The obesity myth revisited

Football

I have an op-ed on the new JAMA meta-analysis which concludes that a BMI between 25 and 35 correlates with a lower mortality risk than that observed among so-called normal weight people. (In America, the former group includes nearly 80% of everyone who public health authorities claim weigh too much).

I’m not under the illusion that a three-million person study authored by five distinguished senior scientists and published in the nation’s leading medical journal is going to actually cause anyone in a position of authority to reconsider anything — for reasons that I allude to in the piece, the actual data still have almost no effect on public policy in this area.

Still, it’s an encouraging sign that the obesity racket continues to be exposed as a product of an invidious combination of cultural obsession, and the economic interests that obsession generates.

A brief note on hazard ratios: Something that ought to tip off the skeptically-minded about the degree to which the focus on weight has nothing to do with mortality risk per se is just how minor the correlations observed in this area are. For example, it’s true that the fattest people in this study — those with a BMI of 35 and above — had a 29% higher mortality risk than the “normal weight” (sic) reference group. But what people tend not to take into account about these sorts of statistics is that, for most demographic groups, baseline mortality rates are extremely low, which means a few extra deaths will produce an impressive-sounding spike in relative risk.

For example, if you compare the risk that a 50-year-old man will die within the next five years to that of a 50-year-old woman, you’ll find that the man’s mortality risk is 71% higher. That sounds pretty bad, especially if you happen to be a 50-year-old man, but what this actually means is that the man has a 2.51% chance of dying over that five-year span, rather than a 1.47% chance. And note that this hazard ratio is nearly two and half times higher than that found among the very fattest people. So among the middle-aged, gender correlates far more powerfully with mortality risk than even the highest levels of “obesity.” (No word yet on what the government plans to do about the masculinity epidemic).

And of course we shouldn’t lose sight of the even more significant fact that we’re talking about correlations in observational studies, rather than any clinical demonstration of real causality. But when you can’t even demonstrate a correlation in the data for your thesis, you should probably reconsider it.

166 comments on this post.
  1. Malaclypse:

    I’m calling the over on 400+ comments.

  2. Timb:

    And for most of those people to either be unaware of he study referenced by Paul or rejecting the idea out of hand. Now, shhh, Mal, I’ve got a plate of hash browns to finish

  3. Timb:

    Stupid reply button

  4. c u n d gulag:

    Ixnay onya ifty ivefay!
    I’ll be turning 55 in less than 2 months!
    Hopefully…

    Churchill, among his many great lines, was once asked if he exercised?
    And he said, to paraphrase, since I probably won’t get the quote right, ‘Yes. And I get all of my exercise being a pallbearer for my friends who exercise.’

    Here in America, we don’t die from hunger, like people in other countries in the world.
    No.
    We die from being obese, due to too many of us eating shitty, low-cost, high-fat, high-carb, high-chemical, high-fructose syrup, foods – because that’s all a lot of us can afford.

  5. Karl:

    As the authors point out, heavier individuals have “greater likelihood of receiving optimal medical treatment,” that is, they are on pills to reduce blood pressure and improve cholesterol ratio. Those two factors are uncontroversially associated with lower mortality. So we have a big government conspiracy to get people to lose weight and avoid the cost and side effects of the pills.

  6. ajay:

    But a study which finds significantly higher mortality among the obese isn’t really destroying the “obesity myth”, though, is it? It’s destroying the “overweightness myth”.

    And the second half of this post is a terrible, horrifying trainwreck of STATFAIL.
    But what people tend not to take into account about these sorts of statistics is that, for most demographic groups, baseline mortality rates are extremely low, which means a few extra deaths will produce an impressive-sounding spike in relative risk. … oh, God, no. This is the kind of thing that drives Nate Silver into an early grave.

  7. Clark:

    So is not dying the only standard we should care about?

  8. Not Nate Silver:

    Care to explain why?

  9. Marc:

    Paul conflates two things. One is a reasonable critique of whether there is a correlation between relatively mild weight gain and increased health risk. This is a difficult thing to establish one way or another. (He tends to assume that “difficult to tell one way or another” automatically means “definitely no health problems”, of course, because he’s attacking a scientific problem as a legal advocate.

    There is also the far less controversial claim that true obesity has serious health consequences. Paul does the global warming denier limbo here – insinuating that there is actually no problem while maintaining deniability.

    And then he claims (in other contexts) that it’s impossible to lose weight – ignoring things like the fact that we used to weigh a lot less than we did, or the fact that people in different countries with different diets have very different mean weights and distributions of weight. The latter is absolutely deadly to a claim that obesity is just some sort of cosmic accident, as opposed to being a direct consequence of public health choices and the marketing efforts of large corporations.

    It’s the latter two tendencies that makes me lose respect and question the “facts” that we’re being given. There are real things like diabetes that are related to obesity. The mean population BMI really did used to be much lower. We don’t need sugar water sold in public schools; kids should walk to school; and we can choose public policies that would lower the mean weight if we deemed it important to do so. I’m reluctant to agree with the more reasonable points when they’re being used to advance a flatly anti-scientific agenda on other fronts.

  10. Kurzleg:

    Paul –

    Since you seem to be pretty heavily engaged in this subject, I have a question I hope that you can answer. My employee rewards employees who are able to get below a certain BMI threshold with discounts on insurance premiums. The implication is that lower BMI results in lower healthcare costs since presumably a person under the BMI threshold (29) will require fewer healthcare services. Have you seen any evidence that this is actually the case?

    (It occurs to me that healthcare services consumed by those considered overweight may be the result of that very categorization. It would be quite ironic if these were indeed the case.)

  11. RedSquareBear:

    No, but it’s probably one of the easier-to-measure health outcomes?

  12. Kurzleg:

    “my employer rewards…”

    Yes, I WOULD appreciate an edit function.

