On Monday, I wrote about a recent study, published in the New England Journal of Medicine ((Adams, K., et al., Overweight, Obesity, and Mortality in a Large Prospective Cohort of Persons 50 to 71 Years Old. New England Journal of Medicine, 2006. 355(8): p. 763-8. )) , which is intended to refute last year’s CDC study (published in JAMA ((Flegal, K.M., et al., Excess deaths associated with underweight, overweight, and obesity. Journal of the American Medical Association, 2005. 293(15): p. 1861-7. There are “Alas” posts discussing this study here and here.)) ) showing that “overweight” people live longer than “normal” weight people.
According to media reports, this new study proved that even a slight weight gain can be deadly. The AP wrote “Being a little overweight can kill you, according to new research that leaves little room for denial that a few extra pounds is harmful.” NPR wrote “being even a bit overweight can potentially kill you.” The Star-Ledger ominously asked “Those few extra pounds won’t kill you … or will they?”
The NEJM study, by Adams et al., is — frankly — a badly conducted study. It uses a non-representative sample (members of The American Association of Retired People, or AARP) with an incredibly low response rate of 18%. Even worse, height and weight were self-reported.
But for the study’s authors, who were determined to come up with a study proving that fat equals death, the data presented a much more serious problem: the people in the “overweight” category were longer-lived than the people in the “normal weight” category. How to spin data showing that overweight people live longer, into their desired message that fat is always deadly?
In Monday’s post, I quoted Linda Bacon extensively. Today I want to expand on one of her points:
…Let’s take a look at the data itself. The authors worked hard for their conclusion. They examined records from over a half million AARP members that had been surveyed over a ten year period. What they found was entirely consistent with the earlier JAMA report: “overweight” people had the lowest mortality risk. But that wouldn’t serve their purposes. NEJM’s press release wouldn’t look nearly as attractive with that headline.
So they subjected their data to numerous manipulations before finally arriving at a suitable conclusion. First they threw out data on people who were smokers or former smokers. Nope, still shows overweight as benign. They hid this with a sleazy method: using only the top (BMIs of 23 to 24.9) of the “normal weight” group compared to the whole of the “overweight” group.
Here’s a table of some of the relative risks reported in the Adams et al. study (if you have trouble reading it, click on the image for a larger version):
The yellow column indicates the relative risk of death for “normal weight” people (the heaviest set of “normal” weight people are used as the baseline; all other risk ratios on this table are in comparison to those folks). The red outlines indicate the areas where the relative risk of death is as low or lower for “overweight” people as it is for “normal weight” people.
- Note that for “all men” and “all women,” the risk of death is less for people who are slightly “overweight” than it is for people in the “normal weight” categories.
- Looking at where the red outlines are located, can anyone seriously argue that the take-away finding of this data is that being a little overweight is generally a killer?
- In nearly every category, being in the “normal weight” BMI ranges of 18.5-20.9 or 21-23.4 carries a higher risk of premature death than being in the “overweight” BMI ranges.
- In most categories, being slightly “obese” (BMI 30-35) is slightly less risky than being in the middle of the “normal weight” category (BMI 21-23.5). In many categories, being solidly “obese” (BMI 35-40) is slightly less risky than being at the thin end of the “normal weight” category (BMI 18.5-21).
- In order to produce the finding that “overweight” is less healthy than “normal weight,” Dr. Adams did a very dishonest statistical manipulation – he compared just one “normal” BMI range, representing the heaviest people in the “normal” range, to the entire “overweight” range. This is because the majority of people in the “normal weight” categories had a greater risk of death than the majority of people in the “overweight” category.
- Race and sex have a strong impact on these findings. The “fat equals death” findings that were reported in the media shouldn’t be taken seriously by anyone, but black women especially would be well advised to ignore the reporting. This study actually found that black women, more than anyone else, have an elevated risk of death if they’re in the “normal weight” category, and live longer if they’re in the “overweight” category.
