To the Editor.
—I was much impressed by the article by Detsky and Sackett in the April 1985 issue of the Archives1 dealing with "negative" clinical trials. The discussion of power in hypothesis testing is ignored far too often; the pragmatic approach offered by the authors one hopes will improve this situation by providing scientists a simplified and straightforward way of dealing with the concept of power in hypothesis testing.However, I would like to point out a potential problem arising from the authors' choices of examples that could cause some confusion to the reader. Specifically, the two examples used by the authors to illustrate the use of their algorithm, the problem of home care of myocardial infarction patients without complications offered by Hill et al2, and the example of the educational program to improve compliance of hypertensive patients offered by Sackett and Haynes3, suggest distinctly different questions
Chester ST. Power in Hypothesis Testing. Arch Intern Med. 1986;146(6):1227–1231. doi:10.1001/archinte.1986.00360180247047