Development of a placebo effect model combined with a dropout model for bipolar disorder

Wan Sun, Thomas P. Laughren, Hao Zhu, Guenther Hochhaus, Yaning Wang

The aim of this study was to develop a placebo model for bipolar disorder to help optimize clinical trial designs for studies targeting manic episodes in bipolar disorder. A bipolar disease database was built based on individual longitudinal data collected from over 3,000 patients in 11 clinical trials for 5 approved bipolar drugs. An empirical placebo effect model with an exponential decay process plus a linear progression process was developed to quantify the time course of the Young Mania Rating Scale total score based on only placebo data from the database. In order to describe the dropout pattern during the trials, a parametric survival model was developed and the Weibull distribution was identified to be the best distribution to describe the data. Based on the likelihood ratio test, it was found that patients with higher baseline score, slower disease improvement and more rapid disease progression tended to dropout earlier, and the trial features such as trial starting year and trial site were also significant covariates for dropout. A combination of the placebo effect model and the dropout model was applied to simulate new clinical trials through Monte-Carlo simulation. Both the placebo effect model and dropout model described the observed data reasonably well based on various diagnostic plots. The joint placebo response and dropout models can serve as a tool to simulate the most likely level of placebo response with the expected dropout pattern to help design a new clinical trial.