HOMA-IR: the formula, the units and why cutoffs vary

HOMA-IR stands for homeostasis model assessment of insulin resistance. It was introduced by Matthews and colleagues in 1985. Their starting point was that fasting glucose and fasting insulin settle into a steady state set by the feedback loop between them. Comparing a patient's fasting values with what a computer-solved model of that loop predicts gives an estimate of insulin resistance and of beta-cell function.

In the original paper, the HOMA estimate of insulin resistance correlated with the euglycemic clamp, a more direct and far more demanding laboratory method (Rs = 0.88). The authors also reported low precision, with a coefficient of variation of 31% for the insulin resistance estimate, and wrote that this limits its use.

So HOMA-IR is not a direct insulin resistance test. It is an estimate built from two common fasting tests, which is why it is cheap, widely used and easy to misread.

The form most papers use is a simple product of the two fasting values, as written for example in the EPIRCE study of a Spanish adult population:

US laboratories usually report glucose in mg/dL. One mmol/L of glucose is about 18 mg/dL, and 22.5 × 18 = 405, so the same calculation with glucose in mg/dL is fasting insulin (µU/mL) × fasting glucose (mg/dL) ÷ 405. Both lines give the same result, so a HOMA-IR calculator only needs to know which glucose unit you entered.

Working it through by hand

1. Confirm both values come from the same fasting draw.
2. Check the insulin unit. The formula expects µU/mL. If your laboratory reports insulin in pmol/L, ask it which conversion factor to use before calculating.
3. Check the glucose unit and pick the matching divisor: 22.5 for mmol/L, 405 for mg/dL.
4. Multiply insulin by glucose, then divide by the divisor.

The literature does not offer one number for a "normal" HOMA-IR. Thresholds depend on who was studied, how the reference group was chosen and which insulin assay was used.

The EPIRCE authors put it plainly: there is great variability in the threshold HOMA-IR levels used to define insulin resistance, and cutoffs differ by ethnicity and clinical method. Their own thresholds had to be adjusted for age in non-diabetic women.

HOMA-IR inherits the weaknesses of its inputs, and insulin is the weaker one. In the Isokuortti study, samples from ten people were analyzed in seven European laboratories. A HOMA-IR of 2.0 measured in Helsinki corresponded to 1.3, 1.6, 1.8, 1.8, 2.0 and 2.1 in the other six. The inter-laboratory coefficient of variation was 25%, driven by insulin (25%) rather than glucose (5%). The authors concluded that standardizing insulin assays would be the first step toward defining normal HOMA-IR values.

Matli and colleagues compared three insulin immunoassays and found unsatisfactory correlation and systematic differences in the calculated HOMA-IR. They concluded that assay-specific reference intervals are required.

For a clinic tracking HOMA-IR over time, this has a practical consequence: a change between two draws at two different laboratories may reflect the assay, not the patient. Keeping a patient on the same laboratory, and noting when that changes, makes a trend easier to read.

• It assumes a fasting steady state. The model was built on basal, fasting glucose and insulin. A non-fasting sample breaks that assumption.
• It is imprecise in individuals. The original authors reported a 31% coefficient of variation for the insulin resistance estimate.
• Some groups need care. A 2010 review in World Journal of Diabetes notes that HOMA-IR may not suit patients whose beta-cell function is deranged, and that its use in patients treated with insulin needs further validation.

Many clinics read HOMA-IR next to its inputs rather than on its own. Fasting insulin and fasting glucose, followed across draws, show which side of the equation moved.

LongevityEHR is an EHR for cash-pay longevity, functional and concierge clinics. It does not interpret HOMA-IR or diagnose insulin resistance. It records the value, grades it against your clinic's band and keeps it on a trend next to fasting glucose and fasting insulin.

• The value your laboratory reports. When a report includes HOMA-IR, AI lab extraction reads it, matches it to your HOMA-IR biomarker and flags anything uncertain. LongevityEHR does not calculate the index from glucose and insulin today.
• Your clinic's band. Because no cutoff is universal, the clinic owner sets the optimal, borderline and critical bands for HOMA-IR, per sex if needed, and documents why. Our guide to optimal vs reference ranges covers how.
• Review before release. Extracted values wait in a review queue, and extracted values reach the patient portal only after a provider releases them.
• Trends that keep their meaning. Each released value keeps the range that applied that day, so editing your band later never regrades past results. See biomarker tracking.