Prevalence of Dropout and Influencing Factors in Digital Psychosocial Intervention Trials for Adult Illicit Substance Users: Systematic Review and Meta-Analysis.

Background: Illicit drug use has become a significant global public health issue, and digital interventions offer new approaches to address this challenge. However, there is a gap in existing research on the dropout rate of adult illicit drug users receiving digital psychosocial interventions. Objective: This study aims to evaluate the dropout rate of adult illicit drug use following digital psychosocial interventions during treatment and the longest follow-up, as well as its predictive factors. Methods: Following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines, studies published up to August 27, 2025, were searched in the Web of Science, PubMed, PsycINFO, Embase, and Cochrane Controlled Trials Register. Randomized controlled trials of digital psychosocial interventions for adult illicit drug users that reported dropout rates were included. Two researchers independently screened studies, extracted data, and assessed bias risk using the Cochrane risk of bias tool (ROB 2.0). A random-effects model in Comprehensive Meta-Analysis software (CMA 4.0) was used for meta-analysis, along with heterogeneity testing, sensitivity analysis, and publication bias assessment. Finally, a moderating analysis was conducted based on the extracted data. Results: A total of 41 studies involving 9693 participants and reporting 48 dropout rates were included. The mean dropout rate in the intervention group after 18 studies was 22% (95% CI 0.13-0.36), which was lower than the control group's 26% (95% CI 0.16-0.39). High heterogeneity was observed between studies (Q=396.18, df=17, P<.001, I²=96%), and moderating analysis revealed that high heterogeneity in dropout rates was associated with four variables across three major characteristics: (1) participant demographic characteristics: employment rate; (2) participant clinical characteristics: baseline clinical diagnosis and baseline drug use type; and (3) intervention characteristics: intervention frequency. In the 30 studies with the longest follow-up period in the intervention group, the dropout rate was 28.2% (95% CI 0.19-0.39), comparable to the control group's 27.8% (95% CI 0.20-0.37). Extremely high variability was observed between studies (Q=1293.13, df=29, P<.001, I²=98%), and moderating analysis showed that high heterogeneity in dropout rates was associated with 4 variables across three major characteristics: (1) participant demographic characteristics: single individuals; (2) participant clinical characteristics: baseline medication frequency; and (3) treatment characteristics: recruitment method and the degree of digitalization. Additionally, publication bias assessment and sensitivity analysis supported the robustness of the study results. Conclusions: This study explored the impact of digital psychosocial interventions on treatment adherence among adult illicit drug users, revealing complex factors affecting dropout rates through mediation analysis. These findings not only emphasize the necessity of further research but also provide important evidence for developing precision interventions, holding significant implications for both theory and clinical practice.