Evaluation of MADRS Severity Thresholds in Patients With Bipolar Depression

BACKGROUND The Montgomery-Asberg Depression Rating Scale (MADRS) is commonly used to assess depression symptom changes in clinical trials; however, the score itself can be difficult to interpret without clinical context. Categories of depression severity corresponding to MADRS total score have not been established for bipolar depression, which was the objective of this study. METHODS Data were pooled from 3 randomized, double-blind, placebo-controlled trials of cariprazine in patients with bipolar I depression; placebo and cariprazine arms were pooled. An anchor-based approach was used to map MADRS total score to the clinician-rated, 7-category Clinical Global Impression of Severity scale (CGI-S). Spearman's correlation coefficient was used to assess associations between MADRS total and CGI-S scores. Optimal MADRS severity thresholds for each CGI-S category was determined via Youden index using receiver operating characteristic (ROC) analyses. RESULTS Using data from 1523 patients with bipolar depression, mean MADRS total scores were positively correlated with mean CGI-S scores at week 6 (r = 0.87; P<.0001). Using ROC curves, MADRS severity thresholds corresponding to each CGI-S category were estimated with high sensitivity and specificity: 0-6 for "normal, not at all ill", 7-12 for "borderline mentally ill", 13-18 for "mildly ill", 19-23 for "moderately ill", 24-36 for "markedly ill", 37-39 for "severely ill", and ≥40 for "extremely ill". CONCLUSIONS Utilizing data from 3 clinical trials of patients with bipolar depression, MADRS severity thresholds were identified. These empirical findings may help clinicians contextualize MADRS results from bipolar clinical research and apply to their practice. TRIAL REGISTRATION clinicaltrials.gov NCT01396447, NCT02670538, NCT02670551.