Obesity Apocalypse
Even by the remarkably mendacious standards of the “obesity” racket some of the claims in this story are beyond belief.
The most laughable is the idea that by 2048 everybody in the US will be “overweight” or “obese.” This result was derived via statistical extrapolation, the crack cocaine of social science analysis (by similar methods one could prove that within a few generations Olympic sprinters will be running at speeds that will hurl them into low Earth orbit and everyone in America will have a plasma TV seventeen miles wide).
In fact there has been no weight gain at all over the past 30 years in the thinnest quartile of the population — whatever (poorly understood) factors have caused Americans to weigh more on average now than they did in the 1970s have had very different impacts across the weight spectrum: thin people have gained no weight, people in the middle weigh 10-15 pounds than they did 30 years ago, while the fattest people have gained a lot of weight, which is exactly what one would expect. Furthermore, as even this story manages to note, there’s quite a bit of evidence that the trend toward weight gain in the populace in the 1980s and 1990s seems to have plateaued.
But this is a side point. The most significant and symptomatic aspect of this story is it’s completely uncritical attitude toward the current definitions of “overweight” and “obesity.” Those definitons are BMIs of 25-29.9 and 30+ respectively. (You can look up your own BMI here, and I encourage you to do so).
I really can’t emphasize enough how utterly without scientific foundation these definitions are. This can be shown in a hundred ways, but here’s one particularly striking illustration.
The best epidemiological data on the U.S. population is the CDC’s National Health and Nutrition Examination Survey (NHANES). This is universally recognized as the gold standard for such surveys, in particular because it’s a nationally representative sample that directly measures its participants. NHANES has been ongoing since the 1960s; the most recent data that allows for significant followup is from NHANES III, which was assembled in 1988-1994.
Now if we’re facing an “apocalypse” because of “overweight” and “obesity,” we should see evidence of this in, at the very minimum, increased relative risk of mortality among people in these categories. Here’s the relevant data from NHANES III on mortality risk. The following statistics use the mortality risk found among supposedly “normal weight” (sic) people (BMI 18.5-24.9) as the referent group. In other words, the mortality risk for this group sets the baseline for comparison to other groups in terms of their mortality risk. A group that has a higher mortality risk than the referent group will have excess deaths over the baseline risk. A group that has a lower mortality risk will have fewer deaths than would be seen in the group if it had the same mortality risk as the referent group of “normal weight” people.
Most recent excess deaths estimates from NHANES III:
Underweight: 38,456
Normal weight: 0
Overweight: -99,980
Obesity Grade I: -13,865
Obesity Grade II and III: 57,515
Underweight less than 18.5 BMI, normal weight 18.5-24.9, overweight 25-29.9, Obesity Grade I 30-34.9, Obesity Grade II and III 35+ What these numbers mean: In the US population at present, we are seeing about 100,000 fewer deaths per year among “overweight” people than we would if “overweight” people had the same mortality risk as “normal weight” people. Note that the majority of people in the US who according to the government’s current classifications weigh too much are in this group. The “overweight” category is to the obesity panic what marijuana use is to the drug war: stories about an “epidemic” of fatness depend crucially on classifying the 35% of the population that’s “overweight” as being at some sort of increased health risk. This is simply false, and is known to be false by the researchers who are quoted in stories like the one linked above.
But the situation is much more egregious than even this suggests. Note that the NHANES III data reveals that most people who are classified as obese have a lower mortality risk than so-called normal weight people. About two-thirds of “obese” Americans have a BMI of between 30-34.9, and currently we’re seeing about 14,000 fewer deaths per year in this group than would be expected if the group’s mortality risk was the same as that of “normal weight” individuals.
Only when one gets to roughly the fattest 10% of the population does the NHANES III data begin to find a relative mortality risk higher than that found among the supposedly “normal weight.” And even here, the relative mortality risk results in about three times fewer deaths per capita than observed among the “underweight” (there are approximately four times as many people with BMIs 35+ than there are people with BMIs below 18.5).
In short, it’s difficult to convey the utter intellectual bankruptcy of the standard discourse surrounding weight and health in this culture.
Update addressing a couple of common themes in these sorts of discussions:
(1) I don’t think that the higher mortality rate among “normal” (sic) or “optimal” (sic) weight people provides any real evidence that someone with a BMI in that range should try to gain weight. The bogus idea here is that a narrow range of weight is optimal for all people. In fact the differences in mortality across an extremely broad range (roughly BMIs from the high teens to the mid-30s) are statistically trivial, and represent the kinds of differences in relative risk that nobody would ever pay attention to if not for cultural considerations that make body mass a subject of great symbolic (though not medical) importance.
