A normal BMI reading at age 70 can be reassuring and wrong at the same time. That is the uncomfortable implication of an analysis published in the Journal of the American Geriatrics Society, which tracked 6,905 Medicare beneficiaries aged 65 and older from 2011 through 2024 and compared what two different body measurements predicted about who died.
One of those measurements is calculated at nearly every medical visit in America. The other one, which performed better, is almost never taken.
How the study was put together
Participants came from the National Health and Aging Trends Study, an annual survey designed to be nationally representative of older Medicare beneficiaries. Inclusion required at least two recorded measurements over the follow-up period, and anyone who died during the first year was dropped from the analysis so that existing terminal illness would be less likely to distort the results.
Rather than fixing each person’s body measurements at their 2011 values, the investigators updated both measures as they changed over the 14 years. They counted someone who gained or lost weight at 78 accordingly from that point forward.
BMI came from self-reported height and weight, sorted into the usual bands: under 18.5 for underweight, 18.5-24.9 for normal, 25.0-29.9 for overweight, 30.0 to just under 40 for class I and II obesity, and 40 or above for class III.
Waist circumference was handled differently, and the difference matters. Trained research staff measured it directly at the waistline between the lowest rib and the top of the hip bone, following a standardized protocol. Thresholds for a high waist followed established consensus values, meaning above 40 inches in men and above 35 inches in women. Each model accounted for age, race, education, income, smoking history, and homebound status, and each body measure was adjusted for the other.
The two measures disagreed with each other
Taken by itself, BMI produced results that will look backward to most readers. Set against men with a normal BMI, men in the overweight range had a 46% lower hazard of dying, and those with class I or II obesity had a 51% lower hazard. Overweight women had a 27% lower hazard. Underweight status carried the steepest risk anywhere in the BMI analysis, at 1.91 in men and 1.97 in women.
Researchers have argued about this pattern, sometimes called the obesity paradox, for well over a decade, and the authors here decline to oversell it. Their own reading is that these numbers reveal the shortcomings of BMI rather than any protective quality of body fat. A ratio of weight to height squared cannot tell whether the weight is muscle or fat, and it does not adjust for the height people lose as they age. When a 74-year-old lands in the overweight band, that may reflect preserved lean mass and metabolic reserve, which is a considerably better situation than the same BMI achieved by a body that has swapped muscle for visceral fat. Undiagnosed illness and survivorship effects likely play a role too.
Waist circumference produced no such ambiguity. After adjusting for BMI and everything else in the model, a high waist carried a 24% higher hazard of death, and the figure was identical in both sexes.
The relationship also scaled. When the researchers sliced waist measurement into 5-inch bands, mortality risk climbed with each step up. Men in the 50.0-54.9 inch range had a hazard ratio of 2.10, and men at 65 inches or more reached 6.40. For women, the progression ran from 1.31 at 40.0-44.9 inches to 1.85, then 2.47, then 3.58, and finally 5.17 at 60 inches or above. Both trend tests were statistically significant. There was no plateau, no safe zone above the threshold, just a steady rise.
What the combined analysis exposed
Putting the two measures side by side is where this paper earns its keep, because it reveals people whom BMI screening simply misses.
Consider someone with a perfectly normal BMI who also carries a high waist. Compared with normal BMI and a low waist, that combination raised the hazard of death by 33% in men and 23% in women. Researchers have a name for this pattern, normal-weight central obesity, and it is completely invisible to any assessment that stops at the scale. These patients get told their weight is fine.
Underweight participants occupied the other end. Their risk was elevated whether their waist was high or low, and the largest hazard in the joint analysis was among underweight women with a high waist (2.77). Low overall body mass paired with abdominal fat likely signals depleted muscle alongside visceral adiposity, a difficult combination in later life. That said, very few people fell into this category, the confidence intervals were correspondingly wide, and the authors treat the finding as a lead worth following rather than an established result.
Among men, being overweight or in class I and II obesity was associated with lower mortality even when the waist was high, with hazard ratios of 0.69 and 0.60. Women showed nothing statistically significant in any of the overweight or obese categories once waist status was factored in. Nobody knows exactly why the sexes parted ways here. Body composition differs, fat redistributes toward the abdomen after menopause, and unmeasured factors such as lifetime socioeconomic position could be doing some of the work.
Where the evidence thins
Self-reported height and weight introduce error, and that error grows in a population steadily losing height. Professional waist measurement is a real strength of this study, though no measurement is immune to variation between examiners.
Excluding first-year deaths reduces reverse causation but does not eliminate it, since this remains an observational design. A few of the combined categories, particularly underweight with a high waist and class III obesity with a low waist, held small numbers of people, which is why several of those specific hazard ratios should be read loosely.
Then there is the question of what a tape measure can and cannot establish. Last year’s Lancet Commission on Clinical Obesity separated preclinical obesity, meaning excess adiposity with organs still functioning normally, from clinical obesity, where obesity-related organ dysfunction or functional limitation can be demonstrated. This analysis measured body dimensions and mortality. It did not assess organ function, so it can point to who deserves closer evaluation without replacing that evaluation.
Putting a number on your own waist
Anyone can do this at home with a cloth tape. Find the midpoint between your lowest rib and the top of your hip bone, wrap the tape level around your waist at that height, breathe out normally, and read the number without pulling the tape tight. Above 40 inches for a man or 35 inches for a woman lands you in the category this study linked to higher mortality, and the numbers above kept getting worse as the measurement grew. Most clinics will not do this unless asked, so ask, or bring your own number to the visit.
Two cautions belong alongside that instruction.
Do not read the BMI results as a reason to relax. Two people with a BMI of 27 can be in entirely different biological situations depending on whether that weight is muscle or abdominal fat, and BMI cannot tell them apart. Waist circumference, grip strength, and body composition testing can.
Weight that falls on its own should also worry you more than weight that stays put. The underweight groups carried the highest hazards in this entire paper. Losing pounds without trying often points to frailty, muscle loss, or an illness that has not yet declared itself, and it deserves a workup rather than a compliment.
BMI became the standard because it was cheap and easy to compute, not because it described anyone accurately. Bodies are muscle and bone and organ and fat, arranged differently in every person, and a single ratio flattens all of that into a number that can mislead precisely when the stakes are highest. Caring faithfully for a body entrusted to us means measuring it honestly, with instruments suited to its season of life.
Bring three things to your next appointment if you are over 65: your weight, your waist measurement, and a straight answer about whether either one has moved in the past year. This article is educational and does not replace individualized medical advice.
