Intensive Insulin Therapy in ICU Lacks Real-World Benefit

By Crystal Phend
SEATTLE, Feb. 29 -- Tight glucose control in the ICU does not reduce mortality overall and may put some patient groups at risk, researchers here found. After a protocol targeting blood glucose levels of 80 to 110 mg/dL was instituted across ICUs at one hospital, mortality in the units increased 26%, particularly among surgical and trauma patients and those in for short stays, Steven A. Deem, M.D., of the University of Washington, and colleagues reported online in Critical Care. Hypoglycemia episodes quadrupled over the same period, they said. The findings add to growing skepticism among intensivists about intensive insulin therapy, Dr. Deem said.
Intensive insulin therapy was widely adopted worldwide based on benefits in the initial trials among critically ill surgical and medical patients.
However, more recent trials have suggested no benefit or even harm for cardiac and other critically ill patients, the researchers noted.
To see what the impact would be for a broader population of critically ill patients, they looked at outcomes for all 10,456 trauma, surgical, neurosurgical, and medical ICU patients seen over a four-year period at a single level I trauma center.
Among them, 2,366 were admitted from March 2001 through February 2002 when the ICUs had no specific glycemic control protocol and hyperglycemia was treated with a general blood glucose target of 120 to 180 mg/dL.
Another 3,322 were admitted from March 2002 through June 2003 when the blood glucose target was 80 to 130 mg/dL.
The final group of 4,768 was admitted from July 2003 through February 2005 when an even tighter glycemic control protocol was used -- targeting blood glucose levels of 80 to 110 mg/dL.
Nurse staffing ratios and the proportion of patients admitted for trauma (about 40%) remained unchanged across the study periods.
The changing protocols had a dramatic impact on clinical practice, though, with the proportion of patients receiving insulin infusion rising from 9% to 25% and then to 43% for the three study periods.
Morning blood glucose dropped accordingly, from 144 mg/dL without a protocol to 139 mg/dL with less intensive glycemic control to 129 mg/dL with tight control.
Nevertheless, glucose levels remained about 20 mg/dL above the target range for tight glycemic control, which "reflects the difficulty in application of clinical protocols to real-world practice, outside the rigid confines of randomized controlled trials," the researchers said.
No mortality benefits were seen with changing insulin protocols.
After adjusting for age, history of diabetes, chronic disease scores, admitting ICU, and mechanical ventilation at ICU admission, hospital-wide mortality rates tended to be elevated for patients admitted during the less-intensive insulin therapy protocol (OR 1.11, 95% CI 0.93 to 1.31) or intensive insulin protocol (OR 1.15, 95% CI 0.98 to 1.32) compared with those admitted during the initial study period.
Additionally accounting for blood glucose levels on admission, the trend for hospital mortality remained in favor of the initial study period compared with tight glucose control (OR 1.16 for period three versus period one, 95% CI 0.99 to 1.37).
Adjusted overall ICU mortality was significantly higher with tight glucose control than in the initial study period (OR 1.26, 95% CI 1.04 to 1.53) and in the surgical and trauma ICU subgroups although not in the medical ICU population.
Although bias was a possibility, the mortality trends were the opposite of what might have been expected from confounding, the researchers said, because the ICUs also implemented protocols for diagnosis of ventilator-associated pneumonia, ventilator weaning, and lung protective ventilation around the same time, which were designed to improve patient outcomes.
The reason for increased mortality could have been related to a direct effect of insulin or hypoglycemia, the researchers suggested.
Moderate to severe hypoglycemic events rose three-fold to four-fold with tight glucose control compared with the initial study period, they said. These events were associated with elevated mortality.
Once Dr. Deem's hospital saw the results, intensivists were given the choice of using a 100 to 140 mg/dL target for glucose levels rather than the lower target.
"I use the higher range," Dr. Deem said. "I think most people are choosing to use the higher range, too."
More data is needed to determine what might be the optimal glucose range for ICU patients, he concluded.
"We really don't know enough about insulin and it's administration in critically ill patients to allow us to truly understand its overall effects on outcomes," he said.