How Might the Relationship Between Early Response and Depression Remission Differ in Naturalistic and Standardized Cognitive Therapy?


Study Aims:

1.  To determine whether the strength of the relationship between early treatment response and ultimate treatment outcome was sufficient to guide clinical decision-making.

2.  To determine whether the best predictor of outcome was change early in treatment (e.g., at week 4) or the severity of the patient’s symptoms at week 4.

3. To test the hypothesis that the relationship between early treatment response and treatment outcome was smaller in a naturalistic sample than in a sample of patients treated using a fixed protocol. We predicted that the relationship between early response and treatment would be smaller when patients were treated in a setting in which therapists monitored progress at every session and made changes when patients failed to show an early response to treatment.

Study Description:

We examined the strength of the relationship between early response and remission in two samples of depressed patients who received cognitive therapy: 124 patients who received naturalistic treatment in a private practice setting, and 41 patients who received protocol treatment in a research setting. We predicted that the relationship between early response and remission would be substantial enough to guide clinical decision-making in both samples, but that the relationship would be weaker in the naturalistic sample due to the clinician’s ability to make changes to the treatment in response to poor early outcome. We also predicted that a simple model of severity would predict remission as effectively as a more complex change score. Results of logistic regressions showed that a simple model based on Week 4 BDI score was as predictive of remission as more complex models of early change. Receiver Operating Characteristics curves showed that BDI score at week 4 was moderately predictive of remission in the naturalistic treatment (area under the curve of 0.78) and almost completely predictive of remission in the protocol treatment (area under the curve of 0.94). As predicted, the relationship between early response and remission was weaker in the naturalistic than the protocol treatment. Our results point to the advantages of modifying treatment when the depressed patient does not show an early response.

Investigator at the Cognitive Behavior Therapy and Science Center:

Jacqueline B. Persons

Research collaborator:

Cannon Thomas, San Francisco Group for Evidence-based Psychotherapy and University of California at San Francisco

Study Status:

Submitted for publication: Click here to access an earlier version of the manuscript. An early version of this study was presented at the annual convention of the Association for Behavioral and Cognitive Therapies in Toronto, Canada in November 2011.


We thank our patients for contributing data acquired during their treatment, our colleagues at the SFBACCT for contributing data provided by their patients, Corey Pallatto Hughan for coordinating the construction of the anonymized dataset, and Howard Liu and Irina Li for help assembling the materials for this paper and the conference presentation.