Insulin resistance assessed by estimated glucose disposal rate and risk of incident cardiovascular diseases among individuals without diabetesfindings from a nationwide, population based, prospective cohort study

You don't have diabetes. Your doctor hasn't flagged your blood sugar. By every standard clinical measure, you're fine. And yet something in your metabolism may already be quietly raising your odds of a heart attack or stroke — years before any diagnosis would catch it. That's the uncomfortable truth sitting at the center of a new study by Zenglei Zhang and colleagues, published from a nationwide Chinese cohort: insulin resistance, running well ahead of diabetes, predicts cardiovascular disease with a clean, linear, dose-response precision. The higher your insulin sensitivity, the lower your risk. There is no threshold and no safe zone. Just a continuous gradient. Insulin resistance is a state in which tissues stop responding efficiently to insulin and fail to clear glucose from the blood properly. That metabolic failure cascades into oxidative stress, endothelial damage, chronic inflammation, inappropriate activation of the renin-angiotensin-aldosterone system — the hormonal network that governs blood pressure and fluid balance — and increased clotting. These mechanisms damage arteries and the heart. The problem is that this process can run for years while blood sugar remains technically normal, invisible on a standard panel. To measure it without a complex lab procedure, Zhang and colleagues used something called the estimated glucose disposal rate, or eGDR. The formula, stated verbally: start with twenty-one point one five eight, subtract zero point zero nine times waist circumference in centimeters, subtract three point four zero seven if the person has hypertension, subtract zero point five five one times their HbA1c percentage — where HbA1c is a measure of average blood sugar over roughly three months. A higher eGDR means better insulin sensitivity. A lower eGDR means the body is struggling to clear glucose, even before diabetes sets in. What makes this useful is that all three inputs — waist circumference, blood pressure status, and HbA1c — are already collected in most routine clinical encounters. No insulin assay is required. No fasting clamp procedure. Most prior research on eGDR and cardiovascular outcomes was conducted in diabetic populations. That's a problem because diabetic cohorts carry so many compounding risks that it's hard to isolate what insulin resistance itself is doing. Zhang and colleagues set out to test the relationship specifically in people who had never been diagnosed with diabetes — and the results are striking. Their cohort came from the China Health and Retirement Longitudinal Study, or CHARLS, a nationally representative survey. After excluding people with prevalent cardiovascular disease, diabetes, cancer, or missing data, they analyzed five thousand five hundred twelve participants with a mean age of fifty-eight point two years, fifty-four point one percent of them female. They followed these participants for a median of seventy-nine point four months — just over six and a half years. During that time, one thousand two hundred thirteen people developed cardiovascular disease: nine hundred twenty-seven heart disease events and three hundred ninety-one strokes. Participants were divided into four quartiles of eGDR, from lowest — indicating the worst insulin resistance — to highest. The crude incidence rates per one thousand person-years tell the story plainly before any statistics are applied: forty-six point three in the bottom quartile, thirty-six point six in the second, twenty-six point eight in the third, and twenty-four point one in the top. That's nearly a two-to-one difference between the worst and best groups, in people with no diabetes diagnosis. To test the shape of that relationship, the team used restricted cubic splines — a flexible curve-fitting technique that lets the data reveal whether the association bends, breaks, or plateaus at any point, without forcing it into a straight line. It found a straight line anyway. For total cardiovascular disease, heart disease, and stroke separately, the spline curves showed significant overall associations and no evidence of non-linearity — with all p-values for non-linearity greater than zero point zero five. The risk declines steadily as eGDR rises, with no U-shaped reversal and no threshold below which the relationship disappears. There is no "safe enough" level of insulin resistance, at least not one visible in this data. The quartile comparisons, adjusted for a full battery of covariates including age, sex, smoking, alcohol use, cholesterol fractions, inflammatory markers, kidney disease, and obesity, put numbers on that gradient. Compared with the lowest eGDR quartile, the fully adjusted hazard ratios for total cardiovascular disease were zero point eight eight for the second quartile, zero point six nine for the third, and zero point six six for the fourth. That top versus bottom comparison is the headline: people with the highest insulin sensitivity had a thirty-four percent lower hazard of cardiovascular disease. When Zhang and colleagues treated eGDR as a continuous variable, each one standard-deviation increase in eGDR was associated with a seventeen percent lower cardiovascular disease risk, a thirteen percent lower risk of heart disease, and a thirty percent lower risk of stroke. Stroke and heart disease diverge in an interesting way. For heart disease, the hazard ratios across quartiles two through four were zero point nine four, zero point seven five, and zero point seven four — a meaningful gradient, but compressed. For stroke, the numbers drop more sharply: zero point six nine, zero point five two, and zero point four two. Moving from the lowest to the highest eGDR quartile was associated with a fifty-eight percent lower hazard of stroke. Insulin resistance appears to carry a particularly heavy toll on cerebrovascular risk, at least in this cohort. Zhang and colleagues also asked a mechanistic question: how much of that cardiovascular risk runs through obesity? Mediation analysis — which quantifies how much of an exposure's effect on an outcome is carried through a third variable sitting on the causal pathway — showed that obesity partially explained the link. In fully adjusted models, obesity mediated fourteen percent of the association between eGDR and total cardiovascular disease, and seventeen point six percent of the association between eGDR and heart disease. For stroke, the mediated proportions were not statistically significant. So roughly one-seventh to one-sixth of insulin resistance's cardiovascular harm appears to move through excess body fat. The rest operates through other pathways. That matters clinically: treating obesity reduces risk, but it does not clear it. Then comes the practical question every clinician will eventually ask: does adding eGDR to an existing risk model actually improve predictions, or is it just telling you what you already know from the variables it contains? Zhang and colleagues tested this by building a standard multivariable model — age, sex, lifestyle factors, cholesterol fractions, inflammatory markers, kidney disease, obesity — and then adding eGDR on top. The C-statistic, which represents the probability that the model will correctly rank a future case above a future non-case, rose from zero point six one to zero point six seven for total cardiovascular disease, from zero point six one to zero point six seven for heart disease, and from zero point six two to zero point six nine for stroke. All three improvements were statistically significant, with p-values below zero point zero zero one. Net reclassification improvement and integrated discrimination improvement — metrics that measure how many people get moved into more accurate risk categories, and by how much — were also significant across all three outcomes. That's a meaningful gain for a measure derived from three variables most clinicians already have on hand. Other insulin-resistance markers like HOMA-IR require an insulin assay, and the hyperinsulinemic-euglycemic clamp — the gold standard — is far too complex for routine use. eGDR avoids both barriers entirely. There are real limitations to consider. This is an observational study — causal inference is off the table. The cardiovascular outcomes were self-reported, which introduces potential misclassification. The sample consists of Chinese adults aged forty-five and older, which limits how far these findings travel to other populations and age groups. HbA1c measured under certain physiological conditions — severe anemia, for instance — can be unreliable, adding another source of noise. But step back from those caveats and consider what the study actually found. In more than five thousand people with no diabetes and no cardiovascular disease at baseline, a simple formula calculated from routine clinic data predicted who would develop heart disease and stroke over the next six and a half years, with a signal that was linear, significant, and incrementally informative beyond existing risk models. The gradient was continuous — meaning action at any level of insulin resistance has potential value, not just action at some clinical threshold that hasn't been crossed yet. That's the window this tool is designed to identify: the years before a diagnosis lands, when the biology is already moving but the chart still says everything is fine. This lecture was created by ennepō. Go to https://ennepo.ai to Discover, Create and Follow the latest research in your field. Read when you can. Listen when you want to.