Survival Methods for Modeling the Time to Drop-Out in Pain Studies

• Activity Number: 256
• Type: Contributed
• Date/Time: Monday, August 2, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Tilda Chuang*+ and Wei Liu and Sandeep Dutta
• Companies: Abbott Laboratories and Abbott Laboratories and Abbott Laboratories
• Keywords: survival analysis ; hazard ; proportional hazard model ; time-to-even data ; pain study ; pain intensity

Abstract

In acute pain studies, drop out rate is often characterized by a flat period (due to encouraged delay of rescue medication use, e.g., for 90 min), followed by a very high drop out rate and then a slower drop out rate. During estimation of the probability of drop out using proportional hazard model, it was found that the correlation between tested covariates and hazard deviated from the log-linear relationship. In order to model the time to drop out, unlike typical method where drop out mainly depends on pain intensity, accumulated pain effect (APE) was added to the model as a covariate. APE was characterized as being 0 for the first 90 min, cumulated to area under time-pain score curve at 90 min, followed by exponentially decline. The model with APE showed significant improvement in the fit and the Kaplan-Meier curve was within the 95% confidence level of the observed survival curve.