Therapeutic community drug treatment success in Peru: a follow-up outcome study
© Johnson et al; licensee BioMed Central Ltd. 2008
Received: 20 August 2008
Accepted: 03 December 2008
Published: 03 December 2008
The purpose of this study was to assess the impact of drug abuse treatment in Peru that used the therapeutic community (TC) model. Program directors and several staff members from all study treatment facilities received two to eight weeks of in-country training on how to implement the TC treatment model prior to the follow-up study.
This outcome study involved 33 TC treatment facilities and 509 former clients in Lima and other cities in five providences across Peru. A retrospective pre-test (RPT) follow-up design was employed in which 30-day use of illegal drugs and alcohol to intoxication was measured at baseline retrospectively, at the same time of the six-month follow-up. In-person interview data were collected from directors of 73 percent of the eligible TC organizations in January and February 2003 and from former 58 percent of the eligible TC former clients between October 2003 and October 2004. Drug testing was conducted on a small sample of former clients to increase the accuracy of the self-reported drug use data.
Medium to large positive treatment effects were found when comparing 30-day illegal drug and alcohol use to intoxication before and six months after receiving treatment. As a supplemental analysis, we assumed the 42 percent of the former clients who were not interviewed at the six month assessment had returned to drugs. These results showed medium treatment effects as well. Hierarchical Generalized Linear Modeling (HGLM) results showed higher implementation fidelity, less stigma after leaving treatment, and older clients, singly or in combination are key predictors of treatment success.
This study found that former clients of drug and alcohol treatment in facilities using the TC model reported substantial positive change in use of illegal drugs and alcohol to intoxication at a six-month follow-up. The unique contribution of this study is that the results also suggest attention should be placed on the importance of implementing the TC drug abuse treatment model with fidelity. Further, the results strongly suggest that TC drug abuse treatment programs should incorporate follow-up activities that attempt to neutralize community negative reactions (perceived stigma) independent of other factors.
Therapeutic community drug treatment research in perspective
Much has been written about the underlying principles of the therapeutic community (TC) drug abuse treatment model (e.g., [1, 2]). Woodhams describes this model as one in which staff members interact with residents in an environment where "community as method" is the organizing principle. This community is assumed to be responsible for one another's treatment, having an equal role in the treatment process rather than staff bearing the primary responsibility for treatment. The staff role is to facilitate the implementation of TC principles by behavior modeling and to reinforce the community concepts and philosophy in the community's decision-making process. Staff and residents share the following concepts and beliefs:
▪ View of the disorder
Addiction problems are centered within the individual; physiological symptoms exhibited are secondary.
▪ View of the person
Treatment plans are individually tailored; however, addicts often share certain characteristics, such as low self-esteem, lack of impulse control, low tolerance for frustration, inability to cope with feelings, dishonesty, poor interpersonal skills, immaturity, and feelings of being a victim.
▪ View of recovery
Residents must learn experientially through feedback from encounter groups and interaction with other residents in the community to recognize and change negative behavior.
▪ View of right living
Residents are encouraged to adopt a philosophy that reinforces moral and ethical beliefs typically held by mainstream society rather then continuing to hold negative, self-serving views. Residents are encouraged to own their feelings and internalize pro-social feelings of doing what is right in a given situation.
In addition to clinical and administrative staff, TCs often employ staff members who are ex-users or graduates of a TC program. The recovering staff members are considered "rational authorities" who use their skills and experiences to guide, teach, coach, and correct negative behavior of clients. The role of the staff is to facilitate the development of clients' treatment plans by coaching, reinforcing corrective and positive behavior, clarifying issues, and lending assistance when needed. Staff interaction with drug addicts in a therapeutic community is more intense than in other treatment milieus; it is more frequent and can be more confrontational.
This article reviews the research on TC drug abuse treatment and training and presents results of a large TC drug treatment follow-up study in Peru. The study focuses on the treatment success of 33 treatment facilities in various locations of the country and predictors of treatment success. Multi-level analysis procedures were employed to take into consideration the bias due to the influence of a facility on all former clients from that facility. This sample of 33 facilities is a subsample of 72 TC facilities that were involved in a study of Daytop TC training impact on staff. The purpose of the current study is to ascertain changes in former clients' use of alcohol and other drugs after treatment in TC facilities and to determine whether predictors of client changes in alcohol and other drug use after treatment could be identified. Evaluation of drug treatment effects in developing countries is scarce. This study is one of the first studies to document what happens to former clients after drug treatment in a developing country.
Studies of TC drug abuse treatment effects on client behavior
Over the past 30 years an unprecedented number of applied addiction treatment outcome studies have been conducted. Major catalysts for this research have been the National Institute on Drug Abuse (NIDA) national research programs: Drug Abuse Reporting Program (DARP) in the 1970s[5, 6]; Treatment Outcome Prospective Study (TOPS) a decade later; in the 1990s, the Drug Abuse Treatment Outcome Studies (DATOS)[8, 9]; and the National Treatment Outcome Research Study (NTORS). A consistent body of research has supported the effectiveness of drug treatment in general (e.g., ) and of TCs in particular[11, 12].
Criticisms of drug treatment outcome studies have pointed to methodological shortcomings that investigators regularly debate. In a review of drug treatment outcome methodology reported in peer-reviewed journals between 1993 and 1997, Ellingstad and colleagues found that less than one-fourth of the articles used a minimum six-month follow-up interval, which is an important consideration because the highest period of relapse has been found to be between three and six months following treatment.
Client outcome studies of the therapeutic community model have focused on specific target populations that the programs are intended to serve (e.g., prison or jail inmates, chronically homeless drug or alcohol abusers, youth/adults in the general population with chronic and debilitating drug/alcohol problems, dual diagnoses of mental illness and drug addiction) and have employed a variety of methodologies (pre, post tests, single case study, comparative studies, etc.). The outcomes that are typically measured are related to the most serious problems associated with these groups (e.g., re-arrest or reconviction rates, unstable living arrangements and employment, lack of reduction or abstinence from drugs or alcohol use).
The single case study method has often been used to examine outcomes of TC treatment clients in a particular program. For example, Wilson and Mandelbrote examined reconviction rates of ex-residents of the Ley Community TC in Oxford, England, and found correlations with historical patterns of criminal behavior, prior drug use, and length of duration in the program. The authors concluded that TC treatment is effective in reducing criminal activity for residents who stay in the program longer than six months. Holland[16, 17] found similar results for residents of the Gateway House in Illinois. In a five-year follow-up study of ex-residents of the Phoenix House program in New York, De Leon also found improvements related to duration in the program. Dekel and colleagues found that fifteen months after leaving the TC program, half of the Israeli heroin addict participants were clean. Those who had lived with a partner before entering the TC and those who had not engaged in theft prior to treatment were more likely to be drug free at follow-up. A 12-month follow-up study of 83 graduates from community-based TC programs found the majority of participants reported being free from alcohol- and illegal drug use as well as experiencing improvements in the areas of employment, living arrangements, family relationships and high-risk negative behaviors.
