SCIENCE
HOW WE KNOW THE OLDEST PERSON WHO EVER LIVED WASN’T FAKING HER AGE
A researcher claims that identity theft was at play in the case of Jeanne Calment, the world’s oldest person, but experts say that evidence is weak.
By Angela Chen@chengela Jan 9, 2019, 12:47pm EST
Illustration by Alex Castro

What if Jeanne Calment, the oldest person who ever lived, lied about her age? What if she wasn’t an astounding 122 years old when she died, but a lowly 99 because she wasn’t even Jeanne Calment?
Such is the theory of Russian mathematician Nikolay Zak, and it has everything: world records, statistics-defying long life, identity theft, tax evasion, and researchers duking it out with each other. In a paper posted to the research-sharing site ResearchGate, Zak claims that Calment actually died at age 59 in 1934, at which point her daughter Yvonne assumed her mother’s identity to avoid paying inheritance taxes. That would have meant that “Jeanne” was not even a century old when she died in 1997.

If true, the Calment story would be a truly spectacular case of fraud; even just the theory has captured international attention. And the same month that Zak released his findings, the journal PLOS Biology published a paper arguing that some exciting conclusions from aging research are caused by statistical error (from bad data if not outright fraud). So how do we know that Calment didn’t lie about her age? How do we know for sure how old anyone is?

The Calment controversy has demographers and non-demographers making different claims. It’s also a case of establishment science versus a less-supported but more titillating hypothesis. Though there continues to be back-and-forth, experts say that, most likely, Jeanne Calment is who she said she was: a woman from the southern French town of Arles who met van Gogh, rode a bike until she was 100, and smoked two cigarettes a day until a few years before she died at 122.

Humans want to know how to live forever — or at least for a little while longer. That’s why people click on headlines about chocolate being the secret to a 102-year-old woman’s longevity even while knowing that, come on, chocolate is not the secret. Spurious connections aside, the past century has seen a big increase in the frequency of really old people surviving, and scientists are still debating the limits of the human lifespan.

“THE PROBABILITY IS EXTREMELY LOW, BUT EXTREMELY LOW PROBABILITY AND IMPOSSIBLE ARE TWO DIFFERENT WORDS.”

Being able to accurately predict how many people will live to very old age is a “really important societal question,” says Daniel Promislow, a gerontologist at the University of Washington who was not involved with either of the recent papers. For example, an accurate understanding of these numbers will affect how much social support we’re going to need for the elderly, and that research would not be very useful if all of these 115-year-olds were actually much younger.

This is exactly what could be happening, says Saul Newman. Newman, a postdoctoral fellow at Australian National University who studies wheat genomics using machine learning, wrote the recent PLOS Biology paper casting doubt on longevity claims. Fraud or bad intentions aren’t necessary. Discrepancies could be as simple as a misrecorded birthdate, especially given that today’s supercentenarians (or people over 110 years old) were born in a time with lower literacy rates and less detailed record-keeping. And because there are so few supercentenarians to begin with, you only need tiny mistakes to throw off calculations and create dramatic statistical results.

Newman says statistical errors undermine the findings of two high-profile (and dueling) papers on the lifespan debate. One, published in Nature in 2016, suggests a maximum lifespan for humans of around 115. The other, published in Science in 2018, claims there might not be such a maximum. As a general rule, the longer we live, the more likely we are to die. The Science paper — which studied 4,000 Italians over the age of 105 — claims that after that age, the chances of dying actually level off, creating a so-called mortality plateau. The possibility of bad data means both of these papers are statistically flawed, Newman says, adding, “For 20 years, scientists have been fighting over an error distribution.”

Yet Kenneth Wachter, a demographer at UC Berkeley and co-author on the Science study, argues that Newman’s critiques are based on a hypothetical model and don’t take into account the actual data the team used. “We have birth certificates matched to age reports and death certificates,” Wachter says. “He has a theoretical exercise, but it’s not one that applies to our data.”

That’s not good enough, Newman replies. “That’s based on the idea that official documentation can never be wrong, and we know that’s not true. How many times are you in the DMV and they’ve made an error?”

So who’s correct?

These ideas aren’t entirely contradictory, says Dmitri A. Jdanov, a mathematician at the Max Planck Institute for Demographic Research who specializes in data collection and processing for the International Database on Longevity. The issue that Newman raises is real, but it’s not new. Demographers have long known that misreporting can create a lot of errors that throw off analyses. Books such as Validation of Exceptional Longevity, Exceptional Longevity: From Prehistory to the Present and Supercentenarians all deal with this methodological question. According to Jdanov, massive errors do exist in population data, but such errors are far, far less likely in the very carefully checked data about supercentenarians.
continued next post