The Impact of Heterogeneity in Individual Characteristics and Motivation to Quit Heroin Use on Contingency Management and Opioid Agonist Treatment Response: A Latent Class Analysis

ABSTRACT Introduction Contingency Management (CM) involves the systematic application of positive reinforcement (financial incentives) to promote behaviour change, to improve treatment outcomes for individuals with opioid use disorder (OUD). To better understand patient heterogeneity, this study sought to identify meaningful subgroups/classes underlying the population based on characteristic groupings, followed by determining whether CM effectiveness varied between these subgroups. Methods Baseline characteristics (evidenced/theorised to impact CM effectiveness) of participants with OUD from the PRAISe trial ( N = 552) were empirically examined using latent class analysis. Model selection was based on fit statistics and interpretability. Class was then evaluated as a potential moderator of CM effectiveness on abstinence from heroin using a mixed effects model (consistent with the PRAISe trial analysis). Results A four‐class solution best described the population: (1) Less clinically complex/low alcohol dependence/high motivation to quit heroin ( N = 115); (2) More clinically complex/low alcohol/high motivation ( N = 287); (3) More clinically complex/high alcohol/high motivation ( N = 73); and (4) More clinically complex/moderate alcohol/low motivation ( N = 77). CM for Attendance, previously shown to increase heroin abstinence in the overall study sample, was more effective than no CM for participants in class (1) and (2) exclusively (OR = 5.21; 95% CI 1.25, 21.78 and OR = 1.87; 95% CI 1.03, 3.40, respectively). CM targeted at abstinence, previously shown to not be effective compared to no CM overall, was not effective in any class. Discussion and Conclusions CM targeted at attendance is more effective for two subgroups of the OUD population characterised by either less complex clinical presentation and/or low dependence on alcohol. Individual differences may therefore inform a targeted CM treatment approach.