Homeostatic Model Assessment of Insulin Resistance (HOMA-IR)

HOMA-IR estimates insulin resistance from a single fasting blood draw. In the original units it is fasting insulin in microunits per millilitre multiplied by fasting glucose in milligrams per decilitre, divided by 405; with glucose in millimoles per litre the divisor is 22.5, each constant chosen so that a lean healthy young adult scores about 1. The companion output of the same 1985 structural model, HOMA-B, estimates beta cell function from the same two numbers. Both are algebraic outputs of a model of the glucose-insulin feedback loop, not measurements.

Against the hyperinsulinaemic-euglycaemic clamp, the reference method, HOMA-IR tracks well at group level and loosely for individuals, and no cut point is agreed across populations. Insulin immunoassays are not standardised between laboratories, so the same serum returns different values on different platforms, and the index inherits that spread. An updated computer model, HOMA2, handles the non-linearity of the original equation better but is used far less often.

The attraction is cost. Two fasting analytes instead of a multi-hour clamp makes the index usable in cohorts of thousands. That is also its limit: it is an instrument for comparing groups or tracking one group over time, not a test that assigns a person a label, and any threshold quoted for it belongs to the population and assay it was derived in.

Two errors recur. The first is treating a fall in HOMA-IR as evidence that a compound is insulin-sensitising, when weight loss of any origin lowers it and the index cannot separate a direct drug effect from that. The second is reading a low value as reassurance in established type 2 diabetes, where failing beta cells secrete less insulin; the numerator falls, the index improves, and the patient has not.