No abstract

T he issue of glucose control for seriously ill patients in the intensive care unit (ICU) is a subject of controversy. Some studies showed improved outcomes with tight glycemic control, but others reported worse outcomes. Still others have focused on blood glucose amplitude variability (BGAV), hypothesizing that increased fluctuation of glucose levels would be associated with increased mortality. The most common measure of variability was SD, although some studies used the presence of both hypoglycemia and hyperglycemia. The Dutch National Intensive Care Evaluation registry uses glucose regulation as 1 of 11 indicators of quality of care in the ICU. All glucose values of all admitted patients are collected prospectively, which permits analysis of the relationship between BGAV and mortality. This cohort study was designed to test the hypothesis that BGAV is independently associated with hospital mortality in medical and surgical ICU patients.Patients had 24 or more hours of stay in the ICU and had 3 or more glucose measurements. Most Dutch ICUs use nursedirected insulin dosing regimens with serum glucose targets between 4.4 and 8.0 mmol/L. The SD was used as the measurement of BGAV. In addition, the mean amplitude of glycemic excursions (MAGE) was used and defined as the mean of the absolute values of any $ glucose from consecutive measurements that were higher than the SD of the entire set of glucose values. The mean absolute glucose change per hour (MAG) was defined as the sum of all absolute glucose changes divided by the time in hours. Glycemic lability index (GLI) was also determined and depends on the order and timing of the glucose measurements. The primary outcome was hospital mortality.From 20,375 patients admitted to 37 ICUs, 855,032 glucose measurements were available for analysis. Median patient age was 67 years; median APACHE IV (Acute Physiology and Chronic Health Evaluation IV) expected mortality rate was 15%; 52% of patients were medical; median glucose level was 7.3 mmol/L, and median SD was 1.73 mmol/L; 78% of patients survived. In the combined cohort of all patients, adjusted hospital mortality was associated with SD (P G 0.0001) of glucose levels and MAGE (P G 0.0012) but not with MAG (P = 0.92) and GLI (P = 0.53). In surgical patients, adjusted hospital mortality was associated with the SD (P G 0.0001), MAGE (P G 0.0001), and MAG (P = 0.04), but not with GLI (P = 0.17). In medical ICU patients, adjusted mortality was associated with SD (P G 0.0001), but not with the other assessed measures of BGAV (MAGE, MAG, GLI [P = 0.61, 0.36, 0.77, respectively]).The association between BGAV and mortality depends on the measure of BGAV and the patient population, whether surgical or medical. Glycemic lability index was not associated with mortality in either type of ICU admission. Most studies, including this one, have found SD to be an independent predictor of mortality. The presence of hypoglycemia or hyperglycemia was associated with hospital mortality, but this measure depends on only the 2 most extre...

No abstract