Author + information
- Received May 4, 2016
- Revision received August 19, 2016
- Accepted August 23, 2016
- Published online December 6, 2016.
- M. Odette Gore, MD, MSCS∗ ( and )
- Darren K. McGuire, MD, MHSc
- ↵∗Reprint requests and correspondence:
Dr. M. Odette Gore, University of Texas Southwestern Medical Center, Department of Internal Medicine, Division of Cardiology, 5323 Harry Hines Boulevard, Dallas, Texas 75390.
Measurement of glycated hemoglobin (HbA1c), the most widely accepted indicator of long-term glycemic exposure, is central for the diagnosis and management of diabetes mellitus. Levels of HbA1c track epidemiologically with diabetic complications, and glycemic control, as reflected by HbA1c reduction, results in decreased risk of microvascular complications, including diabetic kidney disease, neuropathy, and retinopathy. The relationship between HbA1c reduction and cardiovascular disease prevention in patients with diabetes is more complex, with data from large randomized trials published over the past decade providing clear evidence that lowering of HbA1c per se is an inadequate marker for a therapeutic regimen’s impact on cardiovascular outcomes and patient survival. Recent revisions in professional society guidelines moved away from uniform recommendations and toward a more nuanced, patient-centered approach to HbA1c therapeutic targets. The context and key evidence underpinning these recent changes are discussed in this paper, alongside a brief overview of HbA1c contemporary assays and their limitations.
More than 29 million people were estimated to have diabetes mellitus (DM) in the United States in 2012, including approximately 21 million diagnosed and 8 million undiagnosed cases, amounting to a total prevalence of 9.3% of the U.S. population (1). This estimate represents a marked increase from 7.8% in 2007 (2) and is part of a larger trend that is likely to continue in the foreseeable future, with DM prevalence projected to exceed 20% by 2050 (3). Approximately 95% of all DM cases were classifiable as type 2 DM (T2DM) in 2012 (1). Both T1DM and T2DM are major independent risk factors for cardiovascular disease (CVD), including atherosclerotic coronary heart disease, cardiomyopathy, and stroke (4,5). Adults with DM have a 2- to 4-fold higher risk of developing CVD than those without DM and have worse prognosis and higher mortality once clinical CVD is established (4,6).
Measurement of glycated hemoglobin (i.e., hemoglobin A1c [HbA1c]) has become a central pillar of DM diagnosis and management (7). This review provides a brief history of clinical use of HbA1c and an overview of current HbA1c assays and processes, discusses assay limitations, and summarizes the epidemiological associations between HbA1c levels and DM complications, with emphasis on CVD. The review then discusses HbA1c as a prognostic biomarker and its use as a diagnostic screen and a therapeutic target in DM, focusing primarily on CVD considerations, with relevance for both clinical practice and clinical research.
Definition and Relationship With Blood/Plasma Glucose
Approximately 97% of circulating hemoglobin in adults is HbA, containing 2 α- and 2 β-globin chains. In the presence of circulating sugar molecules, HbA can undergo glycation (also known as nonenzymatic glycosylation), whereby sugar molecules are covalently bound to HbA in a process that is essentially irreversible (8). HbA1c is HbA covalently bound to glucose at the amino-terminal groups of the β-chains and is the most abundant glycated hemoglobin, comprising approximately 5% of total HbA in healthy subjects. Notably, although the terms HbA1c and glycated hemoglobin or glycohemoglobin are widely accepted and used interchangeably in the clinical published data, HbA1c is not the only glycated hemoglobin variant present in the circulation; for example, HbA is bound to fructose 1,6-diphosphate in HbA1a1 and to pyruvic acid in HbA1b (8).
The proportion of hemoglobin represented by HbA1c rises in DM because of persistent hyperglycemia and increased availability of glucose as substrate for the glycation reaction. HbA1c levels are proportional to average glucose levels over the preceding 8 to 12 weeks, with this time span determined primarily by erythrocyte turnover (9,10). Compared with glucose, HbA1c has lower biological variability, is not affected by acute factors such as stress and exercise, does not require fasting or timed collections, and provides a better reflection of longer-term glycemic exposure (11).
