HbA1c vs Fasting Glucose: Why the Results Disagree and Which to Trust

Fasting glucose describes the glucose concentration at one moment, after at least 8 hours without food. It depends on the liver's glucose production overnight, on insulin sensitivity at that hour, but also on short-term factors: sleep, stress, infection, the previous evening's dinner. HbA1c is the result of a non-enzymatic reaction: glucose attaches to hemoglobin in red blood cells in proportion to its concentration over time, so the result reflects average glycemia weighted toward the last weeks, including glucose after meals. A person with normal fasting glucose but high post-meal spikes may therefore have an elevated HbA1c, and vice versa.

On top of this comes the different variability of the tests themselves. In the study by Selvin et al. (2007) the within-person coefficient of variation was 5.7% for fasting glucose, 16.7% for glucose 2 hours into a tolerance test and 3.6% for HbA1c, which means HbA1c is the most stable of the three and a single fasting glucose clearly less so. A discrepancy therefore does not have to mean that HbA1c is the 'true' one — it may mean that a single glucose measurement simply caught a more or less typical morning.

The practical conclusion is that disagreement between HbA1c and fasting glucose is a population norm, not an exception. That is why the American Diabetes Association (ADA, 2025) standards provide that when two different tests give discordant results, the one whose result exceeds the diagnostic threshold is repeated, rather than averaging the two or assuming that the 'better' result is enough. In clinical practice an OGTT is often added to the workup when results are borderline or contradictory.

Hemoglobin is glycated more the longer the cell circulates in the blood, so HbA1c depends not only on average glucose but also on the average age of erythrocytes. In a person whose red cells live longer or turn over more slowly, HbA1c will be higher at the same glycemia; in a person whose red cells wear out quickly, lower. This variability in healthy people was shown in a study using biotin labelling.

This individual 'constant' is sometimes called the glycation gap (the difference between measured HbA1c and the value predicted from other measures, such as fructosamine) or the hemoglobin glycation index. In the study by Zafon et al. (2013) of 508 people with type 2 diabetes, in whom measurement was repeated after about 30 weeks, the glycation gap was reproducible and consistent over time, and independently related to creatinine, mean corpuscular hemoglobin concentration (MCHC), glycemia and treatment with metformin — patients taking metformin had a higher HbA1c at similar fructosamine. This result is observational and comes from a group of patients, but it shows that a discrepancy can be a trait of the person, not a coincidence.

What follows from this meta-analysis is practically significant: in a person with iron deficiency, HbA1c can be overstated by about 0.5–0.7 percentage points on average, enough to move a result of 5.3% to 6.0% (the prediabetes zone) with completely normal glycemia. The mechanism is not fully explained; the most often cited are changes in the red cell population, including a longer average age of erythrocytes in iron deficiency anemia, and increased glycation in that population. Importantly, the effect concerned not only overt anemia but also iron deficiency without anemia (see our entries on ferritin and on iron deficiency without anemia).

So if HbA1c is elevated and fasting glucose is normal, one of the first things to check is ferritin and the blood count (hemoglobin, MCV, RDW). Iron deficiency is common especially in women with heavy periods and in people on plant-based diets, so those are the groups in which it is most worth thinking of. After iron deficiency is corrected, HbA1c often falls, as the intervention studies confirm, but it must be remembered that iron treatment is a doctor's decision and requires confirmation of deficiency by tests.

Megaloblastic anemia from vitamin B12 (and folate) deficiency lengthens red cell life and, like iron deficiency, can raise HbA1c. In the study by Aydin et al. (2022), involving 37 people without diabetes who had prediabetes by HbA1c and anemia from B12 deficiency, and 40 healthy volunteers, oral cyanocobalamin 1 mg daily for 3 months raised hemoglobin from 11.31 to 12.26 g/dl, and HbA1c fell from 6.01 to 5.77%. In 35% of patients after 3 months HbA1c no longer met the criterion for prediabetes. The authors conclude that correcting B12 deficiency before making a diagnostic or therapeutic decision based on HbA1c protects against misdiagnosis. This study was uncontrolled for placebo, so its value is moderate, but the direction of the effect is consistent with general knowledge about HbA1c.

