INTRODUCTION: Recent randomized prospective data suggest that early hyperglycemia is associated with excess mortality in critically ill patients, and tight glucose control leads to improved outcome. This concept has not been carefully examined in trauma patients, and the relationship of early hyperglycemia to mortality from sepsis in this population is unclear. The objective of this study was to determine the relationship different levels of early blood glucose elevation to outcome in a trauma ICU population. METHODS: The records of all patients admitted to the ICU over a 2-year period at a Level I trauma center were reviewed for age, injury severity scores (ISS), admission Glasgow Coma Scale (GCS) score, base deficit (BD), blood glucose, and mortality. Three possible cutoffs in defining hyperglycemia were examined (glucose > or =110 mg/dL, > or =150 mg/dL, > or =200 mg/dL) in relation to infection and mortality. Early hyperglycemia was defined as elevated blood glucose on hospital days 1 or 2. Those with diabetes mellitus were excluded. RESULTS: From 1/00-12/01, 516 eligible patients were admitted to the ICU after injury. Early hyperglycemia occurred in 483 at the > or =110 mg/dL level, 311 at the > or =150 mg/dL level, and 90 patients at the > or =200 mg/dL level. Univariate logistic regression demonstrated a significant relationship between ISS and subsequent infection(p = 0.02) and a trend toward such a relationship in GCS score, glucose > or =150 mg/dL, and glucose > or =200 mg/dL (p = 0.06, 0.12, and 0.06). A similar analysis for the relationship of these variables to eventual mortality showed a significant correlation with all examined variables except glucose > or =110 mg/dL. Multiple logistic regression to control for the effect of age, ISS, GCS score, and BD found early glucose > or =200 mg/dL to be an independent predictor of both infection and mortality while no such relationship was found with > or = 110 mg/dL or > or =150 mg/dL. CONCLUSIONS: Early hyperglycemia as defined by glucose > or =200 mg/dL is associated with significantly higher infection and mortality rates in trauma patients independent of injury characteristics. This was not true at the cutoffs of > or =110 mg/dL or > or =150 mg/dL. These data support the need for a prospective analysis of tight glucose control, keeping serum glucose <200 mg/dL in critically ill trauma patients. However, aggressive maintenance of levels <110 mg/dL as reported by others may not be necessary.
INTRODUCTION: Recent randomized prospective data suggest that early hyperglycemia is associated with excess mortality in critically illpatients, and tight glucose control leads to improved outcome. This concept has not been carefully examined in traumapatients, and the relationship of early hyperglycemia to mortality from sepsis in this population is unclear. The objective of this study was to determine the relationship different levels of early blood glucose elevation to outcome in a trauma ICU population. METHODS: The records of all patients admitted to the ICU over a 2-year period at a Level I trauma center were reviewed for age, injury severity scores (ISS), admission Glasgow Coma Scale (GCS) score, base deficit (BD), blood glucose, and mortality. Three possible cutoffs in defining hyperglycemia were examined (glucose > or =110 mg/dL, > or =150 mg/dL, > or =200 mg/dL) in relation to infection and mortality. Early hyperglycemia was defined as elevated blood glucose on hospital days 1 or 2. Those with diabetes mellitus were excluded. RESULTS: From 1/00-12/01, 516 eligible patients were admitted to the ICU after injury. Early hyperglycemia occurred in 483 at the > or =110 mg/dL level, 311 at the > or =150 mg/dL level, and 90 patients at the > or =200 mg/dL level. Univariate logistic regression demonstrated a significant relationship between ISS and subsequent infection(p = 0.02) and a trend toward such a relationship in GCS score, glucose > or =150 mg/dL, and glucose > or =200 mg/dL (p = 0.06, 0.12, and 0.06). A similar analysis for the relationship of these variables to eventual mortality showed a significant correlation with all examined variables except glucose > or =110 mg/dL. Multiple logistic regression to control for the effect of age, ISS, GCS score, and BD found early glucose > or =200 mg/dL to be an independent predictor of both infection and mortality while no such relationship was found with > or = 110 mg/dL or > or =150 mg/dL. CONCLUSIONS: Early hyperglycemia as defined by glucose > or =200 mg/dL is associated with significantly higher infection and mortality rates in traumapatients independent of injury characteristics. This was not true at the cutoffs of > or =110 mg/dL or > or =150 mg/dL. These data support the need for a prospective analysis of tight glucose control, keeping serum glucose <200 mg/dL in critically ill traumapatients. However, aggressive maintenance of levels <110 mg/dL as reported by others may not be necessary.