Being overweight may not extend life after all


As of September 2017, new Sweat Science columns are being published at Check out my bestselling new book on the science of endurance, ENDURE: Mind, Body, and the Curiously Elastic Limits of Human Performance, published in February 2018 with a foreword by Malcolm Gladwell.

- Alex Hutchinson (@sweatscience)


Is it good to be a little bit overweight as you get older? That’s the message from a series of large, reputable studies (including one that made lots of news two years ago). The theory is that having a few extra pounds is good as you get old and frail, so that you don’t waste away to nothing if you get sick or break a hip. And the data seems to back up this theory.

But wait a sec. A new study in the Journal of the American Geriatrics Society from researchers at Loma Linda University in California (abstract here, press release here) shows the opposite:

[M]en over 75 with a body mass index (BMI) greater than 22.3 had a 3.7-year shorter life expectancy, and women over 75 with a BMI greater than 27.4 had a 2.1-year shorter life expectancy.

Why the contradiction? Actually, there are some very good reasons. The authors start by explaining some of the potential flaws with the earlier studies that seemed to show that being a bit chubby helped you live longer. The key problem is that the earlier studies measured weight just once, at the beginning of the study, and then waited to see how quickly the subjects died. But people who were skinny at the beginning of the study might have been skinny because they were already sick, possibly without having being diagnosed yet. If you’re skinny because you’re partway through a dramatic weight loss than ends with your death, that’s not quite the same as being thin with a long-term stable weight and healthy lifestyle.

The new Loma Linda study is able to address this problem, along with a few others. It relies on two large studies of Seventh-Day Adventists in California, one in 1960 and the second in 1976. Weight was measured on both occasions, and mortality was monitored until 1988. By looking only at the people whose weight stayed the same (within 5 kg) between 1960 and 1976, you get to study the effects of having extra weight, rather than the effects of gaining or losing it for whatever reason (both of which could have significant effects, but are separate questions).

Another quirk of the study is that Seventh-Day Adventists are forbidden to smoke. So to further eliminate confounders (e.g. people who were thin because they were puffing on cancer sticks), the analysis was able to completely exclude any past or present smokers, plus anyone with any history of coronary heart disease, stroke or cancer at baseline, and still have a total of 6,030 people aged 25-82. The point here is that we’re looking just at the effects of extra weight, stripped away of as many confounders as possible.

And sure enough, higher BMI corresponded to higher risk of death for 75- to 99-year-olds: above 22.3 was bad news for men, and above 27.4 was bad news for women. Obviously this study can’t tell us why there’s a difference between men and women, but the researchers suggest that women may benefit from having a bit more fat because that’s where estrogen is produced after menopause; too little estrogen may leave them susceptible to everything from hip fracture to insulin sensitivity problems. Here’s how the “hazards ratio” (basically your odds of dying relative to the healthiest group) looks as a function of BMI:

So what does this mean? Well, we’re not going to rewrite the medical textbooks based on a few Seventh-Day Adventists in California. It’s just one study, etc. etc. But given the methodological differences between this study and the previous ones, it certainly seems reasonable to wonder whether the apparent paradox of overweight people living longer is simply an artifact of the confounders this study managed to eliminate. I’m not saying you should panic if you’re overweight — I’m just saying you shouldn’t panic if you’re not.

5 Replies to “Being overweight may not extend life after all”

  1. I’m a bit surprised that the BMI graph starts at 19 (for women). I know some tiny older women who have BMIs lower than that. Thoughts?

  2. @Audrey: Agreed that the cut-offs in those graphs are a bit arbitrary. I think that’s just a function of the population studied in this paper. There were 884 stable-weight women in the lowest BMI quintile (less than 20.6), of whom 478 died within the 29-year follow-up period. Presumably the number of deaths with BMI below 19 was getting low enough that the data was no longer reliable enough to be worth plotting.

    In terms of the health of these tiny older women with BMI less than 19, one key question is how tiny they were earlier in life. What’s unique about this study is that it was able to filter out anyone whose weight had changed by more than 5 kg. The fact that it produced different results compared to earlier studies suggests (though by no means proves) that it’s thin older people who’ve lost a lot of weight as they age that drag down the overall survival numbers for thin people. Which is another way of saying that it’s important to ask: why are they that thin? If something has changed, what is it?

  3. As to the study Chris Robins refers to: Table 3 shows that people in stage 1 of the scale have a slight, but significantly lowered risk of heart disease, and no (significantly) elevated overall risk.

    Indeed, this seems to be in line with the earlier studies that suggest that being merely overweight poses no real health risk. That is, if a BMI between 25 and 30 can be equated with stage 1 of the EOSS scale.

Comments are closed.