The ADAG (A1C-Derived Average Glucose) trial, which included more than 2,700 glucose measurements (including continuous glucose monitoring) over a period of 3 months in each of 507 adults with or without DM and stable HbA1c, reported a strong linear correlation (R = 0.84, p < 0.0001) between estimated average glucose (eAG) and HbA1c (12). That study and a subsequent analysis of a subset of the ADAG cohort with HbA1c in the 5.5% to 8.5% range (13) recently prompted the American Diabetes Association (ADA) to introduce an HbA1c-to- eAG conversion calculator (14), with the recommendation that eAG be introduced in patient care and education. Of note, this conversion was developed as a communication tool to help patients better understand the results of their HbA1c test and is not meant to replace HbA1c for clinical decision making.
Glycated variants of HbA were first described in chromatography studies in the 1950s, with their original nomenclature derived from their order of elution (15), and were later shown to each carry specific sugar molecules (8). The fact that HbA1c is elevated in DM was first recognized in 1968 by Rahbar et al. (16), and a correlation between HbA1c and average glucose levels over the preceding 8 to 12 weeks was suggested soon thereafter by multiple studies (9,10). It did not take long for HbA1c to make its way into clinical guidelines, with the ADA recommending HbA1c measurement for routine monitoring of patients with DM as early as 1988 (17). In the 1990s, the DCCT (Diabetes Control and Complications Trial) and the UKPDS (U.K. Prospective Diabetes Study) demonstrated that glycemic control, as assessed by HbA1c, tracks with diabetic microvascular complications in both T1DM (18) and T2DM (19). These findings spurred the widespread adoption of HbA1c as a valuable and convenient therapeutic target in the management of DM, as well as a key intermediate biomarker endpoint for clinical trials of antihyperglycemic interventions (20), including more recent cardiovascular outcome trials discussed later in this review.
However, the early days of HbA1c clinical use were marred by considerable heterogeneity in the ways glycated hemoglobin was measured and reported and by considerable variability, even among laboratories that used the same assays (21). This problem was avoided in early clinical trials by performing all tests in the same laboratory. As HbA1c assays became more standardized and reproducible in the first decade of the 21st century, primarily as a result of the National Glycohemoglobin Standardization Program (NGSP) (21), HbA1c was also endorsed for DM diagnosis and screening (22). In 2009, an international expert committee jointly organized by the ADA, the International Diabetes Federation, and the European Association for the Study of Diabetes recommended that HbA1c be added to the diagnostic criteria for DM as an alternative to plasma glucose, with the recommended HbA1c cut point of ≥6.5% (≥48 mmol/mol) for diagnosis of DM (23). This recommendation was formally implemented in the ADA guidelines in 2010 (22), with the World Health Organization following suit the next year (24).
Analytical Methods and Standardization
Early clinical use of HbA1c was limited by low intra-assay and interassay reproducibility and by significant disparities between the results reported by different laboratories, in the absence of any standardization (8,21). The NGSP was initiated in 1996 under the direction of the American Association for Clinical Chemistry and established a highly integrated network of reference laboratories that worked with each other and directly with assay and instrument manufacturers to standardize HbA1c assay methods and to ensure their direct comparability with the reference method originally used in the DCCT trial. As of August 2016, there were more than 200 different NGSP-certified commercial assay variants for HbA1c measurement, with the most widely used being either antibody-based immunoassays or chromatography-based tests, primarily ion-exchange high-performance liquid chromatography. The NGSP maintains an excellent website (25), with comprehensive information about the certification process and a regularly updated list of certified assays.
The past decade has also seen the widespread introduction of point-of-care HbA1c testing, highly useful for DM screening but lagging behind in terms of standardization (26), although some point-of-care assays are now NGSP-certified (27). The ADA does not yet recommend use of point-of-care assays for diagnostic purposes, but point-of-care HbA1c testing may be used in DM management because it provides the opportunity for more timely treatment changes (7).
