Insulin resistance

What the research found.

• One study took a random sample of 2,459 Spanish adults and set a HOMA-IR threshold for insulin resistance two different ways. Taking the top tenth of the group gave 3.46. Counting the parts of metabolic syndrome instead gave 2.05, which means a result of 2.5 gets you labelled resistant under one rule and unremarkable under the other.

  Gayoso-Diz and colleagues, BMC Endocrine Disorders, 2013

• Splitting that group by sex and age moved the best value again, so the label depends partly on who is doing the labelling. Men without diabetes came out at 1.85 by one rule. Women without diabetes ran from 2.07 at age 50 to 2.47 at age 70, and every one of those values fell between the 70th and 75th percentiles, meaning 25 to 30 of every 100 ordinary adults sit above the line.

  Gayoso-Diz and colleagues, BMC Endocrine Disorders, 2013

• One study of 413 people with rheumatoid arthritis or lupus found the rheumatoid arthritis patients more insulin resistant than the lupus patients. Insulin sensitivity was lower by 27 points, with the true difference falling somewhere between 9 points and 46 points lower. That range never reaches zero, so the direction is settled.

  Quevedo-Abeledo and colleagues, Journal of Rheumatology, 2020

• In that study the odds of meeting the definition were 2.15 times higher in rheumatoid arthritis than in lupus. The range around it runs from 1.25 to 3.69. That range is the honest part, because the real figure could be a modest quarter higher or close to four times higher, and 413 people aren't enough to settle which.

  Quevedo-Abeledo and colleagues, Journal of Rheumatology, 2020

What the term describes

Insulin's job is to move glucose out of your blood and into the tissue that needs it, and insulin resistance means that process has started working less efficiently than it should. Your body answers by making more insulin, which for years keeps your glucose where it belongs and keeps everything looking unremarkable from outside. What the phrase describes is a range rather than a condition you either have or don't.

It underlies a range of metabolic problems without ever being a disease in its own right, which is part of why the term gets used so loosely. It appears on a rheumatology site for two reasons worth separating. Inflammatory arthritis seems to come with more of it, and heart risk is raised in nearly every inflammatory rheumatic condition, so metabolic measurements belong in that wider picture whether or not anybody mentions your joints.

Measuring insulin resistance directly needs a research procedure rather than a blood draw, so in ordinary practice it gets estimated. The usual estimate is called HOMA-IR, worked out from a fasting glucose and a fasting insulin taken together. Being an estimate built from two numbers, it inherits whatever affects either of them, including how long you'd really fasted before the needle went in.

The number that has no normal value

Here's the most useful thing to know about HOMA-IR, and it surprises nearly everybody who has been handed one. There isn't an agreed threshold. That's not a hole in the literature waiting for somebody to fill it, it follows from how the boundary has to be set in the first place. No outcome defines where insulin resistance begins, so the line has to be drawn statistically, and reasonable people draw it in different places.

One study took a random sample of 2,459 Spanish adults aged 20 to 92 and showed how much that choice changes things. Taking the top tenth of the group put the threshold at 3.46. Counting the parts of metabolic syndrome instead put it at 2.05, so picture a woman whose result comes back at 2.5: under the second rule she's insulin resistant, under the first she's unremarkable, and her blood test never changed.

Splitting that group by sex and age moved the value yet again. Men without diabetes came out at 1.85 by one rule and 2.27 by another, while women without diabetes ran from 2.07 at age 50 up to 2.47 at age 70. People with diabetes came out lower still, at 1.60 in men and 1.58 in women.

What that means for reading your own result

Every one of those best values fell between the 70th and 75th percentiles of the group, and that translation is the part worth having. It means the boundary gets drawn low enough that 25 to 30 of every 100 ordinary adults sit above it. Being called insulin resistant puts you in company with roughly a quarter of the adults around you, which isn't the impression most people take from the phrase.