(2) It really is astonishing how ready people are to accept the most dubious evidence for the proposition that everybody should try to be thin, while engaging in sophisticated arguments about why evidence to the contrary can be explained away. That this blatantly inconsistent attitude is characterized as the essence of science is also rather remarkable.
Any study that uses BMI is a fundamentally flawed.
Good digging into the subject.
We’re waiting for the same Neuskultur which can’t explain how nobody but the rich (statistically speaking) made any gains during a 6 year recovery to make sense of this to us?
Is that measuring BMI at death? Because if so, its not necessarily informative. If for example, being obese increases your chance of cancer, then after the cancer weakens you and the chemo knocks your weight down, then once the cancer kills you, you won’t necessarily still be obese at death.
A nice exercise would be to put together a BMI chart of all the US’s Olympic athletes. Apart from the runners and cyclists, I bet you’d find a fair number of “overweight” folks…
by similar methods one could prove that within a few generations . . . everyone in America will have a plasma TV seventeen miles wide
A few generations?! I want it now!!
Paul, Paul, Paul. You’re never going to get anywhere on this obesity thing by trying to confuse people with the facts.
It’s obvious that us normal weight people die more than the overweight people because all those fatsos we have to look at have upset us so much we get ulcers.
And don’t try to tell me that ulcers are caused by bacterial infections instead of stress, either, because I’m onto that trick. I just happen to know that people who have ulcers cause their own misery because of their lifestyle choices. I see it at my kid’s school all the time.
But fat people are ugly an they’re all going to die.
And don’t try to tell me that ulcers are caused by bacterial infections instead of stress, either
This one bugs me. Helicobacter pylori are apparently a necessary but not sufficient cause of ulcers; for instance, this page says that “most H. pylori infections are “silent” and produce no symptoms.” Which means it’s possible that stress is one of the factors that helps an H. pylori infection become an ulcer. Does anyone know any research on this?
…obviously this has no bearing on the original post.
A nice exercise would be to put together a BMI chart of all the US’s Olympic athletes. Apart from the runners and cyclists, I bet you’d find a fair number of “overweight” folks…
I didn’t think to do Olympic athletes, but I plugged in Mike Tyson’s height and weight when he fought Michael Spinks (5’11″ and 218) and he got a 30.4, which is just over the line for obese. Apparently, Tyson would have been a much better fighter if he would have trimmed down to 150.
I don’t know why that smiley face popped up in my comment. Tyson’s weight was 218.
It’s putting an “8″ next to a “)”. Messes you up every time…
One more: LeBron James: 27.5, chub-by.
It’s slightly misleading to invoke the whole ‘but [insert athlete here] has a BMI that would make him obese’. Professional athletes are almost off the bell curve in terms of their physical parameters anyway, and there aren’t that many of them in the first place (at least as a fraction of the population as a whole). If we randomly pick somebody with a BMI of, say, 35, I don’t think anybody would want even betting odds on whether we’re dealing with a linebacker/boxer/bodybuilder or someone who could probably lose some weight.
Apparently, Tyson would have been a much better fighter if he would have trimmed down to 150.
Well, he would’ve been healthier in the short term, since he wouldn’t have had a guy punching him in the boxing ring.
Seriously, the argument from athletes proves nothing, both because athletes are exceptional and because athleticism doesn’t necessarily translate into long-term health. (See Reggie White, for instance.) Best to stick with Paul’s evidence in the OP.
pwned by Ari!
All the “argument from athletes” is intended to prove is that “it’s more complicated than that.” Which ought to be the automatic response to any sort of One Big Number, whether it’s BMI, GDP or ERA.
This post is pretty crude. For one, the conclusion of the study you’re harshing on is clearly conditional: “The results of the study show clearly how the future situation might become if current trends continue.” You don’t appear to be disputing the study’s conclusions on this score, so what, I wonder, is it that you’re getting at?
Also crude is the claim that “if we’re facing an “apocalypse” because of “overweight” and “obesity,” we should see evidence of this in, at the very minimum, increased relative risk of mortality among people in these categories.” No, we’d expect that if we actually took care to make sure that ceteris were paribus, which you haven’t (Flegal et al do better).
