April 26, 2005
Correlation is Not Causation
Watching David Brooks and John Tierney both race to write the same column extolling the virtues of obesity and mocking liberals for denying themselves cheeseburgers was pretty funny. Did no one warn David that Tierney got there first? Does David not even read his conservative competitor? Seems that the Times token righties need to coordinate a bit better.
But it was also sad to watch two supposedly powerful conservative minds use some of the most read newspaper real estate in the world to misinform their readers in the exact same way and in service on the exact same agenda. So let's get something straight: the study did not tell you to get fat. It did not tell you to get a little fat. It did not, in fact, tell you to do anything at all. We're dealing with observational data that's widely available and the authors are trying to divine a connection between weight and mortality rates from it. We're also focusing on the rate of death, rather than disability and disease (as the Times article on the study -- though not its op-eds -- notes, the connection between excess weight and diabetes, high blood pressure, and high cholesterol in undeniable). And what, exactly, did the study find?
Well, folks whose BMI rests in the normal category provided a baseline mortality, i.e, their death rate is considered the "normal" death rate. Folks in the overweight, though not obese nor extremely obese categories, die less than their "normal" counterparts. The obese, the severely obese, and the underweight die more than everyone else. Why?
First you've got to look at the basic measure used, the BMI. The central mistake of Tierney and Brooks is believing that the BMI is a body fat test. It's not. Instead, it's a purely mathematical calculation that takes your weight, divides it by your height in inches squared, and then multiplies the product by 703. So to give an example, my weight is 188. I'm about 72 inches tall (calculate yours here). So my BMI is 25.5, a shade into the overweight category (which goes from 25-29). But am I overweight?
Now things get interesting. I've lifted weights regularly since my first year of high school, making me a fair bit stronger than most folks. I also have a body fat scale, a relic from when I lost 50 pounds as a high school sophomore (I was a fat kid). My body fat is regularly around 15-16%, which'd put me in the "fitness" category, a good 10% under obesity and 3% under the beginning of "average". So I'm not overweight given my body's composition, but I am a bit overweight if judged via my height-to-weight ratio. To restate, I'm not overweight, but I am if the BMI is used.
This is a long way of saying that for many, many people, the height-to-weight ratio comprising the BMI is wholly inaccurate. And most of those people who throw it off are athletes, or fitter than the norm. So right off the bat you've got a group in the overweight category who're likely living healthier lifestyles than many in the "normal" category, but being categorically penalized by the BMI's inadequacies. Whoops.
Moving right along, folks who struggle with their weight, who've not accepted being overweight and are continually fighting to hold the line on 5-15 pounds of flab, are probably watching their eating much more carefully and exercising more regularly than those who benefit from a genetic predisposition to a good weight. I know that my body's constant threats to get fat, and my father's, and my sister's, have forced us all to live healthier, exercise more, and eat better than our naturally slimmer neighbors. So you've got another category now of people battling the scale, and living healthier lives because of it. As the study found, most people in the underweight category, and probably some at the bottom edges of the average category, are naturally slim and don't work to achieve it. Living a sedentary lifestyle is bad for you no matter what your BMI, and many of the slim and average are probably doing exactly that.
Next is the "obesity paradox". Most deaths occur after age 70, so most of what we're dealing with happens in life's twilight years. As it turns out, a bit of fat has a protective effect in old age, likely due to increased muscle and bone mass from toting it around. There's a real health problem in that so few elderly do any sort of weight training, and in fact often haven't for years and years beforehand. Some participate in aerobics, classes, walking -- but many are sedentary, and few pick up weights. It's a really serious health issue, and anything, even a bit of a pudge, that helps reverse the loss of muscle mass will be a boon. But that doesn't make fat the optimal way to gain muscle, it argues for a wider attention to and adoption of techniques to build muscle and preserve bone.
Now, there's countless more going on here. Cigarettes are an appetite suppressant, but we don't know exactly how many people are slimmer because of them. The BMI might simply be too low, or need to adjust better with age. But the above are a number of potentially confounding factors, and you should immediately judge anyone suggesting that you start bathing in tubs of saturated fat a loon. Tierney and Brooks are misappropriating the study in service of some weird, quasi-libertarian agenda to discredit authorities who advise you on what's good and bad for your health. Why the findings of experts that offer folks better information with which to make their decisions piss some libertarians off, I'll never know. But do yourself a favor and don't read right-wingers dead-set on making life a faith-based endeavor. Health is now, as it's always been, a simple affair. Eat well, exercise regularly, try to avoid stress, and get regular check-ups. If some op-ed columnists want to experiment with an all cheeseburger and decaf coffee diet, well, that's their business.
TrackBack URL for this entry:
Listed below are links to weblogs that reference Correlation is Not Causation:
Nice analysis of some possible factors behind the study's results!
