Treatment re-entry and outcomes of opioid addicts during a four-year follow-up after drug abuse treatment in the United States




Author: D. D. SIMPSON , L. J. SAVAGE
Pages: 1 to 10
Creation Date: 1980/01/01

Treatment re-entry and outcomes of opioid addicts during a four-year follow-up after drug abuse treatment in the United States

D. D. SIMPSON Institute of Behavioral Research, Texas Christian University, Fort Worth, Texas, United States of America
L. J. SAVAGE Institute of Behavioral Research, Texas Christian University, Fort Worth, Texas,United States of America


Post-treatment outcome measures during a four-year follow-up period were examined in relation to readmissions to drug abuse treatment. The sample included 1,174 opioid addicts admitted during 1972-1973 to 26 different treatment programmes throughout the United States that participated in the Drug Abuse Reporting Program (DARP). Admissions to methadone maintenance programmes, therapeutic communities, outpatient drug-free treatments, and outpatient detoxification programmes were studied, as well as a comparison group which completed admission procedures but did not return for treatment at the DARP agency. The examination of temporal patterns of treatment and outcome indicators representing drug use, criminality, and productive activities during the four-year follow-up period showed that there were beneficial effects associated with treatment.


Most evaluation studies of treatment effects focus only on a single treatment episode, and this is particularly true in the field of drug abuse [ 1] . A single treatment episode is frequently of relatively short duration, however, and in some cases it may in fact represent only a minor event in the life of a drug user. Other treatments may also precede or follow the specific treatment on which an evaluation is focused, but their separate impact (and the impact of other factors, such as family and peer influences) on outcome measures is typically ignored because of data limitations and the complexity of such evaluations.

In the long-term treatment evaluation project based on the national Drug Abuse Reporting Program (DARP) [ 2] - [ 5] , multiple treatment experiences of clients have been recorded and analysed. Treatments before and after DARP were examined [ 6] , [ 7] for follow-up samples of 1969-1972 DARP admissions. In addition, other studies [8, 9] focused more specifically on patterns of treatment re-entry over time during the years in the follow-up period and found that there were consistent relationships between post-DARP treatment and outcomes; in particular, the outcomes on drug use, employment, and criminality generally increased favourably immediately following post-DARP treatment. These results were consistent with and complemented other treatment evaluation studies which focused only on the treatment received in DARP programmes [ 4] , [ 10] - [ 14] .

The present study applied the methodology of Simpson and Savage [ 9] to a later DARP cohort (1972-1973 admissions) in an effort to replicate and extend the previous research. In particular, the temporal sequence of readmissions over the first four years in the post-DARP follow-up period were examined in relation to variations in outcome measures during the same time periods. In addition, specific episodes of post-DARP treatments were analysed with respect to outcome improvements for two-month periods before and after the post-DARP treatment.


Sample selection and field-work

Admissions to drug abuse treatment programmes in Cohort 3 of DARP (June 1972 through March 1973) totalled 16,729. A stratified random sample of 2,295 former clients, from 26 different DARP agencies, was selected for the follow-up study as reported in detail elsewhere [ 15] . The follow-up sample included black and white clients of both sexes from methadone hydrochloride maintenance, therapeutic communities, outpatient drug-free treatments, outpatient detoxification programmes, and a comparison (but not control) group, labelled intake-only, that completed admission (intake) procedures but did not return to receive treatment in DARP.

The field-work, carried out during 1978 and 1979, resulted in the location of 77 per cent (N= 1,774) of the target sample of 2,295 clients; 66 per cent (N = 1,519) were interviewed (with informed consent), 5 per cent (N = 115) were deceased, 1 per cent (N = 34) were unavailable for interview (mainly because of military service in a foreign country), and 5 per cent (N= 106) refused to be interviewed. The remaining 23 per cent (N = 521) could not be located within the time and resources allocated for this purpose, but comparisons based on DARP admissions and during-treatment records suggested that this resulted in minimal bias in the completed sample [ 16] .

