3 ms·

There's a difference between how people interested in public health read this research and how people such as us who are interested in our own individual health

by dkarl 16y ago

There's a difference between how people interested in public health read this research and how people such as us who are interested in our own individual health read it. People interested in public health wonder, "What happens when you tell people to follow diet X?" That's what studies like this test. People like us wonder, "What happens when an individual actually follows diet X?" Questions like this are never studied at all, because it's difficult and expensive. Almost everybody lies about their level of compliance, and when people aren't actually lying, they're still underreporting because most people remember eating less than they actually do. Those facts about underreporting are not controversial. I don't know where the misrepresentation happens, whether in the science itself or in the science journalism, but the studies most people read about in the newspaper or on the web are not about diet from an individual point of view. They don't study what happens when somebody actually follows a certain diet. That would require keeping people in an institutional setting and controlling or monitoring their food intake around the clock. That's really expensive, and if you're interested in improving public health, it isn't useful to know what the results of following a particular diet are, because you don't have control over what people eat. You only have influence over the public health message: what people are told they should eat. So that's what is studied. The scientific debate over diet is not about what you should eat to improve your health, but what we should tell the public to eat to improve their health. If 50% of people on a low-fat diet stay up all night eating low-fat cookies and big bowls of pasta with low-fat margarine, then from a public health perspective, low-fat diets make you fat.

wycats 16y ago
It's certainly true that dietary studies suffer from problems of underreporting. So it is possible that the the results of most recent studies (summarized well in http://en.wikipedia.org/wiki/Medical_research_related_to_low-carbohydrate_diets http://en.wikipedia.org/wiki/Medical_research_related_to_low...) all suffer from a reporting bias that leads people consuming low-fat diets to more significantly underreport their consumption than those on low-carb diets. However, that in itself would be a significant finding. If the level of reporting compliance varies widely between different diets, it may also mean that compliance with the diet itself varies. I'd personally be interested in evidence on this point. Also interestingly, studies using animal models of obesity tend to back up the findings using the (admittedly flawed) studies using human reporting. Taubes' most recent book goes into some detail about findings using rat obesity models (such as experiments involving ovary removal and experiments involving Zucker rats, which are genetically prone to obesity). One very interesting result from those studies is that rats prone to obesity (for whatever reason) will often die of starvation while retaining some subcutaneous fat. Obese-prone rats also very clearly exhibit sedentary behavior when the food supply is low, and more active behavior when the food supply is high. They also have more fat when given identical amounts of food as lean-prone rats. In short, while we cannot be certain about the recent experimental results, because of reporting bias, it's hard to believe that the experiments all have no validity, and that animal models of obesity are flawed. Additionally, if there was significant reporting bias, it would likely be the result of compliance problems, which would be extremely relevant to the discussion at hand. It's one thing believe that the jury is still out on some of this; it's quite another to stick to preexisting beliefs on the grounds that virtually all recent studies have some flaw or other.
- dkarl 16y ago
  Compliance levels in studies aren't relevant to our individual choice of diet, though, because we can observe our own compliance for ourselves by recording our intake. Observing our own bodies' response to a diet takes a much bigger commitment, so it's nice to have some generalizations from science. For the public health debate, there's no reason to separate the two, but from an individual perspective it is unnatural not to. It's one thing believe that the jury is still out on some of this; it's quite another to stick to preexisting beliefs on the grounds that virtually all recent studies have some flaw or other. Interesting that you assume I have some particular opinion :-) I don't necessarily disagree with Taube, though I don't think it signifies much that he can persuasively make a point and disparage those who disagree with him when he's talking to lay people. I'm just tired of seeing studies discussed as if the results reflect what happens when people actually follow some particular diet, when that's not what researchers are really interested in. (I.e., researchers take compliance as part of the phenomenon being studied, but journalists and lay people reading and discussing the results assume that compliance is uniformly good or has been factored out somehow.) Edit/PS: If you interpret studies as showing what happens when people actually eat a certain diet, then compliance _is_ a flaw -- a really big one! We don't know what people are actually eating, only that on average it's significantly more than they report. And there's no questions researchers would _like_ to know how actual intake relates to weight gain or loss. It's just rather inconvenient and expensive to find out, and it's arguably less important than the public health question of how instruction/exhortation relates to weight gain or loss.
  - wycats 16y ago
    Compliance problems in studies may be relevant to our individual choice of diet, since they are likely to predict our own personal compliance. In general (as Taubes points out), most people can lose significant weight by reducing their daily caloric intake to starvation levels. However, compliance is so poor that it's not a viable solution. And it wouldn't be correct to say that this problem is irrelevant to those who are looking to start a new diet. I personally find the combination of the (potentially flawed) human studies and the (more scientifically rigorous) animal studies to be fairly persuasive. You can find fault with either of the two, but in sum, the two pieces of evidence seem quite solid. "If you interpret studies as showing what happens when people actually eat a certain diet, then compliance _is_ a flaw -- a really big one!" To me, the studies likely skew in that direction, while also providing stronger evidence when interpreted as showing what happens when people try to eat a certain diet. "I don't necessarily disagree with Taube, though I don't think it signifies much that he can persuasively make a point and disparage those who disagree with him when he's talking to lay people." Sure. On the other hand, he's a researcher who's been investigating this issue for around a decade, and has published a very rigorous (by most accounts) book on the subject. Given the stakes, I suspect that most people would eventually take to the streets in an attempt to improve the situation. Overall, I think we likely agree here :)

