Insulin resistance despite tight glucose control is associated with mortality in critically ill surgical patients.

BACKGROUND: The hyperglycemic state following trauma and surgery is related partially to insulin resistance (IR). The objective is to determine if critically ill surgical patients vary in their extent of IR and is IR associated with mortality.

METHODS: Prospective observational study in trauma and surgical intensive care units. There were 925 ventilated, critically ill surgical patients who were placed on an automated euglycemia protocol. A mathematic multiplier (M) employed by the protocol was used as a measure of IR. Outcome, phenotypic, laboratory, and treatment variables were analyzed.

RESULTS: 54,141 entries for glucose (mg/dl) and M were analyzed. Median glucose was 118mg/dL, with 45% of values between 80-110mg/dL, 81% between 80-150 mg/dL, and 0.2% less than 40 mg/ dL. M varied by 42 fold over the entire population, and by an average of 11-fold among individual patients. The median blood glucose was not different between groups (118 mg/dl for survivors and 118 mg/dl for non-survivors, P = 0.36). The median insulin dose and M were significantly higher in non-survivors (4.1 U/hr versus 3.4 U/hr, P = 0.005; 0.061 versus 0.058, P = 0.02).

CONCLUSIONS: There was a large amount of variation in insulin resistance, as measured by an adapting multiplier, both across the population and within patients. In the setting of tight glucose control measures of glucose control (median blood glucose and percent in range) do not differentiate between patients who lived and died while measures of insulin resistance (median insulin dose and multiplier) do, suggesting that the insulin resistance is a better predictor of outcome.