{"slug":"homa-ir","name":"HOMA-IR","aliases":["Homeostatic model assessment of insulin resistance"],"kind":"biomarker","system":"metabolomics","unit":"index","unitNote":"Fasting insulin (µIU/mL) x fasting glucose (mg/dL) / 405, or insulin x glucose (mmol/L) / 22.5. This is the original 1985 approximation (HOMA1). The Oxford HOMA2 calculator uses the updated computer model and gives different, usually lower, numbers for the same sample.","answer":"HOMA-IR is a calculation, not a measurement, and no guideline defines a normal or optimal value. In the one US cohort that followed adults without diabetes to death, all-cause mortality was lowest in the quarter at or below about 1.4 and 64 percent higher above about 2.8, and cutoffs that separated healthy from at-risk adults in other populations fell between about 1.8 and 2.6, all well under the 2.0 to 2.9 that some laboratories print.","summary":"HOMA-IR multiplies fasting insulin by fasting glucose to estimate how hard the body is working to hold a normal fasting sugar. It was built as a research model in 1985 and tracks the gold-standard clamp reasonably well in groups, less well in one lean person. Most laboratories do not report it; where they do, the cutoff is about 2.0 to 2.9, while the cohorts place the lowest risk at or below about 1.4 and population cutoffs at about 1.8 to 2.6.","whatItIs":[{"text":"HOMA-IR is the product of fasting insulin and fasting glucose, scaled so that the two together describe how much insulin the body is spending to keep the fasting sugar where it is. It was described in 1985 by Matthews and colleagues at Oxford, who built a computer model of the feedback loop between glucose and insulin and published an approximate formula alongside it. In their original comparison, the model's estimate of insulin resistance correlated 0.88 with the euglycemic clamp and 0.81 with the fasting insulin alone, with a coefficient of variation of 31 percent that the authors themselves said limited its use.","cites":["matthews1985","wallace2004"],"grade":"B"},{"text":"The index has been checked against the clamp many times since. In 115 Italian adults spanning normal glucose tolerance to diabetes, HOMA-estimated insulin sensitivity correlated 0.82 with clamp-measured glucose disposal, and the two methods agreed moderately well when people were sorted into insulin-sensitivity categories (weighted kappa 0.63). The authors concluded that HOMA can be relied on in large or epidemiological studies where only a fasting sample exists, which is a narrower endorsement than it is often given.","cites":["bonora2000"],"grade":"B"},{"text":"Because fasting glucose varies little in people without diabetes, HOMA-IR is in practice a scaled fasting insulin: the two correlated 0.81 in the original paper, and the Oxford group's own guidance is that the index is appropriate for cohort and epidemiological work and that its primary inputs, especially the insulin assay, need to be robust before the number means anything. An insulin result in µIU/mL at a glucose of 90 mg/dL maps to HOMA-IR by dividing by 4.5, and the fasting insulin entry carries the cohorts that measured insulin directly.","cites":["matthews1985","wallace2004","widjaja1999"],"grade":"B"}],"labRange":{"typical":"Most laboratories do not report HOMA-IR. Where one does, the cutoff printed is usually about 2.0 to 2.9, and the number depends on which insulin assay the laboratory runs","claims":[{"text":"There is no agreed threshold. In a random sample of 2,459 Spanish adults aged 20 to 92, the cutoff moved from 3.46 when defined as the 90th percentile of the population to 2.05 when chosen to identify people carrying the components of the metabolic syndrome, and to 1.85 in men without diabetes; all of these fell between the 70th and 75th percentiles of the population, and in women the best cutoff changed with age in a way it did not in men. A cross-sectional study, so it describes who the index flags, not what happened to them.","cites":["gayoso2013"],"grade":"C"},{"text":"The cutoff also depends on who was sampled. Among 65 healthy Spanish hospital staff aged 30 to 60 with normal glucose, lipids and