A client came in last spring with eleven years of annual physicals in a folder. Every fasting glucose in the stack was normal. Every one. Her A1c had drifted from 5.2 to 5.9 over that decade, which her physician had described, accurately and unhelpfully, as "still fine." We ordered a fasting insulin. It came back at 24 µIU/mL.
That number is the whole story, and nothing in eleven years of bloodwork had shown it. Her glucose was normal because her pancreas was pouring out three or four times the ordinary amount of insulin to keep it there. The system was compensating beautifully, right up until the point where it would stop — and the standard panel was set up to notice only after it did.
What fasting glucose can and cannot tell you
Blood glucose is a regulated variable. Your body defends it hard, because both directions are dangerous, and it has a powerful lever to defend it with: insulin. So when tissues start responding poorly to insulin — the muscle and liver and fat cells getting a little deaf to the signal — the first thing that happens is not a rise in glucose. It is a rise in insulin. Glucose stays exactly where it was.
This is why fasting glucose is a late indicator. By the time it climbs into the prediabetic range, you have generally been insulin resistant for somewhere between eight and fifteen years, and beta-cell function has already declined measurably. The famous figure from the UK Prospective Diabetes Study is that roughly half of beta-cell capacity is gone at the moment of type 2 diagnosis. That capacity did not disappear in the six months before the appointment.
A normal fasting glucose does not tell you your metabolism is healthy. It tells you your metabolism is still winning.
The three cheap tests that do tell you
None of what follows is exotic, expensive, or new. All three are ordinary blood chemistry that any lab can run, and the reason they are not on your annual panel is convention rather than evidence.
- Fasting insulin. Drawn at the same time as your fasting glucose, from the same tube. Costs a lab a few dollars. Below about 8 µIU/mL is generally reassuring in an adult; above 12 warrants a conversation; above 20 is telling you something loudly.
- HOMA-IR. Not a separate draw — it is arithmetic on the two numbers you already have: fasting insulin multiplied by fasting glucose in mg/dL, divided by 405. Under 2.0 is broadly fine, over 2.9 indicates meaningful resistance. It costs nothing because it is a calculator, not a test.
- Triglyceride-to-HDL ratio. Also free, also arithmetic, also already sitting in your last lipid panel. It is a surprisingly good proxy for insulin sensitivity in people of European descent — less reliable in Black patients, where triglycerides run lower at any given level of insulin resistance, which is a real limitation and one I will not paper over.
What the number is not
Fasting insulin is a single-timepoint measurement of a hormone that is pulsatile, and assay standardisation between labs is genuinely poor — the same sample can return meaningfully different values from two facilities. So do not treat one draw as a verdict. Treat it as an opening. Repeat it at the same lab in three to six months and compare the trend, which is far more informative than any absolute value.
It also is not a diagnosis. Insulin resistance is not a disease you are given at a threshold; it is a spectrum you sit somewhere on, and where you sit is highly modifiable. This is the part I most want people to hear. The reason to look for it early is precisely that early is when it responds best — to resistance training, to protein and fiber at breakfast, to sleep, to reducing the intake of things that spike it hardest and most often.
What we did about that 24
Nothing dramatic, which is the point. Her breakfast — a bagel and juice, eaten in a car — became eggs and a piece of fruit. She started lifting twice a week, badly at first. We moved her fiber from about 12 grams a day to a bit over 30, mostly through beans and a different bread. She did not cut a food group and she did not count a calorie.
Nine months later her fasting insulin was 11. Her A1c was 5.4. Her weight had come down about fourteen pounds, which she cared about more than I did, and which was the least interesting number in the set. The important one was that her pancreas had stopped working a double shift.
809 words · 4 min read · published June 18, 2026
Naomi Okonjo-Bell, MS, RD, CDCES
Founder & Clinical Director
Founder of Copperleaf. Eight years in hospital diabetes education before deciding eleven minutes per patient was not enough.
Get the next one by email.
A fortnightly note from the dietitians here — what we published, one study worth your time, and one thing to cook.
Fortnightly. Double opt-in. Demo note: try an address ending in @example.com to see the failure state.