To the Editor.
—In another retrospective epidemiologic study of possible links between insulin-dependent diabetes mellitus (IDDM) and early infant-feeding history, Dr Norris and colleagues1 have used a novel approach in analyzing not diabetic children, but 18 probands considered to be at risk for diabetes development as defined by the presence of 1 (n=14) or more (n=4) antibodies to insulin (78%), glutamic acid decarboxylase (GAD65), or islet cell autoantigen (ICA) 512 combined with insulinoma-associated islet tyrosine phosphatases autoantibody (IA-2). The authors compared retrospective questionnaires to obtain early infant-feeding histories and found no differences between the probands and 153 unrelated comtrols. They conclude that diabetes risk is not linked to any nutritional antigens derived from either plant or animal sources. Based on the analysis of their very small cohort of children that may (or may not) develop diabetes, the authors reject similar studies that collectively analyzed more than 1000 diabetic
Dosch HM. Association of Exposure to Cow's Milk Protein and β-Cell Autoimmunity. JAMA. 1996;276(22):1799. doi:10.1001/jama.1996.03540220023012