HOMA-IR: what it is, how it is calculated and what the result means

HOMA-IR appears more and more often on laboratory results in Mexico, sometimes because the doctor asked for it and sometimes because the panel included it. It is an insulin resistance index: a calculation that combines two values from the same fasting blood test, glucose and insulin, to estimate how much effort it costs your body to keep glucose in place. The first thing to know about the HOMA index is that it is not a diagnosis but an estimate with wide margins; the second, that there is no single cut-off separating “normal” from “high”, because different populations and studies use different points. This text explains where the number comes from, how to calculate it yourself, which cut-offs research has proposed, including Mexican research, which other data give it meaning and what moves it without your metabolism having changed.

What HOMA-IR is and what it tries to measure

Insulin is the hormone that lets glucose pass from the blood into the cells. Insulin resistance is, in the words of the US National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), “a condition in which your body doesn’t respond to insulin the way it should”: the cells of muscle, fat and liver respond less, and the pancreas compensates by producing more.^[4] While the compensation is enough, glucose stays normal; when it stops being enough, glucose starts to rise. That is why insulin resistance and prediabetes almost never cause symptoms: most people feel nothing.^[4]

HOMA stands for Homeostasis Model Assessment, a model published in 1985 by David Matthews’ group in Oxford. The idea is simple: fasting glucose and insulin are not independent but “are determined by their interaction in a feedback loop”, and from those two figures a model can estimate how much resistance there is and how hard the beta cell of the pancreas is working.^[1] The authors compared their estimate with the reference method, the euglycaemic clamp, and found a high correlation (0.88 in 23 people), but also low precision: the coefficient of variation of the resistance estimate was close to 31 %.^[1] In plain words: the index works well for comparing groups and for research, and is imprecise for one person on one specific morning.

That is why the NIDDK says the test for insulin resistance “is primarily used only for research studies” and that many doctors do not order it.^[4] MedlinePlus adds that the main reason to measure insulin in blood is to investigate low glucose, that the measurement is not used to diagnose diabetes and that an insulin result is “usually not enough by itself”.^[6] If the index is already on your results, use it for what it is: a clue that organizes a conversation with your doctor, not a verdict.

How the HOMA index is calculated

The version used by Mexican laboratories, which report glucose in mg/dL, is this: HOMA-IR = fasting glucose (mg/dL) x fasting insulin (µU/mL) / 405.^[2] The original version of the model uses glucose in mmol/L and the constant 22.5: HOMA-IR = glucose (mmol/L) x insulin (µU/mL) / 22.5.^[1] They are the same formula, because 405 is 22.5 multiplied by 18, and 18 is the factor that converts mmol/L of glucose to mg/dL. Insulin may be written as µU/mL, µIU/mL or mIU/L; all three are the same unit.

An example with round numbers: glucose of 92 mg/dL and insulin of 10 µU/mL. Multiply 92 by 10 and you get 920; divide by 405 and the index is 2.27. If your laboratory reports glucose in mmol/L, 92 mg/dL equals 5.11 mmol/L; 5.11 times 10 is 51.1, and divided by 22.5 gives the same 2.27. The table gathers four cases so you can place yours:

Look at the third row. A glucose of 100 mg/dL sits right on the edge that separates normal from prediabetes, which the guidelines place between 100 and 125 mg/dL fasting.^[5] With an insulin of 15 the index is 3.70; with an insulin of 5 it would be 1.23. The index captures what glucose alone cannot see: how much insulin is being needed to hold that number. And the other way round: a glucose of 85 with an insulin of 5 gives 1.05, and nobody would look at it twice.

What the sample needs in order to count

Insulin is measured after a fast of 8 to 12 hours, without eating or drinking anything but water.^[6] The reference fasting glucose is taken in the morning, after at least 8 hours of fasting.^[5] Before the draw, tell the laboratory if you take biotin (vitamin B7) supplements, because they can alter the insulin measurement and you are usually asked to stop them at least a day before; also if you use insulin medicines or have been found to have insulin antibodies, which interfere with the result.^[6] And one detail that explains many discrepancies: insulin reference ranges vary from one laboratory to another, so two results from different laboratories cannot be compared directly.^[6] We wrote about that in “a reference range is not your personal norm”.

