Claim check: “our patients reversed their biological age by six years”

The claim: a programme reports that its patients reduced their biological age by an average of six years. The measurement is an epigenetic clock. The comparison is before and after.

Three problems, in ascending order of severity.

The instrument moves on its own

Epigenetic clocks are technically reproducible, meaning the same sample run twice gives nearly the same answer, and this is the fact clinics cite. It is the wrong fact. Repeated sampling from the same person under ordinary short-term perturbations, including meals, acute stress and pollution exposure, produces substantial fluctuation in the estimate, with most clocks achieving only moderate reliability at best.1

When the number moves because of when blood was drawn, a before-and-after difference is not evidence about the intervening months.

Selection guarantees the result

Patients who enrol after a discouraging first result are selected for an extreme value. Retest anyone selected that way and the second measurement will on average sit closer to their true value, which is to say better, with no intervention required. This is regression to the mean. It is arithmetic rather than biology, and it is sufficient on its own to produce the headline.

Avoiding it requires a control group. Nobody making this claim has one.

The claim cannot be falsified as stated

No clinic publishing this figure publishes the within-person variance of the clock it used, which is the only thing that would let a reader judge whether six years exceeds the noise. Several do not disclose which clock they used at all.2

Verdict

Unsupported. Note the scope. This is not a finding that the programme’s interventions are worthless, because exercise, sleep and metabolic control have their own evidence independent of any clock. It is a finding that this measurement cannot detect whether they worked, and that presenting it as though it had makes a claim about instrument sensitivity the instrument does not support.

What would make it checkable: the clock used, its published within-person confidence interval, standardised draw conditions, and a comparison group that did not receive the intervention. Any clinic with all four should publish them, and we will grade the result again when one does.