Using a pre- and post-test design, De Leon found that both graduates and dropouts of the Phoenix House program improved significantly on measures of personality disorder and self-esteem at the two-year follow-up but client improvements were still below the "normative" or healthy range. Ravndal collected pre and post data from program applicants, dropouts, and completers of a Phoenix House program in Oslo, Norway, and found that completers had fewer substance abuse problems and better social functioning outcomes at the five-year follow-up than those who never entered or dropped out of the program. Those reporting lower frequency of drug use before applying or entering the program had higher rates of success at the five-year follow-up. In a 12-month follow-up of 83 (70%) graduates of inner-city TC programs, researchers found the majority still abstaining from drug and alcohol use, and experiencing improvements in employment, living arrangements, family relations, criminal and other high-risk negative behavior. In a comparative study of two TCs and two no-treatment groups in the Netherlands, Kooyman found much better outcomes for the TC clients at six-month follow-up. De Leon and colleagues found greater behavioral improvements at the 12-month follow-up among homeless, mentally ill chemical abusers in two modified TC programs than those assigned to the control group.
Predictors of favorable treatment outcomes
Studies have consistently shown the length of time a drug user stays in a treatment program is one of the most important predictors of successful treatment outcomes[25, 26, 11, 27, 28]. The effectiveness of treatment programs is limited by the TC's ability to retain the clients for a period long enough to promote change. Typically, relatively few stay beyond three months. Early treatment follow-up studies [30–32] found that successful client outcomes related to reduced crime and substance use and increased employment were related to time spent in treatment. Gossop and colleagues reported that critical time in treatment (28 days for shorter in-patient programs and 90 days for longer-stay rehabilitation programs) was strongly correlated with improvement in overall drug use, and that those who stay in the programs past the critical times were more than five times as likely to have achieved abstinence from all target drugs at the one-year follow-up than those who left. In a comparison of standard and abbreviated treatment in a TC treatment, De Leon and Nemes and colleagues found that positive outcomes are associated with "graduation" or completion of the entire treatment regimen, regardless of the length of the program. Toumbourou and colleagues concluded that attainment of level progress is a better predictor of treatment outcomes.
Chan and colleagues found that treatment satisfaction is related to pre-treatment problem severity and duration of treatment. More recent studies have found that program dropouts are more likely to have had conflicts with the program's rules and view the program as punishment, while completers tend to view the program as treatment and have more positive evaluations of staff[37–39]. Further, Carlson and Gabriel found that client satisfaction with access and effectiveness was associated with six-month follow-up service utilization, as well as one-year post-treatment abstinence from drugs. Kasarabada and colleagues found that only two perceived characteristics of therapists, nurturance and openness, showed significant correlations with length of stay in treatment. However, positive perceptions of counselors had no significant effect on reducing drug use severity scores at a one-year follow-up.
The extent to which a program adapts the ideal TC model and its essential elements, and how staff actually implements the model, has been of interest to investigators (e.g., [42, 43]). Some attention, although limited, has been given to defining program fidelity operationally. Prendergast and colleagues found in a meta-analysis study that well-implemented TC drug abuse treatment and outpatient drug-free programs were correlated with more positive behavioral outcomes. A number of studies have found positive direct effects of the quality and quantity of program implementation on drug-related outcomes [45–48]. Hansen and colleagues found program fidelity to be a moderator of substance abuse prevention program effectiveness.
Some TCs provide additional staff training and education to build both staff skills in group dynamics and understanding of TC philosophy and ethos. Mistral and colleagues found that applying the principles of the therapeutic community to a high-care psychiatric ward did improve staff communication and staff attitudes. Focus groups with TC staff revealed that even after training on TC theory, methods, and procedures, staff have a vague idea about their roles and responsibilities in a TC setting and felt that experiential learning is important in working with drug addicts.
In a study of the impact of staff training to implementing structural family therapy in an adolescent therapeutic community, Weidman concluded that increased staff confidence and competence may result in fewer dropouts and increased attendance in family therapy. Johnson, Young, Suresh, and Berbaum conducted a three-year intervention study using a social policy experiment design that employed a randomized design with repeated measures to test hypotheses about the effects of TC training conducted in Peru. The study tested for the direct and moderating effects of the training, examined implementation fidelity, and reexamined the underlying theory of the TC training model. The study found that the vast majority of the staff participants reported positive appraisals of the quality of trainers (e.g., trainers explained things), the quality of training content and methods (e.g., training handouts were helpful), the quality of training environment (e.g., training rooms and facilities were comfortable and convenient), and cultural sensitivity (e.g., cultural issues were handled with respect). The researchers concluded that the quality of training implementation was judged as more than adequate since a priori expectations, as set by the trainers, were exceeded for all appraisal criteria. Further, the study found that the training had medium effects on staff behavior outcomes such as implementation fidelity of TC tools and small effects on staff empowerment to use TC methods and tools and actual use of TC principles.
McMillin reports many former drug addicts feel shame and guilt years after their last drink or drug use. He further states that stigma is a primary obstacle to solving our nation's alcohol and drug problems. A few studies have looked at the impact of stigmatization on drug use and treatment. Furst and colleagues suggest that the shame and stigma associated with the label "crackhead" served as a deterrent to potential adolescent users. Falck and colleagues found that stigma associated with cocaine use can serve as a barrier to treatment and use of clinical and non-clinical services. In a study of the relationship between methamphetamine use and depression, Semple and colleagues found that perceived stigma had a significant positive direct effect on depressive symptoms above and beyond that accounted for by methamphetamine use. Although we could find no studies that examined the relationship between outcomes of TC clients and their perception of social "stigma," studies of patients in mental hospitals have suggested that labeling and social stigma are related to treatment outcome variables such as self-esteem, employment status and social networks[57, 58]. Room advocates for both quantitative and qualitative studies that examine potential preventive effects of stigmatization.
Therapeutic communities in Peru and treatment outcomes
As in most developing countries of the world, drug addiction is a serious social problem in Peru. Beginning in the mid-1970s, the United Nations Fund for Drug Abuse Control (UNFDAC) received contributions from the Government of Italy to establish Therapeutic Community (TC) treatment centers throughout Peru. Although a network of TCs were subsequently created in Peru and other South American countries by UNFDAC, treatment reports (outcome evaluations, etc.) in Latin America were scarce, mainly focusing on epidemiological studies. Furthermore, there were no systematic treatment reports in the Peruvian literature, and the initial papers only addressed treatment guidelines. Beginning in 1978, Navarro[62, 63] described the treatment and follow-up of only two patients dependent on coca paste; both patients subsequently stopped using the drug. In 1980, Sanchez published a study on the treatment of 50 drug addicts at the Nana TC center in Lima, of which 15 addicts eventually became abstinent. Several years later, Navarro and colleagues conducted a study on 26 clinical patients at Nana TC who were followed-up from 8 months to 4 years after treatment. From 1982 through 1989, Navarro conducted a larger study of 223 male patients of the Nana TC, who were mainly consumers of coca paste. Over 47 percent were abstinent at follow-up.
Although UNFDAC had assisted in the creation of a network of TCs in Peru, the Peruvian government determined in 1997 that many of these programs were poorly trained, providing inadequate services, and needed to be licensed by the government. In 1997, this urgent need for TC drug abuse treatment training prompted the Peruvian government to make a strong appeal to the United States government for training support. As a result, the U.S. Department of State contracted with Daytop Village, Inc. to conduct extensive drug-free treatment training in Peru that included staff TC drug treatment institutional providers. This training was conducted in Lima, Peru. An extensive evaluation of this training was also conducted and was briefly described earlier.
Question 1. What are the overall changes in illegal drug and alcohol use (to intoxication) of former clients of TC drug abuse treatment facilities in Peru?
Question 2. What are the predictors (treatment processes, capacity-building training exposure, and organizational/client characteristics) of illegal drug and alcohol to intoxication use among former TC drug treatment clients?
Question 3. Are the predictors (identified in question 2) moderated by contextual variables?