There’s a certain lack of humility in the face of what’s we know, and what we don’t, on display here. We know that significant excess weight causally contributes to morbidity, in sundry ways, and there are plausible (and well-documented) biological explanations for how this comes about. There’s some moderately robust evidence that having non-optimal BMI counteracts this (by reducing mortality from lung cancer, for example), but, so far as I know, no well-supported suggestion for any sort of causal mechanism, yet. It’s entirely possible that there is none. In such matters, it at least seems reasonable to lend greater weight to established causal links, less weight to putative causal links that may turn out to be spurious. For some reason, you don’t seem to agree with this.
That’s not to say that there isn’t an interesting and sophisticated debate to be had here. But citing crude statistics about relative excess mortality in various BMI groups isn’t really going to be part of it.
All Obese people should be forced to mate with whales with very large penises.
Nate comments:
“There’s a certain lack of humility in the face of what’s we know, and what we don’t, on display here.”
Speaking of which . . .
“We know that significant excess weight causally contributes to morbidity, in sundry ways, and there are plausible (and well-documented) biological explanations for how this comes about.”
Really? Specific examples, please.
There’s some moderately robust evidence that having non-optimal BMI counteracts this (by reducing mortality from lung cancer, for example), but, so far as I know, no well-supported suggestion for any sort of causal mechanism, yet. It’s entirely possible that there is none.
But as this post notes. the “optimal” BMI definitions have higher mortality rates than the non-optimal ones? What could account this startling paradox?
“In such matters, it at least seems reasonable to lend greater weight to established causal links,”
Which are what again?
“and less weight to putative causal links that may turn out to be spurious. For some reason, you don’t seem to agree with this.”
The reason is that what you’re saying doesn’t address the data that actually exists, but rather keeps using definitions of “optimal” and “non-optimal” that have no basis in the data.
In general, I totally agree. I’m “obese” by BMI standards, but my heart rate, blood pressure, cholesterol etc. are all excellent, and I expect to live to be 100 (my father, who had terrible health habits but great genes, lived to be 94). There’s no question that this issue is being trumpeted a lot to make money for drug companies and the diet industry, and it’s far too complicated to be reduced to BMI. Two caveats: severe obesity is a real problem for certain groups, primarily the poor and First Nations people. Also, the underweight data is potentially misleading because it probably includes a lot of people who are very thin because they’re very sick.
This last is for Matt: you’re quite right that stress can trigger an ulcer, particularly in people who are already infected with H. pylori. H. pylori may be a necessary condition for an ulcer, but it’s definitely not a sufficient one; the population is full of “silent” H. pylori infections.
Snore…..
these posts are a lot like global warming arguments. if you’re on the wrong end of the moral superiority, deny/cherrypick the science.
We know that significant excess weight causally contributes to morbidity,
This word “causally,” I do not think it means what you think it means. (Hint: it is not synonymous with “correlation.”)
BMI is “science”? The claims derived from it are as robust as evidence for global warming? Really? Please explain. (Although at least the invocation of “moral superiority” gives away the show in advance; normally we’re meant to pretend that greatly overstating the independent effects of being overweight derives purely from a rational concern with people’s health.)
Yes, scott, it gives away the show. You’re not interested in what the health effects are; the important thing is to deflate people who have aesthetic values you dislike. There are deep ecologist drawn to claims about global warming and skinny people who are smug because they could get health benefits. Noting that doesn’t really get you far though.
and BMI isn’t ‘science,’ its one of the problems with this study that Campo is always flogging.
If BMI isn’t “science” then there is no “obesity science,” since all claims of an obesity epidemic/crisis etc are 100% dependent on BMI definitions, as are all the studies that people cite on this issue.
And I’m not citing a study here, I’m citing NHANES statistics. There are however hundreds of studies that indicate the lowest mortality rate is found in the so-called overweight category, and that BMI has no socially significant relationship to mortality across a very wide spectrum of weights.
the important thing is to deflate people who have aesthetic values you dislike.
Who says I don’t like them? I don’t care if people find fat people unattractive. I care when people pretend these aesthetic reactions have something to do with health.
I recommend this Circulation paper (which can be freely downloaded), which documents how excess fatty tissue (esp. abdominal fat) does in fact causally contribute to cardiovascular morbidity.
There is no “startling paradox”, Paul. I honestly don’t know what you’re driving at there.
Yes, Scott, I do know what causation is.
