Identifying Treatment Responsive Trajectory Classes in a Depression Trial: An Alternative Statistical Approach

• Activity Number: 256
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Maggie Kuchibhatla*+ and Gerda Fillenbaum
• Companies: Duke University Medical Center and DUMC
• Keywords: Trajectory classes ; growth mixture models ; depression trial ; alternate statistical model

Abstract

In an NIMH sponsored depression trial, the primary outcome score of weekly Hamilton Depression Rating Scale(HAM-D) over 12 weeks analyzed using traditional mixed models did not distinguish effects between the sertraline and placebo groups. We now use generalized growth mixture models(GMM) to determine whether this dataset contains multiple trajectory classes (subgroups) that are responsive to treatment groups. Subgroup characteristics may enable better design, and identify subjects who are likely to most benefit from the drug. A two class piece-wise quadratic model fits the entire sample (n=469) as well as the sertraline group (n=234). The full sample based model consisted of a better outcome and poorer outcome class (82.9% and 17.1% of the total sample respectively). Characteristics that distinguish better and poorer outcome subgroups in treatment groups and, full sample are examined.