- The strongest association between “overweight” and increased mortality were for Latina women, and for the combined category of Asians, Pacific Islanders and American Indians (both male and female). However, the the sample size for these groups were significantly lower than the sample sizes for the other groups. For example, the finding that white men in the BMI range of 35-39.9 have a relative risk of 1.32 was based on the deaths of 1,862 white men in that BMI range; in contrast, the finding that Hispanic women in the BMI range of 35-39.9 was based on the deaths of only 10 Hispanic women in that BMI range. This makes these findings much less reliable.
- None of these risk ratios should be taken seriously. Why? Because they’re all very low ((The sole exception is the 4.12 risk ratio for Hispanic women with BMIs of 40+. This finding — based on the deaths of just 16 women — is such an extreme outlier that it’s almost certainly statistical noise.)) — almost none of the risk ratios rise above two (for smokers the risk ratio of death by lung cancer, compared to non-smokers, is 23.3 for men and 12.7 for women ((The Surgeon General’s “Health Consequences of Smoking 2004,” chapter 7, page 881))). Many of the risk ratios are so small (e.g., 1.07 – remember, 1.00 is a null result) that it’s impossible to imagine that they could have any realistic real-world significance to someone evaluating her own health.
With very small risk ratios – especially in a study with poor data-gathering and many possible factors not controlled for, such as this study – determining causation isn’t really science; it’s guesswork. Maybe these differences are being caused by fat (even though at many BMI ratios fat people live longer); but the cause may also be one of the many factors this study did not account for, such as yo-yo dieting, use of weight loss drugs, body shape, discrimination, and socioeconomic class, to name a few.
So that’s one of the ways Adams and his co-writers twisted the data to produce their desired result. But they weren’t done yet, and so neither am I: watch for a future post on how they used retrospective data (i.e., asking people “do you remember what you weighed when you were 50?”) to twist their data further, and for what they actually found out (but didn’t report): Losing weight is deadlier than gaining weight.
Pingback: Wandering Seven Days Wonder
Pingback: I hate people
Pingback: Liberal Debutante
Pingback: Health Pundits
Pingback: feminist blogs
Ah, statistical analysis, my favorite way to spend an evening (right, Dianne?)
Amp, I think you’re way off base here in your paper analysis (your media comments are spot on).
Simply put, you’re using the wrong stats. I think you’ve lost sight of what the paper was saying in your haste to attack it.
You may THINK this is a problem. And you are correct that it is a potential confounding factor. But–in fact–the authors discuss this. Their group was representative. They also discuss the accuracy of self-reporting. The group was large enough that they can perform the stat analyses for representation.
As they note, their data are limited in certain respects and may not be able to be extrapolated to certain minorites, for example. But this isn’t a psychological survey. If you’re going to attack their data based on a response rate (18% is actually really high for many studies!) then you need to explain why you think the response is a problem.
bullshit. Or, more politely, prove it. The data… just are. Sure, people “massage” data sometimes. But the people who do that are those who work for places with a known bias. Like, um, NAAFA, you know.
See, you may not know this, but going against the grain is a great way to get GOOD PRESS (if you’re right). That accusation of bias needs a lot more support; you shouldn’t be printing it without more basis.
You’re just not understanding the confounding factors, I think. I’ll explain it more below.
Same nasty bias here. What, you think the (neutral) researchers write the damn press release? Enough about the “purposes” already.
These “manipulations” were proper. They performed the necessary tests to figure out what the “real” effect was, and eliminate the confounding factors. That is a GOOD thing, not a BAD thing.
The fact that Linda attacks this, instead of compliments this, makes me really distrust her. You should not be fooled.
The paper goes into a decent explanation of something everyone knows: Smokers are often 1) unhealthy, and 2) thin. They’re not unhealthy because they are THIN, they are thin and unhealthy because they SMOKE.
Understanding this difference, and why this means it is proper to exclude smokers in the analysis, is CRUCIAL to understanding the study itself. So if you disagree with me, stop and let’s argue here before we go on, or we’ll be wasting our time.
I have a few experimental design posts on my page, along with some statistics posts, which I can link to if you do not have a statistics and experimental design background.
Otherwise, let’s start this miniconversation thusly: What part(s) of their statistical analysis do you disagree with, and why? It’s not enough to decry is all as “manipulations”. If you want to raise a cogent argument against it, you have to address the points the authors of the paper raised.
