Predictors of Treatment Adherence in a Randomized Clinical Trial of Digital Therapeutics with Pharmacology for Alcohol Use Disorder (AUD).

Background/Aims: Non-adherence is a significant issue in clinical trials of new therapeutics; and Alcohol Use Disorder (AUD) studies are particularly vulnerable to discontinuation. Moreover, clinical trials involving behavioral intervention over time may be even more difficult to complete. Digital therapeutics make participation in treatment much more accessible and may therefore reduce one of the barriers to study participation. This study aims to investigate predictors of treatment adherence in a clinical trial that collected data over 13 weeks involving a digital therapeutic Cognitive Remediation Therapy (CRT) and donepezil for patients with AUD, focusing on (1) CRT adherence (total hours completed); (2) medication adherence (pill count) and (3) total adherence that combines 1 and 2. Methods: Baseline data including demographics, illness characteristics, cognitive assessments, self-reports of functioning and disability, psychiatric symptoms and personality traits were collected on 52 participants. Exploratory analyses included parametric and non-parametric analyses of baseline variables to predict adherence over 13 weeks. Results: Of the 52 participants who were randomized, 49 completed baseline assessments and entered the study. The mean CRT hours was 29.49 hours (SD=27.90) out of a possible 65 hours (45.37%). Mean weekly medication adherence was 48.78 (SD=38.81) out of 84 pills (58.02%). Medication adherence was correlated with younger age and Total Adherence was negatively correlated with WHODAS 2.0 Self-care indicating that greater adherence is associated with lower self-care dysfunction. Medication adherence was significantly related to physical ability on the VR-12 but with none of the other variables. CRT adherence was not directly related to any baseline variables. A linear regression analysis was performed to predict Total Adherence. A significant model was obtained with WHODAS 2.0 Self-care and Age as significant variables. No relationship was found between Self-care and substance abuse or craving variables, or with cognitive measures, but there were some significant relationships with psychiatric symptoms and relatedness variables. A linear regression of Self-care Predictors produced a significant model with Alienation as the only predictor. Conclusions: This study showed a relatively high rate of adherence: over two-thirds of participants completed the 13-week protocol, with substantial engagement in both digital therapeutics and medication. The findings highlight the importance of Self-care, as a predictor of adherence, whereas illness-specific characteristics did not predict adherence, suggesting that adherence may not vary by AUD characteristics. While Self-care was the best predictor of Total Adherence, we found that psychiatric symptoms and relatedness variables were predictors of Self-care. Regression analysis indicates that Alienation captures the most variance in Self-care, suggesting the importance of therapeutic alliance in clinical trials.