Kentucky and Fort Worth, Texas.
The Bureau's File of Active Narcotic Addicts
This is a national law enforcement file of addicts known to local, state, and federal authorities. Most of the reports sent to the Bureau come from local police departments. On December31, 1967, there were 62,045 active addicts in this file.' Of these,6,417 (or 10.3%) were new addicts reported for the first time during the year. The remainder (55,628) were addicts first reported during the prior four years as well as those rearrested during this period. If not reported for five consecutive years, an addict is removed from the Active File." Thus, the Bureau has both incidence and prevalence data concerning the extent of opiate use in the United States.
The particular epidemiological usefulness of the Bureau's file is that it is national in scope. It includes all fifty states (but not Puerto Rico) and has received reports in recent years from forty-nine states. The obvious limitation of this file is that it is based upon police reports. It does not include medical or agency sources of data pertaining to opiate addiction and it does notinclude voluntary addict admissions to the two federal hospitals.
The New York Narcotics Register
The Narcotics Register of the New York City Department of Health was established in 1966.6 It receives case reports of drug abuse in the city from private physicians, hospitals, social agencies, police, and correctional institutions. During 1967 the Register added 10,870 opiate addicts to their file. For the fouryear period from 1964 to 1967, the Register enumerated 35,822 opiate addicts.'
The particular epidemiological usefulness of the Narcotics Register is that it affords the most complete local enumeration of opiate addicts in the nation. By including medical and agency reports as well as those from police sources, it has greatly expanded our knowledge of the extent of drug addiction in New York City. The Register has, to date, only tabulated incidence data; a procedure for removing addicts from the Register has not been established.
The Lexington and Fort Worth Hospitals
The Lexington and Fort Worth Psychiatric Hospitals accepted voluntary addict patients from throughout the United States until implementation of the Narcotic Addict Rehabilitation Act in 1968. These two hospitals obtained a national sample of patients who were freely admitted and able to leave at any time. The fact that medical treatment was voluntary, confidential, and free resulted in a hospital population of considerable diversity. In 1966, 2,774 addicts were admitted to the two hospitals.'
The particular epidemiological usefulness of the Lexington and Fort Worth data is found in the comprehensiveness of hospital records based upon direct medical examination and drugdiagnosis over police or agency reports, and secondly, with
respect to the present analysis, the completeness of reporting from the Southern states.
The Three Sources of Data Combined
Each of the three sources of information pertaining to opiate addiction has a particular validity as well as definite limitations. The Bureau's file is the most complete national enumeration of opiate users, but it is restricted to addicts known to law enforcement agencies. The Register has effected the most comprehensive case reporting of drug abuse in the nation, but it is restricted to New York City. Addict admissions to the Lexington and Fort Worth Hospitals provide the most valid and detailed epidemiological data about drug use in the nation, but the results are restricted to those who seek medical treatment. Thus, each of these sources has definite usefulness in the study of incidence and prevalence of opiate addiction.
The reasons why we have decided to combine three sources of reporting in devising our method of estimation were both practical and methodological. Practically, we found it necessary to scrutinize and then combine various sources of data because the only operational national file was inadequate. No national register of cases exists. But more significant, perhaps, was our methodological conviction that a statistical combination of sources was superior to reliance upon any one reporting agency. We knew that each of the three sources-the Bureau, the Register, and Lexington-included addicts not known to the others. We knew that enumeration was increased as additional sources of information were added. We also believed that multiple independent correction procedures were superior to reliance upon any one file or register. And finally, we held that a valid method for calculating incidence and prevalence of opiate addiction must integrate the available data from diverse sources according to a documented procedure. For only a verifiable method would enable further improvements to be made in estimation procedures.