A sample of 33 TC institutions and 509 former clients from a subset of the 76 TC facilities that participated in the earlier training impact study in Peru agreed to participate in the follow-up study We included former clients regardless of their discharge status if they stayed 30 days in treatment. While, on average, the attrition rate across the 33 participating TC institutions was 21 percent during the first 30 days, we believe the most accurate measure of treatment success should only include residents who stay 30 days or more. Thus, this 30-day eligibility selection criterion was used to only include clients who were really interested in dealing with their alcohol and/or drug problem. All data collection protocols were reviewed by a U.S.A. government approved Institutional Review Board in the Pacific Institute for Research and Evaluation, Inc.
The study used a retrospective pretest (RPT) design with baseline collected retrospectively at a six-month follow-up assessment. Nimon and Allen conducted an extensive review of RPT literature and found that since the seminal work of Howard, Ralph Gulanick, Maxwell, and Gerber, this evaluation design is reappearing more frequently in the literature. This design's strengths and weaknesses are discussed later. An attempt was made to collect baseline data on 30-day prevalence of alcohol and other drug use from clients' consent forms completed at intake. However, we discovered that there were large numbers of missing cases (about one-third of the data set). Comparable data on two of the five study outcomes (30 day use of any drug and alcohol to intoxication) were available from the consent forms at intake (n = 333) and RPT data at the six month assessment (n = 497) to determine differences in rates of use. Only a small difference for 30-day use of any illegal drug (87% vs. 90%) was found between the two samples. We did find higher percentages reported at intake vs. retrospective self-report at follow-up for nonuse of alcohol to intoxication (52% vs. 33%); however, in using retrospective reports, this is a more conservative estimate of change. Thus, we considered the use of a RPT design as valid for assessing change in use of global and specific drugs and alcohol to intoxication of former clients in the study TC facilities in Peru.
Research setting and sample
This study was conducted in Peru in the capital city of Lima and five important cities of the Provinces (Iquitos, Tarapoto, Chiclayo, Trujillo, and Arequipa). The sample includes 33 Peruvian TC drug treatment facilities that identified themselves as therapeutic communities during the Drug-free Treatment (DFT) Training Evaluation study that was conducted in 1999. This facility sample was 83 percent of eligible facilities for the study and data were collected from TC directors via a self-administrated questionnaire. From 33 TC drug treatment facilities, 879 clients remained in treatment for at least 30 days. Of this sampling frame, 509 clients were tracked six months after leaving treatment and interviewed (study retention rate = 58%). The remaining 370 former clients were difficult to track, because of four main reasons: (1) client address was not available or was incorrect (26%); (2) family reported client relapse and living in the street (21%); (3) family reported client at home and no relapse, but could not make contact (18%); and (4) family could not be not located or had rejected client (13%). Only 22 (6%) clients declined to be interviewed, which yields a cooperation rate of 94 percent.
We do not have comparative drug treatment studies in developing countries such as Peru to determine the adequacy of our response rates. However, as reported below, we assessed treatment success using a before treatment-six month follow-up analysis using three samples and found medium size treatment effects even when assuming that the entire baseline sampling frame including 370 former clients who could not be found at the six month follow-up were using drug and alcohol to intoxication 30 days prior to the six month follow-up period.
Measures and data sources
Study Measures and their Description
Use of all illicit drugs 30 days prior to treatment (retrospective) vs. no use at the six month follow-up
1 = Yes, 0 = No
Use of PCB(Coca paste), cocaine, cannabis 30 days prior to treatment (retrospective) vs. no use at the six month follow-up
1 = Yes, 0 = No
Use of alcohol use to intoxication 30 days prior to treatment vs. no use at the six month follow-up
1 = Yes, 0 = No
Length of Stay (per 100 days)
.3 – 11
TC Model Implementation fidelity scale3
39% – 95%; (7 items; alpha = .65)
1 = Very Satisfied, 2 = Somewhat Satisfied, 3 = Somewhat Dissatisfied, 4 = Very Dissatisfied (2 items; alpha = 0.69)
1 = 6–8 weeks of Daytop training in Peru v. 2 weeks of Daytop training in Peru, 0 = other
Director attended Daytop training in 1999
1 = Yes 0 = No
Organization certified by Pervuvian Ministry of Health?
1 = Yes, 0 = No
Length of operation (years)
2 – 22
Number of paid full-time staff
0 – 32
Number of paid program professionals
0 – 17
Number of paid part-time and contract staff
0 – 11
Planned length of stay
6 – 11 months
Number of clients served in 2002
8 – 580
Number of clients participating in Follow-up
2 – 33
Age of clients served
18 – 77 years
Classification of program
1 = TC, 0 = Non-TC
Number of Treatment Models Used
0 – 10
Percentage of clients that dropped out before 30 days
0 – 64
0 = non-Mestizo, 1 = Mestizo
0 = Female, 1 = Male
30 – 54
1 = Primary School, 2 = Secondary School, 3 = Technical School Incomplete, 4 = Technical School Complete, 5 = Some University, 6 = University 4 year Degree, 7 = Higher than 4 year University Degree
Years in Organization
3 months – 22 years
Director attended Daytop training
1 = Yes, 0 = No
Director attended other TC training
1 = Yes – one or more, 0 = None
Recover Alcoholic or Drug Addict?
1 = Yes, 0 = No
Consult with staff before new policies are implemented
0 = Never, 1 = Rarely, 2 = Sometimes, 3 = Often, 4 = Very Often
Directly involved in clients' treatment plan
1 = Rarely, 2 = Sometimes, 3 = Often, 4 = Very Often
Amount of time spent individually with each client
1 = One or more hours, 0 = Less than one hour
Use of Research
0 = Never, 1 = Rarely, 2 = Sometimes, 3 = Often, 4 = Very Often
Former Client characteristics
0 = non-Mestizo, 1 = Mestizo
18 – 77
1 = Primary School, 2 = Secondary School, 3 = Any Technical School 4 = Some University, 5 = University 4 year Degree or more
1 = Employed Part or Full Time, 0 = Not Employed
1 = Married, 0 = Not Married
Participation in other treatment
1 = One or Two Kinds, 0 = None
0 = low, 6 = high (9 items; alpha = .80)
Client propensity score
Predicted covariate to adjust attrition effect
Treatment success was defined as the presence or absence of 30-day use of illegal drug and alcohol to intoxication use 30 days before treatment with no 30-day illegal drug or alcohol use to intoxication at the six months assessment after treatment, respectively. We created dichotomous change scores (1 = use at baseline [retrospective measure] and no use at follow-up; 0 = other) for change in use of illegal drugs or alcohol to intoxication. There were 48 clients who did not report any drug use for retroactive baseline and follow-up. In addition, 148 clients did not report any alcohol intoxication prevalence for both baseline and follow-up. However, those clients are included in the analysis with outcome measures coded as 0 to be in the reference group, which is a more conservative approach.
