Psychometric Properties of the Substance Use Recovery Evaluator in a Collegiate Recovery Program Sample

Background and Purpose: The Substance Use Recovery Evaluator (SURE) is a 21-item measure of wellbeing in recovery, with 5 domains: (1) Drug use; (2) Self care; (3) Relationships; (4) Resources; and (5) Outlook. The SURE was originally developed and tested in a sample of treatment-engaged individuals who were predominately white (82.4%), male (67.3%), and middle-aged (age M=44), and had high internal consistency (a=0.92), acceptable test-retest reliability, and good validity with similar scales. People engaged in collegiate recovery programs have different demographic and recovery characteristics than the general recovery population, potentially leading to differences in the performance of the SURE. The psychometric properties of the SURE have not been evaluated in the collegiate recovery population, and this study aimed to fill that gap.

Methods: Our sample was 126 student participants in the National Longitudinal Collegiate Recovery Study engaged in collegiate recovery programs from across the US and Canada (age M=31; 47.4% female, 36.4% male, 18.2% transgender or gender expansive; 88.3% white). We conducted confirmatory factor analysis using R lavaan starting with the factor structure identified by the scale developers and modifying the structure based on items loading <.32, modification indices, and empirical and theoretical literature. Validity testing was conducted with correlations between the SURE total and domain scores and the WHO Quality of Life Brief (QOL-8) and the Brief Assessment of Recovery Capital (BARC). Internal consistency reliability was tested with Chronbach’s alpha.

Results: The structure for domains 2-5 was retained as originally specified, with the addition of covariances between two sets of items in domain 2 and one set in domain 3. Domain 1 had substantial changes. The two items related to alcohol and drug use (“I have drunk too much,” “I have used street drugs”) loaded below the cutoff and were not retained, and the item related to experiencing cravings did not load successfully on any factor, but was retained as an individual item in the model. The remaining three items for domain 1 related to coping without substance use, and formed a coherent factor. The final model had good model fit (CFI=0.93; TLI-0.91; RMSEA=0.07; SRMR=0.06) and had significantly superior fit to the original structure (chisq(133)=221.3, p<.001). The overall scale was reliable (a=0.86), and highly correlated with the QOL-8 and BARC (r=.77 for both). Correlations were significant and moderate to high between the QOL-8 and BARC and all SURE domains except domain 1.

Conclusions and implications: Four domains of the SURE—Self care, Relationships, Resources, and Outlook—and the total SURE scale perform well for those engage in collegiate recovery programs. The domain related to substance use did not perform as well, and required modification to reflect ability to cope with problems without substance use. The collegiate recovery population tends to have low rates of substance use compared to the general recovery population due to program requirements of abstinence. This suggests that items related to recent substance use could be dropped from the scale for this population without compromising scale performance.