Dean Eckles writes:
I thought this NPR report was an interesting example of how demand for life advice seems closely connected to low standards of evidence. Like if this didn’t supposedly tell how to live longer, lose weight etc., would anyone bother?
Here they’re looking at so called “blue zones” for nutrition advice etc. These places supposedly have greater longevity. Though as you’ve covered before, they are also generally places with poor record keeping of people’s births and identities.
I have mixed feelings about this one. On one hand, yeah, low standards of evidence. On the other hand, “The five pillars of any longevity diet in the world are whole grains, greens or garden vegetables, tubers like sweet potatoes, nuts as a snack, and then beans. About a cup of beans a day is associated with an extra four years of life expectancy. . . . There’s a little cheese, a little fish, and they cook with lots of aromatic herbs and plants”: that pretty much describes my own diet—except for the bit about sweet potatoes, which I absolutely can’t stand. So I wanna believe, or, at least, I want to believe in four of the five pillars! I guess that’s the problem with this sort of vague recommendation: the theory is weak and it doesn’t give much guidance on how things are supposed to work out if we remove one of the ingredients.
I like sweet potatoes, but I don’t eat nuts as a snack. If we can find 3 more people who are also lacking exactly one out of the five pillars, we can at least do some kind of split plot study!
Do peanuts count?
“About a cup of beans a day is associated with an extra four years of life expectancy.” Does anyone, including whoever wrote the sentence, actually believe that there’s a causal link of this magnitude between beans and life expectancy?
And what does “nuts as a snack” mean? Does the magic of nuts only appear between meals?
What edible plants are not covered in “greens or garden vegetables, tubers like sweet potatoes, … and then beans?” Couldn’t all these be one pillar? I suppose fruits are out. I like fruits, though, so I’ll live dangerously.
The only real science experiment I’ve ever seen on diet basically showed that minimally processed foods were in some sense better than heavily processed foods, and that people seemed to eat until they had “enough” protein (ie. They are a fairly constant level of protein even as total food mass and calories varied)
We discussed this study on the blog actually several times… First in 2019 when it came out
https://statmodeling.stat.columbia.edu/2019/10/28/what-happens-to-your-metabolism-when-you-eat-ultra-processed-foods/
Then there was some concern about the data integrity
https://statmodeling.stat.columbia.edu/2023/07/15/do-ultra-processed-data-cause-excess-publication-and-publicity-gain/
But our discussions at that link suggested that it wasn’t a problem just some complications that were actually handled reasonably
https://statmodeling.stat.columbia.edu/2023/07/15/do-ultra-processed-data-cause-excess-publication-and-publicity-gain/#comment-2238237
There are other worthwhile comments in that post. Especially this one
https://statmodeling.stat.columbia.edu/2023/07/15/do-ultra-processed-data-cause-excess-publication-and-publicity-gain/#comment-2238259
The meaning of “nuts as a snack” is “For a snack, instead of having something like a bag of potato chips, have a serving of nuts instead”. The magic of nuts is that it is healthier than the potato chips. This applies if you are in the habit of having a meal of potato chips too (some people do that!).
‘And what does “nuts as a snack” mean?’
According to the package in my cupboard:
walnuts: “1 serving” = 1oz ~0.25cup (handful) = 180 cal
sliced almonds: “1 serving” = 1oz ~0.33up (good handful) = 160 cal
peanuts: “1 serving” = 1oz ~0.25up (handful) = 160 cal
Statistics, despite R.A. Fisher, proved useful in showing that smoking was harmful and not just a nuisance parameter. Among other contentions, he alleged that causality might go the other way–cancer caused one to smoke. Honest, I was at a lecture of his back in the early 1960s when he offered up such a possibility.
But when it comes to the statistical evidence regarding longevity and diet, it would seem that the situation is hopeless. Furthermore, just ask vegans what they think of vegetarian diets, Mediterranean or not. And, while you are at it, just ask Andrew why his (four or) “five pillars of any longevity diet in the world” suddenly fails to mention celery, a staple of his which he has approvingly mentioned in past postings.
