To the Editor.
—Dr Col and colleagues1 sought to estimate the degree to which the use of HRT influences life expectancy. As 1 step in this process, they had to estimate the effect of HRT on the risk of a variety of health outcomes among subgroups of women defined according to the presence of various other factors (such as family history or cigarette smoking) that bear on the incidence of these outcomes. Their calculations assumed that the relative impact of HRT on the occurrence of a given outcome was the same for each subgroup. For example, current use of HRT was assumed to reduce the rate of coronary heart disease (CHD) by a factor of 0.6, both among women whose rate otherwise would be high (eg, based on a history of diabetes or high blood pressure) and among those whose rate otherwise would be low.Is the assumption of the authors a