On comparing two skewed distributions with unequal variances.

摘要

We agree with Prof Stang that the concept of P-values is often misinterpreted and its misuse may lead to wrong conclusions in clinical research. We also agree with his call for providing estimated effect sizes and estimated precision whenever presenting results. For this reason, in our analysis [1], not only the summary estimates (with confidence intervals) for outcome variables, event rates, and mean values of morbidity and mortality were provided but also the underlying event rates, total umber of events, and number of subjects used for the calculation.