PURPOSE: Changes in insulin resistance (IR) cause stress-induced hyperglycemia after trauma, but the numerous factors involved in IR have not been delineated clearly. We hypothesized that a statistical model could help determine the relative contribution of different clinical co-variates to IR in critically injured patients. PATIENTS AND METHODS: We retrospectively studied 726 critically injured patients managed with a computer-assisted glycemic protocol at an academic level I trauma center (639 ventilated controls without pneumonia (VWP) and 87 patients with ventilator-associated pneumonia (VAP). Linear regression using age, gender, body mass index (BMI), diabetes mellitus, pneumonia, and glycemic provision was used to estimate M, a marker of IR that incorporates both the serum blood glucose concentration (BG) and insulin dose. RESULTS: Increasing M (p<0.001) was associated with age (1.62%; 95% confidence interval [CI] 1.27%-1.97% per decade), male gender (9.78%; 95% CI 8.28%-12.6%), BMI (4.32% [95% CI 4.02%-4.62%] per 5 points), diabetes mellitus (21.2%; 95% CI 19.2%-23.2%), pneumonia (10.9%; 95% CI 9.31%-12.6%), and glycemic provision (27.3% [95% CI 6.6%-28.1%] per 100 g of glucose). Total parenteral nutrition was associated with a decrease in M of 10.3%; 95% CI 8.52%-12.1%; p<0.001. CONCLUSIONS: Clinical factors can be used to construct a model of IR. Prospective validation might enable early detection and treatment of infection or other conditions associated with increased IR.
PURPOSE: Changes in insulin resistance (IR) cause stress-induced hyperglycemia after trauma, but the numerous factors involved in IR have not been delineated clearly. We hypothesized that a statistical model could help determine the relative contribution of different clinical co-variates to IR in critically injured patients. PATIENTS AND METHODS: We retrospectively studied 726 critically injured patients managed with a computer-assisted glycemic protocol at an academic level I trauma center (639 ventilated controls without pneumonia (VWP) and 87 patients with ventilator-associated pneumonia (VAP). Linear regression using age, gender, body mass index (BMI), diabetes mellitus, pneumonia, and glycemic provision was used to estimate M, a marker of IR that incorporates both the serum blood glucose concentration (BG) and insulin dose. RESULTS: Increasing M (p<0.001) was associated with age (1.62%; 95% confidence interval [CI] 1.27%-1.97% per decade), male gender (9.78%; 95% CI 8.28%-12.6%), BMI (4.32% [95% CI 4.02%-4.62%] per 5 points), diabetes mellitus (21.2%; 95% CI 19.2%-23.2%), pneumonia (10.9%; 95% CI 9.31%-12.6%), and glycemic provision (27.3% [95% CI 6.6%-28.1%] per 100 g of glucose). Total parenteral nutrition was associated with a decrease in M of 10.3%; 95% CI 8.52%-12.1%; p<0.001. CONCLUSIONS: Clinical factors can be used to construct a model of IR. Prospective validation might enable early detection and treatment of infection or other conditions associated with increased IR.