| Abstrakt: |
Objectives: Attrition is very common in longitudinal research, including randomized controlled trials (RCTs) testing psychological interventions. Establishing rates and predictors of attrition in mindfulness-based interventions (MBIs) can assist clinical trialists and intervention developers. Differential attrition in RCTs that compared MBIs with structure- and intensity-matched active control conditions also provides an objective metric of relative treatment acceptability. Methods: We aimed to evaluate rates and predictors of overall and differential attrition in RCTs of MBIs compared with matched active control conditions. Attrition was operationalized as loss to follow-up at post-test. Six online databases were searched. Results: Across 114 studies (n = 11,288), weighted mean attrition rate was 19.1% (95% CI [.16,.22]) in MBIs and 18.6% ([.16,.21]) in control conditions. In the primary model, no significant difference was found in attrition between MBIs and controls (i.e., differential attrition; odds ratio [OR] = 1.05, [0.92, 1.19]). However, in sensitivity analyses with trim-and-fill adjustment, without outliers, and when using different estimation methods (Peto and Mantel–Haenszel), MBIs yielded slightly higher attrition (ORs = 1.10 to 1.25, ps <.050). Despite testing numerous moderators of overall and differential attrition, very few significant predictors emerged. Conclusions: Results support efforts to increase the acceptability of MBIs, active controls, and/or RCTs, and highlight the possibility that for some individuals, MBIs may be less acceptable than alternative interventions. Further research including individual patient data meta-analysis is warranted to identify predictors of attrition and to characterize instances where MBIs may or may not be recommended. Meta-Analysis Review Registration: Open Science Framework (https://osf.io/c3u7a/) [ABSTRACT FROM AUTHOR] |
Full text from SpringerLink