Method for Calculating Incidences8
Our method for calculating the incidence of opiate addiction in the United States is as follows:
I. Method for Incidence, using 1967 data .
Part A. For Thirty-eight Heroin States
1. New addicts for heroin states = 6,322 (Bureau)
2. New addicts for New York City = 2,590 (Bureau) (94.4% x 2,743)
3. New addicts in New York City = 10,870 (Register)
4. Correction Ratio for Incidence = 4.20 (10,870 _ 2,590)
5. Calculated new addicts during 1967 in thirty-eight states = 26,552 (6,322 x 4.20)
Part B. For Twelve Southern States
1. Three-year average of new addicts = 129.3 (Bureau)
2. Percent of active addicts in twelve states who are heroin addicts = 33.9 (Bureau) (466 - 1,196)
3. New heroin addicts in twelve states = 44 (Bureau) (.339 x 129.3)
4. Calculated new heroin addicts in twelve states = 156 (44 x 3.54; the latter figure is the heroin correction ratio, see below)
5. Percent of heroin addicts in these states = 3.6 (Lexington) (14 -= 390)
6. Calculated new opiate addicts in twelve states = 4,333 (156-.036)
Part C. Total incidence of opiate addiction for U.S. in 1967 = 30,885. (A+B=C)
II. Method for Prevalence, using 1967 data
Part A. For Thirty-eight Heroin States
1. Active addicts in thirty-eight states = 60,925 (Bureau)
2. New York City five-year prevalence = 46,692 (Register) (35,822 + 10,870)
3. Active addicts in New York City = 30,543 (Bureau)
4. Correction Ratio for Prevalence = 1.53 (46,692 - 30,543)
5. Calculated prevalence for thirty-eight states = 93,215 (60,925 x 1.53)
6. Incidence to Prevalence Ratio = 3.51 (93,215-.'-- 26,552)
Part B. For Twelve Southern States
1. Incidence in twelve states = 4,333 (from Part I. B.6)
2. Calculated prevalence in twelve states = 15,209 (4,333 x 3.51)
Part C. Prevalence of Opiate Addiction in U.S., 1967 = 108,424 (A+B=C)
Note: Heroin correction ratio was derived by comparing the number of new heroin addicts in New York City as reported to the Bureau with new addicts reported to the Register: 9,066 -'.2,562 = 3.54.
Discussion of the Method
The New York City Narcotics Register was used as a basis for correcting the underenumeration of addicts by the Bureau outside of the Southern states. This was done because the Register provides the most complete local enumeration of drug abuse in the United States. In addition, some half of the known opiate addicts in the United States are included in the Register because of the marked concentration of these addicts in our largest city." As a consequence of these two considerations-completeness of reporting and a catchment area which includes the most endemic area of opiate use-we accepted the Register's compilations as our basis for calculating both the incidence and prevalence of addiction from the Bureau's file of active addicts.
It was necessary to devise a separate means of estimating incidence and prevalence for those Southern states in which heroin was not the principal opiate of addiction. There were two reasons for a separate procedure in the Southern states. The first reason was that the entire pattern of opiate addiction is quite different in these states. Heroin is not the principal opiate used, drugs are obtained from legal or quasi-legal sources, and most of the addicts are working-class or middle-class whites. Thus, the epidemiology of opiate addiction in the Southern states is quite different from that found in the northern metropolitan areas of heroin use.10
Secondly, the gross underenumeration of addicts by the Bureau of Narcotics in these nonheroin states required a separate and independent correction procedure for this part of the nation. Most of the southern addicts have not been reported to the Bureau; they are not included in the File of Active Addicts. Thus, the number of first admissions per year to the two federal hospitals from these twelve states has consistently been almost twice the number of new addicts reported to the Bureau from these same states.11 The extent of this underreporting is such that the number of addicts admitted to Lexington and Fort Worth in a given year sometimes exceeds the Bureau's total count of active addicts in a given state.12 In this instance Lexington-Fort Worth admissions for one year exceed the Bureau's prevalencel The reason why the Bureau has not enumerated most of the southern addicts is that their attention has, of necessity, been focused upon the suppression of heroin traffic in our metropolitan centers."
In deriving our estimation method for the twelve southern nonheroin states, we made use of the following data from Lexington and the Bureau. The designation of twelve states as nonheroin states was based upon previous epidemiological research at the Lexington Hospital. 14 The number of active addicts, new addicts, and heroin addicts in these states was obtained from the Bureau's reports and files.'15 The correction procedure employed (in our Incidence Method, Part B) is to utilize the finding that only 3.6 percent of 390 addicts admitted to the two federal psychiatric hospitals from these twelve states were heroin users,'° while the Bureau reports 33.9 percent heroin users.17 What we did for the twelve Southern states, then, was to effect a twofold extrapolation from the Bureau's figures. First, we calculated the extent to which heroin addicts were underreported (using the ratio obtained from our analysis of the Register's reports)."' Second, we calculated the extent to which nonheroin addicts in the twelve states are underreported to the Bureau. This combined extrapolation resulted in a 33-fold increase in incidence and a 13-fold increase in prevalence over the Bureau's tabulation for the Southern states.
