Gender as predictor and moderator of outcome in cognitive behavior therapy and pharmacotherapy for adult depression: an "individual patient data" meta-analysis

Background: It has yet to be established whether gender moderates or predicts outcome of psychological and pharmacological treatments for adult depression because: (1) individual randomized trials typically lack sufficient statistical power to detect moderators and predictors and (2) meta-analyses cannot examine such associations directly.

Methods: We conducted an "individual patient data" meta-analysis with the primary data of 1,766 patients from 14 eligible randomized trials comparing cognitive behavior therapy (CBT) with pharmacotherapy. Five studies also compared CBT and pharmacotherapy with pill placebo. We examined the extent to which gender moderates or predicts outcome, using the Hamilton Rating Scale for Depression-17-item (HAM-D-17), with mixed effects models.

Results: Despite the high statistical power, we did not find any indication (P > 0.05) that gender moderates outcome (i.e., no indication that either men or women respond better or worse to CBT than to pharmacotherapy or vice versa). Gender was neither a nonspecific predictor (indicating whether gender is related to improvement, regardless of comparison or control groups), nor a specific predictor (predicting outcome of CBT and pharmacotherapy compared to pill placebo). The average differences between men and women within three conditions (CBT, pharmacotherapy, and pill placebo) were less than one point on the HAM-D-17.

Conclusions: The lack of predictive relations in a this good sized sample suggests that gender does not moderate differential response to CBT versus medication treatment and that it neither predicts nonspecific response across the treatments nor the specificity of response for either treatment with respect to pill placebo.