Early symptom change prediction of remission in depression treatment.

ABSTRACT

OBJECTIVES: This study investigated hypothesized early symptom changes as differential predictors of long-term remission for duloxetine and escitalopram. EXPERIMENTAL DESIGN: This was a post-hoc analysis from a placebo-controlled, randomized, double-blind study of patients with major depressive disorder treated for 8 weeks with duloxetine 60 mg/day (N = 273) or escitalopram 10 mg/day (N = 274), and for another 6 months with duloxetine up to 120 mg/day or escitalopram up to 20 mg/day. Odds ratios (ORs) for successful treatment (sustained remission), defined as a 17-item Hamilton Depression Rating Scale (HAMD-17) score mood, retardation, and anxiety symptom factor subscales (20% decrease), along with associated positive predictive values (PPVs) and negative predictive values (NPVs). PRINCIPAL OBSERVATIONS For both drugs, 2-week HAMD-17 improvement on all symptom subscales (except sleep for duloxetine) significantly predicted remission with ORs > 2.0. In a follow-up analysis, specific subscale items for psychological anxiety, motor retardation, and suicidality significantly predicted remission for duloxetine, and psychological and somatic anxiety for escitalopram. Notably, high NPVs on the Maier subscale indicated that a lack of 20% improvement on the "core" depression factor by Week 2 was highly predictive of unsuccessful treatment outcome over 8 months. CONCLUSIONS: In accord with hypotheses, early symptom changes were specific to treatment, with early response in the core depression factor (Maier subscale), anxiety, and motor activity for duloxetine, and core factor and anxiety for escitalopram. Lack of early response in depression symptom subscales was highly predictive of lack of sustained remission.