Higher body mass index predicts need for insulin but not hyperglycemia, nosocomial infection, or death in critically ill surgical patients.

BACKGROUND

Strict glycemic control in critically ill patients has been an important advance in surgical critical care, as hyperglycemia is associated with a higher likelihood of death, complications, and nosocomial infections. Insulin resistance is particularly common in obese patients, but the impact of body mass index (BMI) on insulin requirements, ability to achieve euglycemia, and infectious outcomes in critically ill surgical patients has not been studied. We hypothesized that obese patients would not incur a higher likelihood of infection if euglycemia was maintained.

METHODS

Admissions to the surgical intensive care unit (ICU) from October 1, 2004, to October 31, 2006, were identified. Necessary data were available for 946 patients. The main predictor variable was BMI, which was analyzed as both a continuous and a five-level categorical variable. Data on insulin requirements as well as glycemic control were captured. The main outcome variable was the occurrence of at least one nosocomial infection. Additional outcomes were dysfunction of at least one organ system at any time during surgical ICU admission, quantified using the Multiple Organ Dysfunction Score, as well as the ICU length of stay and death. All statistical analyses were performing using SPSS version 11 for Macintosh.

RESULTS

Both the need for insulin infusion (p = 0.0001) and the mean insulin units/day among patients receiving infusions (p = 0.03) increased significantly with increasing BMI. However, periods of euglycemia were similar among BMI groups. A total of 152 patients (16.1%) incurred at least one nosocomial infection, for a total of 169 infections. The majority (n = 107; 63.3%) were ventilator-associated pneumonias. Neither infection (p = 0.99), organ dysfunction (p = 0.14), ICU length of stay (p = 0.22), nor mortality rate (p = 0.09) differed significantly by BMI group. The need for an insulin infusion was associated significantly with nosocomial infection (p = 0.0001). Additional predictors of infection were a higher Acute Physiology and Chronic Health Evaluation (APACHE) III score (p < 0.0001), age-adjusted APACHE III score (p < 0.0001), and emergency admission (0.001). After controlling for the need for an insulin infusion, BMI was not associated with infection.

CONCLUSIONS

Increasing BMI was associated significantly with insulin resistance. Despite insulin resistance, however, obese patients did not incur longer periods of hyperglycemia. Outcomes that have been associated consistently with glycemic control, such as nosocomial infection and mortality rate, did not differ according to BMI. These data suggest that BMI is not associated with infection during critical illness, and that this absence of an association may be influenced at least partially by the ability to maintain similar glycemic control in obese and non-obese patients.