To the Editor: In the January issue, Baqar Husaini, Ph.D, and his colleagues (1) describe racial differences in the incidence of dementia and the costs of treating it. In their discussion, they hypothesize that relatively less common use of "hormone replacement therapy" might account for the overrepresentation of African Americans and women with lower incomes among people afflicted with dementia.
Using similar reasoning, we believed until recently that exogenous hormones administered to peri- and postmenopausal women would prevent cardiovascular disease. Recent findings to the contrary underscore the fact that hormone users are a self-selected population whose relatively better health derives from factors more subtle than income, education, or race. Observational studies are useful for identifying areas of more systematic research but not sufficient for initiating preventive treatment of the general population.