Why Bother Psychometrics? One Example from the Hamilton Rating Scale for Depression (HRSD) in a Depression Trial

• Activity Number: 615
• Type: Contributed
• Date/Time: Thursday, August 2, 2012 : 8:30 AM to 10:20 AM
• Sponsor: Biometrics Section
• Author(s): Chengwu Yang*+ and Wenle Zhao
• Companies: Penn State College of Medicine and Medical University of South Carolina
• Keywords: psychometrics ; factor analysis ; clinical trials ; depression ; instruments ; measurements

Abstract

The Hamilton Rating Scale for Depression (HRSD) is widely used in depression trials as the primary measure of depression despite of its well-known severe psychometrical issues. In a trail that compared continuation electroconvulsive therapy vs. pharmacotherapy for depression relapse prevention, the total score of the HRSD items was used to define relapse, and no positive result was found. But this can be misleading, because the uni-dimensional assumption for the HRSD was wrong. In this study, the original data from that trial were used to investigate psychometrical issues in the HRSD. Confirmatory factor analyses (CFA) were implemented at 4 visits to test the uni-dimensional assumption, and it failed at all visits. Thus the total HRSD score cannot be used as the defining variable for relapse, results based on it was misleading, and may conceal some true positive findings. Exploratory factor analysis (EFA) on all of the HRSD items failed to yield a consistent factor structure across the 4 visits. Item-level analyses found statistically significant differences on 3 important items between the two arms. Psychometrics can play a key role in studies using instruments such as the HRSD.