So, for example, if you think pulling out smokers was wrong, you have to explain why you think the data are more accurate with smokers includes, or less accurate with them excluded.
Given what we know about smoking, health, and smoking-related weight loss, this won’t be easy to show. Certainly Linda does a piss-poor job of actually discussing the all-important data, preferrring to use laymens terms like “manipulation”. But I’m happy to have my mind changed.
I’m looking at Table 4, and they did no such thing in that table. And the graphs, of course, are also matched to each BMI segment.
That is because you are using the wrong table. The study contains multiple tables. This is common, because a good study is one which has data transparency. That does NOT mean that all the tables are equally accurate, or equally citable. In this case you are picking a table from the study and selectively ignoring the explanatoins of why that table is NOT correct in what it shows.
Yup. Well, not from your graph. But from the paper itself: Yup. Which is a bummer, because I am not thin. But it’s no less true merely because I wish it were different.
Don’t let your activism get in the way of the truth. Unlike other areas like feminism/racism/etc there is a “true” answer here.
That’s not true. Again, look to Table 4, which is the most accurate table. It is the most accurate because it accounts for the most confounding factors, while still easily remaining large enough to provide good statistical data.
Unlike you, I don’t know how to paste tables. But everyone’s got the link now so we can all refer to it.
Well, I’ll wait until you and I agree on the proper analysis. But I can’t leave this alone, based on your comment here: You cannot simultaneously bitch about “facotrs not controlled for” and ALSO “manipulation” because that type of manipulation is HOW YOU CONTROL FOR CONFOUNDING FACTORS.
Finally, we agree: Maybe.
But thats not good enough. You can’t assume your own personal “maybe” is true until proven, that’s just not how science works. From the data we have NOW, it seems to be fat. Until we get other, better, data, that is the logical conclusion.
Amp, I have an undergraduate degree in Mathematical Methods and a graduate degree in Statistics. Sailorman is absolutely right. I’m sorry, I know you desperately want it to be different, but it’s not. He’s especially spot-on in the discussion of “controlling for confounding factors” (which good statisticians do automatically, and know how to do well), and that you can’t call it “manipulation.”
Smokers are often 1) unhealthy, and 2) thin.
Sailorman,
Everything that I can find confirms what I had heard about this. That is, the average weight of smokers is around 5 lbs less than that of non-smokers. If I run a calculation on myself, the additional 5 lbs takes me from a BMI of 20.7 to a BMI of 21.5 which does not seem to be the difference between thin and not thin. That increase of .8 in BMI seems to be close to what you get across the range of heights & weights (from my limited research).
Can you provide any links that contradict the info that I currently have? Or maybe simply a more in depth explanation of your statement will make things clearer to me.
Sailorman, I’ve looked at Table 4 of the study, and the accompanying graphs in Figure 2, and it does seem to me that Ampersand has a point — they did use the uppermost range of “normal” BMIs (the one just below the cutoff line for “overweight” at 25) as the reference group, and morality risk for men does seem to increase in both directions from the reference group.
I guess I wouldn’t quite characterize this as “sleazy,” but it does seem like the authors of the study should have foregrounded the fact that if you’re a man within the “normal” BMI range (surely a significant segment of the population), then a few pounds too few are probably more dangerous, mortality-wise, as a few pounds extra. The conclusion on p.1 of their study needs some qualification in this regard. It also would have helped if they had qualified that conclusion with the words “for nonsmokers.”
On the issue of confounding factors, a general point: Sure, smokers and nonsmokers should be disaggregated. Nothing inherently sinister there. But the confounding factors one admits into a statistical model are often as important as the ones that get controlled for. For example, couldn’t one make a case that they should have excluded people who don’t exercise regularly? We know that lack of exercise makes people (1) less healthy and (2) fatter. But the confounding factor represented by smoking gets dealt with while the confounding factor represented by exercise doesn’t. Isn’t this a kind of bias?