Confounding the challenges of HbA1c assay variability, some HbA1c assays are affected by variants of hemoglobin present in patients with hemoglobinopathies, such as β-thalassemia or sickle cell disease in patients carrying the sickle-cell trait (HbS), HbC, HbE, and HbD variants, and in patients with elevated HbF (Central Illustration) (28). The NGSP’s website includes an up-to-date list of the effects of hemoglobin variants on the most commonly used HbA1c assays (29). These are critically important considerations, whereby the choice of a specific assay should be directly informed by hemoglobinopathy prevalence within the population intended for clinical testing. For example, sickle cell trait affects approximately 1 in 12 African Americans; 2% to 3% of African Americans are HbC carriers, and up to 20% of subjects of Mediterranean, African, and South Asian ancestry carry the β-thalassemia trait. Other potential causes of interference with HbA1c assays include naturally occurring chemical modifications of hemoglobin, such as carbamyl-Hb, which is increased in some uremic patients.
In addition, HbA1c will be falsely low (regardless of the assay used) under conditions associated with shortened erythrocyte survival (e.g., hemolytic anemia) or lower erythrocyte age (e.g., acute blood loss) (28). Conversely, HbA1c may be falsely elevated in iron deficiency anemia, an effect that is corrected after iron supplementation (Central Illustration) (30).
Glucose-independent racial differences in HbA1c levels (unrelated to assay interference due to prevalent hemoglobin variants) have been observed across the spectrum of blood glucose values, including in people without DM (31), in those with impaired glucose tolerance (32), and in patients with DM (33); and blacks are reported to have higher HbA1c than non-Hispanic whites by approximately 0.4% to 0.6% for comparable levels of glucose exposure. These race-based differences appear unrelated to CVD outcomes in people without DM, as no significant interaction between race and HbA1c with regard to CVD endpoints and survival was observed among 11,092 adults with no DM or CVD enrolled in the multiethnic ARIC (Atherosclerosis Risk In Communities) study (34). The biological mechanisms underpinning these racial differences remain unclear, and as of the publication date of this review, there have been no recommendations for race-specific HbA1c cutoffs.
The utility of HbA1c measurement in children and adolescents has been questioned (35), and there are currently no specific recommendations for those age groups. There are also no recommendations with regard to use of HbA1c for diagnosis of gestational diabetes mellitus, for which the oral glucose tolerance test remains the gold standard.
Finally, at the risk of stating the obvious, HbA1c measurement does not capture transient episodes of hypoglycemia that may be critically important in the management of patients with DM.
HbA1c and Clinical Outcomes in T1DM
The landmark DCCT trial conducted in the United States and Canada between 1983 and 1993 compared conventional with intensive glycemic control in 1,441 young patients (mean ∼27 years of age) with T1DM who were followed for a mean of 6.5 years. An intensive regimen of glycemic control achieved by 3 to 4 daily insulin injections (or an insulin pump) plus blood glucose monitoring at least 4 times daily was compared with the conventional regimen of 1 to 2 injections per day and once-daily glucose monitoring (18). At the end of the trial, the group undergoing intensive glycemic control achieved a mean HbA1c level of 7.4% compared with 9.1% for the conventional therapy group. This was accompanied by >60% relative risk reduction for microvascular complications, including retinopathy, nephropathy, and neuropathy. Data for CVD were less clear, primarily because of a low number of cardiovascular events in this young cohort, with 0.5 versus 0.8 events/100 patient-years in the intensive versus conventional glycemic control groups, respectively, amounting to a nonstatistically significant relative risk reduction of 41% (95% confidence interval [CI]: −10% to 65%).