I guess I’ll have to phrase it in the simplest way possible. Nate, do you think it makes sense to classify people with a BMI of 25-29.9 as overweight. And if so why?
I recommend this Circulation paper (which can be freely downloaded), which documents how excess fatty tissue (esp. abdominal fat) does in fact causally contribute to cardiovascular morbidity.
If by “does in fact” you mean “may,” then that’s a fair summary of the paper. From the abstract:
“Besides an altered metabolic profile, a variety of adaptations/alterations in cardiac structure and function occur in the individual as adipose tissue accumulates in excess amounts, even in the absence of comorbidities. Hence, obesity may affect the heart through its influence on known risk factors such as dyslipidemia, hypertenstion, glucose intolerance [etc.].” (emphasis added)
And from page 909:
“Abdominal obesity as assessed by waist circumference (independent of ethnicity, gender, smoking status, and age) is associated with increased total healthcare expenditures, especially with the costs of inpatient care. Waist circumference may be a better predictor of healthcare costs than the widely used BMI.” (emphasis added)
Now what were we talking about again?
Paul,
All I’ve been trying to say is that excess fat causally contributes to specific morbidities (and that this is well-documented). If you look at that paper, you’ll find that men and women with high amounts of abdominal fat and elevated BMIs (in the 25-30 range) are at “high” risk for type II diabetes, CVD, and hypertension.
Whether or not this increased risk is offset by decreased risk for non-cardiovascular morbidity is, I understand, an open question. But no one, as far as I know, has an evidence-based account of why excess fat might do this. At this point, it’s a surprising result (given what’s known about the physiology of being fat), but one which may be explained away in future work. At this point, there’s nothing irresponsible about warning people about known adverse-effects of certain body-types.
Except, Nate, for the fact that there’s little people can do to change their body type, and a huge industry that encourages people to do so nonetheless, often with the support of doctors and well-meaning friends and realtives, that can result in far more serious consequences as well as, in many, many case, higher weights over time.
Nate, you’re confusing observational with clinical studies. The arguments for increased or decreased CVD risk and body fat are both based on observational studies. You seem to be under the impression that increased risk associated with higher than average body fat has been confirmed clinically, while decreased risk associated with body fat hasn’t. This is incorrect.
And you still didn’t answer the question.
Last comment. (Feel free to have the last word.)
Obviously I’m making no such confusion, nor did I say anywhere (nor commit myself to accepting) that any of these results have been confirmed clinically (by which I’m presuming you mean confirmed in a controlled study). I should hope that clinical confirmation isn’t required to draw conclusions about what causes what. Unless you’re impugning the conclusions that the studies have drawn (or their basis for having drawn them), I’m not sure what this distinction buys you.
Frankly, I don’t see what your question has to do with my response to your post at all. I’ve exercised my right not to answer it.
What are the units on these statistics? Is that deaths per, say, 100,000, or an absolute number?
Actually, of course I realize deaths per 100,000 is unlikely given the numbers you posted (otherwise almost everyone is dying), but my question still stands – what are the units of these statistics – deaths per (some larger number of people) or are they absolute numbers of total deaths?
Absolute number of total annual deaths in the US.
Nate, you’re confusing observational with clinical studies.
Also, it is impossible to do experimental studies investigating the health effects of “obesity” (and experimental studies are the ONLY way to investigate a causal relationship between two variables). To do that you would have to take a random sample of people from the population, divide it in two random groups, make one of the groups fat while keeping all other factors stable, and then compare the health outcomes for both groups.
Even if one could make everybody in a random sample of people fat (which is impossible due to differences in metabolism in the population) there would still be a number of possible intervening factors (such as the additional social stress fat people are exposed to) that could explain different health outcomes.
I am not arguing that weight does not have negative health consequences independent of the health consequences of being exposed to prejudice and discrimination. In fact, I believe that specifically for people at the the extreme ends of the weight spectrum this might be the case. But I find it frutstrating that possible social explanations why very fat people are on average in worse health are usually ignored – and that although we do know that social factors can have huge health consequences. (Ironically enough the supposed health consequences of being fat are often used as a justification to further increase the social pressure fat people experience; and that although there is not a shred of evidence that this pressure makes people lose weight and keep it off in the long term.)
At 6’2″ 145#, I would be normal? No effing way!
I’m so happy Paul Campos has joined this blog. Any chance for some posts discussing some of the issues of race, gender, and class around the obesity crisis?
