A demonstration of a multi-method variable selection approach for treatment selection: Recommending cognitive–behavioral versus psychodynamic therapy for mild to moderate adult depression

Zachary D. Cohen*, Thomas T. Kim, Henricus L. Van, Jack J.M. Dekker, Ellen Driessen

*Corresponding author for this work

Research output: Contribution to JournalArticleAcademicpeer-review

122 Downloads (Pure)


Objective: We use a new variable selection procedure for treatment selection which generates treatment recommendations based on pre-treatment characteristics for adults with mild-to-moderate depression deciding between cognitive behavioral (CBT) versus psychodynamic therapy (PDT). Method: Data are drawn from a randomized comparison of CBT versus PDT for depression (N = 167, 71% female, mean-age = 39.6). The approach combines four different statistical techniques to identify patient characteristics associated consistently with differential treatment response. Variables are combined to generate predictions indicating each individual’s optimal-treatment. The average outcomes for patients who received their indicated treatment versus those who did not were compared retrospectively to estimate model utility. Results: Of 49 predictors examined, depression severity, anxiety sensitivity, extraversion, and psychological treatment-needs were included in the final model. The average post-treatment Hamilton-Depression-Rating-Scale score was 1.6 points lower (95%CI = [0.5:2.8]; d = 0.21) for those who received their indicated-treatment compared to non-indicated. Among the 60% of patients with the strongest treatment recommendations, that advantage grew to 2.6 (95%CI = [1.4:3.7]; d = 0.37). Conclusions: Variable selection procedures differ in their characterization of the importance of predictive variables. Attending to consistently-indicated predictors may be sensible when constructing treatment selection models. The small N and lack of separate validation sample indicate a need for prospective tests before this model is used.

Original languageEnglish
Pages (from-to)137-150
Number of pages14
JournalPsychotherapy Research
Issue number2
Early online date11 Jan 2019
Publication statusPublished - 2020


Financial support for this work was provided by MQ Foundation to ZC [MQ: Transforming mental health MQ14PM_27]. MQ had no role in the study design, collection, analysis, or interpretation of the data, or in writing the manuscript or the decision to submit the article for publication. The randomized clinical trial from which data were drawn for this study was financed by an unrestricted research grant by Wyeth Pharmaceuticals, The Netherlands. Arkin Mental Health Care, The Netherlands, financially supported research logistics and the contributions of ED, HLV, and JJMD. Vrije Universiteit Amsterdam, Faculty of Behavioral and Movement Sciences, Section Clinical Psychology, The Netherlands, financially supported ED?s contributions to the study. None of the sponsors had a role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; nor in the preparation, review, or approval of the manuscript. The opinions and assertions contained in this article should not be construed as reflecting the views of the sponsors. We thank Robert DeRubeis for helpful comments on the manuscript and support of this work. We also wish the acknowledge the reviewers, whose feedback greatly improved this article and made us feel like we had gained three insightful coauthors. This work was supported by a grant from MQ: Transforming mental health MQ14PM_27.

FundersFunder number
MQ FoundationMQ14PM_27
Wyeth Pharmaceuticals


    • cognitive behavioral therapy
    • depression
    • precision medicine
    • psychodynamic therapy
    • treatment selection
    • variable selection


    Dive into the research topics of 'A demonstration of a multi-method variable selection approach for treatment selection: Recommending cognitive–behavioral versus psychodynamic therapy for mild to moderate adult depression'. Together they form a unique fingerprint.

    Cite this