Body Weight & BMI

What the evidence actually shows

The Global BMI Mortality Collaboration pooled individual data from roughly four million adults across hundreds of studies, and did the one thing that finally settled decades of noise: it excluded smokers and people with known chronic disease at baseline, and it threw out the first five years of follow-up. Those two moves strip out the reverse-causation garbage that had polluted every previous "overweight is fine, actually" headline. What emerged was a clean J-shaped curve with its floor around a BMI of 22 to 25.

Above that floor the hazard ratio climbs in a graded way: mild in the overweight band, roughly 1.5 by the lower obesity grades, and up toward 2.5 to 3 for the most severe obesity. Below the floor there is a milder uptick, but — and this is the whole point of the study's design — much of that low-BMI excess risk is illness-driven weight loss that survives even careful adjustment. The curve is real; the left arm is partly an artefact.

How big the effect really is, in years

Translated into life expectancy, the collaboration's estimates run from a modest cost of about one year in the overweight range to something like eight to ten years at the severe end of obesity. We price the modifier conservatively across that span, roughly −2 to −8 years depending on which obesity band you land in, because the year-figures are sensitive to age, sex, and how much of the association is truly causal.

A number worth internalising: the overweight band (BMI 25 to 30) carries a genuinely small mortality penalty, far smaller than the diet industry implies. The steep costs live in the obesity grades, not in being a few kilos above a chart's arbitrary line. An actuary reads the J-curve and concludes that the actionable message is about the tails, not about vanity.

We hedge the years hard here because BMI is a crude instrument. It cannot tell muscle from fat and it cannot see where the fat sits, and central adiposity carries more risk than the same BMI distributed peripherally. The number is directionally solid and quantitatively fuzzy.

The catch: BMI is a proxy for a proxy

BMI does not measure the thing that actually kills you. Visceral fat, insulin resistance, blood pressure, and lipids are the mechanistic culprits; BMI is a cheap population-level stand-in that correlates with them imperfectly. A muscular individual gets miscategorised as overweight, and a "normal-weight" person with high visceral fat and poor metabolic health — the so-called TOFI phenotype — gets a clean bill from BMI they do not deserve.

This is why we treat BMI as a modifier and not as a verdict. It is included because at population scale it carries real signal and because it is the one body measurement almost everyone actually knows. But if you have waist circumference, blood pressure, and a lipid panel, those tell a truer story, and a future version of this calculator will weight them accordingly.

How this feeds your actuarial age

The bmi modifier applies a banded hazard-ratio adjustment following the J-curve: near-neutral through the normal and overweight range, then increasingly negative across the obesity grades, with a small penalty at the underweight end that we keep deliberately modest precisely because so much of the low-BMI signal is confounded by illness.

Do not over-read small movements. Dropping from a BMI of 27 to 24 barely moves your actuarial age in this model, because the data says that stretch of the curve is nearly flat. The modifier earns its keep at the extremes. If your BMI is in the mid-30s or higher, this is a meaningful lever; if you are hovering a point or two above "normal," the calculator will politely decline to panic on your behalf.