To the Editor.
—The articles by Dr Liu and colleagues1 and Dr Edmonson and colleagues2 and the Commentary by Dr Braveman and colleagues3 raise interesting questions and alluded to several issues without making them explicit. Hidden in the descriptions of the details of the statistical methods and statistical power in various conflicting studies is the underlying fact that the differences sought are insignificant from the standpoint of any one infant or mother. The Commentary by Braveman et al points out that there is more to clinical practice than is available in studies of outcomes and urges that policymakers recognize that fact.To be useful either clinically or as tools for the formation of public policy, studies of outcome have to choose measures appropriately. Certainly, if early discharge were causing severe medical harm to infants or their mothers, that practice would have to be stopped immediately, but it is intuitively obvious