Sailorman:
But there is evidence that thinness itself is the problem. The dramatic upturn in deaths at low BMIs is also seen in when studies exclude smokers from the analysis. For example, it’s apparent in Calle et. al, which involved over 1,000,000 adults , some of whom smoked and some of whom don’t. Special attention was given to the non-smoking group precisely because people suggest what you do: The uptick in death at the low end is due to smoking. In fact, in Calle et all, the uptick in deaths at BMI’s below 20 was more distinct when smokers were eliminated from the population.
Can you cite any studies that show this uptick in the death rate goes away when one corrects from smoking?
Ref:
“Body-Mass Index and Mortality in a Prospective Cohort of U.S. Adults.” 1999, Calle, E.E., Thun, M.J., Petrelly, J.M., Rodriguez, C., Heath, C.W., New England Journal of Medicine. Vol. 341. No. 15. pp 1097-1105
While I would agree that some of their manipulations are good (like the smoker exclusion [As an aside, I read Linda’s criticism of this to be focused on their reasons for doing so, not the fact that they actually do it. I found their paper very suggestive that they were letting the conclusion drive the data. That doesn’t mean the data is wrong, but it is still questionable science.]). Their normalization to the highest 3rd category of the normal weight while ignoring the other two categories is very dishonest however. Table 4 is the pertinent table for their end conclusions following all their manipulations. While overall the overweight categories do have a slightly higher risk than all the normal weight categories, this is really only significant at the highest 3rd category for men. The women do not have as good results, though the CI are rather large (given the scale) to take very seriously.
Assuming the data is in fact true and representative, a more honest conclusion from it would be that men have a slightly higher risk of death if they are approaching, or are, underweight or obese, but otherwise have low risk regardless of being normal weight or overweight. Women on the other hand are more at risk if underweight or the more overweight they are, this being more definite the closer to, and beyond, obese. These are correlations however, causation should not be assumed. And as Amp pointed out, these changes in risk are relatively small to begin with.
There are some reasons I question these data. One is the reliance not upon self-reporting, but self-memory reporting. Table 4 is supposed to be based upon individual’s BMI at 50. To be accurate the memory of the individual’s weight has to be accurate and the BMI must be calculated based upon their height at 50. The paper did not claim to ask for their height, only their weight. This would add another confounding factor, even assuming the participants could recall correctly.
I also am very much against using BMI as an indicator of being fat. While it generally is a good indicator for underweight, there are many other contributions to a person’s total weight besides fat. Muscle is the most notable. This might in fact be an explanation as to why the lower overweight categories do well. Without actually taking up this information though, there is no way to know. Too many invisible confounding factors.
Sailorman,
I’d be more convinced of the value of removing smokers from the mix if they also removed anyone who has ever smoked. Because, while smoking tends to suppress weight, thus elevating the risks of those of normal or below normal weights, stopping smoking tends to cause one to gain more weight than smoking generally suppresses. So it would raise the risks of those in the high end of the normal, the overweight and obese categories to include former smokers, but lower risks for most of the normal and underweight categories to exclude current smokers but include former ones.
Sounds like creative data manipulation to me.
Ah, statistical analysis, my favorite way to spend an evening (right, Dianne?)
Indeed.
I agree that studying smokers and non-smokers separately is appropriate and don’t really have any particular problem with that. Dismissing smokers from consideration altogether seems silly though: there are a lot of smokers out there and knowing how best to help smokers maintain a healthy weight (and defining what that is) is arguably even more important as they have a known risk already.
I don’t like the table amp presents at all. It needs p-values or confidence intervals or something. Without them, how is one to know whether the difference between, for example, 1.00 and 1.07 is signficant or not?
There are at least a couple of confounding factors the Adams study didn’t look at, though. One is exercise. One table in the Adams paper shows that the average amount of exercise is highest in the normal to overweight group and falls in both the underweight and obese groups. It is possible (though not by any means certain given that the analysis was not done) that the differences in exercise account for most or all the differences in mortality. IIRC, this wasn’t addressed at all in the text. Self-reporting is, of course, notoriously inaccurate and self-reports of events or conditions 20 years ago should be treated with extreme caution. (I have no idea what my weight was 20 years ago, for example.) Finally, even assuming that the weights reported for 20 years ago were accurate, weight is not a static characteristic. Did people who were in the overweight catagory at 50 do best if they stayed in that catagory, gained weight, or lost weight? That information might have been useful in terms of making public health policy. If, for example, people who went from the overweight catagory to the normal weight catagory lived longer, then losing weight could be reasonably recommended. If they did worse, then it should not be.
