Approaches for Calculating Power for Case–Cohort Studies

Summary .   An approach for determining the power of a case–cohort study for a single binary exposure variable and a low failure rate was recently proposed by Cai and Zeng (2004, Biometrics 60, 1015–1024). In this article, we show that computing power for a case–cohort study using a standard case–control method yields nearly identical levels of power. An advantage of the case–control approach is that existing sample size software can be used for the calculations. We also propose an additional formula for computing the power of a case–cohort study for the situation when the event is not rare.