Clinical correlates of complete versus partial remission in major depressive disorder: an exploratory analysis of three cohorts.

Abstract

Remission in major depressive disorder (MDD) is conventionally defined as a 17-item Hamilton Depression Rating Scale (HAMD-17) score ≤7, but patients within this range may retain residual symptoms. We explored whether complete remission (HAMD-17 <4) and partial remission (HAMD-17 4-7) differ in baseline clinical features, symptom trajectories and endpoint functioning. Three cohorts were analysed separately: GSRD (N=2583, cross-sectional, tertiary centres), PANDORA (N=215, week 12) and STAR*D level 1 (N=1565). We estimated effect sizes with 95% confidence intervals, false discovery rate (FDR) correction, adjusted and imputed models, alternative cut-offs, individual symptom trajectories and cross-validated discrimination. Baseline differences between complete and partial remitters were small (median absolute standardized difference 0.08-0.15). No baseline variable survived FDR correction; only a concurrent severe-episode rating in GSRD did (odds ratio 2.10, 95% CI 1.43-3.07). In STAR*D, partial remitters had slightly higher baseline HAMD-17 (g=0.15, 95% CI 0.00-0.31), and most item-level associations were explained by overall severity. Baseline features did not discriminate the groups in individual patients (cross-validated AUC 0.42-0.55). Symptom scores of the two groups separated within 4-6 weeks and did not converge during follow-up. In STAR*D, the estimated rate of improvement did not differ detectably, and the model-estimated asymptote was higher in partial remitters. At STAR*D exit, partial remitters had worse work and social adjustment (g=0.45) and quality of life (g=-0.63) despite similar baseline values. Complete and partial remission identify different endpoint states, but baseline clinical features predict this distinction poorly. These exploratory findings require prospective confirmation.