What HOMA-IR value is high: why there is no single cut-off

The original model assigned a resistance value of 1 to “normal-weight normal subjects aged under 35 years”.^[1] That 1 is the model’s starting point, not a clinical cut-off. Since then every population studied has produced its own number, and the number also depends on the method: taking the 75th or 90th percentile of a healthy group is not the same as looking for the point that best separates people with metabolic syndrome, or the one that best matches the clamp. The Mexican authors say it plainly: the cut-off “varies widely across populations”, with values ranging from 1.7 in a Japanese population to 3.8 in a French population, and “more studies are needed to determine cut-off points in the Mexican population”.^[2] A more recent study in Mexican adults agrees: “there is no consensus on optimal cutoff values” and population-specific references are needed.^[3]

Look at what happens with the same Mexican population in the same study: the cut-off is 2.72 if the question is “where does the reference group end” and 4.21 if the question is “which value best separates people who already have metabolic syndrome”.^[3] The 1.22 cut-off from the clamp study is very sensitive but not very specific: with a specificity of 48.8 %, about half of the people without insulin resistance would exceed it, so it serves to avoid missing cases, not to label anyone.^[2] That is why an index of 2.5 can be “high” on one report and “normal” on another without either being wrong.

With the available data, a prudent reading is this: below 2 you are within what healthy reference groups show, both in the original model and in the Mexican study of 765 adults; between 2 and 4 you are in the zone where the cut-offs of the different studies cross and context decides; above 4 the value exceeds even the highest cut-offs proposed for Mexican adults.^[1][2][3] If your report prints a cut-off, ask which study or population it comes from: the Mexican authors agree that validated local cut-offs are still missing.^[2][3]

The result only makes sense in context

Insulin resistance rarely comes alone. It usually appears together with abdominal fat, high triglycerides, low HDL, blood pressure in the upper band and glucose that is starting to rise, a cluster the guidelines call metabolic syndrome and that the doctor evaluates with medical history, physical examination and tests, not with a single number.^[7] The thresholds used by the US National Heart, Lung, and Blood Institute (NHLBI) are these:^[7]

• Waist: more than 40 inches (about 102 cm) in men or more than 35 inches (about 88 cm) in women. The NHLBI notes that professionals may use different values depending on the person’s ethnic background.^[7]
• Triglycerides: consistently above 150 mg/dL.^[7] If yours came out high, see high triglycerides: what it means and what to check.
• HDL cholesterol: below 50 mg/dL in women or 40 mg/dL in men.^[7]
• Blood pressure: consistently 130/85 mmHg or higher.^[7] The seven-day home average weighs more than one reading, as we explain in normal blood pressure.
• Fasting glucose: between 100 and 125 mg/dL is prediabetes, and 126 or more, confirmed with a second test, is the diabetes threshold; HbA1c marks prediabetes between 5.7 and 6.4 % and diabetes from 6.5 %.^[5] The detail is in high fasting glucose.
• History: relatives with diabetes, gestational diabetes, polycystic ovary syndrome, sleep apnea, long-term use of glucocorticoids, age 35 or older and Hispanic or Latino background are on the NIDDK’s list of risk factors.^[4]

Two scenarios with the same number. First: HOMA-IR of 2.8, stable waist of 84 cm, triglycerides of 90, blood pressure of 115/75 and HbA1c of 5.2 %. The index stands alone, it has nothing to talk to, and the reasonable thing is to keep it with its date and repeat it under good conditions later on. Second: the same 2.8 with a waist that grew four centimeters in a year, triglycerides of 190, blood pressure of 134/86 and fasting glucose of 104. Now the index is one more piece of a pattern that already meets several criteria for metabolic syndrome,^[7] and that pattern deserves an appointment without urgency but without postponing it. The same applies if the reason for ordering the index was that you cannot lose weight even though you try or that you get sleepy after meals: the index does not explain the symptom on its own, but together with the rest of the data it organizes the conversation.

