Predicting mortality of critically ill patients by blood glucose levels

Background: The aim of this study is to observe the outcome of critically ill patients in relation to blood glucose level at admission and to determine the optimal range of blood glucose at admission predicting lower hospital mortality among critically ill patients.

Methods: We conducted a retrospective cohort study of a total 1,224 subjects (males, 798; females, 426) admitted to intensive care unit (ICU) from 1 January 2009 to 31 December 2010. Blood glucose levels at admission were categorized into four groups (group 1, <100 mg/dL; group 2, 100 to 199 mg/dL; group 3, 200 to 299 mg/dL; and group 4, ≥300 mg/dL).

Results: Among 1,224 patients, 319 patients were already known diabetics, and 296 patients died in ICU. Five hundred fifty-seven subjects received insulin therapy, and 118 received oral hypoglycemic agents. The overall mortality rate was 24.2% (296 patients). The causes of death and mortality rates of diabetic patients were not different from nondiabetic subjects. The mortality curve showed J shape, and there were significant differences in mortality between the groups of blood glucose levels at admission. Group 2 had the lowest mortality rate (P<0.05).

Conclusion: These results suggest that serum glucose levels upon admission into ICU is associated with clinical outcomes in ICU patients. Blood glucose level between 100 and 199 mg/dL at the time of ICU admission could predict lower hospital mortality among critically ill patients.