Dr Graham claims that our study overestimated the rate of ALF and implies that the sponsor had undue influence over the study design. The essence of ALF is diminished hepatic function occurring over a period of a few months.1 Our criteria for probable ALF were new-onset encephalopathy, hyperbilirubinemia, and coagulopathy. At least one of these criteria of impaired liver function was required for a diagnosis of possible ALF. In our study,2 10 of 26 patients with probable or possible ALF without probable cause other than use of hypoglycemic agents had encephalopathy. If we count only these 10 patients, as Graham suggests, the ALF incidence estimate among patients using hypoglycemic agents would be approximately 4 per 100 000, still higher than the 1 per million rate he cites, and would not substantially alter our conclusion.
Chan KA, Truman A, Gurwitz JH, Hurley J, Martinson B, Platt R, Moseley R, Selby J. Risk of Unexplained Acute Liver Failure in Diabetes Mellitus—Reply. Arch Intern Med. 2003;163(21):2650-2651. doi:10.1001/archinte.163.21.2649-a