  13. Sherm:

    The problem isn’t that there is an obesity paradox, but that obesity is poorly-defined by BMI. At 6’0″, 225, I’m considered obese, although I can run 7 or 8 miles at a time, and have a muscular build, and I need to get below 180 to be considered at a “regular” weight. I was below 180 for a very brief period of time in law school, and It was absolutely impossible for me to maintain, and I looked like shit. If obesity and regular were properly defined, the paradox would not exist. Also, you have to consider not just the mortality rates, but healthcare costs incurred for the medical care used by the obese to treat the conditions caused thereby, such as diabetees and heart disease. While it’s true that we have warped notions regarding body types and obesity, no study is going to convince me that the truly obese are not at a much greater risk of suffering from numerous conditions which lower their quality of life, suck money out of the healthcare system, and lead to premature death.

  14. Marc:

    …and the fact that he’s dismissing epidemiology in the last sentence because you can only demonstrate correlation is straight out of the tobacco lobby handbook. (e.g. you can’t actually prove that smoking causes cancer.)

  15. jon:

    Ah yes, the new new study certainly shows that Fatsos have a far lower mortality rate, per pound.

  16. Sherm:

    It took a whole 15 comments for a judgmental asshole to appear!

  17. ajay:

    Because things like significance tests are specifically designed to take this kind of thing into effect. Just because something is a low-frequency event (and one in eighty is not, really, a very low frequency) does not mean that it is immune to statistical analysis!

  18. Kurzleg:

    I had the same question. Higher healthcare consumption might be linked with high BMI. I don’t happen to know (or have the time to find out) if this is actually the case. As I note below, my employer and its insurance company seem to believe it is.

    Quality of life also seems like it might be an issue. YMMV, but I can’t imagine being unable to participate in sports due to my weight. And for those of you with children, I’d imagine that one’s weight preventing one from engaging with one’s kids might be a downer. Basically, weight limiting freedom of action seems like a problem.

  19. newsouthzach:

    Scattered, smothered, covered, diced, topped, peppered, and chunked, please!

  20. Dave:

    Jeez, take a joke! Fat people are so sensitive. Guess it goes with being immoral scum who have no self-control.

  21. Dave:

    This. BMI is a broken tool.

  22. ajay:

    Sorry, “into account”, of course.

  23. Kurzleg:

    To quote Paul:

    But when you can’t even demonstrate a correlation in the data for your thesis, you should probably reconsider it.

    He’s saying exactly the opposite of what you’re accusing him of.

  24. Kurzleg:

    Seconded. A few years ago when I was pretty fit and racing my bike (even winning a few races), my BMI was in the overweight category.

  25. Paul Campos:

    My point was that an impressive sounding increased risk of a low-probability event is still probably going to be a low-probability event. If one in 100 women die over a five-year stretch while two men out of 100 die, that’s a 100% increase in mortality risk associated with gender. It’s also one extra death among 200 people over five years.

    This has nothing to do with statistical significance and everything to do with social significance. They’re not the same thing, although the word “significance” is being used in both phrases.

  26. ajay:

    That would be “morbidity”, Kurzleg. Strokes, diabetes, skin problems, and so on.

    http://en.wikipedia.org/wiki/Obesity-associated_morbidity

  27. ajay:

    BMI isn’t a broken tool at the population level, but it’s not much use at the individual level – which it wasn’t designed for anyway.

    It’s perfectly possible to have an “obese” BMI and not be any more unhealthy than someone with normal BMI – but it’s not the way to bet. Nor does the existence of an unemployed Harvard graduate destroy the assumption that a Harvard degree helps you get a job, to use an analogy.

  28. MacGyver:

    BMI doesn’t take into account muscle mass. It should not be a measured used by any one person or publication. VO2 MAX and body fat percentage are much more reliable indicators.

  29. xxy:

    The study did in fact conclude that the truly obese (BMI >30) had a significantly higher mortality rate. Only the “overweight” category (BMI 25-30) was associated with a significantly lower mortality rate than a “healthy” BMI (18-25), but as you correctly note it’s common to end up in the “overweight” range due to muscle mass. The results seem to be consistent with a correlation between body fat percentage and mortality rate. The first result for google scholar search of “body fat percentage mortality” shows I’m not the first one to think so.

  30. ajay:

    Your example would make for an extra 300,000 male deaths per year in the US – more or less what the US would see if it were fighting the Axis and the Confederacy simultaneously. That is, in fact, “significant” in the social sense. HIV, to pick something that is definitely of social significance, never got anywhere near that sort of mortality.

  31. xxy:

    My bad, I meant to say truly obese as in BMI >35.

  32. Kurzleg:

    Yup. But those are relatively difficult to measure while BMI is very simple, which is why the latter has gained popularity.

  33. Kurzleg:

    Give me examples of “population level” and “individual level” applications. It’s hard for me to understand how something that’s not reliably accurate in characterizing the particular has any validity in characterizing the general.

  34. timb:

    And, the cost of those meds is pretty staggering. It’s why I want to lose weight. Can’t change family history, but I can save some cash

  35. Marc:

    The sentence that I was responding to:

    “And of course we shouldn’t lose sight of the even more significant fact that we’re talking about correlations in observational studies, rather than any clinical demonstration of real causality”

    This is straight out of the climate denier and tobacco lobby playbook, period.

  36. Paul Campos:

    Yes, observational studies have zero evidentiary value — that’s exactly what I’m saying.

    Other things I’m saying:

    Everybody should try to weigh as much as they possibly can.

    Exercise and good eating habits do nothing to improve health.

    I want to have sex with fat women.

    Hitler was misunderstood.

  37. ajay:

    No one’s collected mortality data for 2012 yet, as far as I know – it’s only three days into 2013! Do you mean 2011? It’s difficult to tell where you’re going wrong without a source.

  38. ajay:

    Huh. That’s weird – the Campos comment I was replying to has vanished. Guess he spotted the error.

  39. ploeg:

    I merely note in passing that 50 year olds typically don’t expect to die within that five-year timeframe, and I know one 50 year old in particular who was told that an operation that he was to have was “fairly routine, only about a one-in-one-hundred chance of death” and promptly proceeded to lose his shit. (Doesn’t mean that he’s ready to quit smoking tho.)

  40. tt:

    So what? Denialists sometimes use arguments with valid form. It’s a valid critique of some epidemiological studies that establishing correlation isn’t the same as establishing causation. It’s a critique actual scientists use all the time. Note that in both the tobacco and climate case we have an extremely plausible mechanism in addition to the correlative evidence, and it’s not clear that exists in the obesity case.

