Modeling individualized trajectories of symptom change to improve feedback procedures in psychotherapy

Abstract

OBJECTIVE: Psychotherapy feedback compares an individual's treatment progress to the averaged progress of all clients to determine whether their progress is sufficient. However, this can invoke the ecological fallacy if the average trajectory combines heterogenous trajectories with clinically meaningful differences. The current study, instead, explored individualized trajectories of change in psychotherapy and examined the feasibility of using these individualized models to predict clients' future trajectories. METHOD: The Outcome Questionnaire-45 was completed at each session of psychotherapy by 398 adults (16-83 years; Mage = 36.01 years) attending two Australian psychology training clinics. Up to seven Bayesian, polynomial curve-linear regression models were fit and compared for each client. For a hold-out sample (N = 50), models were fit sequentially for each client in five-session increments and used to generate tailored predictions of expected progress at the next five sessions. RESULTS: Constant (no change) and linear (steady change) were the most common shapes of change; only 3% of clients experienced negatively accelerating improvement, as per the expected treatment response curve used in current feedback procedures. Three exemplars demonstrated how individualized modeling and predictions could be utilized in clinical practice to provide richer, more nuanced feedback to psychotherapists about client progress and likely prognosis. CONCLUSIONS: This study was the first to model individualized trajectories of symptom change across psychotherapy and in doing so, uncovered substantial heterogeneity in client trajectories. This means that averaged trajectories are likely to be misleading. Individualized modeling could complement current feedback procedures.