The same cohort of patients participating in DCCT was then followed in the nonrandomized, uncontrolled EDIC (Epidemiology of Diabetes and Its Complications) study, with all participants offered the option of intensive management (36). Although the contrast in HbA1c levels between the 2 groups, originally randomized in DCCT, disappeared within 1 year, with mean HbA1c values of 7.8% to 7.9% in both groups, there was a persistent “legacy effect” of prior tighter glucose management on microvascular complications over a mean follow-up of 17 years.
Importantly, a small but statistically significant difference also emerged in the composite primary outcome of CVD events (defined as the time to the first nonfatal myocardial infarction or stroke, cardiovascular death, confirmed angina, or revascularization), with 46 events occurring in 31 patients in the intensive group versus 98 events in 52 patients in the conventional group, amounting to a relative risk reduction of 42% (95% CI: 9% to 63%; p = 0.016). There was also a significant difference in the composite risk of cardiovascular death and nonfatal myocardial infarction or stroke (major cardiovascular adverse events), which decreased by 57% (95% CI: 12% to 79%; p = 0.018) in the intensive glycemic control group. Of note, there were no differences in other major cardiovascular risk factors between groups, including blood pressure, lipids, and body mass index.
Assessment of atherosclerosis during long-term follow-up in subsets of the original DCCT cohort also supported the beneficial “legacy effects” of tighter glycemic control in DCCT, including lower common carotid intima-media thickness (37) and lower coronary artery calcification (38) in the intensive glycemic control group. These differences were no longer statistically significant after adjustment for HbA1c during DCCT.
It is important to note that, although intensive glycemic control in DCCT resulted in a significant HbA1c reduction versus that in the conventional management group, the HbA1c values achieved (mean: 7.4%) were still well above the diagnostic threshold for DM. The potential effects of even tighter glycemic control in T1DM resulting in near-normalization of HbA1c have not been studied, but such aggressive therapy is unlikely to be beneficial, given the increased risk of hypoglycemia, as well as the lack of CVD benefit from more intensive glycemic control in trials of T2DM (discussed in the following text).
HbA1c and Clinical Outcomes in T2DM
Among 3,642 patients with newly diagnosed T2DM enrolled in the UKPDS 35 prospective observational study, each 1% reduction in HbA1c was associated with a 14% reduction of the relative risk of myocardial infarction (95% CI: 8% to 21%) (39). Of note, this relationship was not as strong as for microvascular complications, where each 1% reduction in HbA1c was associated with a 37% relative risk reduction (95% CI: 33% to 41%) (32).
Analyses from the U.S. National Health and Nutrition Examination Survey, including 19,025 adult participants enrolled from 1988 to 1994 with follow-up through 2000, yielded a hazard ratio for CVD mortality of 3.38 (95% CI: 1.98 to 5.77) in a comparison of participants with HbA1c ≥8% versus those with HbA1c <6%, after adjustment for other CVD risk factors (33). The hazard ratio for CVD mortality was lower, at 2.48 (95% CI: 1.09 to 5.64), when analyses were restricted to participants with DM (mostly T2DM) and was no longer significant among participants without DM (40).
More recent studies have suggested that the relationship between HbA1c and CVD events or mortality may be J-shaped, with increased risk at both high and very low HbA1c levels. For example, in a retrospective study of 47,970 patients with T2DM who had undergone intensification of DM treatment between 1986 and 2008, compared with the HbA1c decile with the lowest hazard (median HbA1c 7.5%), the adjusted hazard ratio for all-cause mortality in the lowest HbA1c decile (median HbA1c of 6.4%) was 1.52 (95% CI: 1.32 to 1.76) and was 1.79 (95% CI: 1.56 to 2.06) in the highest decile (median HbA1c of 10.5%) (41). A similar J-shaped curve for all-cause mortality emerged from a large meta-analysis of 7 observational studies (including the study mentioned previously) with a total of 147,424 participants with T2DM (42). These findings may reflect the adverse outcomes of hypoglycemia in patients receiving intensively treated DM, although low HbA1c independent of DM treatment may also be a marker for other advanced disease processes, such as inflammatory and liver diseases, with a similarly J-shaped mortality curve observed in analyses from National Health and Nutrition Examination Survey III, consisting of more than 14,000 persons with no diabetes at study entry (43).
