U-shaped relationship between early blood glucose and mortality in critically ill children

BACKGROUND The aims of this study are to evaluate the relationship between early blood glucose concentrations and mortality and to define a 'safe range' of blood glucose concentrations during the first 24 h after pediatric intensive care unit (PICU) admission with the lowest risk of mortality. We further determine whether associations exist between PICU mortality and early hyperglycemia and hypoglycemia occurring within 24 h of PICU admission, even after adjusting for illness severity assessed by the pediatric risk of mortality III (PRISM III) score. METHODS This retrospective cohort study included patients admitted to PICU between July 2008 and June 2011 in a tertiary teaching hospital. Both the initial admission glucose values and the mean glucose values over the first 24 h after PICU admission were analyzed. RESULTS Of the 1349 children with at least one blood glucose value taken during the first 24 h after admission, 129 died during PICU stay. When analyzing both the initial admission and mean glucose values during the first 24 h after admission, the mortality rate was compared among children with glucose concentrations ≤ 65, 65-90, 90-110, 110-140, 140-200, and >200 mg/dL (≤ 3.6, 3.6-5.0, 5.0-6.1, 6.1-7.8, 7.8-11.1, and >11.1 mmol/L). Children with glucose concentrations ≤ 65 mg/dL (3.6 mmol/L) and >200 mg/dL (11.1 mmol/L) had significantly higher mortality rates, indicating a U-shaped relationship between glucose concentrations and mortality. Blood glucose concentrations of 110-140 mg/dL (6.1-7.8 mmol/L), followed by 90-110 mg/dL (5.0-6.1 mmol/L), were associated with the lowest risk of mortality, suggesting that a 'safe range' for blood glucose concentrations during the first 24 h after admission in critically ill children exists between 90 and 140 mg/dL (5.0 and 7.8 mmol/L). The odds ratios of early hyperglycemia (>140 mg/dL [7.8 mmol/L]) and hypoglycemia (≤ 65 mg/dL [3.6 mmol/L]) being associated with increased risk of mortality were 4.13 and 15.13, respectively, compared to those with mean glucose concentrations of 110-140 mg/dL (6.1-7.8 mmol/L) (p <0.001). The association remained significant after adjusting for PRISM III scores (p <0.001). CONCLUSIONS There was a U-shaped relationship between early blood glucose concentrations and PICU mortality in critically ill children. Both early hyperglycemia and hypoglycemia were associated with mortality, even after adjusting for illness severity.