Characteristics of the sample

The present study focused only on a sub-sample of 1,174 opioid addicts (defined as persons with a history of illicit daily use of any natural or synthetic opiate drug prior to treatment in DARP). They had an average age at the time of follow-up of 31 years (standard deviation of 6.7 and range of 20 to 68); 59 per cent were black and 41 per cent white; 59 per cent were male.

Follow-up interview

Follow-up interviews were conducted face-to-face by trained interviewers who followed strict procedures to protect the confidentiality of data. Self-report data on living arrangements, employment, criminality, drug use, alcohol consumption, and return to treatment were recorded retrospectively on a month-by-month basis from the time the respondent left the DARP treatment programme to the time of the follow-up interview. This interview was conducted an average of about six years after admission to DARP treatment and an average of over four years after termination of DARP treatment. In the follow-up studies of previous DARP cohorts, checks for internal consistency and comparisons of self-report information with criminal justice records of post-DARP incarceration and treatment re-entry records indicated a high level of reliability and validity of the data [ 17] .


Temporal patterns of post-DARP treatment

Temporal patterns of treatment during the four years after DARP were examined by classifying individuals according to whether they received any drug treatment during one or more months in each successive one-year period. Persons who did not have data for each of these four years (either because they were not at risk during one or more years or they had short post-DARP follow-up intervals) were excluded, reducing the sample to 991. Five major patterns of treatment were defined which were mutually exclusive and generally similar to those used in previous research by Simpson and Savage [ 9] . The type of post-DARP treatment was not included in this classification scheme, but the majority (61 per cent) of client readmissions were to methadone maintenance (MM) programmes. The patterns of treatment analysed in this study included (1) NN-persons who had no treatment during the four-year period, (2) TN-persons who had treatment in Year 1 (and possibly in Years 2 and 3) but not in Year 4, (3) NT-persons who had no treatment in Year 1 but were treated in Year 4 (and ossibly in Years 2 and 3), (4) TT-persons who had treatment in each of the four years, and (5) Other-persons who had other mixed patterns between Years 1 and 4 and Years 2 and 3 (e.g., NTTN or TNTT as four-year patterns). These five patterns were not subdivided into more specific categories in order to maintain sufficient sample sizes for analysis.

The prevalence of each of the treatment patterns is shown in table 1 for each of the DARP treatment groups. The therapeutic community (TC) and drug-free (DF) groups were found to have the highest frequency with no post-DARP treatments (they included 44 per cent and 50 per cent, respectively, in the NN pattern). (In this regard, however, it is important to add that only about 74 per cent of the TC and DF clients were dally opioid users at the time of admission to DARP treatment, compared to 84 per cent in the intake-only group (IO) and about 96 per cent in MM and detoxification programmes (DT); all other clients included in this study had a history of daily use sometime before DARP admission.) On the other hand, treatment in all four years was most prevalent in the IO group (18 per cent in the TT pattern), followed closely by the MM and DT groups (17 per cent and 14 per cent, respectively). Log-linear analysis [ 18] showed that the overall differences between DARP treatment groups in terms of post-DARP treatment patterns were statistically significant ( p < 0.01), but treatment re-entry patterns did not differ between males and females ( p > 0.40) and the relationship between DARP treatment groups and post-DARP treatment patterns was independent of sex ( p > 0.40).

