Practice Profile: Contingency Management Interventions for Substance Use Disorders

Prendergast and colleagues (2006) evaluated the effectiveness of contingency management programs for substance-using individuals. To be eligible for inclusion in the meta-analysis, studies had to be outcome evaluations of contingency management programs, delivered to adults or juveniles, and designed to treat dependence on alcohol, tobacco, or illicit drugs. Moreover, studies had to be published in English between 1970 and 2002, have used either an experimental design or quasi-experimental design, and have a total sample size of at least 10. Finally, only studies that included enough information to compute an effect size were eligible for inclusion. To identify studies, a comprehensive search was conducted of a variety of bibliographic databases using search terms that paired a particular technique (e.g., contingency management, token economy, behavioral contracting) with terms referring to the specific problem behavior (e.g., addiction, drug abuse, alcoholism, cocaine, tobacco).

A total of 1,150 studies were identified through this search strategy. After screening these studies across the eligibility criteria, a total of 81 studies were eligible for inclusion in the meta-analysis. However, of these 81 studies, only 47 were used in the final analysis given the limited number of information provided in some of the studies. All 47 studies were conducted in the United States, with 43 using experimental designs, and 4 using quasi-experimental designs. Approximately 70 percent of the included studies were conducted during the 1990s. Moreover, across the 47 studies, sample sizes ranged from 12 to 844, with a median sample size of 69. Finally, although attempts were made to locate unpublished literature, the included studies were all published studies. The target drugs of the included studies were marijuana, tobacco, cocaine, opiates, alcohol, and a combination of one or more drugs. Treatment approaches included cash, methadone take-homes, methadone dosage increase, methadone dosage decrease, graduation to next treatment phase, program discharge, vouchers, fee reduction, and reduction in clinic responsibilities.

To determine whether contingency management programs for substance use disorders are effective, both fixed effects and random effects models were used in the analysis; however, the main outcome of interest (illicit drug use, alcohol use, and tobacco use) was found using a random effects model.

Using meta-analytic techniques, Dutra and colleagues (2008) analyzed the effect of contingency management programs on illicit substance use disorders, including cocaine, opiates, cannabis, and polysubstance abuse and dependence. To identify studies, PsycINFO was used to find articles published between 1840 and March 2005, using a variety of key search terms, including but not limited to cocaine, substance use, substance abuse, treatment outcome, contingency, and voucher. Additionally, MEDLINE was used to identify articles available between 1966 and March 2005, and Cochrane Central Register of Controlled Trials was used to identify studies for the first quarter of 2005. Both PsychINFO and MEDLINE searches were limited to those published in English.   

To be eligible for inclusion in the meta-analysis, studies had to be investigations of the efficacy of individual psychosocial treatments for substance abuse/dependence (not including alcohol or nicotine abuse/dependence), and use randomized controlled trials, including a comparison group. Moreover, studies were limited to adult participants and to investigations on the efficacy of nonintensive outpatient treatments. Nonintensive outpatient treatment was defined as a maximum of three, 2-hour per week treatment sessions. Finally, studies had to include self-report outcomes of interest or toxicology screening outcomes of interest. Self-report outcomes of interest included 1) mean maximum number of days or weeks abstinent throughout treatment, 2) mean percent of days abstinent throughout treatment, 3) percent of sample abstinent for 3 or more weeks throughout treatment, 4) percent of sample demonstrating posttreatment/clinically significant abstinence, and 5) posttreatment scores on the Addiction Severity Index. Toxicology outcomes of interest included 1) mean number of negative drug screens throughout treatment, 2) mean percent of negative drug screens throughout treatment, and 3) percent of sample that demonstrated clinically significant abstinence.

Using this eligible criterion, a total of 34 studies were eligible for inclusion in the meta-analysis. Across the 34 studies, participants were approximately 35 years old and, on average, 62.2 percent male and 61.0 percent white (information about other races/ethnicities was not provided). However, of these 34 studies, 14 pertained to contingency management/voucher programs; the other 20 studies looked at other types of treatment that were not reviewed for CrimeSolutions. Across the 14 studies, the intent-to-treat sample (i.e., the sample size of the treatment condition) included 785 participants (this did not include the sample size of participants in the control condition, which was not provided). In the 14 studies, the treatment condition received contingency management, while the control condition received treatment as usual, motivational enhancement interviewing, 12-step facilitation, noncontingency management, or standard care.

Overall, to determine the impact of contingency management on illicit substance use disorders, an average mean effect size was created using Cohen’s d.