The six key independent variables of interests were selective treatment processes and capacity-building training exposure that were expected to predict change in the drug and alcohol use outcomes. These variables are a combination of single items and unidimensional scales with the alpha reliabilities listed in Table 1 along with a sample item for each scale. These variables include length of stay, implementation fidelity, treatment satisfaction, Peru-based TC training intensity, director's participation in other training, and whether the TC director attended Daytop training. Length of stay was measured by a continuous variable where each unit equals 100 days in treatment – 1 to 10. There were 12 cases that exceeded 1,000 (considered outliers) so they were Windsorized to a value of 10. Implementation fidelity is a scale in which five to seven questions are asked about five important TC tools to ascertain the correctness of implementation. These tools were morning meetings, encounter groups, static groups, learning experiences (sanctions targeting a behavior or attitude change), and vocational skills. Daytop International assisted in constructing this scale. Melnick and De Leon's research developed a survey of essential elements questionnaire (SEEQ) that taps congruence with the ideal TC model. However, the validated TC essential elements instrument was too long for the Peru study. A modified version that was validated for a similar TC treatment evaluation in Thailand found that the implementation fidelity focusing on TC tools was a stronger predictor of treatment success than the modified version of the SEEQ. Treatment satisfaction was a two item scale about the client's self-reported satisfaction with the progress made and the outcome of the treatment experience. TC training intensity was a single dichotomous variable consisting of TC facilities that received six to eight weeks of in-country Daytop training vs. TCs that only received two weeks of in-country Daytop training. Director's TC training other than the in-country Daytop TC training was measured as a dichotomous variable one or more other TC training vs. none. TC-director-attended Daytop training was also a dichotomous variable coded as attended some or all of the training vs. none of the training.
The remaining variables at both the client and organizational level are covariates or moderators included in the multivariate analysis models. A missing value analysis was conducted for multiple items included in scales for client level variables and EM estimates for missing values were imputed[70, 71]. We conducted a sensitivity analysis to determine how much the imputation of missing data in particular might affect the final results. These results with and with out the imputations were similar. More details of these measures, including scale alpha reliabilities and number of items, are reported in Table 1.
Data collection was conducted by a Peruvian partner (SERPA) in collaboration with U.S. investigators of Pacific Institute for Research and Evaluation, Inc. (PIRE) between January 15, 2003 and October 31, 2004. In-person interviews of the 33 directors occurred in January and February 2003 and of the 509 former TC clients between October 2003 and October 2004. Directors were asked to refer to records to answer client related questions, if they were not sure. Eligible follow-up study participants were identified as clients/residents who were already in treatment a minimum of 30 days as of January 15, 2003, or who were admitted after that date and who stayed a minimum of 30 days in the program. Study participants were informed that they were participating in the evaluation and they signed a written informed consent informing them of the goals and procedures of the study. One copy was also given to the client.
All subjects were recruited and interviewed by native Spanish-speaking interviewers who received special interview training, which PIRE staff observed. In-person interviews with item responses on cue cards were conducted at the former client's current place of residence in Lima and surrounding provinces of Peru. We used a drug testing procedure involving a random sample of 8 percent of the sample to increase the accuracy of the self-reported drug use data. We advised subjects that they might be randomly selected for a urine specimen that would be tested for the presence of marijuana or cocaine; but, their participation was completely voluntary. Respondents were not notified of the drug testing results, because the data from the tests were only used as a validity check on the self report data provided by those randomly selected for testing. As such, we treated all respondents equally in not providing drug testing results to respondents.
After completing the interview, a sealed envelope was opened that contained a label indicating whether the subject's ID number had been randomly selected for the urine specimen test. An analysis of the test results found high congruence between the drug test results and the self-reported drug use data collected during the follow-up interview, (i.e., 96% congruence). In addition, the analysis found no significant differences in self-reported drug use between respondents who were tested and those who were not, indicating that strategy was successful.
The final analyses included a combination of data analysis methods, which included tests of significant differences between dependent proportions and Hierarchical Generalized Linear Modeling (HGLM) with appropriate statistical tests. The comparisons between dependent proportions compared the proportion of former clients who indicated using illegal drugs or alcohol to intoxication 30 days prior to treatment in comparison to 30 days prior to the six month follow-up interview. A dependent groups t-test using the binomial approximation to the variance determined statistical significance. We calculated effect sizes by converting proportions with accompanying standard deviations to a Cohen's d, where a small effect size equals .20, a medium effect size equals .50, and a large effect size equals .80. These effects sizes represent differences in standard deviation units using the normal distribution. We also conducted the same analysis on an expanded sample of 76 former clients whose parents reported that they had relapsed (65%) and the entire sample of eligible former clients (879). While assessing change in dichotomous outcomes traditionally uses a nonparametric statistical test, such as the McNemar test, we decided to use the parametric alternative, as we have an adequate sample size and the parametric alternative is justified under the central limit theorem. Further, this inferential test lends itself to the calculation of a Cohen d effect size. Cohen's d has established qualitative effect size intervals of small, medium, and large that we have found policy makers can easily interpret. While this test is traditionally evaluated using the normal standard deviate, we chose the slightly more conservative option of evaluating the test statistic using the t distribution.
In preparation for the HGLM analysis, first, we constructed valid and reliable scales, indexes, and single item measures. Second, we identified statistically significant covariates from a larger set of 26 potential covariates that were entered in a step-wise regression, predicting the outcome variables one at a time, using criteria of p < .20 to enter and p > .25 to be removed
The HGLM analytical procedure is appropriate in this study because data are nested in nature: the individuals (former residents) are nested in the 33 organizations (TC treatment facilities). HGLM adjusts for variation of dependent variables at the organization level, thus providing a more precise estimate of statistical coefficients at the client level. This model technique is explained in detail in Bryk, Raudenbush, and Congdon and Raudenbush and Bryk, including all assumptions, techniques of estimation, and other statistical information. The basic concept behind hierarchical modeling is similar to that of logistic regression. At the client level (also referred to as level 1 in this study), the analysis is similar to that of logistic regression: the outcome variables in Table 1 are predicted by one or more level 1 variables plus an intercept with the treatment variables as the key independent variable of interest. At the TC level (level 2), the level 1 slope(s) and intercepts become dependent variables being predicted by level 2 TC characteristics variables.
Both predictors' main and moderating effects (also referred to as interactions) on the outcomes were assessed through a series of regression analyses to obtain a final equation with stable coefficients for each outcome. The moderating effects were determined by multiplying two variables producing a product term. If one or both of the variables were a continuous metric, they were centered prior to the multiplication. Dichotomous indictors were effect coded. Plots of the interactions against each outcome were assessed to confirm the direction of each statistically significant moderating effect.
Statistical significance was determined using a two-tailed test of significance with the level of significance set at alpha = .05. Effect size for the HGLM results were calculated for the point-biserial correlation with a small effect size equaling .10, medium effect size equaling .24, and large effect size equaling .37.
Description of TCs, directors, and clients
Organizational (facility) Characteristics
A majority (64%) of the facilities had been certified by the Peruvian Ministry of Health; and the average length of time in operation was 10.2 years. The average number of paid full-time staff was 4; the average number of paid program professionals was 2.3 and part-time or contract staff was 2. The number of clients served in 2002, as reported by the DAT directors, ranged from 2 to 580 with a mean of 127 and the number of former clients participating in the follow-up study ranged from 2 to 33 with a mean of 15.4. The vast majority (94%) of the DAT facilities served adult clients and two-thirds (66%) also served clients 17 years of age and younger. The majority of the DAT facilities (58%) reported that their planned length of stay in treatment was 12 months and nearly one-fourth (24.2%) reported a planned length of stay of 6 to 11 months. The proportion of clients reported by the directors as dropping out before completing 30 days of treatment ranged from 0 to 64 percent across the facilities (mean = 21%). A report of no drop-outs during the first 30 days is unusual in the U.S.; however, four Peruvian TC facility directors did report zero dropouts. Thus, on average, 79 percent of the client population was eligible to participate in the study. In addition to the TC treatment model, on average, the DAT facilities employ 6.5 other treatment models (e.g., AA, Tough Love, behavior modification).
