Blood Glucose Variability Predicts ICU Deaths

Wide fluctuation in blood glucose levels in critically ill patients is a predictor of mortality in those patients, suggests a study presented here at the American Thoracic Society 2008 International Conference. Hasan Al-Dorzi, MD, a pulmonary critical care consultant at King AbdulAziz Medical City/King Fahad National Guard Hospital in Riyadh, Saudi Arabia, said he and fellow researchers were prompted to investigate the effect of glycemic fluctuations on the outcomes of critically ill patients.

"We know that hyperglycemia is common in critically ill patients and is not restricted to patients who are diabetic," said Dr. Al-Dorzi. "We also know patients who have hyperglycemia also have worse outcomes, such as higher rates of infection, longer hospital stays, and higher death rates."

Most studies in the area of in-hospital patients and blood glucose have focused on hyperglycemia and managing hyperglycemia in hospital, but there is scant research that has addressed the question of variations in blood glucose levels and how to manage those variations, according to Dr. Al-Dorzi.

"We know that at a cellular level...there is evidence of this up-and-down fluctuation of glucose being harmful to cells," said Dr. Al-Dorzi. "We also know from diabetes research that diabetic patients who have up-and-down fluctuations, even when not in hospital, have a higher incidence of complications."

A total of 523 patients in a surgical intensive care unit were randomly assigned in a prospective fashion to either intensive or conventional insulin therapy to maintain levels of blood glucose between 4.4 to 6.1 mmol/L and 10.0 to 11 mmol/L, respectively. Investigators obtained the mean, highest, and lowest blood glucose as a marker of glycemic variability. Patients were divided into high-variability and low-variability groups according to the median of the glucose range (6.0 mmol/L).

The researchers found that predictors of glycemic variability were age (odds ratio [OR], 1.03 for every additional year; 95% confidence interval [CI], 1.02 - 1.05; P < .0001), diabetes history (OR, 3.00; 95% CI, 1.74 - 5.17; P < .0001), and daily dose of insulin (OR, 1.04; 95% CI, 1.03 - 1.05; P < .0001). Patients in the high-variability group had more than a 2-fold increase in mortality over those in the low-variability group (22.2% and 8.4%, respectively; P < .001).

The researchers statistically adjusted for possible confounding factors that may influence patient outcomes, such as age, diabetes history, and daily dose of insulin, to isolate the effect of glycemic variations on outcomes of critically ill patients. They found that glycemic fluctuations independently predicted intensive care unit and hospital mortality (ICU: OR per millimole, 1.12; 95% CI, 1.03 - 1.22; P = .01; hospital: OR per millimole, 1.09; 95% CI, 1.02 - 1.16; P = .01). Glucose range also independently predicted the incidence of nosocomial infection (OR per millimole, 1.11; 95% CI, 1.02 - 1.21; P = .01).

"Even after we controlled for these variables, we still found glycemic variations to be a predictor of mortality in the ICU [intensive care unit]," said Dr. Al-Dorzi.

Dr. Al-Dorzi put forth that the solution to decreasing the fluctuations in blood glucose that occur in critically patients potentially lies in advances in technology.

"We should try to minimize these fluctuations in blood sugar," said Dr. Al-Dorzi. "If we have devices in the future that allow for continuous monitoring of blood glucose, and we have on hand computerized protocols that can detect changes more rapidly and allow insulin infusion rapidly, we can correct these fluctuations in a short time and improve the outcomes of our patients."

Ellen Burnham, MD, assistant professor of medicine at the University of Colorado in Denver, said that the data point to the challenge in managing blood glucose levels in critically ill patients.

"Perhaps we can't take a 1-size-fits-all approach to managing blood sugar and need to tailor our management to each patient and particular condition," she said in an interview.

The study was independently conducted. Dr. Al-Dorzi and Dr. Burnham have disclosed no relevant financial relationships.

American Thoracic Society 2008 International Conference: Abstract A767. Presented May 20, 2008.

Pearls for Practice
Predictors of glycemic variability in critically ill patients include age (OR, 1.03 for every additional year), diabetes (OR, 3.0), and daily dose of insulin (OR, 1.04).
High glycemic variability compared with low glycemic variability in critically ill patients is associated with higher intensive care unit mortality (OR, 1.12 per millimole) and hospital mortality (OR, 1.09), and glucose range is an independent predictor of nosocomial infections (OR, 1.11 per millimole).

Reviewed by Dr. Ramaz Mitaishvili
Glendale, CA

Color I Color II Color III

Log In or Register