Melancholic features (DSM-IV) predict but do not moderate response to antidepressants in major depression: an individual participant data meta-analysis of 1219 patients

It is sometimes clinically believed that major depression with melancholic features is more responsive to antidepressants than non-melancholic depression. Proper analysis and, therefore, valid evidence to support or refute this common clinical lore is lacking. The sample was taken from three placebo-controlled randomized trials of duloxetine, escitalopram and paroxetine (n = 1219). We conducted a two-step individual participant data meta-analysis to combine linear mixed-effects regressions modeling melancholic features as prognostic factor (variable that predicts overall response regardless of the treatments) and as effect modifier (variable that predict differential response to drug over placebo). Melancholic features represented a statistically significant prognostic factor for greater reduction in depression severity both on antidepressants and on placebo, especially after 4 weeks of treatment. However, they were not an effect modifier of the antidepressant treatment through the acute phase treatment: in other words. The superiority of antidepressants over placebo was not influenced by the melancholic features. The treatment decision-making as to the benefits of antidepressant treatment for patients with major depression should not be influenced by the presence or absence of melancholic features.