Whatever Happened to the Obesity Epidemic?

Half asleep silliness

Recently, Resident Contrarian has been writing a series (Part 1 and 2, plus more to come it seems) picking apart some fellow’s article claiming that, essentially, all this diet and exercise stuff is nonsense, and the reason for the obesity epidemic is that there’s all this lithium and other chemical stuff in the ground, which gets into our food, which gets into our bodies and really messes up our body’s ability to equilibrate our weight without us trying. It has been an interesting read, partly as an exercise in tearing apart pretty sloppy confirmation biased reasoning.

The other interesting part for me was that I hadn’t heard about “the obesity epidemic” for so long that I thought it had stopped being a thing, either because it stopped being real (or was found to never have been so) or because scientists weren’t allowed to say that being obese was bad anymore¹. Thinking about it a bit between installments, I came up with a few reasons why the obesity epidemic might never have been a real thing, regardless of whether or not we are allowed to say that being obese is bad, and I wanted to test them out.

To start with, I trotted over to Our World In Data to look for some numbers. Let’s see… obesity… rate… USA…

Oh, wow, right to deaths. Ok. Let’s add some countries for comparison.

Huh… so age standardized, that’s nice. USA is a bit higher than Scotland and Sweded, but they are all decreasing… interesting… WTF is happening in Bahrain and American Samoa?! Did Coca-Cola get a deal where they replaced all the tap water with Fanta or something?

Better look at the obesity rates… I mean share of men/women who are obese… apparently rates only applies to deaths here.

Huh… so ok I see where the percentages of people being obese is going up and up, the USA going from ~10% of men and women to about 32% since ‘75, although deaths start in 90 so about 17% of men and 21% of US women since 90. Samoa (which I assume is the same as American Samoa?) is pretty darn high still, so that makes sense that they would have more deaths. Bahrain though… the men are much less obese, and the women are not that much above the USA by the end of it, but wow, that’s a lot of deaths for how many obese people they report. It looks like Bahrain should have death numbers closer to the UK or Sweden, unless their ratio of male/female is really skewed towards the women.

Speaking of age adjusted deaths… what is the median age over this period?

Hmmm so Samoa skews really young, Bahrain is drifting older, and Sweden, the UK and the US are all trending older as well. Notably, the US, UK and Sweden are in the “the median person is middle aged” range.

So… what was I checking here? You might have started to guess, but my first thought was this:

1. Assumption: obesity is unhealthy, i.e. it kills you off before you would have died otherwise, likely by a heart attack

2. Assumption: we have gotten better at keeping people from dying from heart attacks in the past 40 years²

3. Expectation: I expect median age to increase along with the obesity rate, as fewer obese people die early and so stay in the sample longer

Which… looks about right. Median age creeps up about 10 years in that 1990-2020³ period, or ~30%, while the share of obese men goes from 17-32% and women 21-32%. The death rate from obesity is pretty slowly decreasing, but I am not really sure how that is measured so it might not matter, but more obese people isn’t causing an uptick in deaths from obesity, so people are apparently dying from it less? In any case, if obese people were being removed from the sample at a younger age before, and then stopped being removed so early, we would expect the rate of obesity to go up without any underlying change in the obese causing factors. There would just be more obese people around to count above ground.

But then there might be something else… obesity is correlated with age, and the age demographics have been skewing old not only because we keep people alive longer, but also because there are fewer young people per old person.

1. Assumption: obesity is correlated with age

2. Assumption: people have fewer kids than they used to, skewing the age distribution older

3. Expectation: I expect median age to increase along with the obesity rate, as the population gets older and fatter on average

Well, that fits the US data, too. Really it fits all the data pretty well, as all the countries are getting older and fatter, albeit at different rates (and with different death rates attributed to obesity). Except the rates of obesity are different, particularly with the western countries… and I don’t know their birthrates, or if they have changed. Let’s see…

Huh… so Sweden, the UK and the US all have pretty similar birthrates now, but the US used to have a higher one and it dropped more sharply. Sweden and the UK both had little rallies around 2000, even. Well, that does explain a bit why the rate of growth in obesity has been slightly slower in the UK and Sweden when compared to the US, as the US age distribution changed a bit more sharply, but it isn’t amazing, no smoking gun. It is pretty telling with Bahrain and Samoa, perhaps, as their birth rate dropped quite a bit, but not enough to really get serious about digging into the numbers more.

