This site complies with the HONcode standard for trustworthy health information:
verify here.

Rewarding for
you and us
Defeat Diabetes Foundation

Defeat Diabetes
Foundation
150 153rd Ave,
Suite 300
Madeira Beach, FL 33708

EASD: Continuous Monitors May Miss Hypoglycemia

Posted: Sunday, October 03, 2010

Continuous blood glucose monitors may not accurately detect low blood sugar, researchers found.

According to Eric Zijlstra, MD, of Profil Institute for Metabolism in Neuss, Germany, stated that, "The devices, which assess blood glucose via interstitial tissue sampling, had a mean deviation of about 18% from samples taken from blood,"

"This accuracy is in a range that has been seen in other studies," Zijlstra said during an oral presentation at the European Association for the Study of Diabetes meeting.

He noted that while many researchers are aware of the questions surrounding these devices, not many studies have focused specifically on their accuracy.

So they conducted a study of a popular brand of continuous glucose monitor -- the Guardian REAL-Time. He said the team only assessed this one type of sensor because it is very popular and has a large market share.

A total of 18 patients, mean age 43, participated in two seven-day in-hospital stays, in which they were connected to the sensors that took a read every five minutes. Blood samples used as reference measurements were taken every four hours.

Blood draws were also done when patients had hypoglycemic symptoms or the monitor alarm went off. Alarms would sound if blood sugar fell below 70 mg/dL. The researchers collected a total of 2,286 data points. They found that the mean relative absolute deviation between blood draws and continuous glucose monitor sampling was 18.3%.

Zijlstra said that deviation was the greatest within the low glucose range, at 26.9%.

He added that frequently calibrating the sensor didn't appear to improve its accuracy, with deviation values one hour after calibration hovering at 18.8%.

The researchers also found that the failure to detect hypoglycemia, as well as false alarms being set off, was quite high. Only 73 of the 145 confirmed hypoglycemic events were also recorded on the monitor, leading to a sensitivity of just 41%.

A total of 104 alarms weren't confirmed by blood reference checks, thus leading to a positive predictive value of just 41%.

"Too many false alarms may decrease patient confidence," Zijlstra said. In the high glucose range, specificity and negative predictive value were both 95%. "So when it does not sound an alarm, that's a good indication that the patient is not having hypoglycemia," he said.

But he emphasized that when the patient "had real hypoglycemia, there was less than a 50% chance of the monitoring system detecting this. And when the monitor sounded an alarm, there was less than a 50% chance that the patient actually had hypoglycemia."

He concluded that continuous glucose monitoring, at least among the brand studied, has suboptimal accuracy, especially in the low glucose range, and sensor calibration doesn't do much to improve that.

"Improvements in continuous glucose monitors are mandatory to further enhance their potential as therapeutic and research tools," he said.

Source: http://www.diabetesincontrol.com/index.php?option=com_content&view=article&id=9872&catid=53&Itemid=8, Heckermann S, et al "Accuracy of a continuous glucose monitoring system (CGMS): Still room for improvement" EASD 2010; Abstract 48.