Historically, the biggest increase in longevity in England occurred when coffee replaced beer for breakfast
https://time.com/3541295/coffee-england-history/
Before I am accused of secretly being paid by the coffee lobby, the last of the three cups of coffee I have ever had in my life was in 1964.
It’s an interesting question about lifestyle choice and credulousness, and there are a lot of possibilities to consider, including whether the association is real or illusory. One thought is that there is a constant stream of decision points about something like diet, and limited evidence can’t stand in the way of deciding which item you’ll reach for at the grocery. Another is that the upside of gravitating toward plausible candidates for dietary health outweighs the downside of mistakenly not eating enough butter, sugar, red meat etc. (I’m excluding the Roganesque meat fanatics here, along with other food cultists, like raw-only etc.) And another is that we get fairly rapid bodily feedback from what we eat, and some foods just make us feel better. For instance, I’ve just finished my Sunday a.m. breakfast of pancakes (whole grain with fruit, but still) and sausage (chicken, but still), and the feedback is not wonderful. If a smidgeon of evidence aligns with this feedback, we’re inclined to believe it. Of course, some form of wishful thinking probably enters in: we *want* to believe there is advice we can follow that will help us enjoy long, healthy lives. I would also give some credence to dietary and other health-related customs in areas where science is weak and which have persisted for centuries; if they really didn’t work there would have been pressure to change them. (Or the societies that rigidly practiced unhealthy customs didn’t make it to the present to pass them on to us.)
Other hypotheses?
The blue zones are spread around the world. Is the diet on Sardinia the same as on Okinawa? Are the beans in Loma Linda from the same plant species as the other sites? What’s the best source of whole grains, wheat, maize, or amaranth? Of course, we could randomly assign a thousand freshly weaned humans to the diet of Nicoya and a thousand controls to a conventional American diet; after only fifty years ten per cent of the males on the American diet will have died, and we could likely be able to have hard evidence.
I have personally surpassed the American median by several years and eat highly processed fats and drink lots of alcohol.
My father-in-law (a rural area primary care physician) smoked (a lot), drank (mostly beer), and made it to 98. And he seemed to like it when I’d visit, since it gave him an excuse to insist on lots of steaks in the menu.
My (SO’s) niece is a cardiologist here. (As I’ve probably mentioned before), her youngest patient was 48, but all the others are over 90, and they all die of cancer before their cardiovascular problems catch up with them.
On the other hand, the stuff that is sold as “bread” here is more like cake than bread and makes wonder bread look healthy. And although we eat brown rice at home, it’s pretty rare and special ordering is easier than looking for it in grocery stores. But male life expectancy in Tokyo is three years longer than average life expectancy in my home state (don’t blame me, I’m from Massachussetts). Go figure.
But the Japanese do like their sweet potatoes. When I was first here, there was this horrific noise outside my dorm. It sounded as though someone was killing a cow with a baseball bat. None of the locals even blinked. It was a baked sweet-potatoe street vender singing his traditional sweet-potatoes are ready-to-eat song over an overly powerful PA system.
One little note about Loma Linda. It wasn’t in the original set of “blue zones”, but was later added by Blue Zones LLC, which was subsequently acquired by the Seventh-Day Adventist health care system, Adventist Health.
The nutrition evidence from “blue zones” is even weaker than is suggested. Those areas frequently have lower life expectancies than other areas, but have large numbers of the super old due to a mixture of inaccurate age measurement and pension fraud. Source: https://www.biorxiv.org/content/10.1101/704080v3
Joseph:
Yes, this issue came up a few years ago.
But that’s ok. A diet of nuts and sweet potatoes might not extend my life, but if it can help me with pension fraud, that ain’t nothing!
The blue zones have poor recordkeeping. The foods of the 5 pillars based on the blue zone diets are relatively inexpensive. If the blue zones have relatively more poverty, that is a factor that could contribute to both poor recordkeeping and inexpensive diets.
As long as a lot of people are ready to believe that eating or excluding a certain kind of food has a major (= many years) effect on life expectancy, this kind of dietary advice will persist.
