Sample size reassessment in a clinical trial to demonstrate a clinical meaningful difference in the relief of back pain

Abstract

The assessment of sample size in clinical trials comparing the means requires a variance estimate of the main efficacy variable. If there is uncertainty on the variance estimate when factors such as the difference between centers/cultures are unknown, the internal pilot study with EM algorithm and the conditional rejection probability (CRP) principles are appropriate to perform a sample-size re-assessment (SSR) without unblinding the treatment status. These two methods fully preserve the Type I error rate. In order to compare these tow methods, both methodologies are applied in a randomized trial dataset and by a simulation study the mean squared error (MSE) of the variance estimator is used to quantify the amount by which the variance estimator differs from the true value of the variance and hence decide for the most appropriate method to perform a SSR. Finally, we conclude that the conditional rejection probability priciple (CRP) is a better approach for re-calculation of the sample- size in the considered clinical trial.