Jennifer: Thanks.
Jake Squid: DId you read the explanation in the paper? I will try to explain it more but you should read theirs first. An additional, short, suggestion: If you believe smokers’ BMIS are so close to normal that their removal would have little effect on the graphs, then you need to explain why their removal DID have a big effect on the graphs. I understand what you’re saying about the fraction of a BMI index, but in prctice it doesn’t seem to be right.
That’s why they chose that reference group: it just makes things easier to read, and their point easier to see. The relative numbers are what is important, and they actual conclusions are not affected by the choice of reference group.
Let me know if I didn’t explain that well enough and I’ll explain more on Monday.
Yes and no. Hmm, this one is always harder to explain.
A real confounding factor is something that is unrelated to what you’re trying to study. So the question “does smoking affect your health?” is not really a subset of “does weight affect your health” even though smoking causes reduced weight. This is because the significant health issues caused by smoking are not–we think–caused by reduced weight. (and if we wanted to find that out, we would perform the sort of data analysis that this paper used,albeit in more detail).
Exercise and weight are somewhat different. Exercise (and eating) are related to weight. Exercising to get thin is part of what makes being thinner healthy. It doesn’t mean it’s not technically a confounding factor, bu it also may be something you WANT to include in your dataset as it is closely linked.
drat, I can tell I’m not explaining this well. Let me try a different tack:
We need to control for smoking because the effects are REVERSED. Smokers are thinner, and also LESS healthy. So they obscure the results. Smoking is a bit like chronic disease.
No, that’s not it either. Heh. Jennifer–help! I’m not explaining this well at all.
Last try lol from a completely different angle: If you control for exercise, then you will still have a range of weights. However, for any given point (let’s use “no exercise”) the weights you have are more likely to be a result of genetic differences in metabolism than of anything else. So your conclusion ends up being “it’s healthier to have a particular type of genes” and while this may be true, or interesting, it doesn’t really address things in your control.
OTOH, if you include exercise, then your conclusion can be “it is healthier to weigh in ____ range.” That means there are things one can control–the atudy has more value.
Hmm. Still not really saying this right. more monday perhaps.
Yup, that would be better, I agree. But you have to use what data you have and I think these conclusions are reasonable based on the existing data. It’s not easy to get good data from hundreds of thousands of people. I am cautious of your inter[pretation that the groups differ in %age of former smokers though. These are pretty big groups and large groups tend to be similar. I think you would need to find some support that %age of previous smoking increases with weight to make that assumption.
Dianne: I gtg, more on Monday–it’s always fun to chat with you on these things. As usual, your comments are pretty accurate. Did you miss their comment on self-reporting accuracy, though? They addressed that in the paper text, as well as the “earlier reporting” issue for the 20 year stuff. That’s certainly a wek spot, but I don’t think it’s fair to say “notoriously inaccurate” without directly addressing the support they gave.
The other important confounding factor which got left out is poverty. It is associated with mortality and a higher average BMI.
I have no objection to controlling for smoking (but don’t think that means that smokers should be eliminated all together), but when that’s the only thing that is controlled for, then it makes conclusions extremely problematic.
They did adjust for exercise. They did not adjust for poverty (although education level is probably a weak proxy for income, and race is a weak proxy for wealth). In the version where they excluded smokers, they also excluded former smokers (rather, they divided up based on current smoker, former smoker, and never smoked).
On page 769:
The confounding effect of smoking was larger than for other variables, so running the never smoked sample as a separate set makes sense. I would speculate that the smoking effect is not so much due to the smokers are thin and unhealthy aspect (given Jake’s point that smokers are only 0.8 BMI thinner than non-smokers, and former smokers are unlikely to be even that much thinner), but rather to the fact that smokers die of very different things than non-smokers, and weight may actually play a different role in speed of death by lung cancer and emphysema than in death from other causes (both that a little extra weight helps, and that weight matters less because it is the cigarettes that kill you, thus the flatter curve with the higher BMI at the inflection point).