Discussion of Concepts and Findings
Among the questions which are often considered in studies of incidence and prevalence are the following: (a) definition of the disease, (b) determination of age at onset, (c) etiology, (d) extent of unreported cases, (e) duplicate reporting of cases, (f) duration of illness, (g) treatment, (h) concepts of prevalence, (i) rates by age and sex, and (j) historical changes pertaining to the disease. Each of these points will be briefly discussed with respect to opiate addiction in the United States.
Scientifically and medically, definition of the disease is not a problem.19 Opiate addiction is a clearly defined biochemical state or disease. An opiate addict is someone who is physically dependent upon one of these drugs, and his addiction can be clinically determined. Two valid procedures for ascertaining current opiate abuse are urinalysis and observation during withdrawa1; 20 the latter is the most conclusive means of establishing addiction status.
The age at which onset of opiate use occurs can be definitely established. Addicts recall quite accurately the circumstances of their initial shot of heroin as well as when they subsequently become hooked (addicted ).21 Still, this crucial question of onset has been neglected. Of the three major sources of data discussed in this paper, only the two federal hospitals routinely obtain age of onset information. And from this source it is known that both heroin use and prior marihuana smoking commence during adolescence--before age twenty.22
In many respects opiate addiction is similar to both juvenile delinquency and venereal disease. As with delinquency it is a peer-group phenomenon pursued in a recreational or street setting. As with venereal disease it is commonly transmitted by personal contacts which are intentional. In this sense opiate addiction is a self-inflicted disease of adolescence. Why only some boys in a family, or neighborhood, become addicts, while others do not, is unknown. As yet, there is no evidence that addicts are psychologically inadequate, or mentally ill, before onset of drug use.
With regard to the extent of unreported cases of opiate addiction, we have found that most addicts are known to either legal or medical sources .23 We have been unable to find any empirical support for either of two contrary viewpoints concerning unreported cases: that either the cases reported to the police are an exaggeration or that a vast number of addicts exist who are unknown to legal or medical authorities.
We have found that the principal problem of underreporting pertains to prevalence rather than incidence. Thus, most cases of opiate addiction come to the attention of medical agency or police sources at some time, but the uncertainty of such contact as well as the problematic duration of illness tends to reduce estimates of prevalence based on current reports. Still, almost any method is preferable to none, so one must accept the arbitrary five-year prevalence period until such time as a more adequate procedure is established.
Duplicate reporting of cases is a practical problem involved in maintaining a file which seeks to identify persons as well as cases. Ordinarily, this problem can be solved by the establishment of a special file which identifies both persons and cases. Each of the three files reviewed-the Bureau, the Register, and Lexington-has such a procedure. In our method of combining these sources, we employed a procedure which avoided duplicate counting of addicts.
Information concerning the duration of illness is meager with respect to opiate addiction. This deficiency in our knowledge especially affects prevalence rates. For this reason further followup studies of addicts are needed to establish yearly readmission and death rates.
Withdrawal from opiate dependence is now routinely accomplished under medical supervision. In this respect, treatment is almost invariably successful. The problem, however, is that the addict usually relapses to drug abuse shortly after withdrawalas occurs with alcoholism. In large measure this relapsing characteristic of opiate abuse has led psychiatrists to regard addiction as a type of mental illness. As yet, no treatment has been found which will prevent former addicts from returning to drug use.
Prevalence has been most commonly defined as the number of active cases existent at a given time and place. The further concept of life-time prevalence has been advanced. The idea of life-time prevalence (whether one has ever had a disease) would have a usefulness in epidemiological research pertaining to opiate addiction if combined with our more usual definition based on physical dependence. In this manner we would know how many persons present in a specified population ever were addicted as well as how many currently were addicted. Such data are not presently available.
Opiate addiction is more prevalent among males than females in the United States. In each of the three files reviewed, the sex ratio was over five hundred. With respect to age, addiction tends to decrease after age forty, although further study and the computation of specific rates is indicated.