Further attesting to the epidemiological link between elevated HbA1c and CVD outcomes, 38% of 2,853 patients hospitalized with acute myocardial infarction in a large U.S. registry-based study had a diagnosis of diabetes on the basis of HbA1c measured before hospital discharge, with nearly 1 in 5 being newly diagnosed (44). Patients with abnormal HbA1c also had higher mortality after 3 years of follow-up. As HbA1c measurement is more convenient than oral glucose tolerance testing and more reliable than fasting glucose, especially in acute settings, hospitalization for acute coronary syndromes may provide a key opportunity for the detection of undiagnosed diabetes using HbA1c (44).
In the UKPDS, intensive glycemic control with sulfonylureas or insulin, achieving a median HbA1c of 7.0% versus 7.9% in the conventional group resulted in decreased risk of microvascular complications in patients with newly diagnosed T2DM at study entry but had no significant effect on macrovascular complications (19). In contrast, a substudy of the UKPDS including 753 patients overweight at trial entry randomized to intensive therapy with metformin versus standard glycemic control (primarily with diet alone), resulted in improved macrovascular outcomes and survival in the metformin group over a median follow-up of 10.7 years, despite achieving a more modest HbA1c contrast (median HbA1c of 7.4% in the metformin group and 8.0% in the conventional group) (45). These UKPDS results formed the basis for the key concepts that not all antihyperglycemic drugs are created equal from the standpoint of CVD outcomes, with metformin gaining first-line therapy status, and that pharmacological HbA1c lowering per se, although clearly beneficial for reducing microvascular complications, may not be an adequate surrogate for CVD risk reduction in T2DM.
Three large randomized trials of intensive versus standard glycemic control published at virtually the same time in 2008 provided further support for the inadequacy of HbA1c as a surrogate for CVD risk reduction (Table 1). The ACCORD (Action to Control Cardiovascular Risk in Diabetes) (46), ADVANCE (Action in Diabetes and Vascular Disease: Preterax and Diamicron Modified Release Controlled Evaluation) (47), and VADT (Veterans’ Affairs Diabetes Trial) (48) trials randomized over 23,000 patients with long-standing T2DM and either prevalent CVD or clustered CVD risk factors to various more intensive versus standard glycemic control regimens, and each failed to demonstrate improved survival or CVD outcomes in the more intensive glucose treatment groups, despite achieving significant HbA1c contrasts (mean achieved HbA1c in the more intensive vs. standard treatment arms: 6.4% vs. 7.5%, respectively, in ACCORD; 6.4% vs. 7.3%, respectively, in ADVANCE; and 6.9% vs. 8.4%, respectively, in VADT). Even more surprisingly, the ACCORD trial had to be terminated prematurely because of increased CVD and all-cause mortality in the more intensive glucose treatment group for reasons that are still unclear. It is possible that the increased mortality observed in the intensive arm of ACCORD was due to the specific antihyperglycemic regimen used, with rosiglitazone cited as the most likely culprit (49), but this remains speculative, as the trial was not designed to evaluate the effects of individual drugs.
A notable finding emerged from the 10-year post-trial follow-up of UKPDS, despite the loss of any difference in HbA1c between study groups within the first year after termination of the initial trial (42). Although no CVD or mortality benefit of intensive glycemic control was detected in the main UKPDS trial (20), as more events occurred over time, a lower risk of myocardial infarction and all-cause mortality emerged in patients originally assigned to intensive versus standard glycemic control (50). In contrast, no such “legacy effect” emerged from the 6-year post-trial follow-up of ADVANCE (51). Although the exact cause for this difference is unknowable, one possible explanation is that younger patients with newly established T2DM (such as those enrolled in UKPDS) may benefit more from early, more intensive glycemic control than older patients who already have high CVD risk (such as those enrolled in ADVANCE). A second possible explanation may be related to the HbA1c differences achieved in the more intensive versus standard treatment arms of the 2 studies (6.4% vs. 7.3%, resepectively, in ADVANCE, and 7.0% vs. 7.9%, respectively, in UKPDS), with higher risk of hypoglycemia in ADVANCE.
