Longevity —
what is actually proven.

A managing director, 54, came to me with a list. Fourteen supplements, a subscription to a biological age test, a cold plunge in the basement and a dashboard tracking his heart rate variability. He wanted to know whether the order was right.

Then he mentioned in passing that he had been sleeping about five hours for two years. On weekdays. He caught up at weekends.

We did not optimise the list. We set it to zero and started with sleep. That is not a jab at him but the most common pattern I meet on this topic: people pouring enormous effort into the last five percent while the first fifty lie wide open.

Longevity means healthspan, not lifespan

The question of what longevity really is resolves fastest through the gap between two numbers. One is lifespan: how long you live. The other is healthspan: how long you live without major limitation. The gap between them is the actual subject.

Medical progress has moved the first number considerably. The second far less. The result is years gained that are often spent in limitation. Anyone who says longevity but means lifespan is working on the wrong end.

Biologically, ageing is not vague wear and tear. The group around Carlos López-Otín summarised the describable mechanisms in 2023 as twelve hallmarks, among them genomic instability, telomere attrition, loss of proteostasis, chronic inflammation and stem cell exhaustion (López-Otín et al., Cell 2023).

That sounds like a list of targets, and it gets sold exactly that way. The uncomfortable part: this model is a framework for research, not a treatment list. Most of what demonstrably acts on these mechanisms acts in mice so far. In humans, the evidence for almost all of these approaches is still open.

The Blue Zones do not survive the audit

Practically every article on this topic tells the same story. Five regions where people reach remarkable ages: Okinawa, Sardinia, Ikaria, Nicoya and Loma Linda. From them come dietary rules, movement patterns and community structures you are supposed to copy.

That story has been taken apart in recent years, and not through the diet but through the data.

Demographer Saul Newman audited age records systematically. His findings: only 18 percent of those considered exhaustively validated supercentenarians, meaning people aged 110 and over, hold a birth certificate at all. In the USA the figure is zero percent. Where comprehensive birth registration was introduced, the number of such age records subsequently fell by 69 to 82 percent.

On top of that comes a pattern that is hard to explain otherwise: the birthdates of supercentenarians cluster on days divisible by five. In England and France, a region's old-age poverty rate alone predicts more than half of the regional variation in reaching a remarkable age. And the designated Blue Zones corresponded to regions with lower incomes, lower literacy, higher crime and shorter life expectancy relative to their respective national averages.

What this finding does not mean

The caveat belongs here, and it matters. Newman's work is still a preprint and has not been through peer review. It received the 2024 Ig Nobel Prize, a satirical award for research that first makes you laugh and then makes you think. His conclusions are contested within demography.

It does not follow that diet, movement or social ties are unimportant. There is solid data for those factors, just from other sources. What falls away is the Blue Zones narrative as a justification for specific rules. Nobody ever tested whether it was the beans, the wine or the stairs. Someone saw a striking number and worked backwards to an explanation.

My approach in coaching has stayed the same since: I derive nothing from a region. I derive it from data where somebody measured what they claim.

What is proven: five factors, fourteen years

The best-known solid figure on this topic comes from two long-running US cohorts, the Nurses' Health Study and the Health Professionals Follow-up Study. Together 123,219 participants, up to 34 years of observation, 42,167 documented deaths.

The research group defined five low-risk factors: never smoking, a BMI between 18.5 and 24.9, at least 30 minutes of moderate to vigorous activity per day, moderate alcohol intake and diet quality in the upper range. Those meeting all five had a hazard ratio of 0.26 for all-cause mortality compared with someone meeting none (Li et al., Circulation 2018).

Translated into years: life expectancy from age 50 was 43.1 further years instead of 29.0. For women that worked out at 14.0 years of difference, for men 12.2.

Four of those five factors cost nothing. None of them is a product. That is the actual finding, and also the reason it rarely takes centre stage in the longevity industry.

Two caveats that belong with it

This is observational data. People who do not smoke, move daily and eat well differ in other ways too from those who do not, in education, income and medical care. The figure shows an association, not a proven cause-and-effect relationship. Add the composition: mostly nurses and health professionals in the USA, with high educational attainment.

And one of the five factors is now outdated. Moderate alcohol intake counted as a low-risk characteristic at the time. More recent analyses, which handle the comparison with lifelong abstainers more cleanly, no longer find that protective effect. Anyone repeating the study today would reach a different result on that point.

And the genes?

"Twenty to thirty percent is genetic" appears in nearly every article on the subject. That figure comes from family and twin comparisons and is probably too high.

An analysis of pedigree data covering several hundred million historical people found the same similarity in lifespan between in-law relatives who share no genes. People choose partners with similar lifestyles, similar education and similar surroundings, and for decades that was booked as heritability. After accounting for this effect, the true heritability of lifespan came out well below 10 percent (Ruby et al., Genetics 2018).

If that unsettles you because it shifts the responsibility: it is the better of the two pieces of news.

