Caution regarding the choice of standard deviations to guide sample size calculations in clinical trials