Standard metrics generated by continuous glucose monitors (CGM), such as average glucose, time-in-range (TIR), and the glycemic management indicator (GMI), may not align with A1c in people who don’t have diabetes, a new study has found.

“Accuracy is within the acceptable range. The issue is how well the values translate to what clinicians usually look at, which is the A1c,” said lead author Jorge A. Rodriguez, MD, of the Division of General Internal Medicine and Primary Care, Brigham and Women’s Hospital, Boston.

“Based on the analysis that we did in people without diabetes, that connection back to A1c was not as robust as we would expect,” and therefore should not be interpreted to reflect A1c, he told Medscape Medical News.

The study was published earlier this month in Diabetes Technology & Therapeutics.

Increasing CGM Use

The use of CGM has been growing among people without diabetes, including those with prediabetes, particularly with the availability of the over-the-counter (OTC) devices Dexcom’s Stelo and Abbott’s Lingo. The accuracy of the OTC devices in reflecting glucose levels is similar to that of the products approved for use in people with diabetes.

However, while studies have looked at the correlation between CGM metrics and A1c in patients with type 1 diabetes (T1D) and type 2 diabetes (T2D), “fewer data exist on prediabetes or normoglycemia,” the study authors noted.

A Diminishing Association

Rodriguez and colleagues used data from 972 individuals who participated in the Artificial Intelligence Ready and Exploratory Atlas for Diabetes Insights cohort, including 43.3% with T2D, 32.8% with prediabetes, and 23.9% without diabetes (normoglycemia). All had A1c measurements taken before wearing the Dexcom G6 for up to 10 days. Data were only considered for those with at least 7 days of wear.

Overall, the association between CGM metrics and A1c was significantly stronger in the group with T2D compared with either the prediabetes or normoglycemia group (P for interaction <.001).

For those with T2D, mean glucose was the metric most strongly correlated with A1c (standardized beta-coefficient or the change in A1c percentage points per one SD increase in each CGM metric = 0.79; P < .001). Each 1-mg/dL increase in mean glucose was associated with a 0.02 percentage point increase in A1c.

Time-based CGM metrics were also closely associated with A1c in the T2D group, including TIR 70-180 mg/dL (standardized beta = -0.73; P < .001) and time in tight range (TITR) 70-140 mg/dL (-0.77; P < .001). Glycemic variability also showed moderate to strong associations.

In the group with prediabetes, the associations between CGM metrics and A1c maintained significance but much less strongly than for T2D. The relationship with mean glucose was reduced so that each 1-mg/dL increase was associated with only a 0.007 percentage point A1c increase (standardized β = 0.22; P < .001). The relationships for TIR and TITR were similarly attenuated (-0.30 and -0.16, respectively; P < .001 for both). Glycemic variability metrics were also attenuated.

However, among those without diabetes, the CGM metrics showed minimal or nonsignificant associations with A1c. The relationship with mean glucose remained significant, although weak (0.10; P = .026), with each 1-mg/dL increase associated with only a 0.003 percentage point increase in A1c. Time-based metrics showed inconsistent relationships with A1c.

Rodriguez pointed to a recent study, led by his colleague Nicole L Spartano, PhD, who was also senior author of the current paper, finding that even among CGM experts, there was little agreement in interpreting potentially challenging CGM reports from people without diabetes.

“For people without diabetes, there’s still work to be done around what the specific glucose thresholds should be and what clinicians should do about it,” said Rodriguez. “We’re still figuring that out.”

An Evolving Landscape

Given an opportunity to respond, Endocrinologist Jessica Castle, vice president of Global Medical Affairs at Dexcom, pointed out that the study was relatively small and unblinded, and that CGM values are not designed to diagnose diabetes or prediabetes.

“The use of CGM in people without diabetes and prediabetes is relatively new and is an area where evidence and consensus guidelines need to continue to develop to best define how to use those data,” she told Medscape Medical News.

“We’ve established that CGM is a powerful tool for behavior change,” said Castle. “How best to have something like an ambulatory glucose profile that is designed for people with prediabetes or without diabetes is not yet available. It’s an evolving landscape.”

Rodriguez reported receiving grants from the National Institutes of Health. Castle reported being an employee of Dexcom.

Miriam E. Tucker is a freelance journalist based in the Washington, DC, area. She is a regular contributor to Medscape, with other work appearing in the Washington Post, NPR’s Shots blog, and Diatribe. She is on X @MiriamETucker and BlueSky @miriametucker.bsky.social.