A Remarkably Inaccurate Comparison of Glucose Management Technologies

I read with great interest the recent article by Drs Salinas and Mendez comparing the results of four electronic glucose management systems used to calculate IV insulin doses for critical care patients. ¹ I felt compelled to respond to multiple inconsistencies and inaccuracies within the manuscript. Notwithstanding any interests we each have in our respective roles with Glytec and Monarch Medical Technologies (EndoTool), we all have an obligation to accurately represent the data and facts.

The authors reported Glucommander (GM) hypoglycemia rates as “percent of patients” and EndoTool (ET) hypoglycemia rates as “percent of blood glucose results,” leading readers to believe ET rates were as much as 13 000% lower than GM. Given hypoglycemia is typically reported as a percent of patient stays, patient days or blood glucose results, comparisons must only be made utilizing equivalent units of measure. The rationale for this discrepancy is curious, given the two GM articles referenced by the authors reported hypoglycemia results as both a percent of patient days as well as a percent of blood glucose results.^2,3

Despite dozens of studies reporting GM’s performance over the 13 years since Glytec has been in business, the authors selected a 77-patient arm from research that predates Glytec and uses an obsolete GM algorithm never associated with Glytec. It would have been more appropriate for the authors to report recent data from the FDA-cleared product actually developed and sold by Glytec. One such study, published in JDST January 2018 by clinicians at Emory University and Grady Hospital, reported hypoglycemia rates (using GM) of 0.01% and 1.27% for BGs <40 mg/dl and <70 mg/dl, respectively. ⁴

The authors also compared vastly different populations of patients with varying glucose targets. The GM study was conducted in a medical ICU with a target glucose range of 80-120 mg/dL. Neither Glytec nor any industry guideline recommends this target as it is associated with high rates of hypoglycemia. Target ranges associated with the studies of Glucostabilizer (2398 ICU patients) and ET (18 000+ patients from various ICUs) were not disclosed. Experience demonstrates it is ineffectual and inaccurate to compare glycemic results among different patient populations with different targets.

Notably, a poster presented by Valerie Garrett et al at the 2016 American Diabetes Association Scientific Sessions, titled “A Comparison of Glycemic Outcomes for Two Computerized Insulin Infusion Algorithms in CV Surgery Patients,” concluded that GM was more effective and safer than ET. Hypoglycemia % BGs <70 mg/dl were 0.7% with GM compared to 1.5% with ET (P < .00001). There was no severe hypoglycemia (BGs <40 mg/dL) with GM compared to 0.04% with ET, and hyperglycemia % BGs >180 mg/dL were significantly lower with GM than with ET, at 3.9% vs 6.0%, respectively (P < .00001).

Finally, the authors’ description of how Glytec’s GM product calculates insulin doses is not valid. The 2005 reference ⁵ reflects on an obsolete product never associated with Glytec. The GM product developed and delivered by Glytec over the past 13 years has four FDA clearances and is vastly more sophisticated.