Glycemic Control w/out Precipitating Life-threatening Hypoglycemia

Despite conflicting evidence regarding mortality and tight glycemic control in the critically ill, the evidence is abundantly clear that the risk of death increases dramatically when blood glucose levels fall below 40 mg/dl, according to R. Harsha Rao, MD, Division Chief, Endocrinology, VA Pittsburgh Healthcare System, speaking today in a session on diabetes mellitus therapies and complications.

“When we started the decade, tight control was worthwhile in spite of the inevitability of hypoglycemia,” Dr. Rao said. “As we ended the decade, tight control was not worthwhile because of the inevitability of hypoglycemia.”

Because no true consensus has been reached on the tight-glycemic-control-hypoglycemia-risk dynamic, Dr. Rao and colleagues took their own unique approach to aggressive insulin therapy using an indigenously developed, fully automated computer-based glycemic expert system for nurse-implemented euglycemia (GENIE), at the VA Pittsburgh Healthcare System.

The logarithms underpinning the system are quite complex (Dr. Rao described it as a “multi-dimensional system” that utilizes a “continuous decision-surface for each independent variable”), but the user-interface only requires four manual inputs — prior glucose level, current glucose level, minutes between prior and current glucose checks, and whether the patient has eaten.

Though there was some push-back from nurses initially (initial adherence was around 30%), “the proof was in the pudding.”

“The problem was that we were changing a culture,” Rao said. “It was very difficult getting them to accept what the program wanted them to do, that when a patient’s glucose level increased to 142 from 135 an hour earlier, the system indicated a bolus of insulin. They were very worried about hypoglycemic events.”

Once the nurses saw the efficacy data, there were no further issues with adoption. The system is now mandated and adherence has increased to 97%. “They all know it and follow it,” Rao said.

It also helps that the system does not increase nurse workload; the average number of finger stick checks increased from 17.6 per day to 18.4. Further, patients are allowed to eat when using GENIE, and the system performs all of the necessary compensatory calculations. However, Rao did caution that there was training required when it came to the minutes-between-glucose-checks parameter. It took some time before nurses were routinely taking the specific time measurements that GENIE required (ie, 121 minutes vs. “about an hour ago”).

Results

Rao and colleagues monitored 609 patients on GENIE between January 1, 2006 and April 20, 2010.

When comparing patients monitored on GENIE with 151 controls one year prior, it was shown that GENIE patients’ mean blood glucose levels were 20 points lower, while time-weighted excess glucose (“calculated as the integrated area under the curve of blood glucose in excess of 140 over time”) was 20 times lower.

Using the GENIE system, there were only three instances of blood glucose levels <40 mg/dL, compared to 6% of controls. These hypoglycemic events, however, were all due to time check errors.

Finally, GENIE had no impact on incidence of surgical site or deep sternal infections, but did eliminate deaths from deep sternal infections.

“With this system, we were aiming for euglycemia in the SICU,” Dr. Rao states. ”We have shown that it can be attained, and it can be maintained.”