  41. rea:

    immoral scum who have no self-control

    And yet, here you are, incapable of stopping making offensive comments on blogs . . .

  42. Manta:

    Paul, if you think that lots of people, quite a few of whom would normally be sympathetic with your side, misunderstand your point, you should write better.

    When I referee an article, if I don’t understand it I ask it to be rewritten: the same advice applies here: rewrite and resubmit

  43. ajay:

    It’s hard for me to understand how something that’s not reliably accurate in characterizing the particular has any validity in characterizing the general.

    That’s what I was trying to get at with the analogy. Say you have a population of 1,000. 100 of them went to Harvard, of whom 70 are employed – 70%. The other 900 went elsewhere, and 450 of them are employed – 50%.

    Now, at the population level, there is a significant correlation between “going to Harvard” and “getting a job”. (Statistically significant, that is.)

    But, at the individual level, you’d be stupid to say to a Harvard grad that you met “of course, you must have a job” because 30% of them don’t.

  44. Sherm:

    I though Dave was joking, while Jon was being obnoxious. But what the hell do I know?

  45. njorl:

    I’d say it’s a broken tool for personal health services, but a functional tool for broad statistical analysis. Attempting to use body-fat percentage measurements would probably be less practical for statistical analysis. Height, weight and mortality data are widely available.

  46. njorl:

    …as ajay pointed out if I read one response further.

  47. catclub:

    Wow, I am on BP and cholesterol lowering drugs, at around 10 cents a day. I am feeling not very staggered.

    Oh, it is each, so $70/yr.

    I must say I am surprised that BMI up to 35 shows little correlation. I am not surprised that BMI up to 30 is lower mortality ( or lower) than BMI 18-25.

    Maybe I should look at the study details.

  48. Richard:

    There was an NPR piece on the study yesterday with pointed criticism from some heavyweights in the field about drawing any conclusions from the study because of problems in the analysis (one was that extremely skinny people are often that way because of severe illnesses that lead to death and this skews the result). I’m no doctor but the critics seemed to be making sense. Paul, no doctor, applauds the study because it bolsters an argument he’s been making for years. More open minded and skeptical folks might want to hear both sides of the medical debate

  49. ajay:

    There was an NPR piece on the study yesterday with pointed criticism from some heavyweights in the field

    …maybe a better term could have been used…

  50. Sherm:

    I think its pretty fucked at the population level as well. Here’s my result from the CDC:

    Height: 6 feet, 0 inches
    Weight: 225 pounds

    Your BMI is 30.5, indicating your weight is in the Obese category for adults of your height.

    For your height, a normal weight range would be from 136 to 184 pounds.

    In what world is 6’0”, 140 pounds normal, whereas 6’0″ 190 pounds is overweight? If I ever get below 175 pounds, it’ll be while wasting away on my deathbed in a cancer ward.

  51. njorl:

    I think it would be easier to explain without analogy.

    If you’re doing a study, a larger pool of data helps. You can get BMI data easily. It’s just height and weight. However, that data pool will contain skinny couch potatoes and heavily muscled athletes. Body fat percentage might be a much more useful statistic to analyze, but you can’t get it as easily. So, you assume in a large pool that the low BMI people with high body fat and the high BMI people with low body-fat are rare and also tend to cancel each other out.

    That doesn’t work for individuals. A doctor doesn’t tell the guy with a 20 BMI and 40% body fat that there is a weightlifter in the next room with a 40 BMI and 5% body fat, so it’s OK.

  52. catclub:

    I will take your word on that,
    but apply Campos argument to 50,000
    men and 50,000 women. One woman dead and two men dead is still double the relative risk for men, but far lower overall numbers of deaths (which you emphasize), still same relative risk number.

    So the relative risk number needs to be related to overall number of deaths. Sounds like the usual sensible, accurate, statistical approach that is applied to diet and BMI studies. ;)

  53. Sherm:

    at around 10 cents a day.

    10 cents a day to whom?

  54. timb:

    some of the drugs comes subsidized by my insurance, but others don’t.

    You are correct that I was using hyperbole on the staggered part (although at 80 bucks per month after insurance, it always feels like there’s something else I could spend that on….after all, isn’t that 16 Big Mac value meals?)

  55. Paul Campos:

    Well yes but there’s also the problem of willful misreading. The proposition that observational studies have important limitations is one that any epidemiologist would consider completely self-evident.

  56. tt:

    (one was that extremely skinny people are often that way because of severe illnesses that lead to death and this skews the result)

    But the interesting result here is the comparison between moderately overweight (BMI 25-30) vs. “normal” weight, which doesn’t include extremely skinny people.

  57. catclub:

    Remember when Hillary and Bill made a commercial to counter the anti-Hillarycare commercials?

    They were reading the fine print of the insurance and said that ‘Here is says: you may die’. It is still too much information, for many people.

    ‘gist’ quotes for illustration value only, not intended as ‘verbatim’ quotes.

  58. Sherm:

    But the cost is still staggering to the healthcare system and is spread through higher premiums for all.

    Timb — I’m in the same boat as you — trying to combat family history of HBP. I’m not on meds and hope to stay that way through exercise and diet (following the DASH diet guidelines). I’ve been tracking my BP rather religiously for several months, and I have found that on the days when I exercise, its in the normal range, and that on days when I don’t exercise, its in the pre-hypertension or borderline hypertension range. So, for what its worth, I have concluded that daily exercise is the best way to go.

  59. catclub:

    Good point. But generic price from Walmart pharmacy is pretty low.
    And these are generics.

  60. Snarki, child of Loki:

    There’s a special button for “Stupid Replies”?

    Sounds like it could come in handy during troll-infestations.

  61. Sherm:

    Normal does in fact include extremely skinny people. As stated above, I am “normal” at 6’0″ 140 lbs. That’s extremely skinny.

  62. Richard:

    I was listening in the car on the way home so wasnt paying very close attention. The point about sick and skinny people was brought up but I dont know how the skinny people point affects the comparison if it is only between moderately overweight and normal. However it was some acknowledged expert (I wont use “heavyweight” again) who made the point so I’m assuming it had some relevance. My real point is that I’m going to listen to the medical debate rather than follow Campos’ lead because he’s far too much of an advocate on this subject.