The use of recollected weight seems like a substantial problem with table 4, and the authors’ handling of the problem in the paper seems to be little more than hand waving. The issue is not whether some overweight people under-report and some underweight people over-report (which the authors claim (without justification) will probably balance out). Rather, the central issue would be whether under and over reporting has any correlation with other issues potentially relating to mortality. This is a compete unknown, and a potentially significant confounding effect.
The fact that the results for recollected weight are hugely different (particularly for the low BMI ranges) than the results from self-reported current weight seems like a significant red flag that the recollected weight information has serious problems.
So it seems like this study may do a better job of separating out the confounding effect of smoking than the CDC study, but uses less reliable data (self reporting and recollection) than the CDC study. Both the CDC study and this study find very small effects on mortality risk from being over weight or low “normal”, but they disagree on the direction of the effect for over-weight. This study finds a larger effect for the top 2 and bottom 1 range of BMI . The CDC study finds a smaller effect. Combined, these studies do not seem to suggest that over weight or low normal weight have a meaningful effect on mortality risk. The combined results from both these studies for the bottom and top ranges are more suggestive of a meaningful effect, particularly for the 40+ BMI range. However, the failure to account for the harmful effects of weight loss attempts (yoyo dieting) may make the meaning of the results in even the top bracket fairly ambiguous.
Charles S,
I’m not clear on how you arrived at the following statement:
You seem to be drawing a quantitative conclusion about the confounding effects of various variables…how is that conclusion reached?
It’s true that the baseline sample characteristics by BMI in Table 1 do seem to show a rather striking inverse correlation between current smoking status and BMI, and it’s also true that the inverse correlation between exercise and BMI does not appear quite so striking. But, needless to say, those impressions are nowhere near what’s needed to even begin to indicate a larger confounding effect for smoking than for any other variables in the study.
The authors’ own justification for segregating nonsmokers was, more or less, that they were surprised by the conclusions of earlier analyses which seemed to show that being a few extra pounds overweight didn’t matter too much, and they wanted to see if by excluding smokers from the analysis they could get a different conclusion (p. 776 — bottom of first column and top of second). This is perfectly legitimate as a matter of statistical inquiry, but it’s weird that they didn’t spend the extra few minutes to try segregating non-exercisers as well.
What’s really odd in relation to exercise is that they don’t even break out by exercise in table 2 (although that may be partly be because exercise is not a binary). So it is actually impossible to tell whether the effect of exercise was more or less significantly than smoking. However, they report that their results are adjusted for exercise level, and it requires a very high level of distrust of the authors to believe that they are hiding something interesting in relation to exercise level that they don’t bother to even mention at all.
Also, exercise level is important to mortality risk, but it is nowhere near as important as whether or not you have ever smoked, so the confounding effects of correlations between BMI and smoking status are going to be much more important than the confounding effects of correlations between exercise and BMI. No exercise is bad for you, but smoking kills.
Smoking effects are complicated. Current smoker correlates with low BMI, former smoker correlates with high BMI, mortality risk is shifted (although not as much as it is for chronic illness, which they also try to control for in table 4) for both former and current smokers. The large effect of smoking, the multiple statuses with opposite correlations, the correlation between smoking and chronic illness, all make it reasonable to accept the author’s claim that controlling for the effects of smoking is difficult, in a way that controlling for the effects of exercise is not.
It is a pity that the authors did not show results broken out by exercise as this data set does represent a decent data set for demonstrating the effect of fitness level on the effects of BMI. Does fitness ameliorate whatever mortality risk increase there is from higher or lower BMI? This dataset could be used to give an answer, but the authors did not bother to do so. Either they didn’t find anything particularly interesting (some slight but not impressive effect), or the authors’ anti-fat bias led them to find it a result not worth mentioning, or they found something interesting and they plan to publish it later (although you’d think it would get a passing mention here).
