People who live past a hundred are studied intensively in the hope of finding what distinguishes them. The design of that research creates several problems that limit what the findings can support.
The group is defined by the outcome
Centenarians are selected for study because they already survived. Every characteristic measured in them is measured after the outcome of interest has occurred.
That inverts the usual logic of a study, where a group is followed forward and the outcome is observed later.
A habit common among survivors may be a cause of long life, a consequence of good health, or simply a feature of the era and place in which that generation grew up.
Age records are less reliable than they appear
Very old ages sit at the edge of what registration systems can verify, particularly for people born before civil registration was universal.
Regions that report unusually high numbers of centenarians often turn out to have had patchy birth records, and stricter verification tends to reduce the count.
Careful studies validate each age against multiple documents, which is slow and shrinks the sample considerably.
Recall covers a very long period
Asking a person of a hundred about diet and activity means asking about decades that ended long ago, often with no contemporaneous record.
Memory reconstructs habits in line with a person's current self-image, and interviewers asking about longevity invite answers shaped by that framing.
Reported patterns of eating and working from such interviews are best treated as description of a way of life rather than measurement of exposure.
Survivors are not a random slice of their cohort
Anyone born a century ago belonged to a birth cohort that lost a large share of its members young, to infection, war and childbirth.
Those who remain differ from the original group in ways that were never recorded, including access to food, occupational hazard and childhood illness.
Comparing them with today's middle-aged population compares two different worlds, which weakens any inference about what habits caused survival.
What the research does establish
Family clustering is one of the more robust findings: relatives of very long-lived people tend to reach older ages themselves, pointing to inherited as well as shared environmental factors.
Another consistent observation is that many centenarians compress serious illness into a short period near the end of life rather than living for years with heavy disease burden.
That pattern is descriptive and does not identify a mechanism, which is why the most quoted claims about centenarian diets rest on far thinner evidence than the family clustering does.