PALM BEACH, FLA. – A commercially available computer program designed to calculate optimal insulin dosages for postsurgical ICU patients halved the rate of hyperglycemic episodes that patients experienced while dropping the rate of hypoglycemic events by 95% in a single-center, U.S. experience with a total of more than 2,000 patients.

The computer program also resulted in a dramatic reduction in the number of blood glucose measures the nursing staff performed, cutting testing down from 7,495 blood glucose measures/month before the system began to 4,072 blood measures/month once it was in place, Dr. Christopher C. Baker, FACS, said at the annual meeting of the Southern Surgical Association.

An additional associated benefit was a substantial reduction in hospital-acquired infections in the postsurgery patients once computer-guided insulin dosing came online, but this was only a temporal association that may have also been driven by other improvements in patient management that happened at about the same time, said Dr. Baker, chairman of surgery at the Carilion Clinic in Roanoke, Va.

"People have shied away from glucose controls systems [like the one introduced at the Carilion Clinic] because the results from the NICE-SUGAR [Normoglycemia in Intensive Care Evaluation–Survival Using Glucose Algorithm Regulation] trial (N. Engl. J. Med. 2009;360:1283-97) showed that hypoglycemia episodes increase when you do this," commented Dr. David Herndon, FACS, professor of surgery at the University of Texas Medical branch in Galveston. "What is important in the Carilion results is that the computerized support system reduced hypoglycemia. If that can be done across the board, they might have a winner. We need to better control sugar in the ICU."

The next step is to prove the efficacy of this computer program in a prospective, randomized controlled trial, said Dr. Herndon, a suggestion that Dr. Baker also endorsed.

"The reduced incidence of hypoglycemia was incredibly impressive," commented Dr. Michael Rotondo, FACS, professor and chairman of surgery at East Carolina University in Greenville, N.C., who also called for results from a prospective, controlled study.

Surgeons at Carilion introduced the computerized support program for insulin dosing in early 2010 into five surgical intensive care departments: surgical ICU, trauma ICU, neurotrauma ICU, cardiac surgery ICU, and the cardiac surgery progressive care unit. Dr. Baker and his associates compared data collected on patients from all five units with the computerized system in place during July 2010-December 2011, a total of 1,682 patients treated using the computer program, with 449 patients treated in the five units during July 2009-December 2009, before use of the computer program started.

When using the program, nurses take an initial blood specimen from a patient newly arrived at the unit and enter the blood glucose level into the program along with the patient’s weight and blood creatinine level. The program then immediately calculates the appropriate bolus insulin dose, the insulin infusion rate, and the time to the next blood draw, said Dr. Sandy L. Fogel, FACS, a surgeon at Carilion who collaborated on the study. At first, the next blood draw is specified for about 15-30 minutes following the first, but subsequently the blood draws are directed to occur at longer and longer intervals as the patient’s blood glucose comes under control, within the target range of 70-150 mg/dL. Eventually, draws occur at 6-8 hour intervals, Dr. Fogel said.

During the 6-month historical control, nurses drew 44,972 blood specimens for glucose measurement, an average of 7,495/month, compared with 73,290 blood draws during the 18-month period with the program in place, an average of 4,072 blood specimens drawn/month.

During the 18 months with the program in use, hyperglycemic episodes, defined as a blood glucose level greater than 150 mg/dL, dropped by 45%-57% across the five units using the system compared with each unit’s historical control. Overall, hyperglycemic events fell by 50%, a statistically significant difference.

The incidence of hypoglycemic episodes, defined as a blood glucose level below 40 mg/dL, fell from a 1% rate during the historical control period, to rates that ranged from zero to 0.12% with the program in place, with an overall rate across all five units of 0.05%, a 95% relative decrease that was statistically significant.

The substantial decline in hypoglycemic episodes "was a surprise for us," Dr. Fogel said. In addition, out of all the blood measures performed using the computer program 17% had glucose levels of 60-69 mg/dL, and 3% were below 60 mg/dL. In short, the computer program "did not increase hypoglycemic episodes by any way you measure it," Dr. Fogel said.

The insulin-dosing program used at Carilion is the EndoTool, marketed by Hospira, Inc.

Dr. Baker, Dr. Herndon, Dr. Rotondo, and Dr. Fogel had no disclosures.

m.zoler@elsevier.com

On Twitter @mitchelzoler

Computer program refines postsurgical insulin treatment