Assessing accuracy of Weibull shape parameter in clinical trials for time-to-event outcome

In single-arm or two-arm clinical trials, sample size calculation is an integral part of the design for time to event outcomes. Weibull distribution became popular in recent literature for sample size calculation of the survival endpoints. However, to calculate the sample size for the current study using the Weibull distribution requires an estimate of the shape parameter from the prior randomized study or historical observational data. The aim of this paper is to assess how reliable is the estimated shape parameter when it is obtained from the published results of the median survival time and/or corresponding interquartile range. Furthermore, we are also assessing given the sample size of the historical study and censoring rate, how many survival quantiles we need from the published results so that we can reliably use it to estimate the shape parameter in our current study. Our simulation result shows that the point estimate of the shape parameter is quite reliable when the historical study is large enough (>= 50) with the censoring rate of less than 20%. If the historical study has a smaller sample size or high censoring rate, we need more information form the KM curves to improve the reliability and accuracy of the estimated shape parameter.

Author: Milind A. Phadnis PhD, Palash Sharma MS, Nadeesha Thewarapperuma MS, Prabhakar Chalise PhD

Download paper here

Recommended citation: Phadnis, Milind A., Palash Sharma, Nadeesha Thewarapperuma, and Prabhakar Chalise. “Assessing accuracy of Weibull shape parameter estimate from historical studies for subsequent sample size calculation in clinical trials with time-to-event outcome.” Contemporary Clinical Trials Communications (2020): 100548.