One overarching conclusion from the studies described in this section (and from others not included because of space limitations) is that HbA1c lowering per se is a poor marker of the impact that a therapeutic regimen has on CVD risk and survival in T2DM (Central Illustration). It is certainly possible for antihyperglycemic therapies to have beneficial effects on CVD and survival, but these effects are going to depend on a multitude of factors beyond HbA1c targets, including the choice of antihyperglycemic drug(s), the timing of the intervention (early vs. late in the course of T2DM), the maintenance of glycemic control over time, and, of course, the control of other CVD risk factors, such as lipids and blood pressure.
HbA1c Treatment Targets: Recent Changes in Professional Guidelines
As a result, in part, of the complex relationship between HbA1c and DM-related complications and outcomes, a common position statement by the ADA and the European Association for the Study of Diabetes released in 2012 represented a major shift from target-based recommendations to a much more nuanced, patient-centered approach to the treatment of T2DM (52), which was reiterated in the 2015 position statement update (53).
As a general guide, the latest ADA guidelines (7) still recommend an HbA1c treatment target of <7% (<53 mmol/mol) for DM in nonpregnant adults, justified primarily by beneficial effects on DM clinical symptoms and on microvascular risk. However, it is now recognized that less stringent targets, such as HbA1c of <8% (<64 mmol/mol) or even higher may be more appropriate for some patients, such as those with a history of severe hypoglycemia and long duration of DM, extensive complications or multiple comorbidities, and importantly, those with moderate to severe CVD. However, a more stringent HbA1c target of <6.5% (<48 mmol/mol) may only be recommended in select patients, such as younger adults with newly diagnosed DM and no significant CVD, especially if such targets can be achieved with lifestyle and metformin only and without significant hypoglycemia.
In just 3 decades, measurement of HbA1c for clinical purposes has evolved from introduction to widespread clinical and research use. Handicapped early on by extremes of interlaboratory and intralaboratory variance of assays and results, contemporary assays have substantially overcome these early challenges with broadly standardized assays presently available across laboratories and, more recently, with the emergence of point-of-care methodologies for HbA1c testing, with advances underpinned by the process of the ongoing NGSP program. The measurement of HbA1c remains challenging due to interference of some assays by hemoglobin variants, and the observation of important race-based variance in the associations between HbA1c and levels of circulating blood glucose, 2 areas that require ongoing attention for research and clinical application.
Given the improved reliability, widespread availability, and low cost of HbA1c measurement in contemporary clinical practice, coupled with the lack of requirement for fasting samples, the adoption of HbA1c criteria to complement prior testing strategies for screening and diagnosis was a major step forward in the clinical space and in the public health domain. Most recently, the revision of professional society guidance with regard to appropriate therapeutic targets for HbA1c in patients with T2DM allowed for more nuanced and personalized care considerations for each patient and for more liberal HbA1c targets for special populations. Key in this evolution from a CVD perspective is endorsement of HbA1c targets of 8% or even higher for patients who have prevalent moderate to severe CVD. In each of these contexts, for prognosis, diagnosis, and therapeutic targeting, HbA1c has rapidly become an invaluable laboratory test in contemporary clinical practice and research programs.
Dr. McGuire has received honoraria for clinical trial leadership from Boehringer-Ingelheim, Janssen Research and Development LLC, Merck Sharp and Dohme, Lilly USA, Novo Nordisk, GlaxoSmithKline, Takeda Pharmaceuticals North America, AstraZeneca, Lexicon, and Eisai; and personal consulting fees from Sanofi, Boehringer Ingelheim, Merck Sharp and Dohme, Novo Nordisk, Lilly USA, and Regeneron. Dr. Gore is uncompensated principal or co-investigator for randomized trials sponsored by GlaxoSmithKline and AstraZeneca; and is supported by a career development award from the National Heart, Lung, and Blood Institute (K23-HL131939).
