Metabolic health

What fasting insulin and HOMA-IR show years before glucose does

By Dr. D. R. Gamalathge, MBBS · SLMC No. 36999 · Published 30 September 2026 · Reviewed 23 October 2026 · 7 min read

In short. Fasting insulin rises years before fasting glucose does. HOMA-IR, calculated from the two, estimates how hard the body is working to keep glucose normal while the glucose itself still reads normal.

A man of forty-two has carried the same line on eight consecutive reports. Fasting glucose 94 mg/dL. Normal, every time.

His fasting insulin has never been drawn. Had it been, there is a fair chance it would have shown a climb across most of those eight years, because insulin is the number that moves first and glucose is the number that moves last.

What the two numbers are

Fasting insulin is measured directly, from the same fasting sample the glucose comes from. HOMA-IR is then worked out from the pair:

HOMA-IR = (fasting insulin in mU/L × fasting glucose in mmol/L) ÷ 22.5

Most Sri Lankan reports give glucose in mg/dL. The same calculation then reads:

HOMA-IR = (fasting insulin in mU/L × fasting glucose in mg/dL) ÷ 405

With a glucose of 94 mg/dL, an insulin of 6 mU/L gives a HOMA-IR of 1.4; an insulin of 18 mU/L gives 4.2.

That is the entire calculation. One extra assay on a sample already being taken, and two isolated values become an estimate of how hard the pancreas is having to work to hold the glucose where it is.

That is what insulin resistance is. Not a failure of glucose control, but the cost of maintaining it, and the cost is invisible on a report that only prints the glucose.

Why glucose is the last thing to move

The Whitehall II study followed 6,538 British civil servants who did not have diabetes and watched 505 of them develop it. Working backwards from each diagnosis, fasting glucose sat at 5.47 mmol/L (98 mg/dL) thirteen years out, and had crept only to 5.79 mmol/L (104 mg/dL) with three years still to go. Then, in those last three years, it went to 7.40 mmol/L (133 mg/dL).

Read that as a timeline rather than as a set of values. For a decade, the fasting glucose of someone heading towards diabetes is a reading no laboratory would flag and no doctor would mention. Insulin sensitivity, measured in the same people, was already falling steeply five years before diagnosis, while the glucose was still normal.

The glucose stays normal because insulin is being raised to keep it there. So the glucose reading is not the early signal. It is the evidence that the compensation has finally run out.

Normal HOMA-IR and fasting insulin ranges

There is no single agreed cut-off for HOMA-IR. Values between 2.0 and 3.0 are in common clinical use, and cut-offs derived in Asian populations generally fall lower.

Fasting insulin is worse served by its printed range. Most laboratories report 2–25 mU/L as normal, a span wide enough to contain a great deal of metabolic dysfunction, largely because it was derived from reference populations that already contained plenty of it. An insulin of 18 mU/L sits comfortably inside that range and is not a comfortable number.

Which is why the direction matters more than the label. A HOMA-IR of 2.1 that was 1.3 two years ago is a more useful piece of information than either value judged on its own against a range.

Who it is worth measuring in

Measuring it matters most wherever insulin resistance turns up at a body weight nobody would describe as excessive. A WHO expert consultation set a BMI of 23, not 25, as the point to act in Asian populations, and the waist limits for South Asians are 90 cm for men and 80 cm for women. Both are lower than the figures most people have in their heads.

Locally the arithmetic is blunt. In a 2025 Western Province survey (Katulanda et al.), 35.6% of adults were pre-diabetic and a further 27.8% already had diabetes. That leaves well under half the adult population handling glucose normally, and almost everyone in the first of those two groups feels entirely well.

It is also the test that explains a lipid profile which does not add up. Where ApoB and LDL disagree, the disagreement clusters in exactly this group: insulin resistance, triglycerides of 150 mg/dL or more, weight carried around the middle.

What the numbers do not do

HOMA-IR is a model, not a measurement. The reference method for insulin sensitivity is the euglycaemic clamp, which takes hours and belongs in research; HOMA-IR estimates it, closely across a population and more loosely in one person.

Insulin assays are also not standardised between laboratories, so a value from one is not interchangeable with a value from another. Serial readings are worth considerably more when they come from the same place.

And neither number diagnoses anything. A raised HOMA-IR is not diabetes and is not treated as diabetes. What it describes is a direction of travel, which is the honest limit of the claim: whether acting on a raised HOMA-IR in someone whose glucose is still normal changes hard outcomes has not been settled by a trial, and the case for measuring it rests on the length of the warning rather than on trial evidence that measuring it saves lives.

HbA1c is later again, because it moves only when glucose has moved. What an HbA1c of 5.7 to 6.4 means is covered separately.

Fasting and timing for a fasting insulin test

It needs a real fast of eight to twelve hours, water only. Insulin responds to food far more sharply than glucose does, so a sample taken after morning tea cannot be interpreted.

It is not part of a standard panel in Sri Lanka and has to be asked for by name, alongside the fasting glucose it is calculated with. One measurement establishes where things stand. A second, some months later and from the same laboratory, establishes which way they are going.

Frequently asked questions

Can you have insulin resistance with a normal fasting glucose?

Yes, and that is the usual sequence rather than the exception. Insulin rises first to hold glucose steady, so the glucose reading stays inside the reference range for years while insulin sensitivity is already falling. In the Whitehall II cohort, insulin sensitivity was declining steeply five years before diabetes was diagnosed, while fasting glucose was still normal.

Does HOMA-IR need a separate blood test?

No. It is calculated from fasting insulin and fasting glucose measured on the same sample, so it adds one assay rather than one appointment. The fast has to be genuine, however: eight to twelve hours, water only.

What counts as a normal HOMA-IR?

There is no single agreed figure. Values between 2.0 and 3.0 are in common clinical use, while cut-offs derived in Asian populations generally sit lower. The value is more useful read against a previous result from the same laboratory than against a printed range.

Is HbA1c not enough?

HbA1c reflects average glucose over the past two to three months, so it moves when glucose moves, which is late. It answers a different question from fasting insulin and does not replace it.

---

If a report has already come back clear, the question worth asking is not whether anything was flagged. It is which numbers were on it. Fasting insulin and HOMA-IR sit outside the standard panel in this country, which means a normal glucose is usually the only thing anybody has looked at — and it is the one that changes last. Fasting insulin, with HOMA-IR calculated from it, is a physician addition on either programme.

This article is general information, not individual medical advice. It does not replace assessment by your treating physician. If you have symptoms — chest pain, breathlessness or any urgent concern — seek medical attention immediately (Suwa Seriya 1990).