More on Labeling of Hypertensive Patients

To the Editor: Sackett and his co-workers recently reviewed nine publications related to the phenomenon of labeling hypertensive patients (Nov. 17 issue), and concluded with an apparent paradox. Hypertension-screening programs that contact the smallest proportion of eligible subjects appear to do the most good, or at least do the least harm. Although we share these investigators' concern that selection bias may distort any clinical trial, we believe that several additional issues confound trials of labeling effects and may have at least as great a role in distorting results. First of all, atypical populations may yield atypical results. Haynes et al. . . .