Table 1

Temporal patterns of treatment during the first four years after DARP by DARP treatment group


Fraction of each DARP treatment group (%)


Treatment pattern






Fraction of total %

37 44 50 30 36 41
19 22 16 22 18 20
16 14 17 18 6 15
17 9 5 14 19 12
11 11 12 16 15 12
Number of persons
298 284 216 112 81 991

Covariation of post-DARP treatment patterns and outcomes over time

The relationships of treatment patterns with outcome criteria during the four years following DARP were examined for evidence of behavioural changes related to drug treatment. Five criterion indicators were defined based on months "at risk" during each separate year (i.e., during months not in a jail, hospital, or residential drug treatment facility). The indicators each consisted of a 3-point scale defined so that higher scores represented more favourable outcomes on each criterion. These indicators were summed together to produce a composite outcome measure ranging from 5 to 15, reflecting the poorest and the best composite outcome scores, respectively (see Simpson and Savage [ 19] for a more detailed account of the composite score definitions). The indicators included criminality (3 = no illegal support or arrests; 2 = illegal support or arrests, but only for so-called victimless crimes such as gambling or prostitution; and 1 = arrests for crimes against persons or crimes of profit, or any illegal support from robbery, burglary, or dealing drugs); opioid use (3 = no use during the year; 2 = any use that was less than daily; and 1 = daily use in 1 or more months); non-opioid use other than marihuana (scored the same as opioid use); alcohol use (3 = average of no more than 4 oz (120 ml) of 80-proof alcohol per day; 2 = 4.1 to 8 oz (121-240 ml) per day; and 1 = over 8 oz (240 ml) per day); and productive activities (3 = engaged in employment, homemaking, or school in over 66 per cent of the months at risk; 2 = 1.66 per cent of the months; and 1 = no productive activities reported during the year).

A three-way profile analysis [ 20] of the composite scores (across the four years) by DARP group (MM, TC, DF, DT, and IO), treatment re-entry pattern (NN, TN, NT, TT, and Other), and sex showed that profiles of the outcome scores over the four years were differentially related to treatment patterns ( p < 0.01). Furthermore, this relationship between treatment patterns and outcomes over time did not differ significantly ( p > 0.05) across DARP groups or sex (although females had more favourable overall composite scores than males, p < 0.01).

The figure illustrates these findings concerning post-DARP treatment patterns and outcomes. The data presented show the percentage of persons in each treatment pattern with highly favourable outcomes during each year (instead of the less descriptive mean composite scores actually used in the profile analysis). Specifically, "highly favourable outcomes" included individuals with scores of "3" (representing the most favourable level) on all five outcome criteria defined above, and the nature of the relationships shown in the figure are the same as those observed for mean scores.

The figure shows that favourable outcome ratings were most prominent in the NN pattern, representing a stable level of 23-27 per cent of this sub-sample in each of the four years after DARP. The TN pattern (persons treated in Year 1 but not in Year 4) had 7 per cent favourable outcome ratings in the first year but increased to 25 per cent in the fourth year, approaching the level of the no-treatment NN pattern. The NT pattern (persons who delayed post-DARP treatment re-entry up to three years) had poor outcomes in Years 1 to 3 (3-5 per cent) but increased to 13 per cent in Year 4 when treatment was received; this group was also lower than those in pattern TT (10-17 per cent) who received treatment in each year. The poorest overall outcomes were for persons with mixed treatment patterns (Other, 3-5 per cent).

Percentage of persons in each post-DARP treatment pattern with the most favourable outcomes during each year (based on black and white males end females, N = 991 )

Full size image: 12 kB, Percentage of persons in each post-DARP treatment pattern with the most favourable outcomes during each year (based on black and white males end females, N = 991 )

As already noted, the association between treatment patterns and trends was the same for males and females, but overall outcomes were more favourable for females. For example, females (in all treatment patterns combined) had 14, 17, 21 and 23 per cent in the most favourable outcome category during the four successive years after DARP, compared to 10, 11, 12, and 15 per cent for males.

As indicated by the profile analysis based on mean composite scores, discussed above, these profiles were not parallel over time. When profiles are non-parallel, the analytic procedure calls for two additional sets of analyses in order to (1) examine the outcome trends of each individual treatment pattern over years, and (2) compare outcomes for treatment patterns during each separate year. The first set of results, using Hotelling's T 2 based on mean composite scores [ 21] , indicated that each treatment pattern except NN showed significant outcome variations over the four years ( p <0.01). In the second set (using analysis of variance and multiple range tests), outcomes in each of the four years were found to be significantly more favourable for persons in the NN pattern than for the other four patterns ( p < 0.01). In addition, outcomes in the TN and TT patterns during Years 2 and 3 were significantly more favourable than for patterns NT and Other. Finally, in Year 4 there were no significant differences between patterns TN and TT, or between patterns TT, NT, and Other.

