Dropout from individual cognitive behavioral psychotherapy for schizophrenia: a systematic review and meta-analysis of randomized clinical trials.

Abstract

Background and hypothesis: Cognitive behavioral therapy (CBT) is a recommended psychological treatment for schizophrenia spectrum disorders (SSDs), often delivered alongside pharmacotherapy, but dropout may limit its effectiveness. In this secondary analysis of a parent systematic review and meta-analysis focused on negative-symptom outcomes, we synthesized dropout among individual, face-to-face, outpatient CBT arms identified within that parent-review dataset. We investigated exploratory associations between dropout and arm-level sociodemographic, clinical, treatment-related, and study-design characteristics.

Study design: This study was a secondary analysis of a comprehensive database from a recent systematic review and meta-analysis of treatments for SSDs, which searched PsycINFO and all Web of Knowledge databases. Eligible studies were randomized clinical trials that included at least one arm delivering in-person, individual CBT in outpatient settings and reported dropout. Thirteen studies, comprising 16 psychotherapy arms and 865 participants, met eligibility criteria. Random-effects models were used to estimate the pooled dropout rate, and mixed-effects meta-regression models were conducted to examine sociodemographic, clinical, treatment-related, and study-design moderators.

Study results: Across this heterogeneous set of individual CBT arms, the pooled dropout estimate was approximately one in five participants, with substantial between-arm heterogeneity. Associations with available arm-level moderators were exploratory and should not be interpreted as causal or as evidence that a moderator has no effect when the analysis was underpowered.

Conclusions: This secondary analysis provides a descriptive estimate of dropout within the individual CBT arms retained by the parent review. The estimate is an average across heterogeneous studies and settings, and the moderator analyses are exploratory.