In Reply.—The issue of a dose-response relationship, brought up by Ms Montague-Brown, is important. Finding a dose-response relationship is strong support that an association is not due to chance. Our data include information about the subjects' ages when various food items were introduced in the diet. We do not, however, have detailed information about the frequency of consumption and amounts of foods consumed over time. Even if such data were available, it would still not be appropriate to combine various exposures into one variable, as suggeted by Ms MontagueBrown, because we would not know how much weight to assign to each factor.