Comparisons of outcome criteria in the two months before and after post-DARP treatment

The results reported above showed a significant relationship between treatment patterns over time and performance indicators. However, since neither the actual duration of treatment nor the sequence of in-treatment and not-in-treatment intervals during each year was specified in these analyses, the data were examined further in order to be more explicit concerning the treatment-to-outcome relationship. In particular, the first post-DARP treatment episode was identified for each treated person and eight different outcome measures were calculated for the two-month periods immediately before and after treatment; these measures included opioid use, non-opioid use, marihuana use, alcohol use, employment, productive activities, illegal support, and jail. Drug use was scored on a 3-point scale (1 = none, 2 = less-than-daily, and 3 = daily), as was alcohol use (1 = none, 2 = up to 8 oz (240 ml) per day, and 3 = over 8 oz (240 ml) per day). The other criterion measures were dichotomized (to reflect "any" versus "none"). If clients were institutionalized or in another treatment during both months, they were considered not at risk for drug use, employment, productive activities, and illegal support and were excluded from analyses involving these variables.

There were 436 persons who were at risk for drug use, illegal support, employment, and productive activities during the two-month periods before and after their first post-DARP treatment episode, and.536 persons were analysed with respect to jail. Drug and alcohol use were analysed using matched-sample t tests, and McNemar's test for a difference in proportions was used for each of the dichotomized criteria [ 22] . The results, shown in table 2, indicate there was significant decrease in mean opioid use and non-opioid use from before to after treatment, and also a significant reduction in the percentage with illegal support. Outcomes on the other variables (marihuana use, alcohol use, productive activities employment, and jail) generally became more favourable, but these changes were not statistically significant (at the 0.01 level of significance).

Table 2

Comparisons of outcomes during the two months before and after post-DARP treatment

Criterion Measure

Before Treatment

After Treatment

Test Results

Mean opioid use a
1.61 2.06 9.90 435
Mean non-opioid use a
2.34 2.44 3.21 435
Mean marihuana use a
2.07 2.09 1.12 435 0.26
Mean alcohol use b
2.43 2.40 1.82 435 0.07
Percentage employed c
36 41 3.4 1 0.66
Percentage with productive activities c
54 59 4.6 1 0.03
Percentage with illegal support c
48 43 9.0 1
Percentage jailed c
21 17 2.3 1 0.13

a Scored 1.00 = daily use, 2.00 = less.than-daily use, and 3.00 = no use.

b Scored 1.00 = over 8 oz (240 ml) 80-proof alcohol per day, 2.00 = 4.1-8 oz (121-240 ml) per day, and 3.00 = 4 oz (120 ml) or less per day.

c Scored 1 = yes, O = no.


The findings provide a convincing replication and extension of the results of a previous follow-up study of admissions to DARP treatment [ 9] . The present study extended the length of the follow-up period to a year longer than the earlier study, but the overall findings are strikingly similar. The data showed that persons with no post-DARP treatment had the most favourable outcomes during each follow-up year examined. Those treated in the first year after DARP but who were out by Year 4 reported the most dramatic changes over time with regard to favourable outcomes. On the other hand, persons who delayed treatment re-entry one to three years after DARP had very poor outcomes, as would be expected, although they tended to improve when readmitted to treatment.

The present study demonstrated further that these relationships among post-DARP patterns and outcomes were the same for females and males. In addition, examination of specific post-DARP treatment episodes in relation to outcomes showed that significant improvements occurred in the period from two months before to two months after post-DARP treatment. Thus, the general relationship between the treatment patterns and outcome changes were directly linked to treatment by these results.

