Background: Both a history of diabetes mellitus and elevated inhospital glucose levels predict death after acute myocardial infarction (AMI). However, only diabetes history (and not glucose levels) is routinely considered in AMI risk assessment.
Methods: We conducted a post hoc analysis of 2 randomized controlled trials of AMI with ST-segment elevation to compare the prognostic value of inhospital glucose levels with diabetes history in 30,536 subjects. Average inhospital glucose (mean of glucose levels at admission, 6 hours, and 24 hours), diabetes history, and death at 30 days (occurring in 2,808 subjects) were documented.
Results: Average glucose predicted 30-day death (OR 1.10 per 1-mmol/L [18-mg/dL] increase, 95% CI 1.09-1.11, P < .0001); this was unchanged after adjusting for diabetes history. In contrast, diabetes history alone predicted 30-day death (OR 1.63, 95% CI 1.48-1.78, P < .0001), but not after adjusting for average glucose (OR 0.98, 95% CI 0.88-1.09, P = .72). The C-indices (areas under the receiver operating characteristic curves) for 30-day death were 0.54 for diabetes history alone, 0.64 for average glucose alone, and 0.64 for glucose plus diabetes. Higher glucose levels predicted death in patients with and without diabetes history, but this relationship was more steep in nondiabetic subjects such that their rate of 30-day death (13.2%) matched that of diabetic patients (13.7%) when average glucose was > or =144 mg/dL (8 mmol/L) (P = .55 after multivariable adjustment).
Conclusions: Although diabetes history is routinely considered in the risk stratification of AMI patients, inhospital glucose levels are a much stronger predictor of death and should be incorporated in their risk assessment. Patients with AMI with inhospital glucose > or =144 mg/dL have a very high risk of death regardless of diabetes history.