chemistry, the 75th percentile of HOMA-IR was 2.6, corresponding to a fasting insulin of 12. In 1,854 adults of Mexican descent in Brownsville, Texas, a population with a heavy burden of diabetes, a machine-learning definition put the best cutoff at 3.80, and 39.1 percent of that population sat above it. The same index, two populations, cutoffs that differ by half.","cites":["ascaso2003","qu2011"],"grade":"C"},{"text":"Whatever cutoff a laboratory prints applies to its own insulin assay. When the American Diabetes Association workgroup ran the same samples through 12 commercial insulin methods from 9 manufacturers, results differed among assays by a median of 24 percent (range 12 to 66), and a common reference preparation did not bring them into line. A HOMA-IR of 2.0 on one assay can be about 1.6 or 2.5 on another, which is why a cutoff derived in one study does not transfer to a different laboratory.","cites":["marcovina2007"],"grade":"A"}]},"evidence":{"headline":"In US adults without diabetes, all-cause mortality was lowest in the quartile at or below 1.4 and 64 percent higher above 2.8, with the excess confined to people of normal weight. Across 65 cohorts, HOMA-IR predicted coronary disease more strongly per standard deviation than fasting glucose or fasting insulin alone. Cutoffs that separated healthy from at-risk adults in population studies ran from about 1.8 to 2.6. No trial has treated to a HOMA-IR target.","scale":{"min":0,"max":6,"ticks":[0,1,2,3,4,5,6]},"bands":[{"label":"Where a laboratory reports it, the printed cutoff is about 2.0 to 2.9","low":0,"high":2.9,"tone":"lab","source":"Typical US laboratories, where reported"},{"label":"Lowest-mortality quartile in US adults without diabetes","low":0,"high":1.4,"tone":"evidence","source":"Ausk 2010"},{"label":"Cutoffs separating healthy from at-risk adults in population studies","low":1.8,"high":2.6,"tone":"neutral","source":"Esteghamati 2010; Ghasemi 2015; Moon 2018; Ascaso 2003; Gayoso-Diz 2013"},{"label":"Second and third quartiles: cardiovascular mortality already higher than the lowest quartile","low":1.4,"high":2.8,"tone":"neutral","source":"Ausk 2010"},{"label":"Top quartile: 64 percent higher all-cause mortality, in people of normal weight","low":2.8,"high":6,"tone":"caution","source":"Ausk 2010"}],"claims":[{"text":"In 5,511 US adults without diabetes from NHANES III (1988 to 1994), followed for a mean of 8.5 years with 643 deaths, HOMA-IR quartiles were 1.4 or less, above 1.4 to 2.0, above 2.0 to 2.8, and above 2.8. After adjustment for age, sex, body mass index, waist-to-hip ratio, alcohol, ethnicity, education, smoking, activity, C-reactive protein, blood pressure, cholesterol and triglycerides, the top quartile had 1.64 times the all-cause mortality of the bottom, and each step up a quartile carried 1.16 times the risk. The association held only in people with a body mass index below the median of 25.2, where the third quartile carried 1.73 times and the top quartile 1.86 times the mortality of the lowest; in heavier people the index added nothing. Cardiovascular mortality appeared uniformly higher in the three upper quartiles rather than rising in a line, and HOMA-IR did not predict cancer deaths.","cites":["ausk2010"],"grade":"B"},{"text":"In the San Antonio Heart Study, 2,569 Mexican-American and non-Hispanic white adults free of diabetes at baseline were followed for 8 years, during which 187 had a cardiovascular event. Risk rose across quintiles of HOMA-IR: the top fifth had 2.52 times the odds of the bottom fifth after adjustment for age, sex and ethnicity, and 1.94 times after further adjustment for LDL, triglycerides, HDL, blood pressure, smoking, alcohol, exercise and waist circumference. The association did not differ by ethnicity, sex, hypertension, glucose tolerance or obesity, and fasting insulin alone behaved the same way.","cites":["hanley2002"],"grade":"B"},{"text":"Pooling 65 cohort and nested case-control studies with 516,325 participants without diabetes, the relative risk of coronary heart disease comparing high with low values was 1.64 for HOMA-IR, 1.52 for fasting glucose and 1.12 for fasting