In general, outcomes reached their most favourable level when a course of treatment was terminated (i.e., either the DARP or other post-DARP treatment) and there were no subsequent readmissions reported; for example, approximately one fourth of this sub-sample had the most favourable scores possible on all five outcome measures (drug use, criminality, and productive activity) following treatment. The poorest post-DARP outcomes were reported by individuals who returned to treatment one to three years later. Therefore, treatment re-entry is in one respect an indication of negative follow-up outcomes, but readmission should also be recognized as an important step toward eventual improvements as illustrated by the TN group.

Comparisons of outcomes between particular treatment modalities were not specifically addressed in the present study, although this has been the focus of other major treatment evaluation studies based on the DARP follow-up information, as summarized by Sells and Simpson [ 4] , [ 5] . These findings have shown that follow-up outcomes of persons treated in DARP methadone maintenance, therapeutic community, and outpatient drug-free programmes were generally more favourable than persons from outpatient detoxification or intake-only groups. Other research based on the most recent follow-up sample of 1972-1973 DARP admissions (on which the current study is based) replicates these overall findings [ 14] . In addition, it is noted that outcomes associated with treatment of opioid addicts in the major DARP modalities (i.e., methadone maintenance, therapeutic communities, and outpatient drug-free treatments) have not been found to be significantly different, and evidence for an optimal match between addict-client types (defined on the basis of pre-treatment background and baseline variables) and DARP-treatment types for maximizing favourable outcomes has been negative [ 23] .

The extent to which treatment per se can be given the credit for post-treatment behavioural improvements can be debated. As discussed in previous studies on these data [ 9] , [ 10] , [ 12] , such causal inferences are complicated by client motivation, family and other social influences outside the treatment setting, and other factors. Nevertheless, evidence from the treatment evaluation studies based on DARP, using different methodological approaches and treatment samples, continues to be highly consistent with the position that drug abuse treatment has a beneficial impact for a significant proportion of treated clients. It is also important to emphasize that poor performance outcomes following treatment in DARP were frequently improved after subsequent treatments. In part, these results might represent some degree of cumulative therapeutic effects, but they could also reflect a tendency for clients to "shop around" until they find a treatment programme better suited to their particular needs and interests.

Effective matches between treatment programmes and client needs and interests, however, are complicated by the fact that treatment admissions are often not "voluntary" and that clients may sometimes be motivated only to resolve acute drug problems or simply to reduce the daily cost of their habits. In any case, it appears that a portion of admissions to each successive treatment episode tends to "clean up", and many of the remaining individuals continue to try other treatments until they are successful. Further study is suggested to document this observation over longer time intervals, and more refined time-oriented analyses of treatment histories and outcomes are also needed.


This research was supported by grant H81 DA 01598-02S1 from the National Institute on Drug Abuse, United States Department of Health and Human Services, as part of a long-term nationally oriented treatment evaluation project. The guidance and contributions provided by S. B. Sells, Ph.D., to this project are gratefully acknowledged, and Michael R. Lloyd also contributed to the data collection and analysis for the present study.



S. Sells, "Treatment effectiveness", in Handbook on Drug Abuse , R. DuPont, A. Goldstein and J. O'Donnell, eds. (Washington, D.C., Government Printing Office, 1979).


S. Sells, ed., The Effectiveness of Drug Abuse Treatment (Cambridge, Mass., Ballinger Publishing, 1974), vols. 1-2.


S. Sells and D. Simpson, eds., The Effectiveness of Drug Abuse Treatment (Cambridge, Mass., Ballinger Publishing, 1976), vols. 3-5.


S. Sells and D. Simpson, "On the effectiveness of treatment for drug abuse: evidence from the DARP research programme in the United States", Bulletin on Narcotics , vol. 31, No. 1 (1979), pp. 1-11.


