Insulin resistance increases before ventilator-associated pneumonia in euglycemic trauma patients

Hyperglycemia caused by stress-induced insulin resistance is associated with both infection and mortality in critically injured patients. The onset of infection may increase stress-induced insulin resistance, leading to hyperglycemia. Hyperglycemia has been shown to precede the diagnosis of ventilator-associated pneumonia (VAP) in critically injured adults and has been suggested to have potential diagnostic importance. However, glycemic control (GC) protocols in critically ill patients limit the development of hyperglycemia despite increasing insulin resistance. Our computer-assisted GC protocol achieves excellent GC, limiting infection-related hyperglycemia while capturing prospectively all glucose values, insulin infusion rates, and the multiplier (M) used to calculate the insulin rate. We hypothesized that surrogate measures of insulin resistance, the insulin infusion rate and multiplier M, would increase prior to the clinical suspicion of VAP, even in euglycemic critically injured patients.

All critically injured patients (2,656) on the computerized glycemic control protocol were included in the analysis and categorized by those developing VAP and those without pneumonia on days 3-10 of their intensive care unit (ICU) stay. Median blood glucose concentration (BG), insulin infusion rate (IDR), and multiplier (M) [Insulin Drip Rate=M*(BG-60)] were determined for VAP patients (n=329) and non-infected ventilated (NIV) patients (n=2,327) on each day of mechanical ventilation. The day of VAP diagnosis according to U.S. Centers for Disease Control and Prevention (CDC) criteria was defined as day zero and VAP patients matched with NIV patients according to ventilator day from -10 to +10. Comparisons were conducted using the Mann-Whitney U test.

Baseline characteristics between VAP and NIV groups did not differ. Measures of insulin resistance increased from the time of injury in both groups. Patients with VAP had significantly greater change in both measures of insulin resistance, IDR and M, in the 48 hours preceding the diagnosis of VAP. These changes occurred despite the fact that the computer-assisted GC protocol achieved lower glucose values in VAP patients for the majority of study days.

Measures of insulin resistance increase in the two days prior to the clinical suspicion of VAP for critically injured patients on the GC protocol. These changes occur despite the protocol maintaining euglycemia. This data suggests that markers of insulin resistance may provide clinically useful information in the early diagnosis of VAP.