  63. catclub:

    When Willie Mays came up to the majors, he was 6-0 and 180 lbs. Why can’t you be like Willie?

    A friend tells a story of being berated by her piano teacher. The teacher was always saying, “Why can’t you be like Lenny, Lenny can play this.” Turns out, Lenny was Leonard Bernstein.

    So, relative to another post, good on you, for exercising!

  64. Malaclypse:

    I think rea is a victim of Poe’s Law.

  65. Richard:

    Well if you are included as normal, then I think its a valid point. Many of the people who are 6′ and 140 lbs are that way because of some medical condition that would have an effect on mortality.

  66. njorl:

    That’s an interesting study, judging just from the abstract. Body fat mass correlated monotonically with higher mortality and non-fat body mass correlated monotonically with lower mortality. The “U” shape of the BMI curve is the result of these effects combined.

    I wonder if a larger study would show negative effects at the extremes – higher mortality with extremely low body fat or with extremely high non-fat body mass. I know there are medical problems which arise from low body fat, but maybe they always co-occur with low weight in general. At the other end, I know those with gigantism or acromegaly have high mortality.

  67. catclub:

    Nope, just took that long to post. I am sure we were here earlier.

  68. Paul Campos:

    From the JAMA study:

    “The results presented herein provide
    little support for the suggestion126
    that smoking and preexisting
    illness are important causes of bias.
    Most studies that addressed the issue
    found that adjustments or exclusions
    for these factors had little or no
    effect.”

  69. Walt:

    No, Paul is completely right about this. Epidemiological studies are super, super questionable evidence, just because it’s so hard to control for everything.

  70. sparks:

    I took it that Dave was joking, too.*

    *I was obese only in my teen years, never before that or since I was 18. Had lifelong thin classmates die or suffer from terrible diseases including cancers, strokes, etc. In fact, one was an extremely fit health nut who ran daily, and was obsessive about same. Didn’t stop her from getting cancer. Twice. Another fit friend had a stroke in his 40s.

    However, I am convinced that obesity causes/exacerbates certain diseases. My guess is stress will be recognized as a major contributor to ill health as well.

  71. Walt:

    Remember, every time you fail to get a joke on the Internet, an angel has its wings ripped off.

  72. Sherm:

    Willie was only 5’10″, so he was “overweight” at 180. Played a pretty good centerfield for a fat guy.

  73. Brandon:

    My brother is 6’1″ and about 140 pounds. It’s entirely due to a very, very high metabolism rate.

  74. JoyfulA:

    BP meds are very cheap, and cholesterol meds are out of patent.

    BTW, about 5 years ago, the guidelines for cholesterol were lowered, to encompass me, for the first time, and also my father, who at 91 has never had a cardiac problem. We are both taking pills to unnecessarily improve our cholesterol levels and to keep our primary care docs happy.

  75. JKTHs:

    I think of the NFL running backs who can be something like 5’9″ and 220 pounds. Their BMI must be off the charts and yet they’re in ridiculous shape.

  76. liberal:

    This has nothing to do with statistical significance and everything to do with social significance.

    I think “clinical significance” (or more generally effect size) is a better phrase.

  77. liberal:

    No, but it is crucial to distinguish between relative risk and absolute risk.

    If I told you that a particular risk might be tripled in a given context, that sounds awful, but if the absolute risk was one in a trillion, then the meaningful risk is negligible, regardless of the statistical significance.

  78. Sherm:

    Aldred Morris is obese at 32.2. Adrian Peterson is merely overweight at 27.9.

  79. liberal:

    Yeah, and IIRC the Cochrane Collaborate says that the evidence for taking statins for primary prevention of heart disease is weak (at least for not-terribly-unreasonable chol levels).

    The NNT for statins isn’t all that different from the NNH (number needed to harm).

  80. liberal:

    Yes, observational studies have zero evidentiary value — that’s exactly what I’m saying.

    False. It depends. Smoking is the canonical counterexample. IIRC smoking increases your chance of lung cancer by a factor of 20. Care to posit a hidden factor which would do that?

  81. RedSquareBear:

    Also, obligatory.

  82. Craigo:

    That’s an overstatement. The authors included many studies for which smoking data was not included at all.

    Not one of the studies actually recorded whether the subjects had a wasting disease.In fact, if a prospective study did so, it was rejected for “overadjustment,” also known as “telling the authors something they didn’t want to hear.”

    The sole adjustment the authors themselves made for health status was excluding studies whose subjects were in hospitals or hospices – because apparently, there are no desperately ill people anywhere else.

    It’s easy to design a study to reach a conclusion you’ve already assumed.

  83. Snarki, child of Loki:

    At a population level, it might be better to look at the per-capita consumption of bacon.

    The correlation/causation debates will be more entertaining, at least.

  84. timb:

    If this gets only 399, are you still gonna count your meta analysis as the 400th?

  85. timb:

    If there were, it still wouldn’t come with an edit function

  86. Malaclypse:

    Comments about the number of comments are central to my point.

  87. Paul Campos:

    The rest of the response makes it clear I’m being sarcastic, but just for the sake of clarity:

    Observational studies are most useful for proving a null hypothesis (Obviously if you have no or extremely weak correlations between an exposure and an outcome then you’ve pretty much proven the exposure has no causal effect). On the other end of the spectrum, an extremely strong correlation can as a practical matter establish causality – if heavy smokers are 20 times more likely to get lung cancer than never smokers, then you really don’t need a clinical study to conclude smoking causes lung cancer.

    The problem in regard to studies that look at the relation between body mass and mortality is that the correlations are either inverse, non-existent, or weak, so they’re much more suggestive of the null hypothesis than anything else.

  88. catclub:

    Ok, the bio I read when I was eleven said 6-0. But the stats website I looked at said 170, so HE WAS NOT A FATSO.

    YOU TAKE THAT BACK!

  89. ajay:

    Indeed. It reminds me slightly of The Truth About Pyecraft – “What you wanted was a cure for Fatness. But you had to call it Weight!”

