Psychotherapy quality control: the statistical generation of expected recovery curves for integration into an early warning system

Abstract The results of the development of expected recovery curves for an empirically driven patient profiling system are presented. Patients undergoing a course of psychotherapy ( N = 11 492) repeatedly took the Outcome Questionnaire‐45 (OQ‐45). Scores across all patients were combined into an aggregate dataset for use in generating expected recovery curves based on severity of symptoms at intake. SAS PROC MIXED was used to create a mixed linear model of recovery curves based on OQ‐45 scores across sessions and the log transformation of session number. Mean estimates were established for each session from one to 20. Tolerance intervals were then created around each estimated mean score. Expected recovery curves were combined with tolerance intervals to create an early warning system capable of identifying patients whose slow progress suggests that they might be expected to have a negative therapy outcome (terminate treatment prior to obtaining a clinically significant benefit). Current efforts to establish a systematic quality improvement procedure using these curves are discussed. Charts of expected recovery values are plotted, and a straightforward system of patient profiling, early identification of treatment failures, and feedback to clinicians is described. Copyright © 2001 John Wiley & Sons, Ltd.