The ADA 2026 rule (Recommendations 2.1a and 2.1b, level B) is short and is often broken in the emergency department. The diagnosis is made by A1C or plasma glucose. In the absence of unequivocal hyperglycaemia (for example, a hyperglycaemic crisis), the diagnosis requires confirmatory testing.
What counts as unequivocal. Classic symptoms plus a random plasma glucose ≥11.1 mmol/L — this is enough: here knowing the glucose also determines the emergency management, and HbA1c helps to understand chronicity but does not "confirm" the acute diagnosis. A hyperglycaemic crisis (DKA, hyperosmolar state) is also not a "come back tomorrow fasting" situation.
In all other cases two abnormal results are needed. They may be:
- from different tests in the same sample (HbA1c and FPG both above threshold at the same time);
- from the same test on two different dates;
- from different tests on two dates.
The ADA example: HbA1c 7.0%, repeat 6.8% — diagnosis confirmed. HbA1c and FPG both above threshold on the same day — also confirmed. If the tests disagree, the one that crossed the threshold is repeated, taking into account factors that distort HbA1c or glucose. The diagnosis is made on the confirmed test. A person with two HbA1c values ≥6.5% and an FPG of 6.8 mmol/L still has diabetes.
The MoH RK 2022 protocol says the same in substance: the diagnosis is always confirmed by repeat measurement of glycaemia on subsequent days, except in unequivocal hyperglycaemia with acute decompensation or obvious symptoms. For HbA1c: in the absence of symptoms of acute decompensation, two values in the diabetic range are needed — HbA1c twice, or HbA1c plus at least one glucose measurement.
Why so strict. Glucose has high within-individual variability, poor pre-analytics (glycolysis in the tube if the plasma is not separated quickly), and is affected by food, stress and acute illness. HbA1c has lower variability, but it is distorted by other aspects of erythrocyte biology. One "almost at threshold" today may become "below threshold" tomorrow without any treatment. ADA advises: if the result is close to the threshold, explain the symptoms of hyperglycaemia to the person and repeat the test in 3–6 months.
Typical reasoning errors that the course treats as exam distractors.
First. "An HbA1c of 6.6% on the clinic's rapid device is already a diagnosis." ADA 2026 (2.2b): point-of-care A1C for screening and diagnosis is acceptable only on devices cleared by the US Food and Drug Administration (FDA) specifically for diagnosis, in laboratories certified under the Clinical Laboratory Improvement Amendments (CLIA) at moderate or higher complexity, by trained staff. A home device "for monitoring" does not make diagnoses.
Second. "An OGTT after three days on a keto diet will show diabetes more accurately." The opposite: carbohydrate restriction falsely raises 2-hour glucose.
Third. "A discrepancy between glucose and HbA1c means the laboratory made a mistake; we take the average." Recommendation 2.3: a persistent substantial discrepancy is a reason to look for a problem with, or interference in, either test, not to average them.
The boundary of the lesson. Confirmation protects a person from a false label. It must not delay treatment of a crisis. If a teaching case begins with Kussmaul breathing and a glucose of 28 mmol/L, a checkpoint that requires "repeat HbA1c in a week before starting therapy" tests not knowledge but harm.