BMI missed half the adults a body-fat scan flagged. Waists caught more.

In 1,900 US adults aged 20 to 59, a waist measurement alone did about as well as a new four-measurement obesity definition, and better than BMI. But the scan's cut-off put 69% of these adults over the line, and where that line sits changes how bad BMI looks. Nobody's health was followed.

11 October 2026 · How we rate evidence

A tape measure costs a couple of dollars. A DEXA scan, the low-dose X-ray that maps how much of the body is fat, lean tissue and bone, costs far more and needs a machine. A team at Rutgers Health asked how close the cheap tools get to the expensive one, using 1,900 American adults who had both.

The claim going around. The study came out on 7 August. Rutgers' press release followed within a week, and ScienceDaily reposted it on 31 August as "Your waist size may reveal health risks BMI misses." Lead author Aayush Visaria is quoted: "Pay attention to the need to loosen your belt or get bigger pants." News 12 and News-Medical carried the release's framing. Medical Daily wrote that "waist circumference alone, or BMI alone, performed about as well as the more complex framework." The paper supports the first half; BMI alone did worst of the five rules tested.

What they did. The occasion is a new obesity definition. In 2025 the Lancet Diabetes & Endocrinology Commission on Obesity proposed that BMI alone shouldn't decide who carries excess fat. Its rule counts someone as having excess fat with a BMI of 40 or more, a BMI of 30 or more plus one raised body measurement, or any two raised measurements from waist size, waist-to-height ratio and waist-to-hip ratio. The last needs a hip measurement, which the authors note is rarely taken in clinics.

The team took everyone aged 20 to 59 who had a DEXA scan in the national survey's 2017 to 2018 round and tested five rules against it: the full Lancet definition, the same without the hip measurement, "BMI or waist" (either one over its usual cut-off), waist alone, and BMI alone. The scan's verdict was the answer key: obesity meant 25% body fat or more for men and 35% or more for women, cut-offs from a 1995 World Health Organization report.

What they found. These are a few of the dozens of figures the paper reports: five rules, each scored nine ways, against two kinds of scan result. By the scan's standard, 69% of these adults had obesity. BMI caught about half of them, 51%. A waist measurement caught 71%. "BMI or waist" caught 74%, and the full Lancet rule 88%. Per 100 adults, that works out like this:

The Lancet rule catches the most, at the price of false alarms: it flagged 29% of the people the scan cleared, against 8% for waist alone. On the summary score, which weighs both errors equally, waist alone and "BMI or waist" came out ahead (0.815 and 0.818 against 0.790). That's the basis for the authors' recommendation of the simpler rules in clinics, where they argue a wrong label does harm. Count total wrong calls instead, and the Lancet rule makes fewer: about 18 per 100 against 21 for "BMI or waist" (our arithmetic; the paper's own error score and national estimates show the same). The Lancet rule also flags about as many people as the scan does, which the authors call near-perfect calibration. The paper prints both views, including the figures that favour the Lancet rule, and every rule was checked against a scan taken at the same visit, in a sample weighted to represent the country. Dropping the hip measurement made the Lancet rule perform like "BMI or waist".

What it doesn't show. It doesn't show which measure predicts disease. The answer key was body fat above a line, measured once, not diabetes, heart attacks or deaths. The paper's findings are about body fat; the release's "health risks BMI misses" goes further than the data.

The line itself matters more than the coverage let on. At 25% and 35% body fat, most working-age Americans count as having obesity. An independent team analysed survey years 2011 to 2018, which include this study's data, in the journal Obesity this February, using 30% and 42%. At that line 36% had obesity, and BMI caught 75% of them. A waist measurement again caught more, 89%, though with more false alarms than BMI. The two studies agree on direction and disagree on size, and they drew the line in different places. The Rutgers paper tested only one line.

The group is specific, too: adults 20 to 59, since the survey scans no one older, measured by trained survey staff rather than in a busy clinic. Waist alone caught about 80% of women with high body fat but closer to 60% of men, and the authors say the male cut-offs likely need recalibrating.

What it could mean for people. Should your doctor reach for a tape measure? For spotting high body fat, this study says a waist measurement adds a lot to BMI and gets close to the more elaborate Lancet rule without a hip measurement. It can't say that a bigger waist, rather than a higher BMI, is the better warning of future illness; that needs studies that follow people over years. The Lancet commission itself treats excess fat as a starting point: whether someone has "clinical obesity" depends on signs of illness, which this study didn't assess.

Why it's still interesting. The Lancet definition is new. This is a direct test against a scan, in a sample built to represent the country, using public data anyone can re-run. Its practical finding is that the hardest part of the new rule, the hip measurement, buys sensitivity at the cost of many false alarms, and that a waist measurement recovers much of the rest. That's a useful thing to know before any switch re-labels millions of people.

Open questions.

Where should the body-fat line be? Re-run the same public data at several cut-offs, or better, set cut-offs by the level of fat at which disease risk actually rises. The answer decides whether BMI misses a third of working-age adults or closer to one in ten, and how many people any switch would re-label.

Which rule predicts who gets sick? Compare BMI, waist and the Lancet rule in groups followed for years for diabetes, heart disease and death. Accuracy against a scan is a stand-in; prediction of illness is the reason to measure anyone at all.

Does it survive a real clinic? Compare waists measured in routine appointments with those taken by trained staff, against the same scan. If clinic measurements scatter widely, the tape-measure advice needs a measuring protocol to go with it.

A note on timing. The paper came out on 7 August 2026. The release followed within a week, and the ScienceDaily repost came on 31 August, more than three weeks later, with more coverage after it. The study hadn't changed.