We propose a sample size calculation method for rank tests comparing two survival distributions under cluster randomization with possibly variable cluster sizes. Here, sample size refers to number of clusters. Our method is based on simulation procedure generating clustered exponential survival variables whose distribution is specified by the marginal hazard rate and the intracluster correlation coefficient. Sample size is calculated given significance level, power, marginal hazard rates (or median survival times) under the alternative hypothesis, intracluster correlation coefficient, accrual rate, follow-up period, and cluster size distribution.
No takes yet. Share an insight, caveat, or question.
Sin‐Ho Jung (2007) studied this question.