You have read that insulin resistance sits underneath most of PCOS, so you asked for it to be measured. One of two things then happened. You were told the test is not useful, with no explanation that made sense at the time. Or you were given a fasting insulin, the number came back, and nobody could tell you what to do with it.
Both of those are the same situation, and it is stranger than it looks: the mechanism is real and the measurement is not recommended. The 2023 international guideline holds both positions at once, deliberately, and once you see why, the refusal stops reading as a brush-off.
What the guideline actually says
Its position on measurement is one sentence, and it is unusually blunt for a document that hedges most things: it is not recommended to clinically measure insulin resistance, because of the limited clinical relevance of current insulin assays.
Note what that sentence does not say. It does not say insulin resistance is unimportant in PCOS — the same guideline treats metabolic risk as a reason to screen everyone with the diagnosis, at any weight. It says the available instruments are not good enough to make decisions with.
The difference matters because the two get confused constantly. “We do not measure that” is heard as “that is not happening to you”. What is being said is closer to: we cannot get a number reliable enough to act on, so we will measure the thing we can stand on instead.
Why an insulin number is harder than a glucose number
Glucose is a standardised measurement. Two laboratories using different machines will agree closely on the same sample, and the thresholds that define impaired fasting glucose and diabetes are agreed internationally and pinned to actual outcomes.
Insulin is none of that. Assays are not harmonised between laboratories, so the same blood can return materially different figures depending on who ran it, and there is no agreed cut-off at which a fasting insulin means anything specific. An index calculated from that number — HOMA-IR is the one you will meet most often — inherits the problem rather than fixing it: it is arithmetic performed on an unreliable input, and the result looks more precise than what went into it.
That is the whole of the argument. It is not that insulin is uninteresting. It is that a number with no agreed threshold, from an assay with no agreed calibration, is a figure you cannot act on and can easily be frightened by.
What each test is, and where the guideline puts it
| Test | What it measures | Where the guideline puts it |
|---|---|---|
| 75 g oral glucose tolerance test | How your blood glucose behaves over two hours after a measured sugar load | First choice for assessing glycaemic status |
| Fasting glucose | Blood glucose after an overnight fast, one reading | Less accurate than the OGTT, usable as an informed choice |
| HbA1c | Average glucose exposure over roughly three months | Less accurate again, usable as an informed choice |
| Fasting insulin | Circulating insulin on one morning | Not recommended for clinically measuring insulin resistance |
| HOMA-IR and similar indices | Calculated from fasting insulin and glucose | Same footing as the input it is calculated from |
The rows in the top half are the ones with agreed thresholds behind them. The rows in the bottom half are the ones people search for.
The test that is recommended, and what it involves
The guideline names the 75 g oral glucose tolerance test as first choice. You fast, a blood sample is taken, you drink a measured glucose solution, and blood is taken again two hours later. It is slower and more tedious than a single tube, which is a large part of why it gets substituted for something quicker.
It is also the test that catches what a fasting sample misses. Someone can hold a normal fasting glucose while handling a glucose load poorly, and impaired glucose tolerance is precisely that pattern. A fasting number alone reports the starting line and says nothing about the race.
Two further points the guideline makes about timing and eligibility are worth carrying into an appointment. Glycaemic status should be assessed at diagnosis and repeatedly afterwards, not once and filed. And it applies regardless of age and BMI — the increased risk of impaired fasting glucose, impaired glucose tolerance and type 2 diabetes is not restricted to women above a weight threshold, which is the assumption that keeps lean women with PCOS unscreened — and one of several reasons “just lose weight” is not a care plan.
What most articles get wrong about this
Search this topic and you will find a great deal of confident material about interpreting a fasting insulin — thresholds, ratios, target figures, what a result “should” be. Almost none of it mentions that the guideline recommends against the measurement, and some of it is selling something that the number is then used to justify.
A figure with no agreed threshold is exactly the kind of number that can be given a threshold by whoever wants one. That is not a conspiracy, it is just what happens to an unregulated measurement: it becomes available for any story, including the ones that end in a purchase.
The honest version is less satisfying and more useful. There is no number here that tells you where you stand. There is a recommended test that does tell you something, and it is a different test.
What to do with this at an appointment
If you went in to ask for a fasting insulin, the more productive question is a different one: has my glycaemic status been assessed, and when is it due again? That question has a recommended answer, a recommended test, and a recommended interval, which the insulin question does not.
If you already have a fasting insulin result sitting in a portal, the reasonable thing to do with it is very little. It is not nothing — it was measured, it exists — but it carries no threshold you can compare it against, and treating it as a score is the error the guideline is trying to prevent.
And if you are told metabolic screening is unnecessary because of your weight, that is the one place the guideline is unambiguous enough to quote back. Screening is asked for at any BMI.
What a record adds that a single test cannot
Screening gives you a point. It does not give you a line.
The things insulin resistance is suspected of doing day to day — the afternoon collapse, the appetite that does not track what you ate, the fatigue that arrives whatever the night’s sleep was like — are patterns across weeks, and nobody reconstructs weeks from memory in a ten-minute appointment. Logged as they happen, they are something a clinician can read. Described from recall, they compress into “I have been tired”, which is where the conversation usually ends.
That is not a substitute for the test. It is the other half: the test says where your glucose handling sat on one morning, and the record says what the months around it were actually like.