What changes the result without changing your metabolism

The index multiplies two values that move from one hour to the next, and any deviation in one of them is multiplied too. This table gathers what most often distorts an isolated result and what to do in each case:

The practical conclusion is that HOMA-IR is the portrait of one morning, not of your life. A value that surprises you is kept with its context (time, hours of fasting, how you slept, whether you were ill or stressed, which medicines you were taking) and repeated under clean conditions before drawing conclusions. Two well-taken points months apart are worth more than one alone, however high or low it is.

What to do with the result and when to repeat it

• A single result and doubtful conditions: conclude nothing. Repeat glucose and insulin from the same sample, in the morning, after 8 to 12 hours of fasting, at the same laboratory and in a week without illness or unusual stress.^[5][6]
• Result above 2 with the rest in order: keep it with its date and add what is missing so it has context: waist, home blood pressure for seven days, triglycerides and HDL if you do not have them. Raise it at your next check-up; we wrote about what to include in that check-up in annual preventive check-up.
• High result with a metabolic context (grown waist, triglycerides of 150 or more, blood pressure of 130/85 or more, glucose of 100 to 125): book an appointment in the coming weeks. Your doctor will decide whether to add HbA1c, a lipid profile or a glucose tolerance test, and which cut-offs to interpret the index with.^[5][7]
• What the guidelines associate with less insulin resistance: in the NIH-funded diabetes prevention program, high-risk people who lost 5 to 7 % of their starting weight reduced their chance of developing diabetes, and the NIDDK notes that keeping a healthy weight and being physically active may help prevent or reverse insulin resistance.^[4] The WHO recommends at least 150 minutes of moderate activity a week.^[8] What to change and how much is a decision made with your doctor, with your data in front of you.
• When to repeat: the guidelines set no fixed frequency. What is useful is to repeat under the same conditions and at the same laboratory, months apart, to see the direction of change and not the noise of one morning.^[6]

Warning signs

Insulin resistance itself causes no symptoms, and most people with prediabetes feel nothing.^[4] What does demand not waiting for another calculation are the symptoms of high glucose described by the WHO: intense thirst, needing to urinate more often than usual, blurred vision, tiredness and unintentional weight loss.^[8] With any of them, book an appointment in the coming days, not at the next check-up.

• If you measure glucose at home and a reading goes above 240 mg/dL, the American Diabetes Association advises checking urine for ketones and not exercising if there are any; it is a value to discuss with the doctor the same day.^[9]
• High glucose together with shortness of breath, fruity-smelling breath, nausea and vomiting or a very dry mouth are signs of ketoacidosis, which “is life-threatening and needs immediate treatment”: emergency.^[9]
• A fasting glucose of 126 mg/dL or more, an HbA1c of 6.5 % or more or a glucose of 200 or more at any time with symptoms are the thresholds at which the guidelines speak of diabetes; not an emergency, but an appointment soon, and HOMA-IR stops being the question.^[5]
• Dark, thick, velvety patches in the folds of the neck, armpits or groin are called acanthosis nigricans; MedlinePlus links it to insulin resistance and advises consulting if it appears. Not urgent, but a concrete reason for the appointment.^[10]

What to bring to the doctor

An appointment about a high HOMA-IR usually lasts as long as it takes to explain the context. If you arrive with the context already organized, the conversation starts where it should.

1. The complete results, with glucose and insulin in their units, date and laboratory. If you have more than one, bring them all: the direction of change says more than the value.^[6]
2. The conditions of the sample: time, hours of fasting, whether you were ill or stressed, how you slept that week.^[5][9]
3. Recent waist and weight, and whether you have noticed your waist growing.^[7]
4. Home blood pressure: the seven-day average with an upper-arm device, if you have it.^[7]
5. Triglycerides, HDL and HbA1c, with dates, if you have them; if not, bring them as a question.^[5][7]
6. History: diabetes in the family, gestational diabetes, polycystic ovary syndrome, sleep apnea.^[4]
7. List of medicines and supplements, including corticosteroids, oral contraceptives and biotin.^[4][6]
8. Three questions: whether it is worth repeating the index and when, whether HbA1c or a glucose tolerance test makes more sense, and which cut-off your doctor uses to interpret the result.