S. Sells and D. Simpson, "The case for drug abuse treatment effectiveness based on the DARP research program", British Journal of Addictions , vol. 75, 1980, pp. 117-131.


G. Joe and M. Gent, Drug Treatment Histories for a Sample of Drug Users in DARP , National Institute on Drug Abuse, Services Research Report No. ADM 78-634 (Washington, D.C., Government Printing Office, 1978).


D. Simpson, L. Savage and G. Joe, "Treatment histories of clients treated for drug abuse", American Journal of Drug and Alcohol Abuse , vol. 7, 1980, pp. 127-140.


L. Savage and D. Simpson, "Illicit drug use and return to treatment: follow-up study of treatment admissions to DARP during 1969-197 I", American Journal of Drug and Alcohol Abuse , vol. 5, 1978, pp. 23-38.


D. Simpson and L. Savage, "Drug abuse treatment readmissions and outcomes: three-year follow-up of DARP clients", Archives of General Psychiatry , vol. 37, 1980, pp. 896-901.


S. Sells, R. Demaree and C. Hornick, Comparative Effectiveness of Drug Abuse Treatment Modalities , National Institute on Drug Abuse, Services Research Administrative Report (Washington, D.C., Government Printing Office, 1979).


D. Simpson, "The relation of time in drug abuse treatment to posttreatment outcomes", American Journal of Psychiatry , vol. 136, 1979, pp. 1449-1453.


D. Simpson, L. Savage and M. Lloyd, "Follow-up evaluation of treatment of drug abuse during 1969 to 1972", Archives of General Psychiatry , vol. 36, 1979, pp. 772-780.


D. Simpson and others, Evaluation of Drug Abuse Treatments Based on First Year Follow-up , National Institute on Drug Abuse, Services Research Monograph No. ADM 78-701 (Washington, D.C., Government Printing Office, 1978).


D. Simpson, L. Savage and S. Sells, Evaluation of Outcomes in the First Year After Drug Abuse Treatment: A Replication Study Based on 1972-1973 DARP Admissions , IBR report 80-8 (Springfield, Virginia, National Technical Information Service, U.S. Department of Commerce, 1980). NIDA/DRD/80/05


D. Simpson, M. Lloyd and L. Savage, Sample Design and Data Collection for the Follow-up Study of 1972-1973 DARP Admissions , IBR report 79-14 (Ft. Worth, Texas Christian University, Institute of Behavioral Research, 1979).


D. Simpson, M. Lloyd and S. Sells, Nonresponse Bias in the Follow-up Sample of 1972-1973 DARP Admissions , IBR report 79-16 (Ft. Worth, Texas Christian University, Institute of Behavioral Research, 1979).


D. Simpson, M. Lloyd and M. Gent, Reliability and Validity of Data: National Follow-up Study of Admissions to Drug Abuse Treatments in the DARP during 1969-1972 , IBR report 76-18 (Ft. Worth, Texas Christian University, Institute of Behavioral Research, 1976).


Y. Bishop, S. Fienberg and P. Holland, Discrete Multivariate Analysis: Theory and Practice (Cambridge, Mass., MIT Press, 1975).


D. Simpson and L. Savage, Treatment Reentry and Outcomes Among Opioid Addicts During a Four-Year Follow-up After Drug Abuse Treatment , IBR report 79-19 (Ft. Worth, Texas Christian University, Institute of Behavioral Research, 1979).


D. Morrison, Multivariate Statistical Methods (New York, McGraw-Hill, 1967).


J. Overall and C. Klett, Applied Multivariate Analysis (New York, McGraw-Hill, 1972).


W. Hayes, Statistics for Psychologists (New York, Holt, Rinehart and Winston, 1963).


D. Simpson and L. Savage, Client Types in Different Drug Abuse Treatments: Comparisons of Follow-up Outcomes , IBR report 80-11 (Ft. Worth, Texas Christian University, Institute of Behavioral Research, 1980).