BMI and Ageing: How the Risk Curve Shifts in Older Adults and Why Sarcopenia Changes the Picture

The single most useful fact for interpreting BMI in older adults is that the J-curve connecting BMI to death risk shifts its lowest point upward after age 65. The Global BMI Mortality Collaboration 2016 meta-analysis of 239 prospective studies across 32 countries found the lowest risk at a BMI of 20.0-25.0 for the general adult population. But a dedicated meta-analysis by Winter et al. in 2014, pooling 32 studies of 197,940 adults aged 65 and older, placed the nadir at a BMI of 27.0-27.9. In that older cohort, risk of dying only began to increase reliably above a BMI of 33.0.

This is not a prescription to gain weight. It is an observational pattern confounded by reverse causation, where undiagnosed illness causes weight loss and inflates the apparent risk of lower BMI, and by smoking, which both lowers weight and raises death rates. The shift is real, but it is a statistical property of populations, not an individual health target.

What complicates BMI interpretation even further in this age group is sarcopenia: the age-related loss of skeletal muscle mass and function. An older adult can carry a normal or overweight BMI and still have critically low muscle mass, a condition called sarcopenic obesity. BMI alone cannot detect it. The European Working Group on Sarcopenia (EWGSOP2) consensus definition requires measurements of muscle strength, muscle quantity, and physical performance. A bathroom scale provides none of these.

The J-curve shift is not in dispute as a finding. The interpretation is what separates useful guidance from harmful advice. The Winter et al. 2014 meta-analysis found that a BMI of 27.0-27.9 carried the lowest all-cause death rate for adults aged 65 and older. Flegal et al. 2013, in a much larger meta-analysis of 97 studies with 2.88 million individuals, reported that the overweight BMI category (25.0-29.9) was associated with fewer deaths compared with the normal weight category (18.5-24.9) in older adults, with a hazard ratio of 0.94.

At first glance, this looks like permission to carry extra weight. The failure case is assuming this is a causal relationship. Two confounders dominate. The first is reverse causation: illness and chronic disease cause unintentional weight loss, so people who are already sick drift into the lower BMI categories. The second is confounding by smoking: smokers have both lower body weight and higher death rates, inflating the apparent risk of leaner body types.

The Prospective Studies Collaboration 2009, a meta-analysis of 57 studies with 894,576 adults, showed that each 5 kg/m² increase in BMI above 25 was associated with a 40% increase in cardiovascular death. That finding was not restricted by age. The J-curve shift in older adults is a real observation, but it describes a statistical pattern in the presence of disease, not a biological optimum reached by gaining weight after 65.

Sarcopenic obesity is the coexistence of excess body fat and sarcopenia. An individual can meet the WHO 1995 criteria for normal BMI (18.5-24.9) or even overweight BMI (25.0-29.9) and still have dangerously low muscle mass relative to their body fat. BMI cannot distinguish between fat mass and lean mass.

How Muscle Loss Accelerates After 50

In older adults, appendicular skeletal muscle mass declines by roughly 1-2% per year after age 50, and the rate accelerates after 65. The EWGSOP2 definition of sarcopenia requires confirmation by low muscle strength, typically measured by handgrip, low muscle quantity measured by DXA body composition or other imaging, and low physical performance measured by gait speed.

Someone with a BMI of 24 and a slow walking speed and weak grip could be sarcopenic. Someone with a BMI of 28 and high muscle mass from lifelong physical activity could have a healthy body composition. BMI misclassifies both.

Why This Condition Goes Undiagnosed

The condition is underdiagnosed because standard clinical practice relies on weight and height alone. DXA body composition is the reference method for quantifying muscle mass, but it is not routine in primary care. The practical consequence: an older adult with a BMI in the normal or overweight range should not assume their body composition is adequate. A gait speed slower than 0.8 metres per second or a handgrip strength below 26 kg for men and 16 kg for women are red flags that warrant a body composition assessment.