  90. Craigo:

    Sorry, I meant to point out that the authors also rejected any study which recorded cardiovascular disease or diabetes, which correlate well with both body weight and mortality.

  91. Sherm:

    Ran the bases well for a borderline fat guy.

  92. ajay:

    Oh, god, NO.

    On the other end of the spectrum, an extremely strong correlation can as a practical matter establish causality – if heavy smokers are 20 times more likely to get lung cancer than never smokers, then you really don’t need a clinical study to conclude smoking causes lung cancer.

    No, no, no. The strength of the correlation between A and B backs up the hypothesis that A causes B, but it also backs up the hypothesis that B causes A, and the hypothesis that A and B are both caused by C.

    The observed correlation between cannabis use and mental illness, for example, could easily be because mentally ill people smoke cannabis to self-medicate, rather than because smoking cannabis makes you mentally ill. (It isn’t, in fact, because longitudinal studies have shown that one tends to precede the other.) So this isn’t a trivial issue!

  93. ajay:

    As is this.

    http://xkcd.com/552/

  94. Scott Lemieux:

    Well yes but there’s also the problem of willful misreading. The proposition that observational studies have important limitations is one that any epidemiologist would consider completely self-evident.

    And certainly none of this willful misreading is related to the fact that people very much want to convince themselves that their belief that fat people are icky is really a product of SCIENCE.

  95. tt:

    Every single sentence of your post is incorrect.

  96. Marc:

    Of course they do. But you factor that into the statistical tests used in your studies. And you don’t dismiss a stronger result while touting a weaker one – which is precisely what you’re doing here.

  97. Scott Lemieux:

    And then he claims (in other contexts) that it’s impossible to lose weight – ignoring things like the fact that we used to weigh a lot less than we did, or the fact that people in different countries with different diets have very different mean weights and distributions of weight.

    1)The fact that people weighed less on average in different cultural conditions does not, in fact, prove that individual attempts to lose weight in the same cultural context is likely to work. (We can’t just easily change things so that people don’t have to drive and engage in much more manual labor.) The fact that most individual attempts to lose significant amounts of weight fail seems more relevant, unless you think that Paul is actually arguing that individual weight loss is literally “impossible” as opposed to “highly unlikely.”

    2)When people on average were thinner, did people have greater life expectancy?

  98. Paul Campos:

    “Of course they do. But you factor that into the statistical tests used in your studies. And you don’t dismiss a stronger result while touting a weaker one – which is precisely what you’re doing here.”

    Um, no. I’m suggesting that HRs of .94 and 1.18 are both very weak evidence for a causal relation. Unlike the obesity zealots, I’m not arguing that a very weak correlation between a particular BMI level and lower mortality is good evidence that this level (BMI 25-34.9) actually lowers mortality per se.

    They of course are more than happy to argue the reverse when it suits their purposes.

  99. Orange Man In Ohio:

    Its pretty clear that fat people from NJ are total jerks, though.

  100. Sy:

    Scott accusing someone else of willful misreading is pretty rich

  101. Marc:

    But the difference is very relevant in deciding, say, whether or not we should serve soda pop in schools. Or whether we should promote exercise, or whether we should subsidize vegetables and tax soda.

    Paul is attacking a scientific problem like a zealous lawyer defends a client. He seizes on evidence that he likes, ignores contrary evidence. For a scientist like me this is incredibly disturbing. I can map his tactics directly onto what climate change deniers do. I’m more sympathetic to his goals than I am to theirs, but that doesn’t change the nature of what he is doing.

    In the first half of his post he trumpets the lowering of risk for the moderately overweight. In the second he does a meatball argument that we shouldn’t pay attention to the much more statistically significant *increased* risk for the truly obese. And then he parrots tobacco industry tactics to dismiss the legitimacy of epidemiology at all.

    Substitute “global climate record” for mortality and “CO2″ for “BMI” and you could xerox his tactics and arguments. That bothers me a hell of a lot.

    Now he could make a real claim about how you can only really establish strong links with epidemiology because it’s such a blunt tool. But that’s not what he’s doing, nor is he acknowledging real health risks from true obesity.

  102. medrawt:

    Honest question – is it really “extremely” skinny in historical terms? I mean, it’s certainly quite thin. But I knew a guy at my old job who was about 6’1″, 145. He was quite thin – had the build of a male model really, broad shoulders and nothing else, and was actually once approached for a photo shoot that we made extensive fun of – didn’t eat much meat, swam regularly. But he didn’t look weak or emaciated. My grandfather was the same height and spent most of his adult life weighing 175 lbs or so, and as a young man he was considered a sturdy strong guy. When I look at photos from the 40s and so forth, I see a lot of folks who look as thin as my friend (or my grandfather). I’m sure there are more men who seriously lift weights today than there were 70 years ago, but even so the slice of the population that either through weight training or genetics carry a lot of healthy bulk on their frame seems pretty small.

  103. Hanspeter:

    6′ 135lbs. No indication of any health issues with family history of similar bodied people living past 95.

    Yes, yes, plural of anecdote is not data, etc.

  104. mpowell:

    Yeah, but trippling a 1% chance of dying in the next 5 years if you are 50 is a big f*cking deal. And frankly, mortality is just never a 1 in a trillion type of thing. Everyone dies eventually. A 70% increase in the mortality rate is a big deal. It’s a continual risk if you are continually obese. We could shrink the window down to the next 10 days and the absolute risk would be vanishingly small, but that doesn’t improve the picture at all.

    And Ajay is absolutely right to point out that if you look at what the gross numbers imply, obesity looks like one of the top causes of death in the US. Dismissing this is really terrible form.

    On the other hand, the point about overweight versus obese is well made and it probably would be worth studying to see if we could find a statistically significant BMI at which mortality rate definitely increases. I suspect it would be somewhere between 30 and 35. The current categories are overly broad and poorly chosen.

  105. A Different John:

    You cannot prove a null hypothesis. You can only reject it, in the “either the hypothesis is wrong or something unusual happened” sense of “reject.”

  106. Paul Campos:

    Marc brings up the excellent point that there are very special people called “scientists,” who remain unaffected by cognitive biases, crass personal and financial interests, etc., and are only interested in the truth.

