|
Abstract:
|
Often clinical trials have one or more interim looks to access early efficacy or futility information. Several methods have been developed to assess efficacy and safety of the study drug during interim. The decision of stopping or continuing a trial, used by these methods, is primarily based on statistical significance, i.e., high/low p-values at interim or high/low probability of rejection of null hypothesis at final analysis given interim data. Most of them do not use prediction of individualized subject responses to convey information regarding potential effect size estimates with trial continuation and associated precision. Evans et al.(2007) introduced predicted intervals as a flexible tool for quantitative monitoring of clinical trials. We propose a Bayesian prediction based method for quantitative monitoring of clinical trials. Proposed methodology is focused on predicting the effect size at future time-points given the interim data. It provides flexibility in the go/no-go decision-making at interim and therefore a very useful tool for Data Monitoring Committee. Monitoring tools will be discussed for binary and time-to-event endpoints with numerical examples using real data.
|