Measurement	Threshold Or Guideline	Issuing Body And Year
BMI reference range (general adult population)	18.5–24.9 kg/m²	WHO, 1995, Technical Report Series 854
BMI mortality nadir (adults aged 65+ meta-analysis)	27.0–27.9 kg/m²	Winter et al., 2014, meta-analysis of 32 studies
BMI mortality risk increase (adults 65+)	≥33.0 kg/m²	Winter et al., 2014
Overweight BMI association with mortality (older adults)	HR 0.94 vs normal weight	Flegal et al., 2013, meta-analysis of 97 studies
Low muscle strength (sarcopenia screening)	Handgrip <26 kg (men), <16 kg (women)	EWGSOP2, 2019
Slow gait speed (sarcopenia screening)	<0.8 m/s	EWGSOP2, 2019
NHLBI high-risk waist circumference	>102 cm (men), >88 cm (women)	NHLBI, 1998
NICE waist-to-height ratio threshold	<0.5	NICE CG189, 2022
WHO Asian-specific overweight BMI threshold	≥23.0 kg/m²	WHO Western Pacific Region, 2004

The WHO 1995 BMI categories are the global default, but they were derived from risk curves in predominantly European and North American adult populations. The WHO Technical Report Series 854 does not provide separate thresholds for older adults. The NICE CG189 guideline, updated in 2022, uses the same 18.5-24.9 healthy weight range for all adults, though it advises using waist-to-height ratio as an additional measure for those with a BMI below 35. The NHLBI 1998 guidelines also apply the same cut-offs across adulthood.

The Gap Between Guidelines and Evidence

No major issuing body has published a separate BMI range for older adults. The shift observed in meta-analyses has not been adopted into formal clinical thresholds. This creates a gap: a clinician applying the WHO 1995 range to an 80-year-old patient will flag a BMI of 27 as overweight, but the epidemiological data suggests that same BMI is associated with the lowest risk of death in that age group.

The practical instruction: avoid treating a BMI of 25-29.9 in an older adult as automatically concerning. It warrants investigation, not alarm. The number that matters more for this age group is body composition.

When Low BMI Is The Real Warning

A low BMI, below 20, is a stronger warning signal for frailty, bone mineral density loss, and risk of osteoporosis, especially when combined with unintentional weight loss. The Mini Nutritional Assessment, used in geriatric screening, flags a BMI below 21 as a risk factor for malnutrition in older adults. The failure case is using a single number from a general adult standard to judge an older person's metabolic health.

Waist circumference is a tape-measured surrogate for abdominal fat that captures risk BMI misses. This matters in older adults where fat distribution shifts toward the abdomen even as peripheral fat and muscle mass decline. The NHLBI 1998 thresholds of 102 cm for men and 88 cm for women mark high risk for cardiometabolic disease, but these were derived from white populations.

The NICE CG189 2022 guideline recommends using waist-to-height ratio alongside BMI for adults with a BMI under 35, with a threshold of 0.5. This has the advantage of being a single dimensionless number that adjusts for stature. In older adults, waist circumference can remain stable or increase while overall weight drops. BMI can stay the same or decrease while central adiposity rises. An older adult with a BMI of 23 and a waist circumference of 94 cm may have normal-weight obesity, carrying excess visceral fat despite a lean appearance on the scale.

How To Measure Waist Circumference Correctly

The measurement protocol matters. Waist circumference measured by a trained observer at the midpoint between the lowest rib and the iliac crest, at the end of a normal expiration, is the standard. Self-measured waist circumference is unreliable, with errors commonly exceeding 3-5 cm. If a protocol-adherent measurement is not possible, the waist-to-height ratio with a tape at the belly button level is a workable heuristic. It is not equivalent to a clinical measurement. The practical step: measure waist circumference using the correct landmark and tape tension, and use the 0.5 ratio as a population signal, not a personal diagnosis.

Body composition cannot be inferred from BMI in older adults. The age-related loss of appendicular skeletal muscle mass, which begins in the fourth decade and accelerates after 65, means that two individuals with the same BMI can have radically different ratios of fat to lean mass. DXA body composition is the reference method for quantifying this. It distinguishes fat mass, lean mass, and bone mineral density in a single scan.