insulin. Per standard deviation the gap was wider: 1.46 for HOMA-IR against 1.21 for glucose and 1.04 for insulin, with no measurable heterogeneity among the HOMA-IR studies. The authors suggested the index may be worth adding to cardiovascular risk prediction; no one has yet shown that doing so changes decisions.","cites":["gast2012"],"grade":"A"},{"text":"In the Tehran Lipid and Glucose Study, 4,942 Iranian adults aged 20 to 86 were followed for 9.2 years, during which 346 (7.0 percent) developed type 2 diabetes. HOMA-IR, HOMA-S and fasting insulin each predicted diabetes independently of other risk factors, and the cutoffs that best separated future cases from non-cases were 1.85 in women and 2.17 in men for the original HOMA1-IR, but 1.41 and 1.18 for the computer-model HOMA2-IR, and 11.13 and 9.16 µIU/mL for fasting insulin alone. The same cohort, the same people, and a cutoff that moves by a third depending on which version of the model is run.","cites":["ghasemi2015"],"grade":"B"},{"text":"Cross-sectional studies that pick the HOMA-IR cutoff best separating adults with and without the metabolic syndrome land in a similar place. In 3,071 Iranian adults aged 25 to 64, the best cutoff in people without diabetes was 1.775, with a sensitivity of 57.3 percent and a specificity of 65.3 percent and an area under the curve of 0.683, which is modest discrimination. In 10,997 Korean adults, the cutoffs were 2.20 in men, 2.55 in premenopausal women and 2.03 in postmenopausal women; in the same sample the triglyceride-glucose index discriminated better (area under the curve 0.877 against 0.737 for HOMA-IR). These describe who the index flags today, not who went on to get sick.","cites":["esteghamati2010","moon2018"],"grade":"C"}]},"levers":[{"name":"Weight loss of about 5 percent","effect":"lowers it","claim":{"text":"In a randomized study of adults with obesity, losing 5 percent of body weight improved insulin sensitivity in liver, muscle and fat tissue and improved beta-cell function, and further loss to about 11 and 16 percent improved muscle insulin sensitivity further. A small study (19 people reached the 5 percent loss, against 14 who held their weight), but one that used clamp and tracer measurements rather than a fasting index, so it shows the physiology that HOMA-IR approximates.","cites":["magkos2016"],"grade":"B"}},{"name":"Exercise, including interval training","effect":"lowers insulin resistance","claim":{"text":"A meta-analysis of 50 studies found that high-intensity interval training reduced insulin resistance, pooled across the indices the trials reported, compared with no exercise (standardized mean difference -0.49) and compared with continuous moderate training (-0.35), and lowered body weight by 1.3 kg. In people at risk of or with type 2 diabetes it also lowered fasting glucose by about 17 mg/dL (0.92 mmol/L).","cites":["jelleyman2015"],"grade":"A"}},{"name":"Carbohydrate restriction","effect":"lowers it","claim":{"text":"Across 149 randomized trials with 9,104 adults in 28 countries, carbohydrate-restricted diets lowered HOMA-IR by 0.54 (95 percent confidence interval 0.33 to 0.75) against comparison diets, with fasting glucose about 3 mg/dL and fasting insulin about 8 pmol/L (about 1.2 µIU/mL) lower. The effect was larger in women, in people with overweight or obesity and in people with type 2 diabetes, and was similar whether or not calories were matched, which suggests the composition itself matters. The pooled trials range from ketogenic to moderate restriction and most were short.","cites":["feng2025"],"grade":"A"}},{"name":"A Mediterranean dietary pattern","effect":"lowers it","claim":{"text":"Pooling 57 controlled trials with 36,983 adults, a Mediterranean diet improved 18 of 28 metabolic-syndrome components and risk factors against comparison diets, HOMA-IR, fasting glucose and fasting insulin among them, and reduced cardiovascular disease incidence (risk ratio 0.61). Heterogeneity was high for several outcomes and the comparators varied, so the direction is firmer than the size.","cites":["papadaki2020"],"grade":"A"}},{"name":"Time-restricted eating","effect":"lowers it