    Now some people might object that there are scientists who do seem to be affected by other considerations, but to save Marc the trouble of replying to this objection, I’d like to note that those aren’t true scientists (just as apparently people with a BMI of 30-34.9 aren’t truly obese, even though most people categorized as obese are actually in that category).

  107. firefall:

    promise?

  108. Richard:

    I understand that is what the authors of this study say. I think you would agree that very few research papers take the position that the conclusions reached are subject to bias because of adjustments or exclusion of other factors. If that were the conclusion, then the paper wouldn’t be published. Very respected research doctors take issue with the conclusions reached by this report and argue that the report IS subject to bias because of adjustment or exclusion of other factors.

    As pointed out above by others, Paul, you are a lawyer making a case for your strongly held position. But there are contrary viewpoints which you, as an advocate, either ignore or downplay. My point is that, at this stage, there are respected medical views on both sides of this debate, that the debate is far from settled and that I prefer to listen to the medical researchers on the subject, rather than the lawyers or press people who have already decided the issue in their mind.

  109. Bolo:

    Metabolism plus body structure (which itself relates to metabolism…).

    I’m 6’1″ and as an adult I have never, ever been below 210 lbs. I’ve had my lean body mass estimated as 190 lbs, and I do not lift weights, etc. My weight can fluctuate 25 lbs without me even noticing. I hit 260 this past year before I realized I was getting heavy and started to work off the weight.

    But again, BMI is not intended to apply to me individually. It applies broadly over a large population–people like me are outliers, just as people who are 6’1″ and 130 lbs are outliers as well.

  110. Richard:

    Absolutely true. But I think it is likely the case that there is a significant percentage of people who are 6′ and 135 lbs who suffer from an ailment or illness which effects mortality. There are probably data and statistics on this

  111. Sherm:

    Does the Framingham Study collect data based upon BMI?

  112. liberal:

    Dismissing this is really terrible form.

    I’m not dismissing anything. I’m stating, flat-out, that paying attention to relative risk only, without considering absolute risk at all, is terribly ignorant.

  113. Marc:

    Straight ad hominem, the rough equivalent of my asking if you’re sponsored by Frito-Lay.

  114. Paul Campos:

    No, I’m not a lawyer (I have a law degree but I don’t act in the professional capacity of a lawyer, that is, I don’t represent other peoples’ legal interests).

    This is not a merely semantic point. A lawyer has a professional obligation to zealously advocate his client’s position. I don’t have a client in this debate, even metaphorically.

    Now perhaps you’re claiming that I’m suffering from some sort of mental disability that “makes” me act like a lawyer representing a client, even when I’m not. In reply to that I can only say that a lot of obesity researchers seem to be much better lawyers than I am, given their willingness to zealously defend positions that lack any factual support.

  115. rea:

    The only blood pressure drug that works for me costs (on a monthly basis) about as much as our monthly electrical bill . . .

    Difficult choice, sometimes.

  116. tt:

    Substitute “global climate record” for mortality and “CO2″ for “BMI” and you could xerox his tactics and arguments. That bothers me a hell of a lot.

    No, you couldn’t, because there are no recent publications in journals with the reputation of JAMA which suggest against a link between greenhouse gases and climate. When climate denialists want to make use of credentialed scientists they have to draw from a small and declining pool who mostly publish in third-tier journals (from which their publications are often retracted). That’s the difference between a settled and unsettled scientific controversy.

  117. ajay:

    Marc brings up the excellent point that there are very special people called “scientists,” who remain unaffected by cognitive biases, crass personal and financial interests, etc., and are only interested in the truth.

    While we’re on the subject of deliberate misreading…

  118. Paul Campos:

    You’re projecting. You dismiss my arguments on the grounds that I’m a “lawyer” rather than a scientist, and therefore I’m supposedly arguing toward a foreordained conclusion rather than pursuing the truth, which is what you’re doing.

  119. liberal:

    The strength of the correlation between A and B backs up the hypothesis that A causes B, but it also backs up the hypothesis that B causes A, and the hypothesis that A and B are both caused by C.

    Yes, but there’s also the fact that it’s pretty straightforward to posit mechanisms whereby smoking causes lung cancer. The reverse, or even a separate factor causing both?

    There’s more to good scientific inference than just the issue of observation vs controlled experiments.

  120. catclub:

    I’ll take the under on 400 and on 300.

    not enough vitriol.

    My time limit is 36 hours.

  121. ajay:

    You dismiss my arguments on the grounds that I’m a “lawyer” rather than a scientist, and therefore I’m supposedly arguing toward a foreordained conclusion rather than pursuing the truth, which is what you’re doing.

    No. Read more carefully – he’s pointing out that you are arguing badly, in a way that a lawyer would. He’s not saying “you’re a lawyer, therefore you must be arguing badly”.

  122. Richard:

    You have a law degree and teach at a law school but dont consider yourself a lawyer? I find that to be a strange distinction.

    I know you don’t have a client but you are arguing the same way as a lawyer who has a client does. You take a firmly held position, garner all the evidence in support of that position, ignore or downplay evidence that is to the contrary and denigrate the advocates for the other side (they “zealously defend positions that lack any factual support”). I dont consider being a lawyer as a mental disability ( I’ve been practicing law for 35 years) but when it comes to this debate, it seems to me you are acting exactly as a lawyer does.

    Your position seems to be that the evidence is undisputed that moderate obesity does not lead to an increased risk of mortality (or any other health risks) and that doctors arguing to the contrary are deluded fools or people with a vested interest in the anti-obesity crusade. My position is that I’m not a doctor, that there seems to be a spirited debate within the medical community about the health risks, that there is no definitive answer yet and I’m keeping an open mind.

  123. joe from Lowell:

    Everybody should try to weigh as much as they possibly can.

    Exercise and good eating habits do nothing to improve health.

    I want to have sex with fat women.

    Hitler was misunderstood.

    Dude! One of these things is not like the others!

    Ain’t nothing wrong with the big, beautiful girls. Nothing at all.