As to the authors’ anti-fat bias, and their tendency to assume their results, I think the use of recollected weight is their worst offense, rather than excluding smokers or not bothering to break out exercise level.
This passage is something of a glaring red flag. Simply because a method that uses lower quality data gives them a result that better matches their expectations, they claim that the lower quality data better accounts for one of the confounding effects of the study. This is very problematic.
Now, it is clear that weight at age 50 has a much larger effect on mortality risk over the following 10 years than weight later in life (note the age effect in fig1b and fig2b), so it does make sense that age 50 weight would have more effect on mortality on beyond the next ten years, so the goal of using age 50 weight seems a reasonable one. However, their source for age 50 weight is much less than ideal. Simply declaring that because using this data gives them the results they are looking for, that this data is the better data to use, shows serious signs of bias.
Sailorman,
You claim that “Their group was representative.”
Looking through the paper, I am not seeing this claim anywhere. Could you quote me the relevant passage? Did you simply mean that they adjusted for race and education level, or am I missing something major in the paper?
Since the inferior representativeness of previous studies is suggested as one of the reasons that previous studies (not using representative samples) found an increased risk, whereas the CDC study (Fleagal et al, citation 29) using a fully representative sample did not, whether the AARP sample is a representative sample seems very significant.
Did they adjust for exercise levels now? Or exercise levels over time? Either way it’s (presumably) self-reporting, which is problematic
Well, let’s see. Who do I get to hate? (And that means hate permanently and forever, since no premanent, safe form of weight loss exists for more than 2% of the population, and any condemnation in that chart is far more an irrevocable declaration of inferiority and deserved stigma than, say, any motivation to embrace life or take reasonable care or one’s health or anything silly like that.)
My “risk,” despite my fasting glucose of 70 (even though diabetes runs in my family), and 6-day a week work out sessions, and resting heart rate of 56 (at 46 years old, thanks), and slightly lower than average blood pressure, and those ubiquitiously fastened seatbelts, on this here chart is 1.95. (Almost one whole “death”. I’m so scared!) So, according to the prevailing hate, anyone less likely to die is a better person, and anyone more likely to die is a worse one, so here’s who I’m better than:
All really skinny people, unless they’re male asians:
OK, all you cancer & AIDS patients, you better feel like absolute, inferior crap whenever you see a fat, middle-aged white woman because we’re your BETTERS! And don’t hand me any crap about using your strength for dealing with your illness. Your business is to conform or else!
Sort-of skinny Hispanic men:
Serves ’em right. Real men have big muscles, and don’t tell me you’re a good dancer, ‘ese. The chart and I have declared you inferior.
Slightly chubby Hispanic women:
(And, yes, these numbers count, because the first rule in any “obesity” research is to always assume the absolute worst about people. We’re all lazy, we’re all liars, we’re all stupid gluttons, and we’re all doomed. The numbers that condemn people are always the right ones.) HA, HA! I WIN! Even though we face a lot of the same personal challenges and hassles, and even though the entire State of California would crumble to a halt without Brown women keeping the joint running, I AM YOUR SUPERIOR. This means, from now on, I get to wear those cute tops with the lace and ruffles, and you have to run around in bullet-proof polyester. BECAUSE THE CHART SAYS SO.
Fat Asian and Hispanic men:
Sorry, Buddha, but I’m BETTER THAN YOU. And, alas, no more flirting in the beer aisle at Raley’s with the big cutie from Hawaii with the great hair and the punk band who’s way too young for me, but tempting anyway. Hm. I wonder if he’s into that whole B&D business. I mean, since I’m his superior and all, I might as well take advantage of the situation.
I’ll be watching the numbers. I’m really hoping fat Black men will edge me out on the risk level, ’cause there’s that one guy at work I’d really like to be able to hate with a clear conscience.
For his own good, of course.
Maia,
Looks like they used exercise levels now (or rather, at the start of the ten year study). As well as being self reported, it also doesn’t distinguish intensity of exercise, nor does it use functional measures of fitness. Also, the paper (frustratingly) only provides the reader with the average hours of exercise, not a distribution of hours of exercise for each BMI range.
