The glucose problem

Standard NHS metabolic screening checks fasting glucose and, in some cases, HbA1c. In the UK, NICE treats a fasting glucose of 5.5–6.9 mmol/L (about 99–124 mg/dL) or an HbA1c of 42–47 mmol/mol (6.0–6.4%) as high risk of type 2 diabetes. Below those, you’re told everything is fine. Come back in a few years.

The problem is that glucose is a late-stage marker. In the Whitehall II study of British civil servants, insulin sensitivity fell steeply over the five years before people were diagnosed with diabetes, while fasting glucose rose sharply only in the last three (Tabák 2009).

What rises first — years before glucose — is insulin.

What insulin actually does

When you eat carbohydrates, your pancreas releases insulin to shuttle glucose into cells. In a healthy metabolic state, a small amount of insulin does this efficiently. There’s no standardised insulin assay and no agreed healthy range. I work to 18–48 pmol/L (3–8 μIU/mL): my own target, not a validated range.

When cells start to become resistant to insulin's signal — when they stop responding as sensitively as they should — the pancreas compensates by releasing more insulin to achieve the same effect. Fasting insulin creeps up. 8. Then 12. Then 18. Then 25.

All while fasting glucose looks completely normal.

This compensation phase can last many years. The pancreas is working harder and harder to maintain glucose in the normal range, and succeeding. So the standard test shows nothing wrong. Meanwhile, elevated insulin is driving inflammation, fat storage (particularly visceral abdominal fat), hormonal disruption, and cardiovascular risk — silently, for years.

Fasting insulin at 15–18 μIU/mL (about 90–108 pmol/L) with normal glucose isn’t a result I’d call normal. It is a compensation result. The system is working overtime to maintain appearances. The question is how much longer it can sustain that effort.

What HOMA-IR actually is

HOMA-IR stands for Homeostatic Model Assessment of Insulin Resistance. It's calculated from two numbers you probably already have on a blood test: fasting glucose and fasting insulin.

The formula is straightforward: fasting insulin (μIU/mL) × fasting glucose (mmol/L) ÷ 22.5 (Matthews 1985).

The result gives you a single number that reflects how much insulin your body is requiring to maintain normal glucose levels — an estimate of insulin resistance. In the original study it tracked the gold-standard clamp measurement closely (correlation 0.88), but a single result is imprecise (Matthews 1985).

HOMA-IR: how I read it
One result
A pointer
There’s no agreed cut-off. Published cut-offs differ between populations, and insulin assays differ between labs.
The trend
The useful signal
Rising over repeat tests, with fasting insulin rising, is what I act on.
Not a diagnosis
Context
The method’s own authors warn against reading HOMA as a diagnostic test (Wallace 2004).

Why isn't it on a standard blood panel?

Two reasons. First, fasting insulin isn't routinely measured by the NHS. It's not in the standard metabolic screening protocol. You can request it, but most GPs will only run it if there's a specific clinical indication — and "your glucose is normal but you have symptoms consistent with insulin resistance" generally doesn't qualify.

Second, even when both fasting glucose and fasting insulin are run, HOMA-IR is rarely calculated. The two numbers sit in different sections of the report. Nobody joins the dots.

This isn't a conspiracy. It's a system designed to diagnose disease, not optimise health. Insulin resistance that hasn't yet manifested as pre-diabetes isn't considered a disease state worth treating. By the time it does manifest, interventions are more expensive, less effective, and considerably more disruptive to the patient's life.

What symptoms suggest insulin resistance before glucose rises?

The classic presentation I see clinically is a cluster of symptoms that appears to have no obvious connection:

Energy crashes 90 to 120 minutes after meals — not hunger exactly, more a sudden drop in energy and concentration that feels disproportionate to what was eaten.

Waking at 2–4am — particularly common when the dysglycaemic pattern involves overnight glucose instability. Glucose drops, cortisol spikes to mobilise energy stores, the client wakes with racing thoughts and can't return to sleep.

Strong carbohydrate cravings — particularly in the afternoon, often for something sweet or starchy. The body is seeking a quick glucose fix to compensate for the crash.

Difficulty losing weight despite eating reasonably — chronically elevated insulin is a fat-storage signal. The body is receiving a constant instruction to store rather than burn.

Brain fog and poor concentration — the brain is highly glucose-dependent. When glucose delivery becomes erratic, cognitive function follows.

Fatigue that isn't explained by sleep quality — cellular energy production is compromised when insulin signalling is impaired.

None of these symptoms would trigger a metabolic screen in a standard consultation. They're more likely to result in a referral for depression, anxiety, or — if the patient is female and of a certain age — an assumption that it's perimenopause. The metabolic component goes undetected.

