Predictors of Drug Court Client Graduation

Background: Substance use disorder in the United States adversely effects society by burdening the justice system with offender incarceration for drug-related crimes, it also strains in the healthcare system with costs in excess of $216 billion dollars for treatment of drug-related mental and physical illnesses. Many offenders of nonviolent crimes with substance use disorder have been diverted to Drug Court (DC) for year-long supervised community-based drug addiction treatment as an alternative to incarceration for non-violent drug-related crimes. Drug Court program outcomes, however, have been studied as a criminal justice intervention, rather than a primary care mental health intervention. The majority of DC program evaluation has focused on admission data and outcomes using univariate and bivariate analyses, rather than longitudinal data using multivariate analyses to identify multivariate predictors of DC graduation. Objective: The purpose of this study is to: (a) describe the Sample Severity for DC clients; (b) discuss the differences between Drug Court graduates and dropouts for Sample Severity, Drug Court Practices, and In-Program Behavior; and (c) develop a prediction model for Drug Court graduation. Methods: This is a descriptive longitudinal design using secondary data analysis of existing DC Shelby County DC data. Data were analyzed from January 1, 2009 through March 17, 2011 for clients admitted to Shelby County DC, and either graduated or dropped out of DC. The MultiSite Adult Drug Court Evaluation (MADCE) Model guided the data selected at three points in time: (a) admission to the DC program (Sample Severity data); (b) during the DC program (DC Practices and In-Program Behavior data); and (c) end of DC program (graduation or dropout data). Results: The sample consisted of 310 Shelby County DC clients, predominately male (80.0%), and African American (60.3%) with a mean age of 29.9 years. Most DC clients had a high school diploma or GED (54.5%) or no high school diploma or GED (41.9%). Thirty-four percent were employed at DC admission and worked an average of 10.4 hours per week. Marijuana (56.1%) and alcohol (15.5%) were the top two primary drugs of choice. To compare differences between DC graduates and dropouts, data were analyzed using t-tests or Chi-squared, as appropriate. There were (48.1%) graduates and fewer male graduates (Χ² = 4.19, p = .041), and fewer African American graduates (Χ² = 4.26, p = .039). There were more graduates who had a high school diploma/GED or a college degree than dropouts (Χ² = 5.21, p = .022), and more DC graduates were employed at DC admission (Χ² = 23.09, p = .001). Of the seven primary drugs of choice, there was only one significant difference with more graduates listing alcohol as their primary drug of choice than dropouts (Χ² = 14.05, p = .002). Of the six DC programs, there were significant differences for four programs. There were fewer graduates who participated in the Outpatient program (Χ² = 4.04, p = .039) and Residential program (Χ² = 8.00, p = .004), more graduates in the Outpatient DUI program (Χ² = 27.5, p = .001), and no graduates in the Early Assessment Intervention Treatment program (Χ² = 5.66, p = 017). Graduates spent more days in DC programs (t-test = 15.17, p = .001), and participated in fewer DC programs (t-test = 2.17, p = .031). Of the ten treatment agencies, there were significant differences for only on agency that had no graduates (Χ² = 4.70, p = .030). Of the 27 candidate predictor variables, there were six significant predictors. Having more diluted urine drug screens (OR = 5.081, p = .002) and greater number of days in the DC programs (OR = 1.019, p = .001) were positive predictors of graduation. Male gender (OR = 0.373, p = 0.47), no high school diploma/GED (OR = 0.214, p = .004), rearrests (OR = 0373, p = .002), and number of jail sentencing sanctions (OR = 0.439, p = .001) were negative predictors of graduation. The Hosmer and Lemeshow Goodness of Fit statistic (Χ² = 11.3724, df = 8, p = .182) documented that the model predicts the data well. The c statistic (0.949) documented highly acceptable predictive ability of the model with 94.9% of all possible pairs of graduates and dropouts predicted correctly. Discussion: The final prediction model suggests that males with no high school education diploma or GED, greater rearrests, and more jail sentencing sanctions are at-risk for not graduating from the Shelby County DC. Education is the only modifiable factor for DC graduation which has implications for DC practice changes and future health literacy research with the DC client population. Drug Court practice changes include: (a) evaluate client literacy and health literacy after drug detoxification; (b) develop and evaluate low literacy DC materials and programs; (c) integrate and require adult reading and GED classes; (d) evaluate need for and design and evaluate programs for men; (e) evaluate and refine exiting programs for women. Future research will: (a) validate the prediction model using cross-validation statistics; (b) develop separate prediction models for men and women; (c) develop a unified data base with continuous variables and MADCE Model variables for DC program reports and evaluation; and (d) use the MADCE Model and Social-Ecological Model to examine Offender Perceptions and Post-Program Outcomes in clients