Six Questions for Judging a Longevity Claim

The reader standard: six questions before accepting a claim

A longevity claim often begins with a true fragment: a measured change occurred in a study. The missing context is whether that change matters to health, whether it applies beyond the participants, and whether the study design can distinguish an intervention from expectation, chance or selective reporting.

This standard is intended for claims about systemic interventions and measurements: medicines, supplements, dietary patterns, fasting approaches, exercise-related claims, and tests said to measure ageing or future health. It does not decide whether a person should start, stop or combine an intervention. That requires clinical context, including diagnoses, medicines, risks and priorities.

Work through the questions in order. A weak answer early in the process can limit what later answers mean. For example, a carefully blinded trial can still be poor evidence for healthspan if it measures only a short-term laboratory marker. Equally, a relevant outcome in a narrow clinical population may not transfer to healthy adults seeking preventive benefit.

Screenshot rule: if the claim cannot answer all six questions, its certainty should not exceed the weakest answer. A favourable result is not, by itself, a recommendation.

1. What was the endpoint: health outcome or surrogate?

Start with the study endpoint, meaning the result the researchers set out to measure. A direct health outcome is something patients experience or that clearly records health status, such as death, a fracture, hospital admission, physical function or onset of a defined disease. In longevity research, direct outcomes are difficult because they may take many years to occur and require large studies.

A surrogate endpoint is a measurement used as a stand-in for a later health outcome. Blood pressure, cholesterol, glucose measures, inflammatory markers, imaging results and molecular measures may all be useful in particular settings. A biological age score is also a measurement, not a direct observation of whether someone will live longer or remain free of disability. The question is not whether a surrogate is real. It is whether changing it through this intervention has been shown to improve the health outcome being claimed.

Surrogates can mislead when a marker sits on one part of a complex disease process, while the intervention affects other pathways in harmful or neutral ways. A marker may also move temporarily without producing durable benefit. This is why language matters. "Changed a marker associated with risk" and "reduced disease risk" are different claims, requiring different evidence.

• Identify the primary endpoint, rather than relying on a list of favourable measurements.
• Ask whether the endpoint was decided before results were known.
• Look for the claimed health outcome in the published results, not just an explanation of biological plausibility.
• For ageing tests, separate a score change from proof that the test predicts, or can alter, an individual's future health.

The existing explainers on surrogate endpoints and on what biological age tests measure provide further context for these distinctions.

2. Who was studied, and do they resemble the intended user?

Every result belongs first to the people who took part. Read the participant description before accepting broad language such as "supports healthy ageing". Age range, sex, baseline health, medication use, disease status, smoking status, activity level and location can all affect both benefit and harm. A study in people with a diagnosed condition is not automatically evidence for people without that condition.

Recruitment matters too. Participants in a trial may receive monitoring, structured advice, regular contact and stronger encouragement to continue than people buying an intervention independently. Those features can improve adherence and sometimes outcomes. They may also mean the study setting differs substantially from ordinary use.

Pay attention to exclusions. Trials often exclude people with multiple conditions, kidney or liver impairment, pregnancy, interacting medicines, frailty, or other factors that complicate interpretation. Exclusions can be sensible for safety and study clarity, but they narrow the group to which findings apply. A result from selected participants should be described as such, not as a universal effect.

A useful reading habit is to write a one-sentence population label: "This was studied in people aged roughly this range, with this health status, for this reason." Then compare that label with the person targeted by the claim. If the article or advertisement does not state who was studied, find the original report before drawing conclusions.

Similarity does not guarantee the same result, and difference does not prove no effect. It defines uncertainty. The larger the gap between participants and intended users, the more cautious the inference should be.

3. How long did the study run, and how long is use proposed for?

Duration has two parts: how long participants received the intervention, and how long they were followed afterwards. These are especially important for longevity claims because the proposed goal is often decades of better health. A result observed over days or weeks may establish that a measurement can change. It usually cannot establish sustained benefit, delayed disease or long-term safety.

Compare the study period with the duration implied by the claim. If an intervention is presented as a daily lifelong practice, ask whether evidence covers sustained use, discontinuation, adherence and delayed adverse effects. A short trial may be adequate for some immediate outcomes, but it is not equivalent to evidence about long-term healthspan.

Also check when outcomes were measured. A result immediately after an intensive intervention may fade after it ends. Conversely, some outcomes take time to appear. Follow-up after treatment can help show whether benefits persist and whether harms emerge later, although it cannot eliminate all uncertainty.

Dropout is part of duration. If many participants leave, especially if departure differs between groups, the final comparison may no longer represent the people originally assigned. Reports should say how many participants began, how many completed the study and how missing data were handled. This is not a technical footnote. It affects whether the reported average reflects tolerable, sustained use.

For a reader, the key distinction is straightforward: a short, controlled change may be interesting evidence of a mechanism. It is not evidence that a person should expect years of improved health from continuing the intervention.

4. Is the effect presented as a relative change or an absolute one?

Relative effects can make a result sound much larger than its practical impact. A relative risk reduction compares proportions between groups. An absolute risk reduction asks how much the actual chance of an outcome changed. Both describe the same data, but absolute numbers are usually more useful for judging what the result means to participants.