But, from random googling, the association of fitness with mortality risk is only a little larger than the association of weight, so there would have to be a very high correlation between fitness and weight for the apparent effect of weight on mortality risk to simply be a placeholder for the effect of fitness, particularly given that they were already adjusting their results for a placeholder for fitness (hours of exercise).
The relationship between fitness and weight on mortality risk would have been an interesting result to have pulled out of this data set, but this data set wouldn’t be ideal for studying it, since it didn’t measure fitness, just self-reported hours of exercise.
“My “risk,” despite my fasting glucose of 70 (even though diabetes runs in my family), and 6-day a week work out sessions, and resting heart rate of 56 (at 46 years old, thanks), and slightly lower than average blood pressure, and those ubiquitiously fastened seatbelts, on this here chart is 1.95. (Almost one whole “death”. I’m so scared!) ”
Huh? The 1.95 shows relative risk. Twice as much risk (and it’s statistically significant). Not “one more death.”
I quit smoking eight months ago and I have gone from 160 pounds to 200 pounds.
I am praying that the weight gain ceases (I am 5 foot 8.5 inches) but given the
KNOWN mortality aspects of cigarette smoking I am willing to take my chances.
I used the anti-smoking drug Zyban to quit smoking. I am getting blood glucose
readings regularly for diabetes control and the readings have been normal up
until now. I am not a statistician (my Ph.D. in physiology was in basic science
research and I never took courses in statistics) so I do not know how to correctly
interpret the reliability of the published material that is presented and I notice
the tons of arguments over how to interpret the data – and I also remember
Benjamin Disraeli’s famous statement about statistics. I am just hoping that this
weight gain at the age of 54 does not take away too much of my future lifespan.
Two times 0.00003 (or whatever. What the hell are the real numbers on this, as in how many people studies and how many dead? I can crunch my own damn ratios.) is still 0.00006.
Nope. Still not scared.
(And, even if I were, there’s a) no proof that “weight loss” makes anyone healthier because b) NO PERMANENT WEIGHT LOSS METHOD EXISTS.)
“despite my fasting glucose of 70 ”
A fasting glucose 0f 70 is very close to being hypoglycemic, so it is dangerous
Pingback: Alas, a blog » Blog Archive » Black Women Live Longer If They’re “Overweight”
I was just discussing this article elsewhere, and somebody pointed something out to me – people put on weight as they age. If you watch ANTM, even Twiggy herself looks to be a healthy, normal body weight now that she’s pushing 60. So the question becomes relevant: at what point in their lifetime was their weight recorded?
I mean, is that the lifespan of somebody who was overweight at 50? Or at 20?
Pingback: Tim Russert, And Why I’m Done With Daily Kos Forever « fat fu
Pingback: Alas, a blog » Blog Archive » Should a 5′9″, 160 lb woman want to lose 35 pounds?
Pingback: Sociological Images » DOES BEING “OVERWEIGHT” CONTRIBUTE TO PREMATURE DEATH?
You all can keep nit picking over the methods and statistics, but even if the study was done perfectly and ethically it doesn’t refute the CDC study. At best this study shows the debate is far from over. At worst, it gets a lot of media attention, keeps people dieting when that might not be the healthiest thing. The most interesting thing to me, which only a few have commented on, is how close the relative risks actually are (with no confidence intervals). This tells me it might make very little difference what you weigh, and a lot more of a difference if you exercise.
The category “underweight” is incredibly vague. Do they mean 17.0-18.4 BMI, or from as low as 10-12 BMI to 18.4, which sort of means its completely unreliable for underweight?
Because I don’t know, at my height, a single pound makes me move by about 0.2 BMI. At 114 lbs I’m underweight, and at 115 lbs I’m in the “normal weight” category. I’m not sure at which point the risks become bad.
For fun I did a search on underweight health risk on google. It gave me dieting tips, and a few BMI things that mention underweight – but nothing about risks. None of them cover it (not meaning there isn’t truth to the higher risk, but it’s not even investigated in most sites).
Pingback: Alas, a blog » Blog Archive » In Defense Of HAES, and of Denialism
Pingback: Sir Mix-A-Lot Was Right