TC Director Characteristics
A majority of the TC directors reported they had attended the Daytop training (70%) and approximately three-fourths (76%) had attended other training on substance abuse treatment. A large majority (78.4%) of the TC directors classified themselves as Mestizo with the remaining 21.6% of the directors representing white (12%), black (6%), Indian (1%), Asian (.6%), and multi-racial (2%) groups. All of the study institutional directors were male (100%), the mean age of the sample was 44 years old, and on average, technical school was the average level of education they had completed. A majority of the TC directors (76%) were recovering alcoholics or drug addicts and had been with the TC facility, on average, for 8.5 years. On a scale of 0 (never) to 4 (very often), on average (2.51), DAT directors reported that they "sometimes" or "often" used research in their institutional decision making. On average, the TC directors reported that they consulted with their staff "often" before new policies were implemented. TC directors reported they were involved in clients' treatment planning "often" to "very often"; on average, 70 percent spent one or more hours per week individually with each client.
Former Client Characteristics
Nearly four out of five (79%) of the former TC clients described themselves as Mestizo and the final sample was 100% male. The ages of the former clients ranged from 18 to 76 with a mean age of 34. On average, secondary school (high school) was the highest level of education completed. Most (77%) were not married. One out of five (20%) had participated in at least one other treatment program since leaving the institution. Excluding the 12 former clients who reported having been in the DAT program for longer than three years, the actual average length of stay in treatment was slightly more than six months. On average, former clients reported being "somewhat satisfied" with both the treatment and services they received, as well as with the progress they made.
Change in 30-day substance use six months after treatment
Question 1: What are the overall changes in drug and alcohol use to intoxication of former clients of TC drug abuse treatment facilities in Peru?
Overall change in 30-day substance use (prevalence) before and six months after treatment (N = 497)
Type of Substance use
6 months after Treatment
Self or Family reported2
Self or Family reported2
Self or Family reported2
Self or Family reported2
Illegal Drug use
PCB (Coca Paste) use
Alcohol use to intoxication
We also conducted an analysis that assumed the 370 former clients not interviewed at follow-up would be using one or more illegal drugs and alcohol to intoxication (n = 879). This conservative analysis strategy showed treatment effects not reported in a table to be medium for illegal use and alcohol to intoxication reductions (d = .68, .46 respectively).
In summary, these results show that former clients of the drug abuse treatment under study improved their quality of life after treatment. While definitive conclusions that the treatment produced positive results cannot be made without a control group, the changes are sufficiently large to conclude that treatment more than likely produced the results. Further conducting the analysis using two additional samples (65% and 100% of the study population) further validated the conclusion that there was treatment success using all five outcomes.
Predictors of successful TC treatment in Peru
Question 2: What are the predictors (treatment processes, capacity-building training exposure, and organizational and client characteristics) of drug use among former TC drug treatment clients?
HGLM regression of illegal drug and alcohol use to intoxication onto treatment processes, capacity-building training, client and organizational predictors (n = 497)
Decrease in Illegal Drug Use
Decrease in Alcohol to Intoxication
Effect Size (r)
Effect Size (r)
TC length of stay
TC Model Implementation fidelity
TC treatment satisfaction
TC training intensity
TC staff participated in drug treatment training
TC director Daytop training in 1999
Client employment status
Client attended other treatment
Percentage of clients that dropped out before 30 days
Institution certified or not
Director collaborated with staff
Director's time spent with clients
Director a recovering addict
Stigma × implementation fidelity
Client age × Implementation fidelity
Some results were consistent with expectations on at least one of the two outcome measures of treatment success. Table 3 shows that the longer the stay in treatment, the more treatment success for illegal drug use (i.e., a larger proportion of former clients not reporting 30-day illegal drug use in comparison with their reported 30-day use at baseline) (See Table 3). There was no significant relationship between length of stay and treatment success for the use of alcohol to intoxication. We also found that as the fidelity of use of TC tools increased, the greater the treatment success (See Table 3). In addition, higher level of stigma (i.e., perceived negative reactions from members of the community – for example, close friends, employees, girlfriends) was a consistent predictor of less treatment success for both illegal drug use and alcohol use to intoxication (See Table 3). This result suggests that reducing negative community reaction (i.e., stigma) may lead to a higher proportion of clients who report a decrease of drug use and alcohol use to intoxication. Client age is shown to have positive impact on decrease in both use of illegal drugs and alcohol use to intoxication (See Table 3). That is, the older the former client, the more successful the treatment.
Regarding inconsistent results, Table 3 also shows that the greater clients' treatment satisfaction, the less treatment success (i.e., a smaller the proportion of former clients who reported not using alcohol use to intoxication (See Table 3); whereas treatment satisfaction among clients with illegal drug problems had no effect. We also found that among TC facilities where the directors reported spending more time with clients, clients were less successful in treatment. None of the training capacity-building training variables were singly related to treatment success.
Question 3: Are the predictors (identified in question 2) moderated by contextual variables?
Table 3 also presents the moderating (or interaction) effects of treatment processes and capacity-building training exposure, and contextual factors on illegal drug use and alcohol use to intoxication. The table shows among those TCs with higher implementation fidelity, former clients with lower stigma have more success in dealing with alcohol use to intoxication (See Table 3). This positive moderating result is in addition to positive effects of the higher implementation fidelity and lower stigma main effects. For the illegal drug use, more treatment success occurred among younger former clients who were in TC facilities with higher implementation fidelity (See Table 3). This is a positive outcome in that, while older clients tend to achieve better treatment success (see main effect of client age variable), the combination of younger age and higher fidelity also contributes to better treatment success. There were no other statistically significant moderating effects on treatment success that were interpretable.
This study found that former clients in Peru who received drug and alcohol treatment in facilities using the TC model reported substantial positive change in use of illegal drugs and alcohol to intoxication at a six-month follow-up. The extent of change is greater than has been reported in other drug treatment program evaluations, including the Treatment Outcome Prospective Study (TOPS) [7, 12], the Drug Abuse Treatment Outcome Study (DATOS), and the National Treatment Improvement Evaluation Study (NTIES). Likewise, the amount of positive change in outcomes in Peru is also larger than National Treatment Outcome Research Study (NTORS)[79, 33]. Because these other studies are not comparable on data collection timeframes or the substance use recall period, we can not conclude greater drug and alcohol treatment success in Peru than elsewhere. However, these results are similar to a more recent evaluation of TC treatment success (30-day use of illegal drugs) in Thailand that used a pre-post design with baseline data collected prior to treatment – reduction = -63 percent.
Regarding predictors of drug treatment success, we found that high implementation fidelity produced more treatment success. These results support the meta-analysis of Prendergast, Podus, and Chang, who found that well-implemented TC drug abuse treatment and outpatient drug-free programs correlated with more positive behavioral outcomes. In the Peru research we also found that the importance of implementation fidelity in predicting treatment success was enhanced among younger clients. That is, while older clients tend to achieve better treatment success (see main effect of client age variable), the combination of younger age and higher fidelity also contributes to better treatment success. Hansen and colleagues also found higher program fidelity moderates drug prevention success.
Clients' higher level of stigma after leaving treatment was a consistent predictor of less treatment success in the Peru study. These results indirectly support the earlier work of Falck and colleagues, who found stigma to be associated with cocaine use, and Semple and colleagues, who found stigma to be a significant predictor of depressive behavior. An additional value of this research in Peru is that TC implementation fidelity moderates the effects of higher stigma on treatment success. That is, among TC facilities that implemented the TC tools more correctly, former clients report more treatment success regardless of higher stigma in the community after leaving treatment.
