American Association for Clinical Chemistry
Better health through laboratory medicine
October 2008 Clinical Laboratory News: The New Glycohemoglobin Standard

CLN Banner Logo

October 2008: Volume 34, Number 10

The New Glycohemoglobin Standard
Will Estimated Average Glucose Boost Patient Understanding?

By John R. Bell


Despite the widespread use of HbA1c as a clinical measurement of patients’ diabetes control, standardization of test results has been a challenge for labs. The National Glycohemoglobin Standardization Program took the first step 12 years ago when it initiated a program to standardize HbA1c measurements on the basis of the Diabetes Control and Complications Trial, results of which were published in 1993. The DCCT method showed stability over many years but did not use pure HbA1c for calibration and was therefore not a true reference method. To fill that need, the International Federation of Clinical Chemistry and Laboratory Medicine formed a collaboration to develop a true reference method. By all accounts, they succeeded, but the transition to the new reference method, accompanied by a requirement to report the glycated hemoglobin results in mmol per mol, presented a new set of challenges that stirred a controversy among the various stakeholders, including the American Diabetes Association and the European Association for the Study of Diabetes. The ADA, which maintains that percentage glycated hemoglobin can be confusing, had long advocated for a simpler way to explain glucose goals to patients.

Now, the ADA says it has the answer. Recently published results of a major international study— the A1c-Derived Average Glucose (ADAG) study—confirm that a linear relationship between estimated average daily glucose and HbA1c exists (Diabetes Care 2008;31:1473–8). Funded by ADA and EASD, the study used continuous glucose monitoring and self-glucose monitoring in 507 patients to determine that HbA1c and estimated average glucose (eAG) are reliably correlated and that a mathematical relationship can be used to report eAG from HbA1c.

“The findings of this large study have confirmed what smaller studies have shown and will give us confidence that HbA1c really does represent an average glucose, because we now have a reliable formula to convert HbA1c into average glucose,” said David M. Nathan, MD, professor of medicine at Harvard Medical School and co-chair of the ADAG study, in an ADA press release. “While eAG will not replace HbA1c, physicians will be able to obtain reports both in HbA1c units of glycated hemoglobin and eAG units of milligrams per deciliter or millimoles per liter, depending on the country, and choose which to use in clinical situations.”

Based on this finding, ADA now recommends that laboratorians provide physicians with estimated average glucose (eAG), in addition to HbA1c. Labs adopting the new HbA1c reporting method will need to work closely with clinicians advised David Sacks, MD, of Brigham and Women’s Hospital and Harvard University Medical School, Boston, who chairs the NGSP Steering Committee and is a member of the IFCC HbA1c working group. “The key advice that I would put forward to laboratorians is to have a dialogue and communicate with your clinicians, to decide how the clinicians want the information reported, so that they can optimally communicate with their patients and improve patient care.”

The Genesis of Estimated Average Glucose

To understand the ADA’s recommendation, it’s helpful to look at some of the history of reference methods and standards for glycated hemoglobin. Those who work closely with the analytical method for measuring HbA1c were well aware that the original DCCT reference method was not specific because other hemoglobin species are included in the HbA1c measurement. While this is not problematic for routine use, some believe that patients’ results should represent their true HbA1c value. In addition, the NGSP standardization protocol was not used in Japan or Sweden, a concern for global harmonization of lab values. The reference standard developed by the IFCC working group (CLN November 2007) does not have this problem, but according to an editorial by Richard Kahn, PhD, and Vivian Fonseca, MD, of the ADA that accompanied the final ADAG report, “The IFCC method is very complicated, requires costly equipment (a mass spectrometer), and is very expensive” (Diabetes Care 2008;31:1704–7).

Most importantly, wrote Kahn and Fonseca, the results produced by instruments calibrated with the IFCC reference standard are 2% lower than those calibrated with the DCCT standard. And there was a fear that reporting these different HbA1c results could cause confusion for both clinicians and patients.

To mitigate those concerns, the working group developed a master equation for use in translating the new reference results back to the original NGSP results, expressed in percentages. But the 2% discrepancy still caused many to believe their reporting the IFCC values could cause confusion for physicians and perhaps patients. According to Randie Little, PhD, coordinator of the NGSP and a member of the IFCC working group, physicians and patients who would use HbA1c values would not be able to easily dinstinguish which HbA1c was being reported—NGSP/DCCT or IFCC. “There were concerns that switching to IFCC numbers would cause complete chaos,” Little said.

Many clinicians already describe HbA1c to patients as an estimated glucose level, and it seemed the opportune time to make the switch. But the big unknown was whether HbA1c correlates well enough to mean glucose levels to derive a standard conversion formula, similar to that now in use for estimated glomerular filtration rate (eGFR), which is based on creatinine levels. To settle the controversy, ADA and EASD funded the ADAG trial, led by Nathan and Robert Heine, MD, of the VU Medical Center, Amsterdam.

