HbA1c: What the Result Measures—and What Can Distort It

HbA1c, also called A1C, measures the fraction of hemoglobin with glucose attached. It is an indirect, weighted summary of glucose exposure over roughly the previous two to three months—not a direct glucose measurement and not a simple average in which every day counts equally. Recent weeks influence the result more than the earliest weeks in that window.

One number can serve different jobs: screening, helping establish a diagnosis, or monitoring diabetes over time. Read it only after naming that job and keeping the assay source, red-blood-cell context, and related glucose evidence attached. The same percentage is not automatically a diagnosis, a treatment target, or a verdict about a trend.

What HbA1c actually measures

Glucose attaches to hemoglobin inside red blood cells, producing glycated hemoglobin. An HbA1c assay reports how much of the relevant hemoglobin is glycated, usually as a percentage or in IFCC units of mmol/mol. Because the circulating red-cell population contains cells of different ages, the result integrates exposure across time.

“Three-month average” is useful shorthand, but it hides three boundaries:

• The window is approximately two to three months, not a fixed 90-day box.
• It is weighted toward more recent glucose exposure. NIDDK explains that the preceding 30 days contribute more than glucose 90 to 120 days earlier.
• It is an indirect summary, not a record of every high, low, or swing inside that period.

The 2026 ADA Standards of Care describe A1C as chronic glucose exposure over about two to three months. The current ADA monitoring guidance adds that A1C does not show real-time glucose, glucose variability, or hypoglycemia. Those limits are why an A1C result and meter or continuous-glucose-monitor data can be related without being interchangeable.

Do you need to fast for an HbA1c test?

The A1C sample itself does not require fasting. Food eaten that morning does not play the same immediate role it does in a fasting plasma glucose test. That convenience is one reason A1C is commonly used for screening and monitoring.

But “A1C is nonfasting” is not permission to ignore the order instructions. The same visit may include fasting glucose, lipids, or another test with different preparation requirements. Preserve the order and follow the instructions for the complete set of tests, not a rule borrowed from one analyte.

Which HbA1c criteria apply in 2026?

For nonpregnant individuals, the ADA's 2026 classification criteria include:

• 5.7% to 6.4% (39 to 47 mmol/mol): the A1C criterion for prediabetes.
• 6.5% (48 mmol/mol) or higher: the A1C criterion for diabetes when the test is performed in a laboratory using an NGSP-certified method standardized to the DCCT assay.

These are classification boundaries, not portable “normal” and “optimal” zones. The ADA states that risk is continuous, including below 5.7%. A result below 5.7% is below that A1C criterion; it does not guarantee that every other glucose test is below its own criterion or that an individual has no relevant risk.

A1C, fasting plasma glucose, and the two-hour glucose result from an oral glucose tolerance test capture different aspects of glucose metabolism. The ADA notes that they identify incompletely overlapping groups. That means apparent disagreement is possible without proving that one result should simply be discarded.

In the absence of unequivocal hyperglycemia, two abnormal results are required to establish a diabetes diagnosis. They may be two different tests collected at the same time or a repeat of the same or a different test at another time. One portal flag is not the whole confirmation process.

For a useful distinction between a laboratory's displayed interval, a classification boundary, and an individual clinical decision, see how lab reference ranges work.

Diagnostic boundaries are not treatment goals

An A1C threshold used to classify screening or diagnostic evidence answers a different question from a goal used to manage known diabetes. Treating those numbers as one universal ladder creates unsafe shortcuts.

The ADA's 2026 glycemic-goals guidance says a goal below 7% (53 mmol/mol) is appropriate for many nonpregnant adults with diabetes who do not have severe hypoglycemia or hypoglycemia affecting health or quality of life. The same guidance explicitly supports more or less stringent goals depending on health and function, hypoglycemia risk, treatment risks and burdens, resources, preferences, and other individual circumstances.

