Uptake, Adherence, and Attrition in Clinical Trials of Depression and Anxiety Apps

Importance Randomized clinical trials (RCTs) support apps as scalable, cost-efficient tools for managing depression and anxiety, but concerns about engagement and attrition remain. With the field’s rapid growth, there is now an opportunity to generate more precise insights on uptake, attrition, and adherence. Objective To examine patterns of uptake (initial activation of the app), adherence (use meeting the trial’s definition of adequate engagement), and attrition (failure to complete outcome assessments) from RCTs of mental health apps for depression and anxiety. Data Sources Studies were identified from a 2024 review of RCTs on depression and anxiety apps (search through January 2024), with an updated search in Medline and PsycINFO conducted from January 2024 to May 2025. Study Selection Trials of mental health apps delivered to participants with depression or anxiety were eligible if they reported rates of uptake, adherence, or attrition. Data Extraction and Synthesis Study characteristics were extracted by 2 independent reviewers. The Cochrane Risk of Bias tool was used. Meta-analyses were conducted using multilevel random-effects models. Main Outcomes and Measures Meta-analyses were conducted on app uptake, posttest and follow-up attrition, and adherence. Absolute rates were estimated using pooled event rates, while relative rates (apps vs controls on attrition) were calculated using pooled risk ratios. Results Seventy-nine trials, including 92 app conditions and 78 control conditions, were included (k = 39 for depression, k = 26 for anxiety, and k = 14 for transdiagnostic samples). The meta-analytic rate of app uptake was 92.4% (95% prediction interval [PI], 48.4%-99.4%) from 56 conditions. The meta-analytic rate of app adherence was 61.8% (95% PI, 53.2%-69.7%) from 21 conditions. The meta-analytic posttest attrition rate was 18.6% (95% PI, 2.4%-68.3%), and at follow-up, it was 28.4% (95% PI, 3.0%-83.7%). Attrition rates were lower among trials that offered reminders, human contact, and no gamification features. Attrition was higher in app conditions than wait-lists (risk ratio, 1.49; 95% PI, 0.34-6.48) but did not differ significantly to other controls. Conclusions and Relevance The findings of this meta-analysis offer benchmarks for future trial planning while highlighting modifiable design features that may enhance user engagement.