If we put all four assumptions together, they do all push in the same direction, however. None are mutually exclusive, or even count the same thing, so that seems legit.

1. Assumption: obesity is unhealthy, i.e. it kills you off before you would have died otherwise, likely by a heart attack

2. Assumption: we have gotten better at keeping people from dying from heart attacks in the past 40 years

3. Assumption: obesity is correlated with age

4. Assumption: people have fewer kids than they used to, skewing the age distribution older

How much of the increase in obesity do those four factors account for since 1990? I have no idea. I expect lower cost calorie dense food, less labor intensive work, less walking everywhere in general, far more time spent watching videos and computer use, and possibly less stigma towards being overweight have all added to it as well. I expect those four factors come into play pretty heavily, however, but since all the countries I peeked at (most of the western ones) had roughly the same patterns with just a few ups and downs of share of obese people, it is hard to suss them out. If only France had outlawed web streaming in 2010 or something…

Wait, come to think of it, obesity is determined by BMI, body mass index… I kind of remember hating that methodology. Let’s see here… yup ok, from the graphs above, a person is considered “obese” if their BMI >= 30, where BMI is weight in kgs divided by height in meters squared. So, let’s see… I am

6 ft = 1.828 meters

205 lbs = 92.99 kg

1.828^2 = ~3.342

So my BMI = 27.82

Whew! Just under… good thing I lost 15 lbs since Christmas, since 221 lbs would have made me obese. Still, according to the CDC I am overweight, but fuck those guys. I am no gym rat, and I definitely gained a good bit of weight in the last 6 years, but then a 36” waist for a 6 foot guy doesn’t seem too excessive.

But what if I were a gym rat? If I were to get swole, how much should I expect to weigh? Are there 220 pound 6’ guys out there due to sick gains in the gym?

Ok, let’s toss “I am 6' 220 lbs” into Duck Duck Go’s image search and see what comes up…

uhm ok… I don’t know what I was expecting, but apparently lots of people post before and after pictures going from lots down to ~195-220 at the gym, and being pretty well muscled. Some of the guys that look heavier don’t seem to list weights, but they look significantly larger than even the 220 lb guys.

Still, I don’t think I can add “More people are working out a lot to get swole, and it is screwing up the BMI numbers because it is making very healthy people look obese.” I recall that used to be an issue when I was in college, tall people really screwing up BMI indexes especially if they were not rail thin, but that must have gotten fixed? I dunno.

So, why bring this up? Well, has anyone made the argument that the obesity epidemic is in large part due to a population that is getting older, both because obese people do not get selected out earlier and because there are fewer younger people being born? I have never seen that claim, but maybe it is common somewhere in the literature. That alone doesn’t explain why the US obesity share is higher than European countries that are equally old, but it does explain at least a fair bit of the increase over the past 30 years or so.

That is, if you are wondering “What changed in the US that so many people started being obese?” a large part of the answer is probably “They started being older.” Somethings to control for when doing your regressions, at least.

1. Is it a sign that the American left has become more extreme when that last bit is a reasonable possibility?

2. Statistica tells me 163.6 deaths per 100k in 2018 vs 412.1 in 1980. That matches my anecdotal priors, so sure, why not.

3. In retrospect, I don’t know why I normalized the start date to 1990… that’s the start of the “death rate from obesity” range, but that’s also the one I care about least. It doesn’t change things much as the trends are all pretty constant, but just in case you were wondering why I did that. (I also have a vague notion that people start the obesity epidemic in the 1990’s, but I don’t know why I think that.)