Certainty in science went out of fashion early last century following Heisenberg's postulates. We now see ranges and confidence intervals around any given mode, mean, or median. Truth, in short, is a dependent variable. Yet the American Diabetes Association (ADA) seems to be set on recommending the introduction of an estimated average glucose (eAG) value based on A1C levels, as an ADA Web site carries the formula for conversion of A1C to eAG.
However, three factors have led representatives of 18 professional organizations in the U.K. to reject the introduction of this same eAG conversion based on current evidence (1). These representatives, including the current two authors, are concerned with the validity of the basic observation, its lack of validation, and the level of uncertainty associated with this conversion formula.
First, the critical conversion formula, proposed by Nathan et al. (2), is probably flawed because the patients who comprised their study group had been nonrandomly selected (and included individuals without diabetes), so there is, therefore, no way of knowing whether this conversion would apply to the full breadth of patients that clinicians treat.
Second, validation within fields of biochemistry requires that findings from one study population be validated by a second phase that uses different and unselected patient groups. This validation step has not been done, in either a separate group of Caucasian adults or other patient groups, such as children or patients of Afro-Caribbean descent. Further, other studies have shown that A1C can vary with hemoglobinopathy, race, age, pregnancy, and renal dysfunction (3).
Third, there is uncertainty associated with the eAG value itself, acknowledged by the authors but not evident in the formula promulgated on the ADA Web site. That uncertainty is such that two patients with the same true mean glucose level of 170 mg/dl could have an A1C value anywhere from 6.5% (which is as low as the A1C values of the intervention groups of the VA and ACCORD [Action to Control Cardiovascular Risk in Diabetes] studies) to 9.0% (which was adversely high in the Diabetes Control and Complications Trial) (2). Clinicians are already aware of ways to reconcile this degree of uncertainty, but we—as a diabetes community—are now planning to offload this task onto our patients. As a result, based on possible conversions of A1C to eAG, patients with the same true mean glucose level of 170 mg/dl could be told that their eAG is anywhere between 140 and 212 mg/dl—with predictable consequences when they try to compare this laboratory measurement with their own meter readings. The authors of the original article (2) and the editorial (4) seem anxious to reject this between-individual variation in A1C, which compounds the inherent error in their method, despite having been previously demonstrated in evidence from twin, racial, and prospective studies, as well as our knowledge of the genetics of glucose metabolism (5).
Here, then, are three reasons why the ADA should be as wary as we were in deciding whether there is currently sufficient experimental evidence to support the introduction and use of eAG values. Or, at the very least, the ADA could reflect on the Principle of Uncertainty and present the degree of uncertainty around their estimations of the truth.
Article Information
No potential conflicts of interest relevant to this article were reported.
References
Readers may use this article as long as the work is properly cited, the use is educational and not for profit, and the work is not altered. See http://creativecommons.org/licenses/by-nc-nd/3.0/ for details.