Aims: The aims of this study were to examine the relationship between admission blood glucose and mortality in a large, unselected cohort of acutely ill medical patients and to assess the impact of diabetes on this relationship.
Methods: We studied the broad pattern of acute medical admissions over an eight year period and the impact of admission serum glucose on in-hospital mortality. Significant predictors of outcome, including acute illness severity and co-morbidity, were entered into a multivariate regression model, adjusting the univariate estimates of the glycaemic status on mortality.
Results: There were 45,068 consecutive acute medical emergency admissions between 2005 and 2012. The normoglycaemic (>4.0 ≤7.0 mmol/l) cohort (86%) had a 3.9% in-hospital mortality. Both hypoglycaemia (OR: 3.23: 95% CI: 2.59-4.04; p<0.001) and hyperglycaemia (OR: 2.1; 95% CI: 1.9-2.4; p<0.001) predicted an increased risk of an in-hospital death. Neither of these increased risks were fully adjusted nor explained by a highly predictive outcome model, using multiple acute illness parameters. Hyperglycaemia did not carry similar adverse prognostic implications for patients with diabetes.
Conclusion: In patients without diabetes, an abnormal serum glucose is independently predictive of an increased mortality among the broad cohort of acute emergency medical patients. Similar disturbances of glucose homeostasis for patients with diabetes do not confer equivalent adverse prognostic implications.
Keywords: Acute illness; Glucose; Hospital admission; Risk prediction.
Copyright © 2013 Elsevier Ireland Ltd. All rights reserved.