slightly","claim":{"text":"Across 23 randomized trials with 1,280 participants, an 8-hour eating window lowered HOMA-IR by a standardized mean difference of 0.16, fasting glucose by 0.25 and fasting insulin by 0.22 against a control diet: real but small effects, with trials longer than 6 months also lowering hemoglobin A1c. The subgroups reported were sex, activity level and trial duration, not starting weight, so the effect in lean adults with a low HOMA-IR to begin with is not shown.","cites":["wong2026"],"grade":"B"}},{"name":"Sleep","effect":"short sleep raises insulin resistance","claim":{"text":"When 11 healthy young men were restricted to 4 hours in bed for 6 nights, glucose tolerance fell, evening cortisol rose and sympathetic nervous activity increased, all of which reversed with recovery sleep. The study measured glucose tolerance rather than HOMA-IR, and it was small and short, but it shows that a week of poor sleep is enough to move the inputs to this index.","cites":["spiegel1999"],"grade":"C"}}],"caveats":[{"text":"HOMA-IR is not portable between laboratories. The insulin assay is the dominant input, and commercial insulin assays disagreed by a median of 24 percent in the American Diabetes Association's comparison, so a change in the index after switching laboratories is often a change in method, not in you. The Oxford group that built the model warns that the primary input data need to be robust and interpreted with care. Compare HOMA-IR values only from the same laboratory, and read the fasting insulin and glucose that produced them.","cites":["marcovina2007","wallace2004"],"grade":"A"},{"text":"A single HOMA-IR is noisy, because a single fasting insulin is noisy. In 12 healthy adults sampled on 12 consecutive mornings, fasting insulin varied by 26 percent within a person from day to day, against 4.8 percent for glucose; the original model's authors put its coefficient of variation at 31 percent. When 11 postmenopausal women without diabetes were sampled 10 times at 4-day intervals, their mean HOMA-IR was 2.11 with a standard deviation of 0.79, and across that study a repeat value had to rise by more than 90 percent or fall by more than 47 percent to count as a real change. A move from 1.6 to 2.2 between two draws is inside the noise.","cites":["widjaja1999","matthews1985","jayagopal2002"],"grade":"B"},{"text":"The index fits lean people worst, which is also where the mortality signal was found. Among 485 healthy volunteers tested with a direct measure of insulin-mediated glucose uptake, HOMA-IR correlated 0.36 with the direct measure in people of normal weight, 0.55 in overweight and 0.60 in obese people; fasting-based indices explained no more than 13 percent of the variation in insulin action in the normal-weight group. In 90 Korean adults, the correlation with the clamp was weaker below a body mass index of 25 (0.44 against 0.62), weaker with lower beta-cell function and weaker at higher fasting glucose. In a lean person a HOMA-IR near a cutoff says less than it would in a heavier one.","cites":["kim2004","kang2005"],"grade":"B"},{"text":"The index assumes a pancreas that still responds. As beta-cell output falls in established diabetes, insulin drops while glucose rises, and the product can understate resistance or sit at a value that looks reassuring; the clamp correlation was 0.70 in people with diabetes against 0.75 without in one study, and the index should not be used to assess beta-cell function on its own. HOMA-IR is a tool for people without diabetes, read alongside the glucose and insulin it was made from.","cites":["bonora2000","kang2005","wallace2004"],"grade":"C"},{"text":"There are two HOMAs. The 1985 formula (HOMA1) is what laboratories and most papers use; the updated computer model (HOMA2), available from Oxford, corrects for several simplifications and has not been as widely adopted. They give different numbers for the same sample: in the Tehran cohort the diabetes-predicting cutoffs were 1.85 and 2.17 by HOMA1-IR and 1.41 and 1.18 by HOMA2-IR. A cutoff from one cannot be applied to the other, and a reader should know which version a calculator or a report