Sample size calculation for multicenter randomized trial: taking the center effect into account.

In multicenter trials, data from the same center are more similar than those from different centers. These similarities induce a correlation between data, known as the center effect, which is assessed by the intraclass correlation coefficient (ICC). Here, we derive a sample size formula for continuous data that takes into account this center effect. Our analytical developments lead to an elementary formula different from the classical one by a (1-rho) factor, where rho is the ICC. This work allows for adjusting and reducing the sample size according to the magnitude of the center effect and leads to a better consistency in the conduct of multicenter randomized trials.