In addition to the piranhas critique (it is not possible for so many things to have large effects), I always wonder about feasibility. All these wonder diets seem to focus on boiling down the secret to a few simple rules, but following a few very simple rules will either be too vague, or very specific and thus very boring. Eg I am ok with both sweet potatoes and beans occasionally, but if I had to eat a cup of beans every single day I would opt for the shorter lifespan so that my suffering ends earlier.
Andrew, incidentally I am wondering if you like (plain vanilla) potatoes? I checked the nutritional facts in the USDA tables and the only difference seems to be that sweet potatoes have a lot more of simple sugars (I guess that’s why they are sweet). Otherwise, macro and micronutrients are pretty much the same (neither seems to be an outstanding source of any micronutrient). I don’t get the hype about them (and I also dislike the sweet taste).
So you can eat the tuber of your choice, and still live forever (and do pension fraud, up to you). Or carrots, bit healthier and also very tasty.
I am often confused about the sarcasm on this blog!
Specifically: why do you say the theory is weak? I thought the theory was quite consistent and satisfying. Almost all of the positive associations between certain foods and longevity boil down to:
1. The foods that are cited in these type of articles (fruits, veggies, whole grains and beans) inadvertently induce caloric restriction by promoting satiety. Caloric restriction is linked to lower body fat. This is by and large the one factor that confounds “causal” links in most of the epidemiological literature on food. This includes that recent “you are what you eat” documentary on veganism with the pairs of twins (results in that experiment are confounded because the vegan group ate less calories and lost weight: a trite classic!).
2. The fiber in those foods has positive qualities on its own that exceed caloric restriction; e.g. it lowers LDL levels, and leads to lower chance of colon cancer.
3. Limitation of saturated fats – also indirectly linked to those foods – decreases LDL (decreases total cholesterol up to ~20% in RCTs).
4. Variety in food sources prevents deficiencies in specific micronutrients.
From what I know, that’s the theory that has reputable evidence. And if true, it is perfectly consistent with all of the apparently conflicting information we get from news outlets.
Aldo,
The diet advice does seem reasonable to me, perhaps because it’s similar to what my doctor advises. I was just amused by the specificity of the recommendations (“About a cup of beans a day is associated with an extra four years of life expectancy”) and that sweet-potato thing.
In the recent big Harvard epidemiological study on diet, specific food groups, and “healthy aging” the two categories sweet potatoes could fall into did terrible. Potatoes came out below pizza and butter, and just one spot above sugar sweetened beverages, and starchy vegetables came out below sweets and desserts. Sorry, if I’m bring up this subject and paper too much.
https://www.nature.com/articles/s41591-025-03570-5
Richard:
Sweet potatoes are not the same thing as potatoes, right?
Yes, but the starchy vegetables category seems like it would contain sweet potatoes. I was just saying that those were the only two categories they had where one could possibly imagine they would place sweet potatoes, even though you and I would never consider the potatoes category to contain them.
Still, it is interesting that the only place you could very likely see sweet potatoes placed did worse than sweets and desserts! and that potatoes, which seem to be the worst of the starchy vegetables, or at least worse than sweet potatoes, did better, although still bad. The study really has some serious questions of statistical interpretation.
Looking back at that video I linked to with one of the authors, he says:
“Starchy vegetables here, we usually usually include the potatoes, yams,
32:11 and also peas, and also corn. So, these usually are high in starch…We assessed the intake of sweet potato with I think, with yams.”
Also, the category, “Fast and fried foods” did surprisingly well. He responds to a question on this with:
“so I checked the FFQ we used in our cohorts, the exact question to measuring the fast fried food is,
43:07 how often do you eat deep fried chicken, fish, shrimp, clams, or onion rings away from home?”
So, perhaps it did so well because people who are high in this category, when they eat fast food, it tends to be the chicken and shrimp rather than hamburgers. The thing is, there looks to be a great deal of opportunity to misinterpret this study, and it would be great to hear explanations from a top statistician. Also, I wasn’t able to find the complete FFQ? Is there no easy way to get important information like this?
https://www.youtube.com/watch?v=YiNK8hDk5M4
If that study really implies that potatoes are worse than pizza, and sweet potatoes are worse than candy, then I can safely ignore studies coming from Harvard and published in Nature.