HOMA-IR in the context of the full picture

HOMA-IR is most useful not in isolation but in the context of other markers. In functional diagnostic nutrition practice, I look at the cluster:

Fasting insulin — the raw fuel gauge. I work to 18–48 pmol/L (3–8 μIU/mL), my own target; there’s no agreed range.

Fasting glucose — the later-stage marker. High risk starts at 5.5 mmol/L (99 mg/dL) for NICE and 5.6 mmol/L (100 mg/dL) for the ADA 2025. A normal glucose with a raised insulin is the pattern worth noticing.

HbA1c — the three-month average. High risk starts at 42 mmol/mol (6.0%) for NICE and 39 mmol/mol (5.7%) for the ADA; 48 (6.5%) is diabetes. In EPIC-Norfolk, risk of death rose steadily across the whole range in men, and was lowest below 31 mmol/mol (5%) (observational; Khaw 2001).

Triglycerides — often raised with insulin resistance. The metabolic syndrome definition counts 1.7 mmol/L (150 mg/dL) or above (Alberti 2009). The triglyceride-to-HDL ratio is useful as a trend, but there’s no agreed cut-off, and the number depends on whether the lab reports in mmol/L or mg/dL.

HOMA-IR — the calculated bridge between the glucose and insulin story.

Together, these five numbers tell a metabolic story that no single marker can tell alone. A client with fasting insulin of 14 μIU/mL (about 84 pmol/L), fasting glucose of 5.2 mmol/L (94 mg/dL), HbA1c of 36 mmol/mol (5.4%), triglycerides of 1.8 mmol/L (159 mg/dL) and a HOMA-IR of 3.2 isn’t the clean picture a standard panel suggests: the insulin is well above my band, and the triglycerides are above the metabolic syndrome cut-off.

What to do if your HOMA-IR is elevated

The intervention hierarchy matters here. Before reaching for supplements, the foundational variables need addressing:

Meal composition and sequencing — eating protein and fat before carbohydrate at each meal significantly blunts the insulin response. Not a diet, a sequencing strategy.

Movement timing — a 10 to 15 minute walk after meals is one of the most evidence-backed interventions for post-meal glucose management available. Simple, free, and immediately effective.

Sleep quality — a single night of poor sleep measurably impairs insulin sensitivity the following day. Chronically disrupted sleep drives insulin resistance independent of diet.

Stress management — cortisol raises blood glucose. Chronic stress chronically elevates cortisol. The metabolic and stress systems are not separate.

Reducing refined carbohydrate load — particularly liquid calories, refined sugars, and ultra-processed foods. The goal is not zero carbohydrate but appropriate carbohydrate relative to individual metabolic capacity, which varies significantly between people.

Only once these foundations are addressed does targeted supplementation become relevant. Berberine, magnesium, alpha lipoic acid, and inositol all have evidence in insulin resistance contexts — but they're Layer 7 on the intervention hierarchy, not Layer 1.

Want to know your HOMA-IR?
The TDG five-test programme includes a comprehensive blood chemistry panel covering fasting glucose, fasting insulin, HbA1c, and all lipid markers — with each range labelled by its source and cross-referencing against DUTCH hormone data, GI-MAP, and organic acids results. The metabolic picture doesn't exist in isolation.
Read Test, Don't Guess →

The bottom line

HOMA-IR is not a difficult test. It's not expensive. It requires two blood markers that are already available from any laboratory. The calculation takes thirty seconds.

The reason it isn't routinely run is not technical — it's systemic. Standard medical practice is designed to catch disease at the point it becomes diagnosable, not to identify the decade of dysfunction that precedes it.

Even after glucose crosses the threshold, the window is still open: in the Diabetes Prevention Program, people who already had raised glucose cut their risk of diabetes by 58% with lifestyle change (DPP 2002). But earlier is easier. A fasting insulin of 12 μIU/mL (about 72 pmol/L) with a glucose of 4.8 mmol/L (86 mg/dL) gives you a head start that waiting for the glucose to rise doesn’t.

That’s why I read fasting insulin alongside glucose: not to pathologise normal variation, but to see the trajectory before it becomes the destination.

Sources

Stephen Duncan
FDN-P · MSc · Working With People Since 1988 · Detective Health
Stephen Duncan is a Functional Diagnostic Nutrition Practitioner based in Edinburgh. He runs the TDG five-test programme through Detective Health and is the author of Test, Don't Guess. His clinical approach integrates blood chemistry interpretation, using guideline thresholds and his own clearly labelled working ranges, with GI-MAP, DUTCH Plus, organic acids, and food sensitivity testing.