|
Abstract:
|
Blinded sample-size re-estimation allows for modifying the sample size without compromising the integrity of the trial. One challenge in this is the blinded estimation of the variance while the trial is ongoing. Existing methods are often biased, based on untenable assumptions or applicable to only two-treatment trials. We propose a simple unbiased method, which uses subject enrollment order and randomization block size to estimate the variance. It can be applied to normal or non-normal data, trials with two or more treatments, equal or unequal allocations, and fixed or random block sizes. The proposed variance estimator performs best when the randomization block size is the smallest. Simulations and data analyses show that for many commonly used randomization block sizes, the proposed method performs well in re-estimating the variances and sample sizes in a blinded manner.
|