used.","cites":["wallace2004","ghasemi2015"],"grade":"C"}],"seeAPhysician":["A fasting glucose of 100 mg/dL or higher, or a hemoglobin A1c in the prediabetes range, on standard testing, whatever the HOMA-IR works out to.","A high HOMA-IR driven by a high fasting glucose with a low fasting insulin, which can mean the pancreas is no longer keeping up rather than that tissues are resistant, and needs evaluation rather than a lifestyle plan.","A HOMA-IR that has climbed steadily across repeated draws from the same laboratory, even while each value sat under the printed cutoff.","Increased thirst, frequent urination, blurred vision or unexplained weight loss.","Any insulin or glucose result during pregnancy, in polycystic ovary syndrome, or while taking steroids or other medications that affect blood sugar."],"openQuestions":["There is no agreed cutoff for insulin resistance by HOMA-IR, and no guideline has adopted one. Values that separated healthy from at-risk adults ranged from about 1.8 to 2.6 across populations and from about 1.4 to 2.8 across the quartiles of a single US cohort, and every one of them depends on the insulin assay in use.","No randomized trial has treated people toward a HOMA-IR target and measured outcomes, so every threshold on this page describes where risk sat in observational data, not a tested goal.","Whether HOMA-IR adds anything to the fasting insulin it is built from is unresolved: the two correlated 0.81 in the original paper, one meta-analysis found HOMA-IR a stronger predictor of coronary disease than insulin alone, and the San Antonio cohort found them interchangeable.","The mortality signal in US adults was confined to people of normal weight, which is also where the index tracks the clamp least well. Whether a high HOMA-IR in a lean person is a different disease, a measurement artifact, or both, is not settled.","No healthy-reference distribution for HOMA-IR has been published for US adults on the insulin assays laboratories run today; the NHANES III quartiles come from a 1988 to 1994 radioimmunoassay.","HOMA2 is the better model and the less used one. Whether reports and calculators should move to it, and what its cutoffs would be, has not been worked out."],"questions":[{"q":"Is a HOMA-IR of 2.5 insulin resistance?","a":"It depends on who you ask and which assay produced it. Some laboratories would print 2.5 as normal, and some population studies set their cutoff above it. In the US cohort summarized here, 2.5 sits in the third quartile, where cardiovascular mortality was already higher than in the lowest quartile, and it is above the 1.8 to 2.2 cutoffs that predicted later diabetes in a long Iranian cohort. It is a reason to look at the fasting insulin and glucose that produced it, at triglycerides, HDL and waist size, and at the trend on repeat from the same laboratory, not a diagnosis."},{"q":"What is an optimal HOMA-IR?","a":"No guideline defines one. The lowest-risk quarter of US adults without diabetes sat at or below about 1.4, and healthy-population cutoffs fell between about 1.8 and 2.6. At a normal fasting glucose, a HOMA-IR in that lowest quarter is the same thing as a fasting insulin in the single digits. Those are descriptions of where risk was lowest in cohorts, not a target proven by a trial, and the number moves with the insulin assay."},{"q":"How do I calculate HOMA-IR?","a":"Multiply fasting insulin in µIU/mL by fasting glucose in mg/dL and divide by 405. If glucose is in mmol/L, divide by 22.5 instead. For example, an insulin of 8 and a glucose of 90 give 8 x 90 / 405, which is about 1.8. Both values must come from the same fasting draw. That gives HOMA1-IR, the version most studies use; the Oxford HOMA2 calculator will give a different, usually lower, figure."