Posted by: Aron | Apr 26, 2005 2:56:36 PM
Best read of this I've seen yet Ezra. BMI is a horrible measure of a person's fitness. I'm in the same camp as you, 6'2" 195. By their definition I'd be a fatty, but then of course I rep 210 on the bench and am pretty close to abs myself. What this study shows is not so much that overweight people live healthier but how BMI is a poor measure of health. I could eat regular balanced meals/work out or just eat mcdonalds everyday and have the same BMI, yet vastly different health.
I actually took the time monday to read the study behind this and it's pretty worthless. They basically followed a few thousand people for a while and saw how quickly they died. They then divided them into 4 categories based on their BMI. That tells you less than nothing, I'm surprised they got a grant for this.
The thing that really disgusts me is that 20% of the kids in our schools are becoming obese and these guys use a misinterpretation of an epidemiological study to take pot shots at the people who hurt their self esteem. Pathetic.
Posted by: slapshot57 | Apr 26, 2005 3:13:02 PM
I will tell you why "public health" types piss libertarians off. It is because they are preachy and can't distinguish someone who chooses to smoke and gets lung cancer from someone who gets coughed on in the subway and gets TB. There is a difference.
Libertarians realize that the health experts out there goes far beyond giving useful advice. Health experts these days spend more time stigmatizing behavior and creating alarmist propoganda than they do giving sound, balanced and measured advice. No matter what you do, it is likely you will live into your seventies. Accordingly, we can all relax a little about the "epidemics" of smoking and obesity related disease.
Slapshot, I am willing to bet that more damage to heavy people's self-esteem has been done by public health types than by libertarians. Why would a libertarian accuse some fat person of causing a national catastrophe? For libertarians there is practically no such thing as a national catastrophe.
Posted by: Neil Paul | Apr 26, 2005 4:15:28 PM
No, correlation is not causation. Which is precisely why the heavily funded campaign against fat people was so ill-founded in the first place. This study just says what all studies on weight and mortality have said. Pick up Paul Campos' "The Obesity Myth" published last year and you'll find that his examination of the underlying data revealed precisely what these studies reveal.
Certainly, it is absolutely wrong to conclude that people should now gain weight to be healthier. I suspect Brooks and Tierney reach that conclusion in part to mock it. But we do need to ask ourselves why the exact reasoning has long been used to urge weight loss when the statistical justification was precisely as strong as the belief that being "overweight" is healthiest. The most pessimistic ready of this study still reveals that Bush's CDC had miscalculated the actual risks of fatness by 400%. That is an enormous overstatement, but also a predictable one.
This study still makes no effort (and indeed I do not know how this could be done) to control for the effects of weight cycling, something that's been shown to have a negative impact on health irregardless of what weight the person ends up at. Given that weight cycling is the extraordinarily likely result from the "treatment" virtually every fat people will have engaged in for most of their life, this is a striking omission.
A wealth of evidence indicates that weight is a poor measure of health and that a healthy lifestyle will provide enormous benefits though may not effect weight much or at all. If health is our goal, that should be what is persued. Not a wrong-headed fixation on weight that can be genuinely counterproductive to encouaging actual improvements to people's health.
Posted by: BStu | Apr 26, 2005 4:58:50 PM
My hypothesis- cholesterol and blood-pressure lowering drugs have recently been shown to have other unexpected benefits- there are some calls for everyone to be on a statin (and not just from the companies that make the drugs). People who are slightly overweight have higher baseline BP and cholesterol, so they go on these drugs more often, and get the side benefits. For obese people, the extra weight overrides the side benefits of the drugs, but for the overweight, the drugs are enough to control BP and also make them live longer than people who aren't on the drugs. This is easy to test, give the drugs to regular-weight people and see what happens, and I'm guessing this study is already in progress.
Posted by: SP | Apr 26, 2005 5:46:05 PM
My reading of it was that there just aren't that many people who stay "moderately overweight." It's basically a transition phase (although Ezra did a good job of showing why its also just a bad metric). Sometimes I'm moderately overweight (as finals approach and I spend less time in the gym and less effort trying to eat healthy -- although I know in my heart that's precisely the time that I should be focusing on eating healthy and exercising) but I then drop back to my normal --- 6'2", shade under 190. The overweight part is a transition. For others, it's a transition from skinny to obese. I think that the study is just showing that it's tough to hit a moving target -- people aren't dying when they're moderately overweight because its hard to pin them down at that weight for very long.
Plus, as death approaches, it's quite possible that people drop quite a bit of fat. I know my grandparents did.
Posted by: J.C. | Apr 26, 2005 7:07:59 PM
i try to do myself a favor by not reading david brooks
Posted by: pat | Apr 26, 2005 10:21:32 PM
Something that helps reverse the loss of muscle mass, the Testosterone, Trenbelone, and Equipoise used to build all that lean muscle in such a short time...
Posted by: steven davies | Jul 22, 2007 9:08:09 AM
Posted by: weight training | Aug 8, 2007 5:57:10 AM
Posted by: peter.w | Sep 15, 2007 7:19:44 AM
The comments to this entry are closed.