Blinded sample size reestimation in clinical trials with recurrent event data and time-dependent event rates

Welcome to the project

The determination of the number of patients to be included in a clinical study is a key aspect of the study's design. However, the risk of sizing the study inappropriately is often considerable due to uncertain a priori information on variances or event rates. To overcome this problem so-called internal pilot studies have been proposed which allow to reestimate the required sample size during the course of the study based on accruing data. If the interim sample size adjustment is done without breaking the treatment code in blinded, randomized trials, then the procedure is called blinded sample size reestimation (BSSR).

This project considered such adaptive designs for recurrent events and aimed to (a) systematically compare existing BSSR methods for recurrent event data; (b) extend existing methodologies to include covariates such as baseline disease severity and intermittent missingness patterns; (c) develop statistical methodology for recurrent events with time-varying rates; and (d) explore the feasibility of applying these methodologies to multi-regional and multi-national trials with complex recruitment schemes.

The project was funded by the German Research Foundation (DFG).