The intersection between crime and drug dependence: Establishing the clinical utility of the Severity of Dependence Scale (SDS) with a sample of federally incarcerated, male offenders

Between 70% to 80% of Correctional Service Canada's (CSC) general offender population and over 90% of its Aboriginal offender population has an identified substance abuse problem requiring intervention. Ensuring that these offenders receive the most effective treatment is a major challenge that is best addressed through the application of assessments that are shown to be reliable, accurate, and useful for client-treatment matching and correctional planning. Aim. The main objective of the study was to establish the Severity of Dependence Scale (SDS) (Gossop et al. 1995) as a suitable measure for client-treatment matching, and as a predictor of recidivism and relapse to substance use. Setting. The SDS and the Drug Abuse Screening Test (DAST) (Skinner, 1982) were administered to a sample of 3350 adult, male inmates from CSC between 2002 and 2007. A total of 1667 inmates were eventually released from custody and available for 24 months of follow-up. Measurements. Cronbach's coefficient alpha provided a measure of internal consistency (reliability), and canonical correlation analysis quantified the dimensional relationship between the two instruments. With DAST as the reference standard, Receiver Operating Characteristics (ROC) analyses established the optimal cut-off score for a classification of psychological drug dependence on the SDS. A number of multivariable logistic regression models uncovered the dimensions of the classification, while a series of Cox proportional hazards models examined SDS's ability to predict the rates of revocation and relapse to substance abuse over a maximum of 24 months of follow-up into the community. Findings. Large Cronbach's coefficient alpha values confirmed the internal consistency of both the DAST and SDS. The canonical correlation analysis revealed 11 linear combinations of DAST and SDS items that were highly correlated along a single dimension that closely approximated the dependence syndrome as defined by the Diagnostic and Statistical Manual of Mental Disorders-IV. The results from the logistic regression and Receiver Operating Characteristics (ROC) analyses underscored the strong relationship between DAST's classification of drug dependence and the SDS. The cut-off value of ≥6 for a classification of psychological drug dependence produced the best trade-off between sensitivity and specificity. The individual logistic regression models and the significant unconditional associations between indicators within a number of life domains and psychological drug dependence uncovered a host of deficits that are important for client-treatment matching and correctional planning. The SDS was also predictive of post-release outcomes. After adjusting for the effects of other predictors within a series of Cox proportional hazards models, offenders who were classified as psychologically drug dependent had higher hazards of revocation and relapse to substance abuse. However, exposure to the high intensity program and community-based maintenance reduced the hazard of revocation and relapse to substance. Conclusions. The SDS was a reliable measure of psychological drug dependence, and useful for differentiating offenders for treatment and for predicting post-release outcomes. The findings underscore the importance of accurately matching offender criminogenic need to appropriate levels of service delivery, and reinforce the importance of community aftercare in mitigating the risk of recidivism and relapse to substance abuse.