So “below 7%” is neither an ideal for every person nor the definition of a normal screening result. A personal goal belongs to a shared clinical plan and may change as circumstances change. This article cannot select one.

A diagnostic boundary classifies evidence. A treatment goal belongs to a person-specific care plan.

Two different ways HbA1c can become unreliable

“Interference” is often used as a catch-all, but there are at least two distinct problems.

The assay can be affected

Some hemoglobin variants, elevated fetal hemoglobin, or chemically modified hemoglobin can affect particular assay methods. The NGSP interference review, updated June 23, 2026, stresses that the effect depends on both the specific factor and the specific method. Its method-by-method table exists because a variant is not a universal instruction to add or subtract a fixed amount.

If a report identifies the assay method, preserve it. If it does not and method interference is a real question, that is something for the clinician and laboratory to resolve. Do not calculate a corrected A1C from a generic internet list.

Red-cell biology can change the relationship

A method can perform as designed while the result still fails to represent glucose exposure in the usual way. Conditions or events that change red-cell turnover, hemoglobin availability, or the circulating cell population can alter the A1C-glycemia relationship.

Current ADA and NIDDK sources identify contexts such as some anemias, recent blood loss or transfusion, erythropoietin treatment, hemodialysis or kidney failure, pregnancy, glucose-6-phosphate dehydrogenase deficiency, HIV treatment, liver disease, and some hemoglobin variants. The list is context to review, not a way to diagnose a cause from the A1C or predict the direction and size of an effect for one person.

The NIDDK hemoglobinopathy guide also notes that reliable methods are available for most heterozygous variants. Known variant information and the actual method matter more than assumptions based on race or ancestry.

What discordance with glucose evidence means

HbA1c may not line up with a laboratory glucose result, fingerstick record, or continuous-glucose-monitor summary. “Discordance” describes that mismatch; it does not explain it.

Possible questions include whether the measurements cover comparable time periods, whether glucose varied substantially within the A1C window, whether either measurement has a technical limitation, and whether red-cell or hemoglobin context affects A1C. The ADA recommends evaluating for a problem or interference when glucose values and A1C show consistent, substantial discordance. It also directs clinicians to use plasma-glucose criteria for diagnosis when a condition alters the relationship between A1C and glycemia.

Do not average unlike results, choose whichever is more reassuring, or infer that one proves the other is wrong. Preserve each measurement with its own source and timing so the clinician can decide what evidence answers the actual question.

How to review an HbA1c result in five steps

1. Name the result's role. Was it ordered for screening, diagnostic confirmation, or monitoring? Do not apply a treatment goal to a screening result or a screening boundary to an established care plan.
2. Preserve the source. Keep the exact result, percent or mmol/mol unit, laboratory, collection date, report comments, and assay or point-of-care label if shown.
3. Add reliability context without assigning an effect. Keep known hemoglobin-variant, anemia, red-cell, transfusion, pregnancy, kidney, liver, and relevant therapy context visible. The clinician decides what applies.
4. Compare compatible evidence. Place related laboratory glucose, meter, or CGM evidence beside the A1C with its own dates. Mark disagreement instead of silently reconciling it.
5. Prepare bounded questions. Ask which criteria apply, whether the method and context make the result usable, what may clarify discordance, and which individualized goal applies.

The broader blood-test review workflow can help you preserve result names, units, source ranges, dates, and context without turning a portal flag into an interpretation.

How to compare HbA1c results over time

A series can be useful, but a line between two percentages does not prove a biological direction. Before describing a change, compare:

• the exact result name and unit;
• the laboratory and assay or point-of-care label, if reported;
• the collection dates and how much time each result summarizes;
• report comments, flags, and any reliability context;
• related glucose, meter, or CGM evidence from the relevant periods; and
• changes in clinical context that a clinician says matter.

The ADA/AACC laboratory-analysis guideline notes that small A1C changes may reflect assay variability rather than a true change in glycemic status. That is not a universal “ignore changes smaller than X” rule. It is a reason to preserve provenance and ask whether two results are meaningfully comparable. For records from multiple sources, use the cross-laboratory checklist.