Consistent with prior research [26, 11, 28], length of stay in the treatment program correlated with treatment success for impacting illegal drug use, although this relationship was weak. When defining treatment success as reduced use of alcohol to intoxication, we found no relationship with length of stay. That is, if a client stayed 30 days, which was part of the eligibility requirement of this study, length of stay beyond 30 days made no difference in the proportion of former clients reporting 30 day alcohol use to intoxication in comparison with retrospective baseline 30 day use. It may be when heavy alcohol use is a problem, client stay in treatment as long as is needed.
Further, unlike prior studies that found a positive correlation between treatment satisfaction and success after treatment, the Peru study found that less treatment satisfaction led to higher treatment success. This result in Peru may be plausible in that lower satisfaction may result from more rigorous implementation of the TC model, which contributes to more treatment success. It is interesting to note that TC staff informed the Daytop team during the course of training that they were experiencing resistance from clients as they made changes to the implementation of TC tools and methods, making the program more challenging for the clients.
The treatment study in Peru is not without methodological controversy. First, the response rate of 58 percent is less than is reported in the U.S. However, since there has been little treatment evaluation conducted in developing countries, we believe the U.S. standards should not be applied. Data collection in developing countries is much more difficult with limited treatment facility infrastructure to support baseline data collection and a satisfactory tracking system. While generalization from a treatment sample to a treatment population is important to determining success, we did conduct comparative analyses that included former clients not interviewed in the sample and assumed they were users. Thus, they were counted as treatment failures in the analyses. These results of drug and alcohol use reductions were lower than those that only included self reports, but still they showed treatment success.
Second, we used a retrospective pretest treatment (RPT) only design in place of a traditional pre-post treatment only design. Lamb declared the RPT design as imperfect but useful. These authors and others have discussed the strengths and weaknesses of this design. For this study, the major strength is the design allows collection of data when pre-testing is impossible. The major weakness is problems associated with memory and recall. We believe the comparison of our retrospective pretest data with the consent data at intake demonstrates the validity of our collection of baseline data retrospectively. That is, our comparative analysis showed that retrospective reported baseline data were, in most cases, lower prevalence than consent reported use at intake.
The unique contribution of this study is that the results suggest attention should be placed on the importance of implementing the TC drug abuse treatment model with high fidelity, especially in connection with lower client stigma as perceived by the former resident. The results also strongly suggest that TC drug abuse treatment programs should incorporate follow-up activities that attempt to neutralize community negative reactions (perceived stigma) independent of other factors. For example, in Peru and Brazil, there are drug abuse treatment programs among street children that incorporate a follow-up strategy that may positively impact stigma. High implementation of TC tools and principles, along with implementation of stigma reducing follow-up strategies is not a panacea for treatment success of its clients. However, this research suggests that these organizational actions may help improve the quality of life of TC treatment former clients.
This research was supported by the U.S. Department of State, Bureau of International Narcotics and Law Enforcement (INL) through a contract SINLEC-02-F-0033 with Conwal Division of Axiom Resource Management, Incorporated. We wish to offer a special acknowledgement to our Peruvian core research team (Rodolfo Salinas, Luis Tapia and Virgilio Chavez) and the project interviewing team for their excellent work in instrument translation, data collection, and symposium coordination. The assistance of Chris Bayer and Susan Squires was also invaluable to the final preparation and editing of the manuscript.
- De Leon G: The therapeutic community: theory, model, and method. 2000, New York: SpringerGoogle Scholar
- Woodhams A: The staff member in the therapeutic community. Therapeutic communities for the treatment of drug users. Edited by: Rawlings B, Yates R. 2001, Philadelphia, PA: Jessica Kingsley, 224-240.Google Scholar
- Sacks S, Sacks JY, De Leon G: Treatment for MICAs: Design and implementation of the modified TC. Journal of Psychoactive Drugs. 1999, 31: 19-30.View ArticlePubMedGoogle Scholar
- Johnson KW, Young LC, Suresh G, Berbaum ML: Drug abuse treatment training in Peru: A social policy experiment. Evaluation Review. 2002, 26: 480-519.View ArticlePubMedGoogle Scholar
- Simpson DD, Sells SB: Effectiveness of treatment for drug abuse: An overview of the DARP research program. Advanced Alcohol Substance Abuse. 1982, 2: 7-29.View ArticleGoogle Scholar
- Simpson DD, Sells SB: Opioid addiction and treatment: a 12 year follow-up. 1990, Malabar, FL: Robert E. KriegerGoogle Scholar
- Hubbard RL, Marsden ME, Rachal JV, Harwood HJ, Cavanaugh ER, Ginzburg HM: Drug abuse treatment: A national study of effectiveness. 1989, Chapel Hill, NC: University of North Carolina PressGoogle Scholar
- Simpson DD, Brown BS: Special issue: treatment process and outcome studies from DATOS. Drug and Alcohol Dependence. 1999, 57: 81-174. 10.1016/S0376-8716(99)00089-7.View ArticleGoogle Scholar
- Simpson DD, Curry SJ: Special issue: Drug abuse treatment outcome study (DATOS). Psychology of Addictive Behaviors. 1997, 11: 211-337. 10.1037/0893-164X.11.4.211.View ArticleGoogle Scholar
- NTORS: Bulletin 4. Two year outcomes of the National Treatment Outcome Research Study. 2000Google Scholar
- Condelli WS, Hubbard RL: Relationship between time spent in treatment and client outcomes from therapeutic communities. Journal of Substance Abuse Treatment. 1994, 11: 25-33. 10.1016/0740-5472(94)90061-2.View ArticlePubMedGoogle Scholar
- Condelli W, Hubbard R: Client outcomes from therapeutic communities. Therapeutic Community: Advances in Research and Application. Volume NIH Publication No. 94-3633. Edited by: Tims F, DeLeon G, Jainchill N. 1994, Washington, DC: U.S. Government Printing Office, 80-98. National Institute on Drug Abuse Research Monograph Series #144].Google Scholar
- Ellingstad TP, Sobell LC, Sobell MB, Planthara P: Drug treatment outcome methodology (1993–1997) strengths, weakness, and a comparison to the alcohol field. Addictive Behaviors. 2002, 27: 319-330. 10.1016/S0306-4603(01)00261-1.View ArticlePubMedGoogle Scholar
