LingeringQuestions Concerning Specific Dietary Fats andMortality

factors,1,2 but no consistent relation with the risk of CHD has been found.3 Thus, in our study4 we did not think about ferritin level as a potential confounder. An important interpretation of our results is that the risk in women seems to be unrelated to menopausal status or to any factors related to menopausal status. There appears tobenoprotectionof being premenopausal vs postmenopausal above what can be explained by a difference in age. The declining relative risk for sex with increasing age seems to be a consequence of a more pronounced flattening of risk level changes in middle-aged men, approaching the risk level for women. This is in contrast to previous beliefs that the risk in women approximate the risk level for men at a certain age. As Mascitelli and Goldstein point out, the absolute risk of MI in women increases about a decade after normal menopausal age. However, an even steeper increase in risk is seen in men, leading to increasing sex differences in absolute risk with increasing age. It is thus difficult to explain the age incidence patterns in view of sex differences in ferritin levels. In a Danish population study5 with follow-up through 1991, standardized age incidence curves were rather parallel for men and women (log-linear scale), in apparent contrast to the ageand sexrelated changes in body iron stores. In addition to increasing oxidative stress, increased iron stores have been suggested to promote CHD by alteration of endothelial function, decreased vascular reactivity, and reperfusion injury by iron-induced free radicals.6 Different markersofbody ironstatushavebeenused for testing the ironheart hypothesis in humans. Despite initial positive findings from experimental and preclinical research, conflicting results have emerged fromepidemiological studies.3,6 Theonly large randomized clinical trial that assessed the effect of ironstore reduction by phlebotomy in patients with peripheral arterial disease found no significant association with allcause mortality or MI.7 Mascitelli and Goldstein refer to results from age-specific analyses of data from this study. A positive effect of reduction in body iron stores were found only among persons in the lowest age quartile. Although the ferritin hypothesis is fascinating and has generatedmuch research, there is no strong evidence for an association betweenbody ironstoresandriskofCHD. It is therefore less likely that body iron can explain the contrast in risk between sexes. However, body iron has been suggested to act synergistically with hypercholesterolemia.3,6 Thus, it is possible that ferritin levels act as effect modifier on the association with established CHD risk factors, and that sex differences in ferritin levels lead to sex heterogeneity in the association with these factors unless adjusted for in the analysis. Interaction effects involving body iron are not yet fully explored.