In the opposite direction act states that shorten red cell life or speed up their turnover: hemolytic anemia, recent blood loss or transfusion, treatment with erythropoietin, and also some hemoglobinopathies. HbA1c is then often understated relative to true glycemia and can mask diabetes. The review by Radin (2014) describes these scenarios and alternative markers, which include fructosamine and glycated albumin. In chronic kidney disease, HbA1c is affected by shortened red cell life, anemia and treatment with erythropoiesis-stimulating agents, which makes interpreting HbA1c in advanced kidney disease harder and requires caution.

A hemoglobin variant is a genetically altered form of the hemoglobin molecule. The most common are sickle cell trait (HbS), HbC and HbE, and thalassemias are also counted in this group. The effect of such a variant on the HbA1c result depends on the method the laboratory uses (chromatography, immunoassay, capillary electrophoresis): some methods are resistant to the most common variants, others give falsely raised or lowered results. Laboratories using certified methods usually have procedures for detecting variants and annotate the result, but the patient does not always know about this.

In practice it is worth asking the laboratory (or your doctor) what method was used to measure HbA1c, especially when the result strikingly departs from glucose, anemia, thalassemia or sickle cell disease runs in the family, or you come from a region with a high frequency of hemoglobinopathies. When a variant is suspected or HbA1c is unreliable, ADA standards advise relying in such conditions on plasma glucose-based criteria (fasting glucose, OGTT) rather than HbA1c.

Fructosamine and glycated albumin reflect glycation of serum proteins, and hence glycemia over a shorter period, about 2–3 weeks instead of 2–3 months, and do not depend on red cell lifespan. For this reason they are sometimes used as additional markers when HbA1c is unreliable (e.g. in anemias and hemoglobinopathies). The review by Ribeiro et al. (2016) indicates that glycated albumin and fructosamine can provide additional information or a reliable measure when HbA1c does not hold up, and the shorter exposure period can be useful for detecting rapid changes. In a small study of 34 people with diabetes (Desouza et al., 2023), glycated albumin correlated with CGM time in range better than HbA1c and fructosamine. It must be remembered, however, that these markers have their own limitations (disorders of albumin and protein concentration, liver and kidney disease, thyroid hormones), do not have such well-established diagnostic thresholds as HbA1c, and are less available.

This result matters for CGM users without diabetes: the GMI shown in the app is not a 'home HbA1c' and in healthy people can be more than half a percentage point higher than the laboratory result. The index was developed and validated mainly in people with diabetes, so it should not be the basis for diagnostic decisions in healthy people. The GMI–HbA1c discrepancy arises from the same mechanisms as the HbA1c–glucose disagreement: differences in red cell lifespan and individual glycation rate, as well as the sensor's measurement error.

It is important to resolve a discrepancy systematically and not pick the result that seems more favorable. If one test exceeds the diagnostic threshold for diabetes, it must not be ignored just because the other is 'nice'. So the standard is to repeat the test that gives a result in the disease range and, where there is no agreement, to extend the workup with an OGTT. Nor should one make independent decisions to change diet or supplementation (especially iron, which is harmful in excess) solely on the basis of a discrepancy in results.

A discrepancy between HbA1c and fasting glucose is one of those results that is easy to take as a contradiction and that in reality carries additional information: about how a person responds to glucose after meals, about their iron and B12 status, about the lifespan of their red cells. It is therefore worth starting not with the question 'which to believe' but with 'why do they differ'. A simple blood count with ferritin and a B12 measurement often explains far more than further repeats of glucose alone, and when in doubt the most reliable arbiter is the OGTT.