Evaluating Endpoint Determination and Approaches for Handling Missing Data in Pain Trials

• Activity Number: 256
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Lindsey Lian and Benjamin Luke Peterson*+
• Companies: PPD, Inc. and PPD, Inc.
• Keywords: missing data ; pain trials ; longitudinal data

Abstract

We explore selection of the primary endpoint and a strategy for handling missing data in pain trials. Longitudinal visual analog scale (VAS) data are simulated based on known pain trial results, where data are assumed to be missing at random (MAR). We evaluate change from baseline (CFB) and percent change from baseline (PCFB) as endpoints. Baseline observation carried forward (BOCF), last observation carried forward (LOCF) (with/out baseline), worst observation carried forward (WOCF) (with/out baseline), observed case data, multiple imputation (MI), and repeated measures (RM) are used to handle missing data. The results indicate that RM has the highest power followed by MI, while WOCF with baseline has the lowest. CFB is more powerful than PCFB, except for BOCF and LOCF/WOCF when baseline can be carried forward. Distributions are examined to explain differences between CFB and PCFB.