Physical Activity & Cardiorespiratory Fitness

What the evidence actually shows

Cardiorespiratory fitness is one of the few longevity variables where the observational signal is so large that the usual hand-wringing about confounding does not make it disappear. In the Mandsager cohort of over 122,000 patients referred for treadmill testing, mortality fell monotonically with measured fitness, and the gap between the least-fit and most-fit groups was wider than the gap you get from being a smoker versus a non-smoker. That is not a typo. Being in the bottom fitness quintile was associated with a hazard ratio of roughly 4 against the top performers.

The important structural fact is that the curve is steepest at the bottom. Most of the benefit is captured by not being sedentary at all. Lee and colleagues, pooling prospective data in the Lancet, estimated that physical inactivity accounted for a mortality burden on the order of tobacco, and that eliminating inactivity would raise life expectancy in the population by something like 0.6 to 0.7 years averaged across everyone — which, spread over the minority who are genuinely sedentary, concentrates into a much larger individual effect.

So the honest framing is: fitness is measured, not self-reported, which is why it behaves better than most lifestyle variables. When you measure the thing directly on a treadmill instead of asking people how virtuous they feel, the noise drops and the dose-response tightens.

How big the effect really is, in years

We price the sedentary-to-fit transition at roughly +4 years of life expectancy for a middle-aged adult, and we want to be clear that this is a hedged central estimate, not a promise. The confidence interval is wide, the effect shrinks with age because you have fewer years left to save, and a large chunk of the apparent benefit at the very top of the fitness distribution is people who were already healthy enough to run fast.

Notice what we are not claiming. We are not claiming that going from fit to elite-athlete adds another four years. The dose-response flattens hard once you are past average fitness; the marginal year bought by your fifth weekly training session is a rounding error next to the year bought by your first walk. Diminishing returns is the whole shape of this curve.

The data supports about ±a couple of years of uncertainty on that +4, not the decade some longevity influencers attach to their zone-2 protocols. The mechanism is real, the direction is not in doubt, and the magnitude is still genuinely good — it just is not magic.

The catch: reverse causation and the sick-quitter problem

The obvious confounder is reverse causation. People with undiagnosed cancer, early heart failure, or incipient frailty are less fit precisely because they are already ill, and they die sooner for reasons that have nothing to do with exercise. This drags the low-fitness group's mortality up and inflates the apparent benefit of fitness. Good studies handle this by excluding early deaths and adjusting for baseline disease, and the effect survives that adjustment — but it does shrink, and anyone quoting the unadjusted number is selling you something.

The second issue is that fit people differ from unfit people on a hundred other axes: income, smoking, diet, conscientiousness, access to care. Randomised exercise trials exist but they are short and powered for surrogate endpoints like VO2max, not for the decades-long mortality signal we actually care about. So we are stitching together strong observational data plus a plausible mechanism plus RCT evidence on intermediate markers. That is a grade A body of evidence by the standards of this field, which tells you more about the field than about the certainty.

How this feeds your actuarial age

In the calculator this maps to the activity modifier. We treat your self-reported activity band as a proxy for where you sit on the fitness distribution, and we apply a hazard-ratio adjustment that is deliberately more conservative than the raw treadmill studies, because self-reported activity is a noisier signal than a measured VO2max and because we assume you are already partway up the curve.

The modifier moves your actuarial age most when you tell us you are sedentary, and barely moves it when you go from "active" to "very active" — which mirrors the real dose-response rather than rewarding heroic training volume you may or may not sustain. If you want the single biggest lever on your number, this is usually it. If you are already active, do not expect the calculator to hand you another four years for buying a fancier bike.