Studying treatments and outcomes over time in substance abusers: Persistent Effects of Treatment Studies (PETS)

There has been increasing awareness that for many individuals addiction can be a chronic disorder, characterized by multiple relapses and additional treatment episodes over time (Anglin et al., 1997Anglin M.D. Hser Y.-I. Grella C.E. Drug addiction and treatment careers among clients in the Drug Abuse Treatment Outcome Study (DATOS).Psychology of Addictive Behavior. 1997; 11: 308-323Crossref Scopus (182) Google Scholar, McLellan et al., 2000McLellan A.T. Lewis D.C. O'Brien C.P. Kleber H.D. Drug dependence, a chronic medical illness: Implications for treatment, insurance, and outcomes evaluation.Journal of the American Medical Association. 2000; 284: 1689-1695Crossref PubMed Scopus (1895) Google Scholar). However, most studies of addiction treatment have focused on the evaluation of a single treatment episode by examining outcomes over periods of a year or less. Recognizing this discrepancy between the nature of addiction and the methods that have usually been employed to study treatment effects, the Substance Abuse and Mental Health Services Administration's Center for Substance Abuse Treatment (CSAT) provided funding in 1997 for a family of studies that were collectively referred to as the Persistent Effects of Treatment Studies, or PETS. The purpose of this initiative was to obtain and examine longer-term outcome data (i.e., 30 months or more) from “real world” treatment programs. This was done by extending the follow-up periods of several CSAT-funded treatment studies. Two large-scale studies of adults that had been initiated under the Target Cities grant program were included; the Cuyahoga County, OH (Cleveland) site collected data over a 36-month period while the Chicago site conducted follow-up interviews for up to 5 years. Seven studies of adolescent treatment program outcomes were extended to a total duration of 30 months; four study sites were from the Cannabis Youth Treatment Study and three sites were from the Adolescent Treatment Models grant program. The primary goal of these studies was to develop a better understanding of the dynamic relationship between episodes of substance abuse treatment and substance use and related behaviors over time. This included evaluations of the effects of the index treatment as well as additional episodes of treatment on outcome trajectories over the course of the follow-ups, and identification of pre-, during-, and post-treatment factors that predicted outcome trajectories. A number of related questions were addressed, including the costs and economic benefits associated with treatment, changes in homelessness and criminal behavior over time, and the policy implications of the findings. Many of these studies used newer data analytic techniques, such as mixed effect and growth curve analyses, which are better able than traditional data analytic approaches both to take advantage of the strengths and to address the weaknesses in longitudinal data sets. The PETS family of studies has generated an impressive number of empirical reports and reviews. Previous groupings of PETS-based articles have appeared in a special issue of Evaluation Review in April of 2001 and in two special sections of the journal Evaluation and Program Planning in 2003. This special issue contains eight empirical articles that examine baseline moderators of outcome, during- and post-treatment predictors and correlates of long-term substance use outcomes, and factors associated with important non-substance use outcomes (e.g., homelessness and costs). Another article identifies and discusses ethical issues in conducting longitudinal research with substance abusers. Some of these topics were addressed in previous PETS publications, but most of the articles in this special issue make use of data from the entire PETS follow-up periods, including the most distal follow-up points. This collection of articles presents the most complete compilation of findings available from the PETS longer-term adult studies. Here, we briefly describe each article included in the special issue. Two of the articles address baseline moderators of long-term outcomes. Using data from the PETS Chicago study, Grella, Scott, and Foss examined gender differences in alcohol and drug use, employment, criminal justice involvement, self-help participation, and family and social functioning over a 36-month follow-up. Cacciola, Dugosh, Foltz, Leahy, and Stephens compared participants who were in treatment for the first time to those with prior treatment experiences on 30-month outcomes in the PETS Cuyahoga study. Predictors of outcome within each of these two groups were also examined. Two other articles in the special issue examine treatment outcomes other than drug and alcohol use—specifically the course of homelessness following treatment and the economic benefits associated with treatment. Orwin, Scott, and Arieira looked at transitions in and out of homelessness over the 36-month follow-up and identified predictors of achieving and sustaining residential stability. Koenig, Siegel, Harwood, Gilani, Chen, Leahy, and Stephens estimated the economic costs of treatment and compared those costs to economic benefits obtained during the follow-up period. These investigators also studied the costs and benefits of continuing care and additional episodes of care obtained during the follow-up. Four articles focus on the identification of predictors and correlates of long-term outcome. Dennis, Scott, Funk, and Foss examined the duration and correlates of addiction and treatment careers and years to recovery in the PETS Chicago sample. In this sample, the median time from first use to last use was 27 years, and the median time from first treatment episode to last use was 9 years. Using the same sample, Scott, Foss, and Dennis characterized participants at each assessment as in the community using, incarcerated, in treatment, or in the community not using. The probabilities of transitioning between each state and the correlates of pathways between these states were then estimated. Over 83% of the participants had at least one transition between states during the follow-up. Using multivariate analyses, McKay, Foltz, Stephens, Leahy, Crowley, and Kissin examined the relation of various factors assessed at each follow-up in the PETS Cuyahoga study and substance use over the subsequent follow-up period. Several differences in the pattern of predictors of alcohol and crack cocaine use were observed, and further analyses identified predictors of the two factors that were most strongly associated with subsequent substance use. Finally, Sridharan, Kawata, Campbell, and Tseng examined the relation between treatment received during the follow-up and simultaneous poly-substance use. Results indicated that over the course of the follow-up, higher durations of treatment were associated with decreases in poly-drug use. The findings from these articles all point toward the need to adopt a chronic care approach to treating addictions in individuals with substance use disorders who have entered the treatment system. Finally, Scott and White identified seven areas that present ethical challenges in conducting longitudinal studies of addiction treatment. This article then describes a number of case studies that illustrate ethical dilemmas within each area, and provides suggestions for managing these issues. If, as we hope, longitudinal studies come to be more widely conducted, the observations and lessons from the Scott and White article may be important ones for researchers and practitioners in the years ahead.