Converting HbA1c to
Estimated Average Glucose

Based on the results of the A1c-Derived Daily Glucose (ADAG) study, ADA is recommending the use of a new term in diabetes management, estimated average glucose, or eAG. This new method is intended to help healthcare providers report HbA1c results to patients using the same units (mg/dl or mmol/l) that patients see routinely in blood glucose measurements.

There are two formulas for deriving eAG from percentage HbA1c or the reverse, using either metric or SI units:
eAG (mg/dL) = 28.7 x HbA1c – 46.7
eAG (mmol/L) = 1.59 x HbA1c – 2.59

On its website, ADA has also provided the following equivalents for HbA1c percentages:

HbA1c (%)
eAG (mg/dL)
eAG (mmol/L)

A previous formula exists in the literature, noted ADA’s Matt Petersen. The ADA intends for the new eAG formula to replace the older conversion method, which he said produced higher estimates of average glucose. However, the organization does not intend for eAG to replace HbA1c at any point. “We do not foresee ever saying ‘Don’t use A1c, but we believe you can have an easier, more productive conversation with your patients if you use eAG,” he said.

Additional resources are also available on ADA’s website.

Source: ADA website

ADAG Findings

The ADAG study group compared HbA1c with mean glucose in patients with either type of diabetes at 10 centers in the US, Europe, and Africa (Cameroon). Diabetic participants were required to have stable glycemic control, and those with any potentially confounding conditions or treatments were excluded. Nondiabetic subjects were also included for comparison. To measure glycemia, the researchers used continuous interstitial glucose monitoring, as well as self-monitoring of blood glucose and seven-point fingerstick capillary glucose monitoring. HbA1c was measured at baseline and monthly for 3 months. The final report included 507 patients with type 1 diabetes (268 patients) or type 2 diabetes (159 patients), as well as 80 participants without diabetes. Average glucose was derived by multiplying the continuous interstitial glucose level by 1.05. The study was observational, and there was no blinding of participants or clinicians.

The a priori criterion that eAG had to meet to establish a linear relationship with HbA1c was for 90% of the estimates to be within 15% of the regression line. The study found that 89.95% of patients did so. The investigators also conducted regression equations to examine any effect by diabetes type, sex, age, ethnicity, or smoking status; in each case, the difference did not reach statistical significance.

HbA1c: Confusing for Patients?

At a symposium titled “Use of the A1c-Derived Average Glucose in Patient Care” during the AACC Annual Meeting in July, Sue Kirkman, MD, vice president of clinical affairs at the ADA, explained why the new way to report glucose will benefit patients. She attributes the problem to the difference in the units used to report blood glucose and HbA1c. “It’s confusing for everyone, and people have a hard time remembering it,” she said. She cited a study in which 75% of patients with an HbA1c test in the last 6 months could not report their HbA1c within 1%. In addition, “the concept of percentages is difficult for patients to understand,” Kirkman noted.

Nathan said that despite efforts to educate patients as to the meaning and significance of their HbA1c levels, it’s something that clinicians routinely struggle with. “There continues to be a disconnect between the information supplied by their clinicians and the long-term metabolic goals that have been established, and the patients’ day-to-day management of their diabetes,” he explained. “By providing an estimated average glucose value, we hoped that some of this disconnect would be fixed. Specifically, the patients would be provided their level of chronic glucose control in the same units as the day-to-day measurements that they perform.”

Another researcher believes that some patients don’t understand decimals, and therefore find percentage HbA1c too complicated. Philip Home, MD, of the Newcastle (U.K.) Diabetes Centre and Newcastle University, wrote in Diabetic Medicine, “The concept of the decimal place is not well understood by a proportion of people with diabetes.” Compounding this problem, he noted, is that HbA1c is a relatively small number, and one study from Sweden reported that small numbers were associated with less patient control of their blood glucose (Diabetic Medicine 2008;25:995–8). However, that study was in children, its design was flawed, and its results have not been replicated, said Garry John, PhD, Clinical Director of Laboratory Medicine Norfolk and Norwich (U.K.) University Hospital, a member of the NGSP Steering Committee and Chair of the IFCC working group.

The idea that patients find the different units used by HbA1c and their daily glucose measurements confusing is at the heart of the ADA’s push for eAG. “For newly diagnosed patients and those who haven’t really been paying attention, skipping that whole conversation about glycated hemoglobin and just talking about eAG makes sense. It’s one set of numbers to learn,” explained Matt Petersen, director of information resources for the ADA.