}],"citations":[{"id":"matthews1985","authors":"Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC","title":"Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man","journal":"Diabetologia","year":1985,"pmid":"3899825","doi":"10.1007/BF00280883","type":"laboratory","finding":"The original model: insulin-resistance estimate correlated 0.88 with the euglycemic clamp and 0.81 with fasting insulin, with a coefficient of variation of 31 percent.","url":"https://pubmed.ncbi.nlm.nih.gov/3899825/"},{"id":"wallace2004","authors":"Wallace TM, Levy JC, Matthews DR","title":"Use and abuse of HOMA modeling","journal":"Diabetes Care","year":2004,"pmid":"15161807","doi":"10.2337/diacare.27.6.1487","type":"review","finding":"The model's authors on appropriate use (cohort and epidemiological studies), the under-used computer model, and the need for robust input data.","url":"https://pubmed.ncbi.nlm.nih.gov/15161807/"},{"id":"bonora2000","authors":"Bonora E, Targher G, Alberiche M, et al.","title":"Homeostasis model assessment closely mirrors the glucose clamp technique in the assessment of insulin sensitivity: studies in subjects with various degrees of glucose tolerance and insulin sensitivity","journal":"Diabetes Care","year":2000,"pmid":"10857969","doi":"10.2337/diacare.23.1.57","type":"laboratory","finding":"In 115 adults, HOMA correlated 0.82 with the clamp overall, 0.75 in people without diabetes and 0.70 with diabetes; weighted kappa 0.63 for categories.","url":"https://pubmed.ncbi.nlm.nih.gov/10857969/"},{"id":"kim2004","authors":"Kim SH, Abbasi F, Reaven GM","title":"Impact of degree of obesity on surrogate estimates of insulin resistance","journal":"Diabetes Care","year":2004,"pmid":"15277430","doi":"10.2337/diacare.27.8.1998","type":"cross-sectional","finding":"HOMA-IR correlated 0.36 with directly measured insulin action in normal-weight volunteers, 0.55 in overweight and 0.60 in obese; fasting indices explained at most 13 percent of variation in the lean.","url":"https://pubmed.ncbi.nlm.nih.gov/15277430/"},{"id":"kang2005","authors":"Kang ES, Yun YS, Park SW, et al.","title":"Limitation of the validity of the homeostasis model assessment as an index of insulin resistance in Korea","journal":"Metabolism","year":2005,"pmid":"15690315","doi":"10.1016/j.metabol.2004.08.014","type":"cross-sectional","finding":"Clamp correlation weaker below a body mass index of 25 (0.44 versus 0.62), with lower beta-cell function and with higher fasting glucose.","url":"https://pubmed.ncbi.nlm.nih.gov/15690315/"},{"id":"ascaso2003","authors":"Ascaso JF, Pardo S, Real JT, Lorente RI, Priego A, Carmena R","title":"Diagnosing insulin resistance by simple quantitative methods in subjects with normal glucose metabolism","journal":"Diabetes Care","year":2003,"pmid":"14633821","doi":"10.2337/diacare.26.12.3320","type":"cross-sectional","finding":"In 65 healthy adults with normal glucose metabolism, the 75th percentile of HOMA-IR was 2.6, matching a fasting insulin of 12.","url":"https://pubmed.ncbi.nlm.nih.gov/14633821/"},{"id":"gayoso2013","authors":"Gayoso-Diz P, Otero-González A, Rodriguez-Alvarez MX, et al.","title":"Insulin resistance (HOMA-IR) cut-off values and the metabolic syndrome in a general adult population: effect of gender and age: EPIRCE cross-sectional study","journal":"BMC Endocrine Disorders","year":2013,"pmid":"24131857","doi":"10.1186/1472-6823-13-47","type":"cross-sectional","finding":"In 2,459 Spanish adults the cutoff was 3.46 by 90th percentile, 2.05 by metabolic-syndrome components and 1.85 in men without diabetes; age-dependent in women.","url":"https://pubmed.ncbi.nlm.nih.gov/24131857/"},{"id":"qu2011","authors":"Qu HQ, Li Q, Rentfro AR, Fisher-Hoch SP, McCormick JB","title":"The definition of insulin resistance using HOMA-IR for Americans of Mexican descent using machine learning","journal":"PLoS One","year":2011,"pmid":"21695082","doi":"10.1371/journal.pone.0021041","type":"cross-sectional","finding":"In 1,854 Mexican-American adults in Brownsville, Texas, the best cutoff was 3.80 and 39.1 percent of the population sat above it.","url":"https://pubmed.ncbi.nlm.nih.gov/21695082/"},{"id":"esteghamati2010","authors":"Esteghamati A, Ashraf H, Khalilzadeh O, et al.","title":"Optimal cut-off of homeostasis model assessment of insulin resistance (HOMA-IR) for the diagnosis of metabolic syndrome: third national surveillance of risk factors of non-communicable diseases in Iran (SuRFNCD-2007)","journal":"Nutrition and Metabolism","year":2010,"pmid":"20374655","doi":"10.1186/1743-7075-7-26","type":"cross-sectional","finding":"In 