Objective: Mounting evidence suggests a role for glucose variability in predicting intensive care unit (ICU) mortality. We investigated the association between glucose variability and intensive care unit and in-hospital deaths across several ranges of mean glucose.
Design: Retrospective cohort study.
Setting: An 18-bed medical/surgical ICU in a teaching hospital.
Patients: All patients admitted to the ICU from January 2004 through December 2007.
Measurements and main results: Two measures of variability, mean absolute glucose change per hour and sd, were calculated as measures of glucose variability from 5728 patients and were related to ICU and in-hospital death using logistic regression analysis. Mortality rates and adjusted odds ratios for ICU death per mean absolute glucose change per hour quartile across quartiles of mean glucose were calculated. Patients were treated with a computerized insulin algorithm (target glucose 72-126 mg/dL). Mean age was 65 +/- 13 yrs, 34% were female, and 6.3% of patients died in the ICU. The odds ratios for ICU death were higher for quartiles of mean absolute glucose change per hour compared with quartiles of mean glucose or sd. The highest odds ratio for ICU death was found in patients with the highest mean absolute glucose change per hour in the upper glucose quartile: odds ratio 12.4 (95% confidence interval, 3.2-47.9; p < .001). Mortality rates were lowest in the lowest mean absolute glucose change per hour quartiles.
Conclusions: High glucose variability is firmly associated with ICU and in-hospital death. High glucose variability combined with high mean glucose values is associated with highest ICU mortality. In patients treated with strict glycemic control, low glucose variability seemed protective, even when mean glucose levels remained elevated.