- Marlatt GA, Gordon JR: Relapse prevention: maintenance strategies in the treatment of addictive behaviors. 1985, New York: Guilford PressGoogle Scholar
- Wilson S, Mandelbrote B: The relationship between duration of treatment in a therapeutic community for drug abusers and subsequent criminality. British Journal of Psychiatry. 1978, 132: 487-491. 10.1192/bjp.132.5.487.View ArticlePubMedGoogle Scholar
- Holland S: Gateway houses: effectiveness of treatment on criminal behavior. International Journal of the Addictions. 1978, 13: 369-381.PubMedGoogle Scholar
- Holland S: Evaluating community based treatment programs: a model for strengthening inferences about effectiveness. International Journal of Therapeutic Communities. 1983, 4: 285-306.Google Scholar
- De Leon G: The therapeutic community: Study of effectiveness. Treatment Research Monograph Series. 1984, Rockville, MD: National Institute on Drug AbuseGoogle Scholar
- Dekel R, Benbenishty R, Amram Y: Therapeutic communities for drug addicts: prediction of long-term outcomes. Addictive Behaviors. 2004, 29: 1833-1837. 10.1016/j.addbeh.2004.01.009.View ArticlePubMedGoogle Scholar
- Carroll JF, McGinley JJ: An agency follow-up outcome study of graduates from four inner-city therapeutic community programs. Journal of Substance Abuse Treatment. 2000, 18: 103-118. 10.1016/S0740-5472(99)00036-7.View ArticlePubMedGoogle Scholar
- De Leon G: The therapeutic community for substance abuse: perspective and approach. Therapeutic communities for addictions. Edited by: DeLeon G, Zeigenfuss JTJ. 1986, Springfield, IL: ThomasGoogle Scholar
- Ravndal E: An outcome study of therapeutic community based in the community: a five-year prospective study of drug abusers in Norway. Therapeutic communities for the treatment of drug users. Edited by: Rawlings B, Yates R. 2001, Philadelphia, PA: Jessica Kingsley, 224-240.Google Scholar
- Kooyman M: The therapeutic community for addicts: intimacy, parent involvement and treatment success. 1993, Amsterdam: Swets and ZeitlingerGoogle Scholar
- De Leon G, Sacks S, Staines G, McKendrick K: Modified therapeutic community for homeless mentally ill chemical abusers: treatment outcomes. American Journal of Drug and Alcohol Abuse. 2000, 26: 461-480. 10.1081/ADA-100100256.View ArticlePubMedGoogle Scholar
- De Leon G: The therapeutic community: status and evolution. Int J Addict. 1985, 20 (6-7): 823-844.PubMedGoogle Scholar
- De Leon G: Therapeutic communities for substance abuse: overview of approach and effectiveness. Psychology of Addictive Behaviors. 1989, 3:Google Scholar
- Gerstein DR, Harwood HJ, (Eds): Treating drug problems: study of the evolution, effectiveness, and financing of public and private drug treatment systems. 1990, Washington, DC: National Academy PressGoogle Scholar
- Zhang Z, Friedmann PD, Gerstein DR: Does retention matter? Treatment duration and improvement in drug use. Addiction Biology. 2003, 98: 673-10.1046/j.1360-0443.2003.00354.x.View ArticleGoogle Scholar
- De Leon G: The therapeutic community: study of effectiveness.Google Scholar
- De Leon G, Wexler H, Jainchill N: The therapeutic community: success and improvement rates 5 years after treatment. Int J Addict. 1982, 17 (4): 703-747.PubMedGoogle Scholar
- Simpson DD: The relation of the time in drug abuse treatment to post-treatment outcomes. American Journal of Psychiatry. 1979, 136: 1449-1453.View ArticlePubMedGoogle Scholar
- Simpson DD: Follow-up outcomes and length of time spent in treatment for drug abuse. 1980, Fort Worth, TX: Institute of Behavioral Research, Texas Christian UniversityGoogle Scholar
- Gossop M, Marsden J, Stewart D, Rolfe A: Treatment retention and 1 year outcomes for residential programmes in England. Drug and Alcohol Dependence. 1999, 57: 89-98. 10.1016/S0376-8716(99)00086-1.View ArticlePubMedGoogle Scholar
- Nemes S, Wish ED, Messina N: Comparing the impact of standard and abbreviated treatment in therapeutic community: findings from the District of Columbia Treatment Initiative Experiment. Journal of Substance Abuse Treatment. 1999, 17: 339-347. 10.1016/S0740-5472(99)00009-4.View ArticlePubMedGoogle Scholar
- Toumbourou JW, Hamilton M, Fallon B: Treatment levels progress and time spent in treatment in the prediction of outcomes following drug-free therapeutic community treatment. Addiction Biology. 1998, 93: 1051-1064. 10.1046/j.1360-0443.1998.937105110.x.View ArticleGoogle Scholar
- Chan M, Sorensen JL, Guydish J, Rajima B, Acampora A: Client satisfaction with drug abuse day treatment versus residential treatment. Journal of Drug Issues. 1997, 27: 367-377.Google Scholar
- Inciardi JA: The therapeutic community: An effective model for corrections-based drug abuse treatment. The dilemmas of corrections: Contemporary readings. Edited by: Haas KC, Alpert GP. 1995, Prospect Heights, IL: Waveland, 406-417. 3Google Scholar
- Stohr MK, Hemmens C, Shapiro B, Chambers B, Kelly L: Comparing inmate perception of two residential substance abuse treatment programs. Int J Offender Ther Comp Criminol. 2002, 46 (6): 699-714. 10.1177/0306624X02238163.View ArticlePubMedGoogle Scholar
- Strauss SM, Falkin GP: The relationship between the quality of drug user treatment and program completion: understanding the perceptions of women in a prison-based program. Substance Use & Misuse. 2000, 35: 2127-2159. 10.3109/10826080009148252.View ArticleGoogle Scholar
- Carlson MJ, Gabriel RM: Patient satisfaction, use of services, and one-year outcomes in publicly funded substance abuse treatment. Psychiatric Services. 2001, 52: 1230-1236. 10.1176/appi.ps.52.9.1230.View ArticlePubMedGoogle Scholar
- Kasarabada ND, Hser YI, Boles SM, Huagn YC: Do patients' perceptions of their counselors influence outcomes of drug treatment?. Journal of Substance Abuse Treatment. 2002, 23: 327-334. 10.1016/S0740-5472(02)00276-3.View ArticlePubMedGoogle Scholar
- Rohrbach LA, Graham JW, Hansen WB: Diffusion of a school-based substance abuse prevention program: predicators of program implementation. Preventive Medicine. 1993, 22: 237-260. 10.1006/pmed.1993.1020.View ArticlePubMedGoogle Scholar
- Sechrest L, Redner R: Strength and integrity of treatments in evaluation studies. 1979, Washington DC: National Criminal Justice Reference ServiceGoogle Scholar
- Prendergast ML, Podus D, Chang E: Program factors and treatment outcomes in drug dependence treatment: An examination using meta-analysis. Substance Use & Misuse. 2000, 35: 1931-1965. 10.3109/10826080009148246.View ArticleGoogle Scholar
- Botvin GJ, Baker E, Dusenbury L, Tortu S, Botvin EM: Preventing adolescent drug abuse through a multimodal cognitive-behavioral approach: Results of a 3-year study. Journal of Consulting and Clinical Psychology. 1990, 58: 437-446. 10.1037/0022-006X.58.4.437.View ArticlePubMedGoogle Scholar
- Botvin GJ, Batson HW, Witts-Vitale S, Bess V, Baker E, Dusenbury L: A psychosocial approach to smoking prevention for urban black youth. Public Health Rep. 1989, 104: 573-582.PubMed CentralPubMedGoogle Scholar
- Emshoff JG, Blakely CH, Gray D, Jakes S, Brounstein PJ, Coulter JB: An ESID case study at the federal level. Am J Community Psychol. 2001, 32 (3-4): 345-357.Google Scholar
