Explainer · Blood

HbA1c, fasting glucose, and insulin, read together

Each marker alone can look normal while the pattern drifts. How the three combine into an early read on metabolic health.

October 8, 20266 min readBy the layers team

Blood sugar is a defended quantity. Your pancreas raises insulin as tissues become resistant to it, and for years that compensation holds glucose in the normal range. Reading glucose alone therefore tells you how the defense is doing, not how much it costs.

The three markers of a metabolic panel each sample a different layer of that system, on a different timescale. Read together, they can place you on a trajectory; read alone, each can reassure falsely.

Three clocks

Fasting glucose is the shortest clock: hours. It mostly reflects overnight glucose output from the liver, and it carries day-to-day noise from sleep, stress, and illness. HbA1c is the long clock: red blood cells live about 120 days, and glucose glycates their hemoglobin in proportion to exposure, so the value integrates roughly eight to twelve weeks, weighted toward the most recent month. Fasting insulin is not a clock so much as a load gauge: the concentration required, this morning, to hold your glucose where it is.

The cutoffs

The American Diabetes Association draws the lines at fasting glucose of 100 to 125 mg/dL or HbA1c of 5.7 to 6.4% for prediabetes, and at 126 mg/dL or 6.5% for diabetes, with any diagnosis confirmed by a second abnormal result. The two tests disagree in a meaningful fraction of people, which is expected: they are not measuring the same thing.

When they disagree

Discordance usually points at the red cells rather than the sugar. Anything that shortens red cell lifespan, such as hemolysis, recent transfusion, or late pregnancy, lowers HbA1c falsely because cells are replaced before they glycate. Iron deficiency skews the population of red cells older and raises HbA1c falsely; it often drops after iron repletion with no change in glucose. Hemoglobin variants can interfere with specific assays. The ADA's instruction for these cases is simple: trust the plasma glucose.

The earliest signal

In the Whitehall II cohort, 6,538 adults were followed for up to thirteen years before some developed diabetes. Fasting glucose drifted only slightly for most of that time, then rose steeply in the final three years before diagnosis. Insulin sensitivity, by contrast, was falling steadily five years out, while the pancreas compensated with more insulin until compensation failed. The population pattern is the point: by the time glucose moves, the process is old.

That is why fasting insulin, for all its imperfections, belongs on the panel. It rises during the years glucose still looks fine, so it is the number most likely to change what you do this year rather than next decade.

Reading HOMA-IR carefully

HOMA-IR multiplies fasting glucose by fasting insulin against a healthy-reference constant, and it correlates well with the gold-standard clamp in validation work. Its limits are real: insulin assays differ between labs, the index mostly reflects liver rather than muscle insulin sensitivity, and single values are noisy. It works best the way we use every marker here: same lab, same conditions, tracked as a trend rather than judged as a one-off verdict.

Sources

Keep reading