Department of Biostatistics Seminar/Workshop Series

The utility of Bayesian predictive probabilities for interim monitoring of clinical trials

Benjamin R. Saville, PhD

Assistant Professor, Department of Biostatistics, School of Medicine, Vanderbilt University

Wednesday September 18th, 2013, 1:30-2:30pm MRBIII Room 1220

---

Bayesian predictive probabilities can be used for interim monitoring of clinical trials to estimate the probability of observing a statistically significant treatment effect if the trial were to continue to its predefined maximum sample size. The use of predictive probabilities for interim monitoring is contrasted with alternative strategies such as Bayesian posterior probabilities, p-values, conditional power, and group sequential methods. The benefits of predictive probabilities are discussed in the context of futility and efficacy monitoring, where it is shown that only predictive probabilities properly account for the amount of data remaining to be observed in a clinical trial.

-- AudreyCarvajal - 09 Aug 2013