Why DXA Is The Reference Standard

DXA is not a routine screening tool, but it is available in many sports medicine and geriatric research centres. In its absence, BIA devices are common. Their individual error against DXA is ±3-8 percentage points for body fat percentage, and the error is larger in older adults because hydration status is more variable and the algorithms were often derived from younger populations. Skinfold measurement is operator-dependent and loses accuracy in people with high body fat.

Functional Screening When Scanning Is Not Available

The practical alternative when DXA is not available: use functional screening for sarcopenia. Handgrip strength measured with a dynamometer. Gait speed over a measured distance. The EWGSOP2 2019 consensus defines sarcopenia as a condition confirmed by low muscle strength, low muscle quantity, and low physical performance. A person who meets these criteria has sarcopenia regardless of their BMI score. The failure case is assuming an older adult with a normal BMI is metabolically healthy, or that an older adult with an overweight BMI is overfat. Neither assumption holds without direct body composition or functional assessment.

The obesity paradox is the epidemiological observation that in certain disease cohorts, such as heart failure or chronic kidney disease, a higher BMI is associated with fewer deaths. In older adults, this pattern is particularly pronounced because of the high prevalence of chronic disease. The Flegal et al. 2013 meta-analysis found that overweight BMI was associated with a 6% lower risk of dying compared with normal weight in the general population, and the effect was stronger in older subsamples.

This is not a recommendation.

The paradox is a statistical pattern confounded by reverse causation and selection bias. People with chronic illness lose weight, so the reference category of normal BMI includes many sick individuals. The healthy survivors at higher BMI are not healthy because of their weight. They are healthy despite it, and they lived long enough to be studied. Mendelian randomisation studies, which use genetic variants as a proxy for adiposity, suggest the causal effect of higher body fat on death risk is more linear than the observed J-curve.

The practical instruction: do not use the obesity paradox to justify weight gain or discourage weight loss in older adults who are metabolically compromised. The paradox describes a pattern in observational data, not a biological mechanism. The person who needs to lose weight for metabolic health should lose weight. The person who is stable at a higher BMI with good function, muscle mass, and no metabolic disease should not be alarmed by the number on the scale. They also have no evidence that gaining weight would be protective.

Unintentional weight loss in an older adult is a clinical warning sign independent of BMI category. The Mini Nutritional Assessment, a validated geriatric screening tool, flags weight loss of 3 kg or more in three months as a risk factor for malnutrition. The frailty index, which counts deficits across domains of physical function, cognition, and nutrition, includes weight loss as a core criterion.

A person with a BMI of 28 who has lost 5 kg unintentionally in six months is more clinically fragile than a person with a BMI of 23 whose weight has been stable for a year. BMI cannot capture this trajectory. The Global BMI Mortality Collaboration 2016 showed that the J-curve flattens or disappears when restricting analyses to healthy, never-smoking individuals. Much of the apparent protective effect of higher BMI in older adults is driven by the inclusion of people whose low weight is a consequence of disease.

The practical action: ask about weight stability. A stable BMI above 20 in an older adult is a better risk signal than the absolute number. A downward trend, even from a BMI of 25, warrants investigation for sarcopenia, chronic disease, or depression. The failure case is ignoring the trend because the absolute number falls within a supposedly healthy range. BMI is a screening tool for population risk, not a diagnostic for individual health. In older adults, the trend matters as much as the static value.

The single most common failure is treating a BMI of 25-29.9 in an older adult as a health problem that requires correction. The WHO 1995 classification labels this overweight, but the epidemiological data for people over 65 consistently shows that this range carries the lowest risk of dying in that age group. Jumping from the classification to a weight loss prescription without assessing body composition, muscle strength, and unintentional weight loss history is a clinical error.

The person with a BMI of 27 and good handgrip, a normal waist circumference, and stable weight is not a case for intervention. The person with a BMI of 23 who has lost weight unintentionally, walks slowly, and has weak grip needs investigation for sarcopenia and underlying illness, not reassurance that they are in the healthy weight range.

Before acting on a BMI number, ask three questions. Is the weight stable? Can the person walk four metres in under five seconds? Can they carry a shopping bag or open a jar? If the answer to those is yes, the BMI number is a secondary concern. If the answer to any is no, the BMI number is irrelevant. What matters is the underlying body composition and function.