The relation between therapy quality, therapy processes and outcomes and identifying for whom therapy quality matters in CBT and IPT for depression

Background: Research on which therapy processes are crucial for whom is necessary to optimize treatment outcomes for major depressive disorder (MDD) but may be impeded by a lack of variation in therapy quality and because of individual differences in therapy response. The present study used a dataset with large variation in therapy quality to investigate the impact of therapy quality on change in therapy processes and outcome. Machine-learning techniques were used to explore individual differences in these relationships. Method: Data come from a multicenter trial that randomized patients with MDD into weekly versus twice weekly sessions of cognitive behavioral therapy (CBT) or interpersonal psychotherapy (IPT). Correlations between quality of therapy, change in therapy processes and depressive symptoms were calculated. Using elastic net regression, a prognostic model was developed that investigated individual differences. Results: There were no significant correlations between therapy quality and change in therapy process variables or between therapy quality and therapy outcome, except between therapy quality and dysfunctional thinking and IPT skills. Improvement in therapy processes played a larger role in patients whose therapy outcome was poorly predicted by their baseline characteristics. Conclusions: In this study, in which therapy quality varied considerably, we found that therapy quality was not related to change in therapy processes or outcome. In addition, this study provides a first demonstration of proposed methods to identify individual responses to change in therapy processes. Results suggested that the importance of certain therapy processes might differ between patients. Future research into the relation between therapy quality, change in therapy processes and outcome should take into account the variation of therapy quality, focus on improved measurement of therapy quality and use a combination of machine learning techniques and experimental lab studies to determine to which extent and for whom therapy quality and change in therapy processes matter.