- Abbreviations and Acronyms
- cardiovascular disease
- diabetes mellitus
- glycated hemoglobin A1c
- type 1 diabetes mellitus
- type 2 diabetes mellitus
- Received May 4, 2016.
- Revision received August 19, 2016.
- Accepted August 23, 2016.
- American College of Cardiology Foundation
- Centers for Disease Control and Prevention
- National Diabetes Information Clearinghouse
- de Ferranti S.D.,
- de Boer I.H.,
- Fonseca V.,
- et al.
- Mozaffarian D.,
- Benjamin E.J.,
- Go A.S.,
- et al.,
- for American Heart Association Statistics Committee,
- Stroke Statistics Subcommittee
- American Diabetes Association
- Sacks D.B.
- Sacks D.B.
- Nathan D.M.,
- Kuenen J.,
- Borg R.,
- et al.,
- for A1c-Derived Average Glucose (ADAG) Study Group
- Wei N.,
- Zheng H.,
- Nathan D.M.
- ↵eAG/A1C Conversion Calculator. American Diabetes Association Diabetes Pro. Available at: http://professional.diabetes.org/diapro/glucose_calc. Accessed September 14, 2016.
- ↵(1989) Standards of medical care for patients with diabetes mellitus. Diabetes Care 12:365–368.
- Little R.R.,
- Rohlfing C.L.,
- Sacks D.B.,
- for National Glycohemoglobin Standardization Program Steering Committee
- American Diabetes Association
- International Expert Committee
- ↵(2011) Use of Glycated Haemoglobin (HbA1c) in the Diagnosis of Diabetes Mellitus: Abbreviated Report of a WHO Consultation (World Health Organization, Geneva, Switzerland) Available at: www.who.int/diabetes/publications/report-hba1c_2011.pdf. Accessed September 14, 2016.
- ↵National Glycohemoglobin Standardization Program. NGSP Web site. Available at: http://www.ngsp.org. Accessed September 14, 2016.
- Whitley H.P.,
- Yong E.V.,
- Rasinen C.
- Bry L.,
- Chen P.C.,
- Sacks D.B.
- ↵National Glycohemoglobin Standardization Program. HbA1c Assay Interferences. NGSP Web site. 2106. Available at: http://www.ngsp.org/interf.asp. Accessed September 14, 2016.
- Herman W.H.,
- Ma Y.,
- Uwaifo G.,
- et al.,
- for Diabetes Prevention Program Research Group
- Kirk J.K.,
- D'Agostino R.B. Jr..,
- Bell R.A.,
- et al.
- Nowicka P.,
- Santoro N.,
- Liu H.,
- et al.
- Polak J.F.,
- Backlund J.Y.C.,
- Cleary P.A.,
- et al.,
- for DCCT/EDIC Research Group
- Cleary P.A.,
- Orchard T.J.,
- Genuth S.,
- et al.,
- DCCT/EDIC Research Group
- Stratton I.M.,
- Adler A.I.,
- Neil H.A.,
- et al.
- Saydah S.,
- Tao M.,
- Imperatore G.,
- et al.
- Carson A.P.,
- Fox C.S.,
- McGuire D.K.,
- et al.
- Inzucchi S.E.,
- Bergenstal R.M.,
- Buse J.B.,
- et al.
- Inzucchi S.E.,
- Bergenstal R.M.,
- Buse J.B.,
- et al.
- Central Illustration
- Definition and Relationship With Blood/Plasma Glucose
- Analytical Methods and Standardization
- HbA1c and Clinical Outcomes in T1DM
- HbA1c and Clinical Outcomes in T2DM
- HbA1c Treatment Targets: Recent Changes in Professional Guidelines