The J-curve
There is a long history, from well before the “benefits of red wine” nonsense, of the studies, often funded by the alcohol industry, that purported to show that moderate alcohol consumption actually causes better health results than teetotalling. But because these, like the NHANES data, were from observational studies, they could show only correlation, not causation. The teetotalling groups would, on careful analysis of the data, always be found to be weighted down with people too sick to still be drinking. Of course these people died at higher rates, but it was their bad health causing the teetotalling, not the alcohol abstinence causing bad health.
Not having made that careful study of the NHANES data, I can’t give a firm conjecture on which direction the predominant causal arrows run in the various weight classes studied. But, given the strong known association of conditions that cause mortality and morbidity also causing low weight, I think it is quite rash to assume that the excess mortality/morbidity in the low weight groups doesn’t derive from these conditions, rather than lower, recommended weight, which is well known to decrease the risk of any number of unhealthy conditions. It is certainly uncalled for to call people on the other side of this dispute liars or fools based on what will probably prove to be just another J-curve study.
Paul and Scott are attempting a legalistic, discussion. This is completely the opposite of what a responsible scientist would do: namely present the state of knowledge whether or not it agrees with their point. And this is the root of why I find the exercise ultimately unpersuasive. It is simply too easy to fool yourself if you dismiss contrary information or look for the one way to see the numbers that fits what you want.
For example, any weight vs. mortality study has to account for the severe weight loss that accompanies many terminal illnesses. Not doing so is a legalese trick – failing to mention an inconvenient datum that undercuts your case. As a scientist this makes me simply bristle. That’s just not how the game is played.
You know what makes me bristle? People who don’t know what they’re talking about spouting off on this subject with a bunch of sanctimonious bullshit about what “scientists” do, as if scientists are some unique breed of human uncontaminated by economic, cultural and ideological factors.
The occult wasting syndrome argument — that thinner people have higher mortality rates than all “overweight” and many “obese” people — is a favorite of of these very unlegalistic, utterly disinterested scientists, who, when the crudest epidemiological associations favor their argumemts, treat epidemiology as if it were high energy physics. But as soon as the data doesn’t support them, they instantly turn into a bunch of postmodern skeptics who make Nietzsche look like a logical positivist.
In fact the NHANES data remains substantially unaltered if you exclude everybody who dies within a year or three years of entry into the study, thus disposing of the occult wasting hypothesis.
Absolute number of total annual deaths in the US.
If that’s the case, these numbers are meaningless without context. Absolute numbers tell you nothing about relative risk. A decreased total number of deaths in a particular class doesn’t mean a decreased risk of death if that particular class is smaller by the same percentage. That is, if the overweight class is, say, 40% of the population, and risk of death were equal, you’d expect 40% of the deaths to be in the overweight class and 60% in the normal/underweights. That would result in a substantial apparent decrease in absolute numbers, but it wouldn’t mean any decrease in risk.
So what I’m asking is, what are the numbers when expressed relative to population frequency for these groups?
Above should read “that thinner people have higher mortality rates because they’re already sick when their weight is measured.”
The numbers are adjusted for percentage of the population represented by each group. The 100,000 excess deaths among “normal weight” people are the 100,000 excess deaths associated with their higher relative risk of mortality when compared to “overweight” people. If the two groups had the same relative risk there would be no excess deaths in either category, no matter how many total deaths there were in each category.
“whatever (poorly understood) factors have caused Americans to weigh more on average now than they did in the 1970s”
I’m putting my money on overeating. However, you are welcome to continue believing that someone sprinkled magic fattifying dust over the US, Paul.
Fine, but that’s exactly the opposite of your answer to my initial question. You said it was an absolute number, when it apparently isn’t. This is what I wanted to know, because it’s not clear in the post, at least not to me.
Marc – I would be careful what you say. You are not the only “scientist” around, and you are certainly not the only one who understands that for correlational studies to be meaningful you have to control for all obvious intervening factors (and even then you cannot show causality with a correlational design).
Concerning your argument: This is not a topic that’s part of my field, but as far as I know there are also a number of diseases that at least initially cause weight gain (through corticosteroid treatment for example, or through water retention). In a well controlled study these factors also would have to be taken into account – yet, as far as I know researchers are far more likely to control for weight loss due to disease. (And I am not going to believe that weight gain due to disease is much more unusual that weight loss until someone has shown me data supporting that.) Also, practically no study I have ever seen controls for social stress, which could very well have a significant negative influence on the health of fat people.