Alternative Viewpoints

But not everyone agrees with the conclusions of the ADAG trial. One criticism was raised by Eric S. Kilpatrick of the Hull (U.K.) Royal Infirmary. In an editorial, he said that the ADAG authors’ a priori criterion—90% of the estimates being within 15% of the regression line—allowed too great a degree of variance in HbA1c. This means a targeted eAG could actually be higher than the daily average glucose that many patients are shown on their own glucose monitors, he noted, and that “approximately 12,000 U.K. diabetic patients could have mean glucoses at least 60% higher than 12,000 others, and yet all potentially have the same eAG.” In addition, a patient’s own glucose meter may be calibrated to whole blood rather than plasma, Kilpatrick noted, and so would report a level 10%–15% lower than that shown by eAG. Likewise, patients who measure before meals may also be used to seeing an average glucose much lower than what their eAG will report, he wrote.

Nathan says that the 90% criterion was chosen “on the basis of our need to take into account the imprecision and inaccuracy of the methods we were using”—SMBG, continuous glucose monitoring, and the HbA1c assay itself. “This was kind of a calibration study,” he explained. “We knew that, for example, we weren’t going to get a complete capture of chronic glucose, because we couldn’t practically do that in this kind of international study. We knew that there were going to be holes in the data we collected. We knew that the methods we were using were not going to be the same that a lab would have. We picked a number that we thought was reasonable, given the methods we were using, and one we thought most clinicians would actually accept.”

Nathan acknowledges that patients may notice a difference between the average glucose value they see on their own glucose monitors and laboratory-calculated, HbA1c-derived eAG. “Invariably, the reason is that they’re missing a glucose period; for example, they might be testing three times per day, but they never check at bedtime.” Given this, clinicians will likely need to ensure that their patients are not omitting a certain time of day when checking their glucose. He noted that in his own clinic, anomalies in SMBG-derived average glucose are almost always explained by such “sampling errors.”

In their publication, the investigators for the ADAG study group noted that ethnic minorities were under-represented in the study population, given their increased incidence of diabetes. Of the 507 participants included in the final analysis, 38 (8%) were of African descent, and 39 (8%) were Hispanic. Only eight patients were of another non-white ethnicity. One reason for this was that an originally planned Indian study site was not included, Nathan said, because blood samples were corrupted due to storage difficulties in transporting samples to a central lab in Europe.

Regarding potential differences in the relationship between HbA1c and eAG among different races, “the data are not conclusive,” Nathan acknowledged. “There was no significant difference in the relationship between A1c and average glucose across racial groups. However, the minority groups were relatively small, and more study is necessary.” He noted that no one has suggested a credible physiologic mechanism by which any purported interracial difference in eAG/HbA1c correlation. “Glycation is a very fundamental chemical process. One could argue that maybe there’s a difference in red blood cell turnover between races. That could do it. But absent hemoglobinopathies, which we have ruled out, that doesn’t seem to be right.”

Petersen said that although some reports have found that HbA1c is slightly different in African Americans from Caucasians, this doesn’t mean the new reporting method is inaccurate. “You’re just as right or wrong using the eAG numbers as the A1c numbers,” he said. “So we may not be getting closer to accuracy and knowledge with eAG, we’re not getting any further away.”

Nathan agrees that any such difference that may exist would have “probably a fairly minor impact.” He believes that although ADAG did not achieve its goal in terms of ethnic representation, the ADA’s recommendation to calculate eAG for all patients, regardless of ethnicity, is logical. “I think that the difference between the equations, if there is one, translates to a slight difference. I think that for the time being, eAG is the best thing we have.”

Adopting eAG: The Global Picture

The ADA recommends that labs report all three measurements: eAG, HbA1c in mg/dL, and HbA1c in SI units, as mmol/mol (see Box, p. 3). Some lab information systems can do the calculations, although these can be performed manually as well, using the published formula.

According to ADA’s Petersen, the organization has no plans to recommend or advocate for elimination of HbA1c reporting. He said that ADA consciously decided not to pursue legislation that would mandate the inclusion of eAG in lab reports, based on opposition engendered by such legislation about eGFR.

However, it’s too early to tell how eAG will be implemented by labs around the world. Petersen noted that EASD is “officially supportive of eAG term and values. At the same time, there’s no legal sense in which its member countries are obligated to use eAG,” he said. According to Little, the U.K., Australia, and Sweden will not use eAG for now but will report NGSP and IFCC numbers. A report from a U.K. consensus meeting noted “there was a general agreement that reporting a single measurement in three different formats was considered to be unacceptably complicated” (Annals of Clinical Biochemistry 2008;45:343–4).

Japan, Germany, and much of the rest of Europe is unlikely to adopt eAG, according to Garry John, PhD. “And I am unsure Canada will,” he added. “It’s not that we’re saying it’s wrong or that it’s not the right thing to do; we’re just saying the evidence isn’t there yet,” he explained. However, Sacks said clinicians in Italy are “quite enthusiastic about eAG.”

Clearly, labs are front and center in making eAG a valuable educational tool for diabetic patients. “This should improve patient care,” Sacks said. “Communication is the key. What do the clinicians want? What do they believe will improve patient care? I think the whole concept of eAG was initiated because it was thought that it would enhance communication and help patients understand their care. And obviously, labs have an essential role.”