  124. Malaclypse:

    Fat-bottomed girls, they make the rocking world go round…

  125. chris:

    And then he claims (in other contexts) that it’s impossible to lose weight – ignoring things like the fact that we used to weigh a lot less than we did

    I don’t think he claimed it was impossible to *gain* weight. There’s a big difference between trying to lose weight and never gaining it in the first place.

    or the fact that people in different countries with different diets have very different mean weights and distributions of weight. The latter is absolutely deadly to a claim that obesity is just some sort of cosmic accident, as opposed to being a direct consequence of public health choices and the marketing efforts of large corporations.

    None of this has any bearing whatsoever on the feasibility of, as an individual, losing weight once you have gained it. Changing your diet or lifestyle at that point *might* help — or it might just be closing the barn door after the horse is gone.

    I happen to agree that obesity isn’t a cosmic accident, but what does that have to do with what (if anything) can or should be done for people who are already obese? Measures like rebuilding American urban and suburban areas to be more walking-friendly might, in the long run, be useful enough to justify the cost and political difficulty of pushing them through, but that doesn’t necessarily mean they will have much impact on people who are already obese right now.

  126. joe from Lowell:

    I can map his tactics directly onto what climate change deniers do.

    Marc brings up the excellent point that there are very special people called “scientists,” who remain unaffected by cognitive biases, crass personal and financial interests, etc., and are only interested in the truth.

  127. Sherm:

    Perhaps your op-ed could have been clearer for the skeptics, but it seems to me that these findings are not particularly controversial to anyone who understands that BMI is complete nonsense.

    The findings are that people who are considered “overweight” (which includes many people who are not in fact overweight) and people who are considered mildly obese (which includes many people who are not in fact obese) have lower mortality rates than people considered “normal” (which includes many people who are much thinner than normal). For whatever reason, people assume that the overweight and mildly obese people in question all look like Gilbert Grape’s mom, although the group actually includes fit and muscular people such as NFL running backs. For what its worth, I think your point would have been better made if you conceded that the study confirmed higher mortality rates for the truly obese and focused more clearly on the problems with the BMI calculator as confirmed by this study.

  128. RedSquareBear:

    If you’ve been practicing law for 35 years but still don’t know the meaning of the word “lawyer” (as opposed to “professor at a law school who has a JD”) you should probably practice harder. You seem to need it.

  129. KadeKo:

    “You may die.”

    (worriedly) “Even Leon Panetta?”

    (sadly) “Yes, even Leon Panetta.”

  130. xxy:

    Yeah, judging from the study Paul is definitely right that the national obsession with weight is all wrong and like other forms of shaming cruel, harmful, and counterproductive, but he’s wrong when he says there is an obesity myth or suggests a correlation between fat and mortality is hocus pocus.

  131. Richard:

    And if the op-ed piece was just about the problem in using the current BMI guidelines in determining who is obese , I would have no problem with it.

  132. Sherm:

    That sucks. And it also sucks to put your health in the hands of the pharmaceutical industry. But its often a necessary evil. Chances are that my wife is going to kill me for counting every milligram of sodium that enters my body before the blood pressure catches up to me.

  133. Richard:

    My professors at law school, most of whom did not practice law and in many cases had never practiced, considered themselves to be lawyers nonetheless. Paul makes a distinction. Really doesn’t matter since it seems to be one of semantics. I concede that in his position at the university, he is not practicing law as that term is commonly used.

    My point is that, despite whether Paul considers himself to be a lawyer or not, he is acting like a zealous advocate for a point of view – that is, acting like a lawyer.

  134. joe from Lowell:

    36-24-36? Only if she’s five foot three.

  135. catclub:

    “although the group actually includes fit and muscular people such as NFL running backs.”
    I suspect that NFL running backs die earlier than average, so are actually a bad example. I am very sure that NFL players overall die earlier than average.

    Now Willie Mays, on the other hand, is over 80, and is not fat.

  136. Sherm:

    Why the Willie Mays’ infatuation? And I suspect that he was quite “overweight” when I first saw him as a New York Met.

  137. snurp:

    I have no particular dog in the obesity research fight, this is just curiosity: are lifespan-limiting disorders linked to weight gain via, say, mobility impairment, or as a medication side-effect, already corrected for in the study? Were the critics assuming they were rare relative to wasting diseases?

  138. ajay:

    But does the US really have a national obsession with weight? No, it doesn’t. All these high-BMI football players and weightlifters and so on that everyone mentions aren’t getting criticised for their weight. No one said that Arnold Schwarzenegger was too heavy, though I bet his BMI used to be close to Chris Christie’s.

    No, the US has a national obsession with fatness and that, however undesirable from other points of view, seems to be medically justified. It also has a national lack of awareness of the best way to measure fatness (thinking that BMI is the best way to do it).

  139. Malaclypse:

    I’m a bit concerned – I had not expected Scott to open a second thread up.

    Needless to say, I blame Ralph Nader for Scott’s perfidy.

  140. Sherm:

    Plus, the trolls haven’t appeared to mock the “fatties” and Campos’ love of them.

  141. Tom in Raleigh:

    Now that’s meta.

  142. elm:

    And then he puts up a Charlie Weis post. If he had managed to work in steroids and Tebow-bashing into that post, this one would be quickly forgotten, but even so, I have a hard time seeing this get to 400 unless you start spamming it, someone blames Obama and joe starts defending him, brad decided that this is the post to go off on some weird accounting discussion, or JenBob shows up. We may actually need all 4.

  143. mpowell:

    In general this is true. But in this case Paul’s attempt to diminish the significance of the mortality rate of the obese by bringing this point up is a dismissal of a very real and substantial population wide health risk.

    My comment was not just a reply to yours.

  144. xxy:

    You must not be a woman. Or talk to women. Or consume media. Or glance at magazine covers. Or work in high fashion.

  145. Anonymous:

    very special people called “scientists,” who remain unaffected by cognitive biases, crass personal and financial interests, etc., and are only interested in the truth.

    Why are the scientists who support your position any more reliable than the ones you dismiss? Do they not have biases or personal or financial interests?

    And lets talk about those alleged financial interests. If it were true that scientists could be so easily bought and sold, there way more money to be had from the side that would be interested in downplaying health risks from being overweight (food/agriculture/restaurants/etc.) than from the side that would be interested in promoting such risks (weight loss industry, diet products, etc.).

