The Discordance Between the Glucose Management Indicator and HbA1C

Evaluating the discordance between laboratory and CGM glucose management indicator estimated HbA1c.

Many experts find the estimated HbA1c (eA1C) in the clinical management of diabetes useful. Estimated HbA1c, referred to as the glucose management indicator (GMI), is the conversion of mean glucose from continuous glucose monitoring (CGM) into an estimate of laboratory HbA1c, using a population-based formula. Ideally, mean glucose is derived from at least 14 days of CGM data. The GMI may be higher than, lower than, or similar to the laboratory A1C. To date, there are not any extensive “real world” studies looking at laboratory and GMI HbA1c discordance outside of a clinical trial or organized database. A retrospective review evaluated discordance between laboratory and CGM estimated HbA1c.

The review was conducted on patients with diabetes who use CGM. The study included 641 office visits and subjects with a mean age of 45.6 ± 15.3 years. 52% of the patients were female and 93% Caucasian. 91% had type 1 diabetes, and the majority had diabetes for greater than 20 years. These patients had been on prescribed insulin for at least three months and were required to have a documented HbA1c value no later than 30 days before or after the associated office visit and CGM upload. There were no exclusion criteria. From 2012 to 2019, the patients were seen at the University of Washington (UW) Diabetes Care Center. This study included 641 separate office encounters. The patients were required to use CGM, HbA1c, and sensor data obtained fewer than four weeks apart. Dexcom G5 was the most common CGM device used. The mean duration of CGM wear was 24.5 – 8 days. In addition, 16% of the patients received noninsulin diabetes medications.

The mean HbA1c among encounters was 7.3% ± 1.0 %, and the average mean CGM glucose was 162 ± 29 mg/dL. HbA1c-GMI discordance was calculated for each subject and assessed for any impact of comorbidities. HbA1c-GMI discordance was defined as the absolute difference between laboratory and eA1C. The analyses show in the Bland-Altman plot that GMI was higher on average than HbA1c when the average value was lower, and GMI was lower than HbA1c when the average value was higher, showing a negative correlation. The population of 641 individuals showed surprising levels of discordance between GMI and HbA1c. Only 11% of encounters showed discordance of <0.1%; 50% and 22% had discordances of >0.5% and >1%, respectively. It was also evident that patients with various comorbidities affected the accuracy of HbA1c. The only comorbidity to have a statistically significant effect on discordance was CKD. Overall, the levels were considerably higher than what is reported in previous studies.

This study showed substantial discordance between GMI laboratory and eA1C in real-world patient encounters, which may cause detrimental errors in diabetes management. Clinicians must be aware that HbA1c may not as accurately reflect mean glucose as previously vouched to do. Acknowledging this discordance will allow clinicians to provide the best possible patient care regarding diabetes management.

Practice Pearls:

• The relationship between HbA1c and mean glucose needs to be understood for each patient individually.
• There is substantial discordance between GMI laboratory and eA1C in a real-world setting.
• It is essential to consider CGM metrics as an alternative assessment of glycemic levels.

Perlman, Jordan E et al. HbA1c and Glucose Management Indicator Discordance: A Real-World Analysis. Diabetes technology & therapeutics; April 2021.

Author: Brenda Oppong, PharmD Candidate, LECOM School of Pharmacy