- Pentz MA, Trebow E, Hansen WB, MacKinnon DP, Dwyer JH, Johnson CA, Flay BR, Daniels S, Cormack C: Effects of program implementation on adolescent drug use behavior: the Midwestern Prevention Project (MPP). Evaluation Review. 1990, 14: 264-289. 10.1177/0193841X9001400303.View ArticleGoogle Scholar
- Hansen WB, Graham JW, Wolkenstein BH, Rohrbach LA: Program integrity as a moderator of prevention program effectiveness: results for fifth grade students in the adolescent alcohol prevention trial. J Stud Alcohol. 1991, 52 (6): 568-579.View ArticlePubMedGoogle Scholar
- Mistral W, Hall A, McKee P: Using therapeutic community principles to improve the functioning of a high care psychiatric ward in the UK. International Journal of Mental Health Nursing. 2002, 11: 10-17. 10.1046/j.1440-0979.2002.00220.x.View ArticlePubMedGoogle Scholar
- Tsiboukli A, Wolff K: Using focus group interviews to understand staff perceptions from training in the therapeutic community model. Journal of Drug Education. 2003, 33: 143-157. 10.2190/FUK5-BQVL-R53W-4RG7.View ArticlePubMedGoogle Scholar
- Wiedman A: Family therapy and reductions in treatment dropout in a residential therapeutic community for chemically dependent adolescents. Journal of Substance Abuse Treatment. 1987, 4: 21-28. 10.1016/0740-5472(87)90006-7.View ArticleGoogle Scholar
- McMillin S: Stigma can kill – be a stigma buster!. Addiction Letter. 1995, 11: 4-6.Google Scholar
- Furst RT, Johnson BD, Dunlap E: The stigmatized image of the "crack head": a sociocultural exploration of a barrier to cocaine smoking among a cohort of youth in New York City. Deviant Behavior: An Interdisciplinary Journal. 1999, 20: 153-181.View ArticleGoogle Scholar
- Falck RS, Wang J, Carlson RG, Eddy M, Siegal HA: The prevalence and correlates of depressive symptomatology among a community sample of crack-cocaine smokers. Journal of Psychoactive Drugs. 2002, 34: 281-288.View ArticlePubMedGoogle Scholar
- Semple SJ, Patterson TL, Rant I: Methamphetamine use and depressive symptoms among heterosexual men and women. Journal of Substance Use. 2005, 10: 31-47. 10.1080/1465989042000271264.View ArticleGoogle Scholar
- Link BG: Understanding labeling effects in the area of mental disorders: An assessment of the effects of expectations of rejection. American Sociological Review. 1987, 52: 96-112. 10.2307/2095395.View ArticleGoogle Scholar
- Link BG, Cullen FT, Struening E, Shrout PE, Dohrenwend BP: A modified labeling theory approach to mental disorders: An empirical assessment. American Sociological Review. 1989, 54: 400-423. 10.2307/2095613.View ArticleGoogle Scholar
- Room R: Stigma, social inequality and alcohol and drug use. Drug and Alcohol Review. 2005, 24: 143-155. 10.1080/09595230500102434.View ArticlePubMedGoogle Scholar
- Drogas CdIyEplPdAd: The impact of drugs on Peru. 1996, Lima: CEDROGoogle Scholar
- Navarro R: Development of a program of behavior modification directed to the rehabilitation of drug-dependent patients: treatment and follow-up of 223 cases. The International Journal of the Addictions. 1992, 27: 391-408.PubMedGoogle Scholar
- Navarro R: Teripia y modificacion de la conducta adictiva: tratamiento y sequimiento de 2 casos de addiccion a la pasta basica de cocaine. Anales del Quinto Congreso Nacional de Psiquiatria. 1978, 226-231.Google Scholar
- Navarro R: Modificacion de la conducta adictiva: Tratamiento y sequimiento de 2 casos de addiccion a PBC. Rev Neuro-Psiquiatr. 1978, 41: 83-91.Google Scholar
- Sanchez E: Sistemas de tratamiento y rehabilitacion de farmacodependencia. Anales del Sexto Congreso Nacional de Psiquiatria. 1980, 80-93.Google Scholar
- Navarro R: Terapia de la conducta en el comportamiento dependiente a drogas: tratamiento y seguimiento de 26 casos clinicos. Rev Peruana Psiquiatr "Hermilio Valdizan". 1983, 1: 29-38.Google Scholar
- A review of the retrospective pretest: Implications for performance improvement evaluation and research [Electronic Version]. http://voc.ed.psu.edu/projects/publications/books/Spring2007/WEF_spring2007.3.html
- Howard GS, Ralph KM, Gulanick NA, Maxwell SE, Gerber SK: Internal validity in pretest-posttest self-report evaluations and a re-evaluation of retrospective pretests. Applied Psychological Measurement. 1979, 3: 1-23. 10.1177/014662167900300101.View ArticleGoogle Scholar
- Melnick G, De Leon G: Clarifying the nature of therapeutic community treatment: The survey of essential elements questionnaire (SEEQ). Journal of Substance Abuse Treatment. 1999, 16: 307-313. 10.1016/S0740-5472(98)00036-1.View ArticlePubMedGoogle Scholar
- Johnson KW, Young L, Pan T, Zimmerman RS, Vanderhoff KJ: Therapeutic Communites (TC) drug treatment success in Thailand: a 2006 follow-up study. Research Monograph. US Department of State's Bureau of International Narcotics and Law Enforcement Affairs. 2007, Louisville, KY: Pacific Institute for Research and Evaluation – Louisville CenterGoogle Scholar
- Schafer JL: Analysis of incomplete multivariate data. 1997, New York: Chapman & HallView ArticleGoogle Scholar
- SPSS: SPSS advanced statistics 7.5. 1997, Chicago, IL: AuthorGoogle Scholar
- Bruning JL, Kintz BL: Computational handbook of statistics. New Jersey: Pearson Education, 4Google Scholar
- Cohen J: Statistical power analysis for the behavioral sciences. 1988, Hillsdale, N.J.: L. Erlbaum Associates, 2Google Scholar
- Hannan PJ, Murray DM: Gauss or Bernoulli – A Monte Carlo comparison of the performance of the linear mixed-model and the logistic mixed-model analyses in simulated community trials with a dichotomous outcome variable at the individual level. Evaluation Review. 1996, 3: 338-352. 10.1177/0193841X9602000306.View ArticleGoogle Scholar
- Bryk AS, Raudenbush SW, Congdon RT: Hierarchical Linear and Nonlinear Modeling with the HLM/2L and HLM/3L Programs. 1992, Chicago, IL: SSIGoogle Scholar
- Raudenbush SW, Bryk AS: Hierarchical linear models: applications and data analysis methods. 2002, Thousand Oaks, CA: Sage, 2Google Scholar
- Hubbard RL, Craddock SG, Flynn PM, Anderson J, Etheridge RM: Overview of 1-year follow-up outcomes in the Drug Abuse Treatment Outcome Study (DATOS). Psychology of Addictive Behaviors. 1997, 11: 261-278. 10.1037/0893-164X.11.4.261.View ArticleGoogle Scholar
- Substance Abuse Mental Health Services Administration (SAMHSA): The National Treatment Improvement Evaluation study: final report. 1997, Services USDoHaH ed.: Center for Substance Abuse TreatmentGoogle Scholar
- Gossop M, Marsden J, Stewart D, Edwards C, Lehmann P, Wilson A, Segar G: The National Treatment Outcome Research Study in the United Kingdom: six-month follow-up outcomes. Psychology of Addictive Behaviors. 1997, 11: 324-337. 10.1037/0893-164X.11.4.324.View ArticleGoogle Scholar
- The retrospective pretest: An imperfect evaluation exchange, XI, No. 2. http://www.gse.harvard.edu/hfrp/eval/issue30/spotlight.html
- Johnson KW, Young L, Suresh G: Evaluation study of street children institutions in Brazil and Peru. 2000, Louisville, KY: Community Systems Research Institute, Inc. & Pacific Institute for Research and EvaluationGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.