Julio Leey, MD, of the Malcom Randall VAMC Center and the University of Florida, Gainesville, FL
GAINESVILLE, FL — Nearly 1 in 4 veterans receiving care through the VA lives with diabetes—more than twice the prevalence in the general U.S. population—putting them at risk for cardiovascular complications such as stroke, peripheral artery disease and amputations as well as chronic kidney disease, cognitive impairment, diabetic retinopathy and glaucoma, among other conditions.
Veterans face heightened risk of developing Type 2 diabetes due to service-related chronic pain and joint damage, according to the VA. Both can limit movement, increasing the risk of obesity, a major risk factor for Type 2 diabetes (T2D). In addition, diabetes is a presumed service-related condition for veterans exposed to herbicides such as Agent Orange during their time in uniform, as well as for those who develop the disease in the year after separation. It is also a presumptive condition for active duty servicemembers.
Randomized clinical trials have consistently demonstrated that continuous glucose monitoring (CGM) improves glycemic control, reduces hypoglycemia and enhances quality of life for patients on intensive insulin regimens. Over the past two decades, CGM technology has advanced substantially in accuracy, ease of use and integration with insulin pumps.
Unlike finger-stick glucometers, CGMs measure interstitial glucose via a small sensor inserted just beneath the skin. The device records glucose values every few minutes and transmits data wirelessly to a smartphone, tablet, computer or other device. The result is a continuous stream of information that captures fluctuations and trends that conventional testing cannot.
For veterans, this capability is especially valuable. VA’s Share My Health Data app allows CGM users to automatically upload readings directly into the electronic health record (EHR). If a veteran opts in, the data are shared with their VA care team in real time.
“It’s ultimately about giving veterans the power to share their health information so that their VA care team can not only look at the data but then use the data to target their care and provide really on-point interventions for them,” explained Rebecca Bixler, program analyst for VA’s Patient-Generated Health Data (PGHD) team.
Recent work has demonstrated the feasibility and impact of such integration. In an analysis of more than 100,000 CGM and EHR records within the VA healthcare system, Tomoki Okuno and colleagues at the Phoenix VA Health Care System and VA Boston Health Care confirmed that CGM use remains high and consistent over time across age, race and diabetes type.¹ These findings suggest broad acceptance of CGM among veterans and the potential for longitudinal monitoring that can inform personalized treatment decisions.
Understanding GMI and Its Limitations
CGMs generate an enormous volume of data, but translating that data into a tool that can guide treatment more specifically than “that’s too high” poses some challenges, particularly because clinicians and patients are accustomed to interpreting blood glucose through the lens of the hemoglobin A1c blood test (HbA1c, also abbreviated as A1c).
A1c reflects the nonenzymatic glycation of hemoglobin over approximately three months and remains the gold-standard biomarker associated with microvascular and macrovascular complications of diabetes. An indirect measure, it is affected not only by average glucose exposure but also by nonglycemic factors such as red-blood-cell turnover, hemoglobin variants, stress, acute illness and other comorbidities. These influences can weaken the correlation between A1c and true mean glucose levels, particularly in older or medically complex populations such as veterans.
One widely used summary metric is the glucose management indicator (GMI), which estimates the A1c value from mean glucose levels recorded by a CGM. GMI was designed to help clinicians and patients translate continuous data into the familiar A1c framework used for decades to assess long-term glycemic control, but discrepancies between GMI and laboratory A1c values are common and sometimes clinically meaningful. Differences greater than 0.5% are considered significant and have been observed in 26% to 68% of published studies.
GMI–A1c Discrepancies
To better understand these inconsistencies in veterans, researchers Thao Nguyen, MD, and Julio Leey, MD, of the Malcom Randall VAMC Center and the University of Florida, both in Gainesville, FL., conducted a retrospective study of 110 veterans who used insulin pumps and either the Medtronic Guardian 3 or Guardian 4 CGMs in 2023–2024.² The results were presented in a poster session at the 85th American Diabetes Association Scientific Sessions in Chicago and simultaneously published in Diabetes.
Among 36 veterans with paired A1c and 90-day GMI data, yielding 122 lab-sensor pairs. Across the cohort (mean age 64 years; 97% male), CGM wear rates ranged from 94% at 14 days to 85% at 90 days. Ten participants had T2D, and 26 had Type 1 diabetes (T1D).
Among all pairs, discrepancies between A1c and GMI were negatively correlated with CGM wear percentage at the 90-day interval (R² = 0.0153, p = 0.0034). In veterans with T2D, the correlation was particularly pronounced (R² = 0.466, p = 0.0102), while it did not reach significance in those with T1D.
Leey told U.S. Medicine that the association was “possible because more hyperglycemia occurs when patients are off their CGM.” Many studies have demonstrated that the rapid feedback from CGMs leads to better glucose control.
Multiple linear regression confirmed that 90-day CGM wear percentage was the only statistically significant predictor of A1c–GMI discrepancy (p = 0.032).
Both the high rate of CGM use and the lack of demonstrated correlation in between use and A1c-GMI discrepancy in veterans with T1D could be related to selectivity in pump use, Leey said. “As a retrospective study, there could be some unintended biases. Those who were able to count carbs were prescribed a pump, whereas those who could not (education level, motivation, understanding of disease process, etc.) were not offered one. This could be a marker for compliance or ‘simply being a good patient.’”
Neither anemia nor renal function was linked to variation, likely reflecting the generally normal laboratory values in the study population. Kidney function was largely preserved (45% stage 1, 45% stage 2, 10% stage 3 chronic kidney disease), and anemia was present in 14%.
“This is also potentially a selection bias,” Leey said. “We only prescribe pumps to those with mild-moderate renal failure, thus the correlation or lack thereof could not be captured.” In addition, he noted that. While there was no clear profile for veterans who had high CGM use and significant discrepancies between A1c and GMI, “at first glance, it seems more prominent in those with more compromised renal function.”
The investigators concluded that consistent CGM wear enhances the reliability of GMI as a proxy for HbA1c. They recommended further research to clarify how CGM-derived metrics should be interpreted in routine practice and how best to reconcile discordant results.
These findings have important implications for VA clinicians managing diabetes in complex patients. GMI can be a helpful adjunct for understanding trends between laboratory visits, but it should not replace HbA1c as the standard for assessing long-term glycemic control—particularly when discrepancies arise. Providers also should consider nonglycemic influences on HbA1c and the potential impact of inconsistent CGM wear or data gaps on GMI accuracy.
- Okuno T, Macwan SA, Miller D, Norman GJ, Reaven P, Zhou JJ. Assessing Patterns of Continuous Glucose Monitoring Use and Metrics of Glycemic Control in Type 1 Diabetes and Type 2 Diabetes Patients in the Veterans Health Care System: Integrating Continuous Glucose Monitoring Device Data with Electronic Health Records Data. Diabetes Technol Ther. 2024 Nov;26(11):806-813. doi: 10.1089/dia.2024.0083. Epub 2024 Jun 11. PMID: 38768417.
- Nguyen T, Leey J. 982-P: Factors Associated with Discrepancy Between A1c and GMI in Veteran Population on Insulin Pump. Diabetes. 2025;74 (Supplement_1):982-P.