by dkarl 16y ago

wycats 16y ago

It's certainly true that dietary studies suffer from problems of underreporting. So it is possible that the the results of most recent studies (summarized well in http://en.wikipedia.org/wiki/Medical_research_related_to_low-carbohydrate_diets http://en.wikipedia.org/wiki/Medical_research_related_to_low...) all suffer from a reporting bias that leads people consuming low-fat diets to more significantly underreport their consumption than those on low-carb diets. However, that in itself would be a significant finding. If the level of reporting compliance varies widely between different diets, it may also mean that compliance with the diet itself varies. I'd personally be interested in evidence on this point. Also interestingly, studies using animal models of obesity tend to back up the findings using the (admittedly flawed) studies using human reporting. Taubes' most recent book goes into some detail about findings using rat obesity models (such as experiments involving ovary removal and experiments involving Zucker rats, which are genetically prone to obesity). One very interesting result from those studies is that rats prone to obesity (for whatever reason) will often die of starvation while retaining some subcutaneous fat. Obese-prone rats also very clearly exhibit sedentary behavior when the food supply is low, and more active behavior when the food supply is high. They also have more fat when given identical amounts of food as lean-prone rats. In short, while we cannot be certain about the recent experimental results, because of reporting bias, it's hard to believe that the experiments all have no validity, and that animal models of obesity are flawed. Additionally, if there was significant reporting bias, it would likely be the result of compliance problems, which would be extremely relevant to the discussion at hand. It's one thing believe that the jury is still out on some of this; it's quite another to stick to preexisting beliefs on the grounds that virtually all recent studies have some flaw or other.

dkarl 16y ago
Compliance levels in studies aren't relevant to our individual choice of diet, though, because we can observe our own compliance for ourselves by recording our intake. Observing our own bodies' response to a diet takes a much bigger commitment, so it's nice to have some generalizations from science. For the public health debate, there's no reason to separate the two, but from an individual perspective it is unnatural not to. It's one thing believe that the jury is still out on some of this; it's quite another to stick to preexisting beliefs on the grounds that virtually all recent studies have some flaw or other. Interesting that you assume I have some particular opinion :-) I don't necessarily disagree with Taube, though I don't think it signifies much that he can persuasively make a point and disparage those who disagree with him when he's talking to lay people. I'm just tired of seeing studies discussed as if the results reflect what happens when people actually follow some particular diet, when that's not what researchers are really interested in. (I.e., researchers take compliance as part of the phenomenon being studied, but journalists and lay people reading and discussing the results assume that compliance is uniformly good or has been factored out somehow.) Edit/PS: If you interpret studies as showing what happens when people actually eat a certain diet, then compliance _is_ a flaw -- a really big one! We don't know what people are actually eating, only that on average it's significantly more than they report. And there's no questions researchers would _like_ to know how actual intake relates to weight gain or loss. It's just rather inconvenient and expensive to find out, and it's arguably less important than the public health question of how instruction/exhortation relates to weight gain or loss.
- wycats 16y ago
  Compliance problems in studies may be relevant to our individual choice of diet, since they are likely to predict our own personal compliance. In general (as Taubes points out), most people can lose significant weight by reducing their daily caloric intake to starvation levels. However, compliance is so poor that it's not a viable solution. And it wouldn't be correct to say that this problem is irrelevant to those who are looking to start a new diet. I personally find the combination of the (potentially flawed) human studies and the (more scientifically rigorous) animal studies to be fairly persuasive. You can find fault with either of the two, but in sum, the two pieces of evidence seem quite solid. "If you interpret studies as showing what happens when people actually eat a certain diet, then compliance _is_ a flaw -- a really big one!" To me, the studies likely skew in that direction, while also providing stronger evidence when interpreted as showing what happens when people try to eat a certain diet. "I don't necessarily disagree with Taube, though I don't think it signifies much that he can persuasively make a point and disparage those who disagree with him when he's talking to lay people." Sure. On the other hand, he's a researcher who's been investigating this issue for around a decade, and has published a very rigorous (by most accounts) book on the subject. Given the stakes, I suspect that most people would eventually take to the streets in an attempt to improve the situation. Overall, I think we likely agree here :)