Fluoxetine response: endpoint vs pattern analysis

A 6-week double-blind trial of fluoxetine treatment of unipolar major depressive disorder in 49 patients was evaluated in terms of improvement in the Hamilton Depression Rating Scale (HDRS) and Clinical Global Improvement (CGI) scores. A relationship was found between change in HDRS (absolute and percentage decrease from baseline) and final CGI ratings. However, because of a few placebo-responders, endpoint analysis with both the HDRS and CGI showed no differences between active drug and placebo. Pattern analysis of persistent response did successively separate active drug from placebo (less than 0.05).