3,071 Iranian adults the best cutoff for the metabolic syndrome in people without diabetes was 1.775, with sensitivity 57.3 percent and specificity 65.3 percent.","url":"https://pubmed.ncbi.nlm.nih.gov/20374655/"},{"id":"ghasemi2015","authors":"Ghasemi A, Tohidi M, Derakhshan A, Hasheminia M, Azizi F, Hadaegh F","title":"Cut-off points of homeostasis model assessment of insulin resistance, beta-cell function, and fasting serum insulin to identify future type 2 diabetes: Tehran Lipid and Glucose Study","journal":"Acta Diabetologica","year":2015,"pmid":"25794879","doi":"10.1007/s00592-015-0730-3","type":"cohort","finding":"In 4,942 adults over 9.2 years, the diabetes-predicting cutoffs were 1.85 (women) and 2.17 (men) by HOMA1-IR, 1.41 and 1.18 by HOMA2-IR, and 11.13 and 9.16 µIU/mL by fasting insulin.","url":"https://pubmed.ncbi.nlm.nih.gov/25794879/"},{"id":"moon2018","authors":"Moon S, Park JH, Jang EJ, et al.","title":"The Cut-off Values of Surrogate Measures for Insulin Sensitivity in a Healthy Population in Korea according to the Korean National Health and Nutrition Examination Survey (KNHANES) 2007-2010","journal":"Journal of Korean Medical Science","year":2018,"pmid":"30008630","doi":"10.3346/jkms.2018.33.e197","type":"cross-sectional","finding":"In 10,997 Koreans the cutoffs were 2.20 (men), 2.55 (premenopausal women) and 2.03 (postmenopausal women); the triglyceride-glucose index discriminated better (AUC 0.877 versus 0.737).","url":"https://pubmed.ncbi.nlm.nih.gov/30008630/"},{"id":"ausk2010","authors":"Ausk KJ, Boyko EJ, Ioannou GN","title":"Insulin resistance predicts mortality in nondiabetic individuals in the U.S.","journal":"Diabetes Care","year":2010,"pmid":"20200308","doi":"10.2337/dc09-2110","type":"cohort","finding":"In 5,511 NHANES III adults over a mean 8.5 years, the top HOMA-IR quartile (above 2.8) had 1.64 times the all-cause mortality of the bottom (1.4 or less), only in people with a body mass index under 25.2.","url":"https://pubmed.ncbi.nlm.nih.gov/20200308/"},{"id":"hanley2002","authors":"Hanley AJ, Williams K, Stern MP, Haffner SM","title":"Homeostasis model assessment of insulin resistance in relation to the incidence of cardiovascular disease: the San Antonio Heart Study","journal":"Diabetes Care","year":2002,"pmid":"12087016","doi":"10.2337/diacare.25.7.1177","type":"cohort","finding":"In 2,569 adults without diabetes over 8 years, the top HOMA-IR quintile had 2.52 times the odds of a cardiovascular event, 1.94 after full adjustment.","url":"https://pubmed.ncbi.nlm.nih.gov/12087016/"},{"id":"gast2012","authors":"Gast KB, Tjeerdema N, Stijnen T, Smit JW, Dekkers OM","title":"Insulin resistance and risk of incident cardiovascular events in adults without diabetes: meta-analysis","journal":"PLoS One","year":2012,"pmid":"23300589","doi":"10.1371/journal.pone.0052036","type":"meta-analysis","finding":"Across 65 studies and 516,325 people, coronary risk per standard deviation was 1.46 for HOMA-IR, 1.21 for glucose and 1.04 for insulin.","url":"https://pubmed.ncbi.nlm.nih.gov/23300589/"},{"id":"marcovina2007","authors":"Marcovina S, Bowsher RR, Miller WG, et al.","title":"Standardization of insulin immunoassays: report of the American Diabetes Association Workgroup","journal":"Clinical Chemistry","year":2007,"pmid":"17272483","doi":"10.1373/clinchem.2006.082214","type":"laboratory","finding":"Among 12 commercial insulin methods from 9 manufacturers, among-assay variation ranged from 12 to 66 percent with a median of 24 percent.","url":"https://pubmed.ncbi.nlm.nih.gov/17272483/"},{"id":"widjaja1999","authors":"Widjaja A, Morris RJ, Levy JC, Frayn KN, Manley SE, Turner RC","title":"Within- and between-subject variation in commonly measured anthropometric and biochemical variables","journal":"Clinical Chemistry","year":1999,"pmid":"10102917","type":"laboratory","finding":"In 12 healthy adults on 12 consecutive days, within-person variation was 26 percent for fasting insulin and 4.8 percent for glucose.","url":"https://pubmed.ncbi.nlm.nih.gov/10102917/"},{"id":"jayagopal2002","authors":"Jayagopal V, Kilpatrick ES, Jennings PE, Hepburn