Suppose an outcome occurs in 2 out of 100 people in a comparison group and 1 out of 100 in an intervention group over the study period. That is a 50 per cent relative reduction, but an absolute reduction of 1 percentage point. The example is purely illustrative, not evidence for any intervention. Its purpose is to show why a large percentage does not tell the full story without the starting risk.

Baseline risk changes interpretation. An intervention with the same relative effect can produce a larger absolute difference in people at higher risk than in people at lower risk. That is one reason results in clinical populations may not translate neatly to younger or healthier users.

Look for event counts, denominators and confidence intervals. A confidence interval describes a range of effect sizes compatible with the data under the study's statistical model. A result may be described as statistically significant yet still have an effect too small, uncertain or short-lived to support a broad healthspan claim. Conversely, a study with few events may be unable to rule out a meaningful effect in either direction.

• Ask: compared with what baseline chance?
• Ask: how many participants had the outcome in each group?
• Ask: over what time period was the difference measured?
• Ask: does the claim report a percentage while omitting the absolute result?

The existing explainer on absolute versus relative risk addresses this reading problem in more detail.

5. Was there a control group, sound allocation and blinding?

A control group provides the comparison needed to ask whether the intervention, rather than time or outside influences, caused the observed difference. Depending on the question, a control may receive placebo, usual care, an alternative treatment or a different intervention. Without a comparator, improvement can reflect regression to the mean, natural fluctuation, changes in behaviour after enrolment or participants joining a study when symptoms are at their worst.

Random allocation is designed to make comparison groups similar on average before treatment begins, including factors researchers did not measure. It does not guarantee perfection in every small trial, but it is a central protection against selection bias. A report should explain how participants were allocated and whether the groups were similar at baseline.

Blinding means that participants, clinicians, outcome assessors or analysts do not know which intervention was assigned. Blinding is particularly important when outcomes involve symptoms, self-reporting or judgement. It can be difficult or impossible for interventions such as exercise programmes, dietary changes or time-restricted eating. That limitation does not make such research useless. It means readers should examine whether outcomes were objective, assessors were blinded, and the comparison was credible.

Claims based on before-and-after changes, testimonials or open-label studies should therefore be treated as preliminary signals. They can identify questions worth testing. They cannot reliably separate the intervention's effect from expectation and other changes occurring at the same time.

Finally, check whether all assigned participants were included in the analysis where possible. Analysing people according to their original group is one safeguard against a misleading comparison created by excluding those who did not continue.

6. Who paid, who published, and can the full record be checked?

Funding and competing interests do not automatically invalidate research. Commercial, charitable, academic and public funding can all support valuable work. But financial or professional interests can influence which questions are asked, which comparators are chosen, how outcomes are emphasised and whether inconvenient findings receive equal visibility. They belong in the reader's assessment of risk of bias.

Look for a funding statement and conflict-of-interest declaration in the full paper. Consider whether the funder supplied the intervention, controlled data access, participated in analysis, or had a role in writing the report. These details help show how independent the research process was. A sparse disclosure leaves an important question unanswered.

Publication status matters as well. A full paper generally offers methods, participant flow, adverse events and numerical results that are not available in a press release, conference abstract or social-media post. Peer review can identify some problems, but it is not a guarantee that a result is correct. It is one stage of scrutiny, not a final certificate.

Prospective trial registration is another useful check. A registry record can show the planned primary outcome, participant number and timing before results are known. Differences between a pre-specified plan and a final report may have reasonable explanations, but they should be clear. Selectively presenting a favourable secondary result as though it were the main test can inflate confidence.

Ask whether findings have been replicated by researchers working independently and whether adverse events are fully described. A single favourable study may justify further investigation. A durable healthspan claim needs a body of evidence that survives repetition, transparent reporting and close examination.

Limits of this standard and when to seek individual advice

This standard is a tool for judging the strength and relevance of a claim. It does not calculate an individual's risk, diagnose a condition, interpret a test result, or choose treatment. It cannot replace discussion with an appropriate clinician, particularly for prescription medicines, pregnancy, existing disease, eating disorders, frailty, significant weight change, or use of multiple medicines.

It is also not a substitute for a formal systematic review. A reader can use the six questions to identify missing information and overstated conclusions, but a full evidence appraisal additionally considers all relevant studies, search methods, risk-of-bias assessment, certainty of evidence, consistency and applicability. One well-presented trial may still sit within a conflicting literature.

The framework does not assess product quality, label accuracy or regulatory compliance. In the United Kingdom, medicines are regulated differently from food supplements, and the legal status of a product does not by itself establish that it extends healthspan. For the regulatory context around supplements, see the publication's existing explainer on supplement regulation in the UK.

Nor does this piece grade a particular intervention. The same six questions can be applied to a supplement, drug, dietary pattern or test claim, but the resulting conclusion may be "insufficient information", "evidence for a marker only", or "evidence limited to a specific population". Those are useful conclusions. They prevent a marketing line from becoming a promise that the underlying study did not make.