Questions worth bringing to a clinician

• Was this result being used for screening, confirmation, or monitoring?
• Do the 2026 nonpregnant A1C criteria apply in this situation?
• Was the diagnostic result produced with an appropriate certified laboratory method, and is the assay method available?
• Does any known red-cell, hemoglobin, transfusion, pregnancy, kidney, liver, or therapy context affect how this result should be used?
• If A1C and glucose evidence disagree, which possibilities need review and what evidence would clarify the question?
• If this is monitoring established diabetes, what individualized goal applies and what tradeoffs shaped it?
• Are results from different dates, laboratories, or methods comparable enough to discuss a change?

These questions preserve the clinical decision for the person who has the full history. They do not prescribe a test, timing, diagnosis, or treatment.

What Libby can organize—and what it cannot decide

You can use Libby to keep the original HbA1c report, units, laboratory, date, related glucose records, known reliability context, and questions in one reviewable record. That organization can make it easier to show where two sources agree, differ, or leave a gap.

Libby does not perform the assay, certify the method, verify that two results are comparable, determine why values disagree, establish a diagnosis, set a personal goal, or recommend testing or treatment. Those decisions stay with a qualified healthcare professional who can review the complete context.

FAQ

Does an HbA1c test require fasting? The A1C sample itself does not require fasting. Other tests ordered for the same visit may have different preparation instructions, so follow the instructions for the complete order.

Does an HbA1c of 5.7% mean prediabetes? In the 2026 ADA criteria for nonpregnant individuals, 5.7% to 6.4% is the A1C criterion for prediabetes. Whether A1C is reliable and applicable in the specific context still matters, and the result is not a personal treatment target.

Does an HbA1c of 6.5% diagnose diabetes by itself? For nonpregnant individuals, 6.5% or higher is an ADA diagnostic criterion when measured by an appropriate certified laboratory method. Without unequivocal hyperglycemia, two abnormal results are required to establish the diagnosis.

Does an HbA1c below 5.7% rule out diabetes? No. It is below the ADA A1C prediabetes boundary, but risk is continuous and A1C, fasting plasma glucose, and an oral glucose tolerance test can classify some people differently. A clinician applies the appropriate evidence and context.

Can HbA1c change before three months have passed? Yes. HbA1c is weighted toward more recent glucose exposure, so a meaningful change can appear before a full red-cell lifespan has passed. That does not define when any individual should be retested.

Can anemia or a hemoglobin variant alter HbA1c? It can in some circumstances. Red-cell biology can alter the relationship between glucose and A1C, and some variants interfere with some assay methods. The direction and size cannot be assigned from the condition name alone.

Why can HbA1c disagree with glucose or CGM data? The measurements cover different aspects and time windows. Glucose variability, timing, measurement limitations, assay interference, or altered red-cell biology may contribute. Consistent, substantial discordance deserves clinical and laboratory review rather than a do-it-yourself correction.

What is a good HbA1c goal for someone with diabetes? ADA guidance uses below 7% as an example for many nonpregnant adults with diabetes, but goals can be more or less stringent. Health, function, hypoglycemia risk, treatment burden, resources, and preferences make the goal person-specific.

References

• ADA Standards of Care 2026: Diagnosis and Classification
• ADA Standards of Care 2026: Glycemic Goals
• NIDDK: The A1C Test and Diabetes
• NIDDK: Sickle Cell Trait, Hemoglobinopathies, and Diabetes
• NGSP: Factors That Interfere With HbA1c Results
• NGSP: HbA1c Assay Interferences
• ADA and AACC: Laboratory Analysis Guideline

---

Educational content, not medical advice. This article is for general information and personal record-keeping. Classification boundaries, treatment goals, report intervals, assay methods, and reliability differ by context. Discuss results with a qualified healthcare professional.