DA, Atkin SL","title":"Biological variation of homeostasis model assessment-derived insulin resistance in type 2 diabetes","journal":"Diabetes Care","year":2002,"pmid":"12401750","doi":"10.2337/diacare.25.11.2022","type":"laboratory","finding":"Sampled 10 times at 4-day intervals, women without diabetes had a mean HOMA-IR of 2.11 (SD 0.79); a repeat had to rise more than 90 percent or fall more than 47 percent to be a real change.","url":"https://pubmed.ncbi.nlm.nih.gov/12401750/"},{"id":"magkos2016","authors":"Magkos F, Fraterrigo G, Yoshino J, et al.","title":"Effects of Moderate and Subsequent Progressive Weight Loss on Metabolic Function and Adipose Tissue Biology in Humans with Obesity","journal":"Cell Metabolism","year":2016,"pmid":"26916363","doi":"10.1016/j.cmet.2016.02.005","type":"rct","finding":"Five percent weight loss improved insulin sensitivity in liver, muscle and fat and improved beta-cell function; further loss improved muscle sensitivity further.","url":"https://pubmed.ncbi.nlm.nih.gov/26916363/"},{"id":"jelleyman2015","authors":"Jelleyman C, Yates T, O'Donovan G, et al.","title":"The effects of high-intensity interval training on glucose regulation and insulin resistance: a meta-analysis","journal":"Obesity Reviews","year":2015,"pmid":"26481101","doi":"10.1111/obr.12317","type":"meta-analysis","finding":"Across 50 studies, interval training reduced insulin resistance versus control (SMD -0.49) and versus continuous training (-0.35).","url":"https://pubmed.ncbi.nlm.nih.gov/26481101/"},{"id":"feng2025","authors":"Feng S, Liu R, Colwell B, Chung S, Barry A, Wang H","title":"Carbohydrate-restricted diet types and macronutrient replacements for metabolic health in adults: A meta-analysis of randomized trials","journal":"Clinical Nutrition","year":2025,"pmid":"41004974","doi":"10.1016/j.clnu.2025.09.005","type":"meta-analysis","finding":"Across 149 randomized trials and 9,104 adults, carbohydrate restriction lowered HOMA-IR by 0.54 (95 percent CI 0.33 to 0.75), similar whether or not calories were matched.","url":"https://pubmed.ncbi.nlm.nih.gov/41004974/"},{"id":"papadaki2020","authors":"Papadaki A, Nolen-Doerr E, Mantzoros CS","title":"The Effect of the Mediterranean Diet on Metabolic Health: A Systematic Review and Meta-Analysis of Controlled Trials in Adults","journal":"Nutrients","year":2020,"pmid":"33143083","doi":"10.3390/nu12113342","type":"meta-analysis","finding":"Across 57 controlled trials and 36,983 adults, the Mediterranean diet improved HOMA-IR, glucose and insulin among 18 of 28 metabolic outcomes, and lowered cardiovascular incidence (RR 0.61).","url":"https://pubmed.ncbi.nlm.nih.gov/33143083/"},{"id":"wong2026","authors":"Wong PS, Wan K, Dai Z, Yu AP, Wong SH, Poon ET","title":"Effect of 8-Hour Time-Restricted Eating (16/8 TRE) on Glucose Metabolism and Lipid Profile in Adults: A Systematic Review and Meta-Analysis","journal":"Nutrition Reviews","year":2026,"pmid":"41351878","doi":"10.1093/nutrit/nuaf206","type":"meta-analysis","finding":"Across 23 randomized trials and 1,280 participants, an 8-hour eating window lowered HOMA-IR by a standardized mean difference of 0.16, glucose by 0.25 and insulin by 0.22.","url":"https://pubmed.ncbi.nlm.nih.gov/41351878/"},{"id":"spiegel1999","authors":"Spiegel K, Leproult R, Van Cauter E","title":"Impact of sleep debt on metabolic and endocrine function","journal":"Lancet","year":1999,"pmid":"10543671","doi":"10.1016/S0140-6736(99)01376-8","type":"rct","finding":"Six nights of 4 hours in bed lowered glucose tolerance and raised evening cortisol in 11 healthy young men.","url":"https://pubmed.ncbi.nlm.nih.gov/10543671/"}],"written":"2026-10-10","updated":"2026-10-10","reviewed":false,"related":{"essays":["fasting-insulin-of-12-what-it-means"],"glossary":["HOMA-IR","Fasting insulin","Insulin resistance"],"symptoms":["weight-loss-resistance","food-cravings"],"entries":["fasting-insulin","fasting-glucose","hba1c","triglycerides","hdl-cholesterol","alt","uric-acid"]},"url":"https://taggehealth.com/labs/homa-ir","author":"Daniel Tagge, MD","publisher":"Tagge Health","license":"Free to read and cite with attribution to the URL above."}