In reply
I would like to thank Dr Fujita for his thoughtful comments about our editorial. He questions whether it would be best to use the effect size, odds ratio, or number needed to treat for calculating the sample size needed for a clinical trial with binary data outcomes. Cohen's book Statistical Power Analysis for the Behavioral Sciences1 is an excellent source for sample size calculation. It is easy to read and provides numerous sample calculations that facilitate performance of these estimates by investigators who have relatively modest backgrounds in statistical analysis.