Discovering How Youth Psychotherapies Work: Three Approaches to Identifying Mediators of Treatment Outcome

Abstract

Although a lot is known about which psychotherapies work, surprisingly little is known about how they work. Understanding the processes and mechanisms through which evidence-based psychotherapies (EBPs) improve outcomes can enrich theories of change and inform treatment refinement. I conducted three studies intended to identify candidate change processes and mechanisms that mediate the effects of EBPs for youths, and also to illustrate the use of quantitative methods that can advance research on change processes and mechanisms, but are rarely used in psychotherapy research. In Study 1, I conducted a systematic review and meta-analysis of mediation effects in EBPs for youth depression. Using meta-analytic structural equation modeling, negative thinking, and social skills and relations were identified as robust mediators, and family dysfunction was not a significant mediator, across multiple randomized trials. Problem solving, pleasant activities, cognitive skills, and avoidant coping also did not significantly mediate outcomes, but analyses involved fewer trials; more data are needed on these candidate mediators as well as others (e.g., role engagement, therapeutic alliance) that had insufficient data for analysis. In Study 2, I examined youth coping and therapeutic alliance as candidate mediators in a randomized effectiveness trial comparing cognitive behavioral therapy and usual care for youths with internalizing disorders. Bootstrapping of indirect effects with multiply imputed datasets revealed that improved youth-reported secondary control coping (i.e., adjusting oneself to adapt to existing conditions) may be a potential change mechanism underlying the effectiveness of usual care, and improved parent-reported youth secondary control and primary control coping (i.e., influencing existing conditions to fit one’s goals) may be common factors predicting better outcomes across therapies. In Study 3, I investigated whether parent-identified “top problems”—a measure of youths’ achievement of personalized treatment goals—as a candidate change process that might mediate overall reduction of youth internalizing and externalizing symptoms in a randomized effectiveness trial comparing modular EBPs to standard EBPs. Bivariate latent change score models suggested that top problems did not predict subsequent symptom change, but that symptoms predicted subsequent change in top problem severity. Thus top problems was not supported as a change process through which modular EBPs outperformed standard EBPs.