Admission hyperglycemia and other risk factors as predictors of hospital mortality in a medical ICU population

Background: Tight glycemic control is recommended for patients in the ICU, as hyperglycemia is associated with increased morbidity and mortality. Design: Observational cohort of patients admitted to a 12-bed, inner-city, medical ICU (MICU). Subjects: A total of 1,185 of 1,506 patients from July 1, 1999, to December 31, 2002, selected based on a diagnosis other than diabetic ketoacidosis or glycemia > 280 mg/dL or < 80 mg/dL. Purpose: To determine if the highest serum glucose level within 24 h after ICU admission is associated with increased hospital mortality when adjusted for confounders. Measurements: Age, gender, race, worst values within 24 h after ICU admission to construct the acute physiology and chronic health evaluation (APACHE) II score, and highest glucose within 24 h after ICU admission. Hospital mortality was the primary outcome. Admitting diagnosis, MICU length of stay (LOS), and hospital LOS were obtained. Glucose, albumin (n = 867), and lactic acid (n = 319) were stratified for analysis. Analysis: Univariate analysis identified factors included in the multivariate model. Results: Patients were predominantly African-American (79%) and men (56%; mean age, 49.2 years). The mean ICU admission highest glucose level was 139 +/- 43.7 mg/dL (+/- SD). MICU LOS and hospital LOS were 6.2 days and 12.9 days, respectively, and 50% of patients received mechanical ventilation. MICU and hospital mortality were 18% and 20%, respectively; standardized mortality ratio was 66%. On univariate analysis, survivors (n = 945) and nonsurvivors (n = 240) showed APACHE 11 score, mechanical ventilation, hypoalbuminemia, lactic acidemia, and logistic organ dysfunction system score to be hospital mortality predictors; however, the highest admission serum glucose level was not. Logistic regression estimated APACHE 11 score/per point (odds ratio, 1.06; 95% confidence interval, 1.02 to 1.11), mechanical ventilation (odds ratio, 3.06; 95% confidence interval, 1.34 to 6.96), severe hypoalbuminemia (< 2 g/dL) [odds ratio, 2.98; 95% confidence interval, 1.3 to 7.02], and severe lactic acidemia (>= 8 mmol/L) [odds ratio, 7.3; 95% confidence interval, 2.14 to 24.9], but not ICU admission hyperglycemia, to be associated with hospital mortality. Conclusions: Conventional factors of disease severity, but not highest glucose value during the first 24 h after ICU admission, predict hospital mortality in an inner-city MICU.