There seems to be little doubt that opiate addiction has markedly decreased in the United States since the early part of this century. Terry and Pellens reported that there were some 200,000 opium users in the 1915 to 1920 era. Our figure of 108,424 for 1967 is considerably below this, despite the increase in population from 105,710,620 in 1920 to 179,323,175 in 1960. Current prevalence, then, is some one-fourth of what it was fifty years ago. Improved medical practice, increasing public awareness of the dangers of abuse, and stricter law enforcement would appear to have been responsible for this decline in opiate use.
We have advanced a method for estimating the extent of opiate addiction in the United States. It is based upon a combination of the three most comprehensive sources of data pertaining to addicts-that of the Bureau of Narcotics, that of the New York City Narcotics Register, and that of the two NIMH hospitals. A separate method for calculating the incidence and the prevalence of addiction was derived. These estimation procedures can be employed for any recent time period, and they are subject to verification. In 1967 we calculated that there were 108,424 known opiate addicts in the United States. Of this number, 30,885 were first reported during 1967.
1. See Chapter 3.
2. On the one hand, the Bureau's enumeration has been regarded as an overestimate-Chein, Isidor, et al.: The Road to H. New York, Basic Books, Inc., 1964, pp.7-9. On the other side, the Bureau's figures are considered as a gross underestimate; for example, one author holds that there are some one million opiate addicts in the nation-Nyswander, Marie: The Drug Addict as a Patient.New York, Grune and Stratton, Inc., 1956, p.13. Another point of view is that the extent of the problem cannot be estimated-Lindesmith, Alfred R.: The Addict and the Lam. Bloomington, Indiana, Indiana University Press, 1967. Ch.4
3.'Active Narcotic Addicts as of December 31, 1967. Annual Report, Bureau of Narcotics.
4. Addicts who receive sentences in excess of five years are removed from theActive category, as they are not at risk while incarcerated.
5 Kavaler, Florence, Densen, Paul M., and Krug, Donald C.: The narcotics register project: early development.British Journal of Addictions, 63:75-81, 1968.
6 Amsel, Zili, Erhardt, Carl L., Krug, Donald C., and Conwell, Donald P.: The Narcotics Register: Development of a Case Register. Paper presented at the 31st Annual Meeting of the Committee on Problems of Drug Dependence, National Academy of Sciences-National Research Council, February25, 1969, Palo Alto, California, Table 111.
7 Of the 2774 patients82.6 percent were voluntary admissions, and17.4 percent were federal prisoners.
8 For further technical details concerning the procedure followed for calculating incidence and prevalence, see Englander, David M.: Incidence and Prevalence Calculation. Unpublished paper, July 21, 1989. Available from authors.
9. Annual Report, Bureau of Narcotics: op. cit
10 See Chapter 5.
11. See Chapter 7, Table 7-11; Annual Report, Bureau of Narcotics: op. cit., Table 4.
12 O'Donnell, John A., and Ball, John C.: Narcotic Addiction. New York, Harper and Row, 1966, p.9.
13 "Maas, Peter: The Valachi Papers. New York, G.P. Putnam's Sons, 1968, Ch. 11-13.
14."See Chapter 7.
15 Annual Report. Bureau of Narcotics: op. cit.
16 See Chapter 7.
17 ''Bureau of Narcotics and Dangerous Drugs, June, 1969.
18 Above, Method for Incidence, Part B. 4 and Note.
19 'Isbell, Harris: Medical aspects of opiate addiction. Bulletin o f the New York Academy of Medicine, 31:886-901, 1955.
20 Elliott, Henrv W., et al.: Comparison of the Nalorphine test and urinary analysis in the detection of narcotics use. Clinical Pharmacology and Therapeutics, 5:405-413, 1964.
21 See Chapter 10.
23 "Robins, Lee N., and Murphy, George E.: Drug use in a normal population of young Negro men. American Journal of Public Health, 57:1580-1596, 1967; O'Donnell, John A.: Narcotic Addicts in Kentucky. Washington, D.C., U.S. Government Printing Office, 1969, p.259.
24 "Lapouse, Rema: Problems in studying the prevalence of psychiatric disorder. American Journal o f Public Health, 57:951, 1967.