The three measures with the strongest association

Sort the cohort data by strength of association with mortality and three measures stay at the top. Two of them you can train, one you have to tend.

The aerobic capacity figure deserves a second look, because it makes the scale understandable. In that analysis, the risk added by poor cardiorespiratory fitness was comparable to smoking and larger than diabetes. And the authors found no upper limit: even very high versus merely high fitness still carried an advantage, including in people over 70 (Mandsager et al., JAMA Netw Open 2018).

The caveat belongs here too: this was a retrospective analysis of people referred for an exercise test. On average they are sicker than the general population. The direction is supported by many other cohorts, the exact size of the effect is not.

On muscular strength, grip strength deserves a note, since it is becoming popular as a longevity marker. In the PURE study with 139,691 participants, every 5 kilogram decrease came with a 16 percent higher all-cause mortality, making grip strength a stronger predictor than systolic blood pressure. But grip strength indicates the state of the whole system; it is not a lever. Train your forearms and you change the marker, not what it stands for. How to build sensibly instead is set out in the guide to strength training after 40.

The third measure is the one least discussed in this context, because you cannot buy it. In the Holt-Lunstad meta-analysis, the effect of social connection sits in the same range as established medical risk factors.

What does not deliver what it promises

This section is the reason I wrote the article at all. In few health topics is the gap between what gets sold and what has been shown this wide.

For dietary supplements, not one has a controlled human trial showing extended lifespan. That does not mean useless: a documented deficiency should be corrected, and individual compounds have documented effects on individual functions. For lifespan, the evidence is entirely absent. Where the real risks sit I have written up separately, and which form of a given compound is even absorbed is shown clearly enough by the example of magnesium.

The biological age tests, meaning the epigenetic clocks, measure methylation patterns and do say something about disease risk at group level. For you as an individual the benefit is open. One point decides it, and it never appears in the advertising: to this day there is no evidence that a lowered test score goes with a longer or healthier life. You are optimising a number whose link to the outcome is untested.

As drugs against ageing, rapamycin and metformin are the most frequently named candidates. In mice there are solid findings on life extension. In humans the decisive trial has not reported yet, and until then any recommendation is a bet, not evidence.

The rest of the toolkit holds cold plunges, red light, infusions and breathing protocols. Some of it has measurable short-term effects; for almost none of it has the transfer to lifespan been studied. The detailed assessment sits in the article on whether biohacking is dangerous. One exception with comparatively decent data is creatine, though for cognitive performance and muscle mass, not for life expectancy.

Honesty requires the addition: not proven is not the same as disproven. Much of this has simply not been studied. One just should not pretend otherwise.

Longevity is a system, not a stack of measures

The managing director from the start had no shortage of measures. He had fourteen. What he lacked was the insight that these measures do not work alongside each other but with each other.

Five hours of sleep lowers glucose tolerance, raises appetite, worsens recovery after training and makes any dietary change harder to sustain. Four of the five factors from the Circulation analysis therefore hang on a single one that does not appear on that list at all. That is not an argument against the list but for working through it in the right order. How this connection works in detail is set out in the guide to recovery and in the article on sleep quality.

The same holds in the other direction. Anyone under sustained pressure who also raises training volume is loading the system on two fronts and gets the bill weeks later. What goes wrong there and how to spot it is covered in the piece on overtraining.

In practice that means the question is never which measure is best. It is which one closes your largest open gap right now. For most people that is not the fourteenth capsule. It is sleep, training volume, or a stress level nobody has looked at seriously in years.

That assessment is exactly what the Blueprint Session is for: 90 minutes in which training, sleep, nutrition, stress and blood values sit on the table together rather than separately. If you would rather work through it yourself, the structure for that is in Blueprint Digital.

Where to start if you start today

The appeal of this topic is that there is endlessly much to do. That is also its main problem. So four steps in this order, not in parallel.

The first step is to measure what you have. For aerobic capacity a twelve-minute run test or a fixed distance with a stopwatch is enough. For strength, a hand dynamometer and two compound lifts with the load written down properly. Together that costs an hour and an afternoon, and both give you a baseline to compare against in six months.

The second step is to name the largest gap, not the most convenient one. If you sleep seven hours but never get out of breath, the lever is aerobic capacity. If you train three times a week but sleep five hours, it is sleep. Tackling both at once sounds efficient and usually ends with neither sticking.

Then you take one thing for four weeks. Long enough for something to change, short enough to stay manageable. After that the next one, without dropping the first.

And at some point the blood values deserve a look. Not as a four-figure longevity panel but as basic diagnostics read by somebody who knows what to look for. Which values make sense and what the standard package typically leaves out is in the article on the full blood panel.

One last part, which I had to be told myself. Years ago I had a drawer of tubs, and for half of them I no longer remembered why I had bought them. The drawer is considerably emptier today and the numbers are better. Correlation is not causation, but I have my suspicion.