    If you want to critique individual studies by saying, hey, this was funded by Jenny Craig, fine, but dismissing ALL of them just makes you look really, really foolish, and it makes it impossible to take anything you say seriously.

  146. wengler:

    A Campos BMI post on January 3rd? We are getting this year started out right!

  147. brad:

    Once again, the fact that BMI is a bad tool, often even at the broad population group level, proves that there are no negative consequences to the additional weight many Americans carry as a result of a cheap, easy, far too high in calories and overprocessed carb, especially various forms of sugar, rich diet combined with a sedentary lifestyle?
    Or, everyone has a natural body shape, we all should accept and enjoy this, and the food and… physical cultures someone lives in also have a sometimes significant bearing on an individual’s weight and/or health. These aren’t contradictory ideas, I’ll never understand why Campos seems unable to hold both in his mind at the same time.

  148. evodevo:

    So did Babe Ruth ….

  149. Bloix:

    The reason people use BMI in the first place is that you don’t have to collect any special data to calculate it. If a study has gathered height and weight data, the BMI can be calculated. Every study gathers height and weight data. So you can use data sets that were collected without overweight in mind into sources of data for obesity studies.

    And everyone knows their height and weight, so you can calculate BMI from data sets that were created from oral interviews – cheap! – without any clinical measurements being taken – expensive!

    This is why BMI is so widely used even though it’s recognized as being very crude.

  150. Sammy:

    Oof. I’m a lawyer too, and normally I’m a big fan of all the authors on this blog, but reading your posts here is a trainwreck, Campos. I’m not going to bother with the ‘why’ of it, but there’s some poor arguing (but classic internet-arguing techniques!) going on here on your part.

    I think the study being referenced makes sense, but it’s also not the whole picture, and I’m not convinced by you in the slightest that this study means being overweight is good for you.

  151. Ahistoricality:

    You can be thin, or fat, but you’re still gonna die.

  152. ajay:

    You must not be a woman. Or talk to women. Or consume media. Or glance at magazine covers. Or work in high fashion.

    I’d love to see an example of one of those criticising a thin person for weighing too much, rather than for being too fat. Hell, men’s fitness magazines specifically advertise food supplements that are supposed to help you gain weight – muscle mass.

  153. Substance McGravitas:

    Diet and weight-loss emphasis does not come with disclaimers saying “Thin people need not read this.”

  154. Substance McGravitas:

    Something I was not aware of: fat days.

    http://www.google.co.nz/search?q=“fat+days”&

  155. Chris:

    Why are the scientists who support your position any more reliable than the ones you dismiss? Do they not have biases or personal or financial interests?

    This. One of the most frustrating things about such debates is how often science is biased until it says what you want it to say.

    I wonder to what extent Paul thinks of himself as unbiased, by the way. Also, he should talk to some statisticians and epidemiologists. He clearly has some basic misunderstandings that would be pretty easy to clear up (maybe he could audit a couple stats courses at his U?), not only about methods and concepts, but about the basic philosophy of hypothesis testing and statistics.

  156. skeptonomist:

    The main conclusion of the metastudy seems unassailable – weight short of gross obesity does not affect mortality very much. I do wonder somewhat about the affect of serious illnesses which cause weight loss – there is a cause-effect problem here. A really good individual study would try to eliminate this problem, for example starting only with people who are healthy, but I doubt if a meta-study can screen out all studies which do not control for this. This is the sort of thing which could potentially account for the small differences between “normal” and “overweight” groups.

  157. elm:

    Mal, there’s no way this is getting to 400. But you should probably just transfer your over/under to the Michelle Obama post from today. That’s gonna get there unless Scott once again posts a follow-up!

  158. actor212:

    BMI is a notoriously poor way of judging obesity. Body fat percentage is far, far more accurate. This study is a tale told by idiots.

    For instance, my BMI is about 27, borderline mildly obese.

    My body fat percentage, however, is around 20%, which is actually low-average.

    I’m a huge man. I’m 6’3″, weighing 225. Former athlete, could have turned pro in any of three sports, yadayadayada. And I still spend time in the gym at age 55.

    By this study’s criteria, I’d be judged as being overweight (I could stand to lose some pounds) but I can get on my bike today and ride 50 miles in somewhen north of 3 hours.

    This is the critical piece of information that these morons are missing. People with BMIs like mine are the sweet spot of people who still work out but aren’t committed enough to it to give up things like sugar and bread.

    They need to redo this study using body fat analysis. Obesity is, and will be, the defining health problem for the next century. These morons are merely giving succor and comfort to people to grab an extra roll when they damned well shouldn’t.

  159. actor212:

    Body fat percentage?

    You grab two handles of a battery operated device that shoots a mild electrical charge through you and you have a very nice approximation of the BFP. Costs you ten bucks. Granted, it’s not the “dstick ‘em in a sling and dip ‘em in a pool” buoyancy test but it’s close enough for science.

  160. actor212:

    I was 6’0″, 175 once.

    It was called “puberty”, I believe.

  161. actor212:

    So all those Nutrisystem commercials with Terry Bradshaw? I just made those up in my dreams?

    Cuz he doesn’t say “fat” at any time. He repeats, ad nauseum, “weight”.

  162. actor212:

    Oops. Did not see that this entire discussion was raised below in comments.

  163. actor212:

    Two words: grapefruit juice.

  164. Craigo:

    Nope. It’s fairly stupid of you to write that when I included a link which backs up those claims.

  165. South Africans 3rd fattest in the world | Vuma News:

    [...] The obesity myth revisited (lawyersgunsmoneyblog.com) Share this:LinkedInFacebookTwitterGoogle +1EmailPrintDiggRedditStumbleUponTumblrPinterestLike this:LikeBe the first to like this. This entry was posted in obesity and tagged Africa, Conditions and Diseases, Health, National Health Service, Nutrition and Metabolism Disorders, Obesity, South Africa, Southern Africa by post. Bookmark the permalink. [...]

  166. Geoffrey Sallach:

    Bed Bugs Chicago, Treatment of Bed Bugs Chicago, Removal of Bed Bugs Chicago Contact our professional extermination company at http://www.bedbugstreatmentchicago.com/ to get rid of those bed bugs!

Leave a comment