The practical result follows from that. A HOMA-IR figure quoted without the rule behind it isn't telling you much, so when somebody says your result is high, the two questions worth asking are high against which threshold, and whether that threshold came from people like you. Anywhere between 2.05 and 3.46 your label depends on who is doing the labelling rather than on anything happening in your body.

None of that makes the calculation useless, and I don't want to leave you thinking it does. HOMA-IR is a reasonable tool for describing a population and a rough one for describing a person, which is an ordinary state of affairs for any estimate. The error worth avoiding is treating one figure as a verdict about you.

Why it appears on a rheumatology site

One study in our sources compared insulin resistance between two autoimmune diseases seen in one clinic, covering 413 people, of whom 227 had rheumatoid arthritis and 186 had lupus. The rheumatoid arthritis patients were more insulin resistant. Insulin sensitivity was lower by 27 points, with the true difference falling somewhere between 9 points and 46 points lower, at a P value of 0.004.

That range is worth understanding, because it's how these studies tell you what they do and don't know. The best estimate is 27 points, and the honest answer is that the real difference lies somewhere between 9 and 46. Since the range never reaches zero, the direction is settled: rheumatoid arthritis patients really were less sensitive, even though how much less is still open.

Those patients also more often met the definition of insulin resistance outright, at odds 2.15 times higher, across a range from 1.25 to 3.69. Beta cell function went the other way, coming out higher in rheumatoid arthritis by 38 points, from 23 to 52, at a P value below 0.001. Everybody in that study was measured at a single moment rather than followed over years, so it shows you a real difference between two groups of patients and leaves the reason to a rheumatologist who knows the person in front of her.

The problem with that odds ratio

Our own record for that study flags one thing as critical, and it's the difficulty the threshold data already described. The definition applied to those 413 patients was imported from general adults rather than worked out in people with inflammatory arthritis. It was HOMA2-IR at or above 1.85 in men, and in women it was above 2.36 at age 30, above 2.07 at age 50, and above 2.47 at age 70.

A line drawn in one population and applied to another can invent a difference or bury one, which is why this page raises it rather than quoting the headline and moving on. People with inflammatory arthritis may differ from general adults in body composition, activity, or medication, and if they do, a line drawn elsewhere won't sit where it should on their spread. The figure of 2.15 depends entirely on where that line was put.

That isn't a reason to dismiss the finding, and I'd rather you didn't. The continuous measures pointed in that direction too without depending on any threshold, so insulin sensitivity lower by 27 points is a difference in what was measured rather than a difference in which box somebody was sorted into. Quote the continuous result alongside the odds ratio, because the odds ratio on its own tells you less than it appears to.

Before you have the test

The question nearly everybody brings is what moves insulin resistance, and this site holds no treatment source for it in autoimmune disease. There's no honest figure here for what a particular diet does to a HOMA-IR result in rheumatoid arthritis, or what an exercise program does, or what a supplement does. That absence is recorded rather than filled in with an estimate, because an estimate dressed as a finding is the thing this site exists to avoid.

The threshold moves between 2.05 and 3.46 depending on the rule chosen, and no outcome study in our sources ties a HOMA-IR value to anything that happens later in autoimmune disease. A measurement with no agreed boundary and nothing attached to it downstream is genuinely hard to act on. You deserve to know that before the blood is drawn rather than afterwards.

So here's the question worth asking first. What would change as a result of the number? If the answer is a conversation about activity and diet that would be worth having anyway, then the test adds confirmation rather than information, which is fine as long as everybody knows that's what it's doing. If somebody has promised you a specific drop from a specific thing, asking them for the reference is entirely reasonable.

Common misconceptions.

Related terms.

• HOMA-IR, the calculation itself and what goes into it.
• HOMA-IR calculator, which shows what your own fasting numbers produce.
• Rheumatoid arthritis and blood tests, for where this sits among the other measurements.
• Inflammation, and the difference between the kind you can measure and the kind you have.

Questions patients ask.