Should We Bury HbA1c?

Widely available, easy to measure and interpret, and relatively inexpensive, providers continue to depend on HbA1c values to guide diabetes care. Glycated hemoglobin (Hb) was first detected more than 50 years ago. ¹ The path toward widespread use of glycated Hb for diabetes management began with Koenig and Cerami's case series correlating HbA1c and mean glucose in 1976. ² It continued in the 1990s when the Diabetes Control and Complications Trial (DCCT), followed by the United Kingdom Prospective Diabetes Study (UKPDS), established the epidemiologic relationship between HbA1c and prevention of microvascular complications in type 1 and type 2 diabetes, respectively. ³ However, glycated Hb is not without well-known limitations. Altered red blood cell turnover, hemoglobinopathies, chronic kidney disease, chronic liver disease, iron deficiency, race, and ethnicity can all impact glycated Hb performance (see Table 1 for a comprehensive list). Discordance between HbA1c and mean blood glucose can also occur in the absence of known risk factors. ⁴ The ADAG study confirmed this reality on a large scale in its comparison of glycated Hb and mean glucose. ⁵

There is less variation in HbA1c within than between individuals, making it an effective tool to monitor individual glycemic control, despite its deficiencies. ⁴ Even so, clinically significant discordance between A1C and glycemic management indicator (GMI) has been shown in 28%–50% of patients. ^6,7 There remains potential for over- or undertreatment of patients when glycated Hb values are used without understanding glycemia. A common example of this is an individual with advanced renal disease where we know HbA1c often reads low. If the only data available is a HbA1c of 7%, a provider may not change treatment. Adding data from a continuous glucose monitor (CGM) would reveal hyperglycemia, with an elevated GMI of 8.5%, thereby prompting treatment intensification. ⁶ Conversely, in those for whom HbA1c values are higher than mean glucose, also referred to as positive discordance, there is a risk of overtreatment leading to hypoglycemia. This is more than a theoretical risk. Data from Hempe et al.'s ⁸ subgroup analysis of the ACCORD trial showed higher risk of hypoglycemia for those with increasing positive discordance, as quantified by hemoglobin glycation index (HGI).

Analysis of CGM data has magnified this issue of HbA1c misinformation. Some would interpret this as a trip to the graveyard, without eulogies, for glycated Hb. Others have argued for the continued relevance of HbA1c due to extensive epidemiologic evidence for its association with diabetes complications. ⁹ Randomized controlled trial evidence of this type is not and may never be available for CGM due to ethical concerns. Regardless of these limitations, several recent, large, prospective studies have established associations between time in range (TIR) from CGM and diabetic microvascular complications. ^{10 –13} Furthermore, retrospective analysis of the DCCT data has confirmed the associations between TIR and both diabetic retinopathy and microalbuminuria. ¹⁴ In light of this data, risk for microvascular complications can be inferred from CGM glucometrics. For many patients using CGM, quarterly HbA1c measurements are no longer medically needed, but they are often required for payors or a health care system's “quality metrics.”

For clinical care, this is great news for those using CGM: one less blood sample; however, most patients with diabetes in the United States are not using CGM. In 2021, there were an estimated 37 million people in the United States living with diabetes and an estimated 2.4 million using CGM. ¹⁵ Health care disparities in technology use in the United States are pervasive and well documented. ^{16 –18} Patients with type 1 diabetes using government-funded insurance are two to five times less likely to use CGM than those with private insurance. ^16,17 Even with continued expansion of CGM coverage, in 2023, five states had no published Medicaid coverage guidelines and three restricted coverage to the pediatric population. ¹⁸ Racial and ethnic disparities in access to diabetes technology in type 1 and type 2 diabetes are all too common. In a recent analysis of data from a safety-net hospital, white patients with type 1 diabetes were more than twice as likely (55%) to use CGM as Black (21%) and Hispanic (28%) patients. ⁶⁶ These trends are similar in pediatric populations. ⁶⁷

For too many primary care providers, HbA1c is the metric of glycemic control out of necessity. Interpretation and use of glucose meters can be limited by many barriers, including the requirement for physical download of glycemic data. Point of care HbA1c, while imperfect, provides invaluable and at times, the only data on glycemic control. Too often, no infrastructure exists to support CGM or glucose meter downloads. A recent survey of attitudes toward CGM use echoes these experiences. Primary care providers reported insurance coverage and training as barriers to prescribing for patients. ⁶⁸ Success with CGM implementation requires infrastructure for authorizations, patient education, and data review. This is an all-too-common scenario in medicine: over-burdened physicians struggle to implement guideline-based care and are effectively underserved by their own systems.

Policy change, especially at the state level, is needed to broaden access to technology, but it is not the only avenue for change. Two recent studies improved equity in CGM prescribing in type 1 diabetes. ^69,70 Schmitt et al. used a quality improvement approach to increase CGM access for children. Overall, CGM prescriptions increased from 34% to 85% with a significant decrease in disparities in both Black patients and those on Medicaid. ⁶⁹ Mathias et al. described a multifaceted practice transformation that improved CGM prescribing equally in Black, Hispanic, and White adults. ⁷⁰ Interventions like these offer hope for practice level improvement in health care equity.

There are other more academic reasons it may be too soon to bury HbA1c. At the 2023 ADA meeting, our group presented data correlating HbA1c discordance with risk of diabetes complications. ⁷¹ This pilot study quantified HbA1c discordance using a ratio of GMI to HbA1c. Positive discordance was associated with the complications of diabetic retinopathy and diabetic kidney disease. Recently, two groups reported similar analyses of the HGI (derived from GMI) and A1C. It is quite possible that knowing both GMI and HbA1c may improve prediction of risk for microvascular complications. ^72,73

CGM has revolutionized diabetes care and improved the lives of millions of patients living with diabetes. For those with CGM, these metrics have become and should continue to be our major metric for treatment decisions. However, it is clearly not time to bury A1C. This extensively validated test has improved diabetes care for millions of patients. Pervasive inequities in our health care system will continue to limit access to CGM. Equitable care will remain out of reach for many without policy change. For the foreseeable future, HbA1c will remain the backbone of glycemic control for many with diabetes, and when paired with CGM metrics may offer additional precision in prediction of diabetes complications.