the observation that fat people have significantly greater mortality than thin people
That’s not how I read that chart and the many similar ones showing mortality as a function of body mass index.
If, for the sake of argument, we make the (unreasonable and wrong!) assumption that the variance in mortality is caused by the variance in body mass index, it looks to me more like being fat is much less dangerous than being thin. Look at the shape of the curves as you move away from the minimum mortality trough around BMI 19-26 or so (which is slightly higher than many official guidelines for ‘normal’ weight). Sure, mortality increases steadily as BMI increases from the minimum, but it shoots up much more steeply as BMI decreases. Indeed, for all subjects, the BMI=17.5 bucket (the only one thinner than the minimum mortality trough) has higher mortality than the BMI=42.5 bucket (significantly fatter). To put that in concrete terms, the mortality risk for a six-foot high person (1.83m) is higher if they weigh 130 lb (59 kg, 9 stone 3 lb, BMI 17.5) than if they weigh 310 lb (142 kg, 22 stone 6 lb, BMI 42.5).
There aren’t any data points in the buckets below BMI 17.5, and my guess is that’s because people tend not to have BMIs in that range—or if they do it’s not for very long, because they die. Of course, that’s often because—contra the assumption above- causality there often clearly runs the other way: people who get terminally ill often lose an awful lot of weight before they die (cachexia).
Curves like the one you link to are a common finding. Another common finding is that the curve shifts to the right as people age—i.e. the lowest-risk BMI increases over time for an individual, and being officially ‘overweight’ (BMI 25-30) has lower all-cause mortality than being officially ‘healthy weight’ (BMI 18-25). (Can’t instantly put my hand on a good example, but this study found the mortality minimum was 27.1 for people aged 70-75.)
That seems to be to be at least some evidence against the idea that to maximise your lifespan you should make your weight decrease as your life progresses.
This is very difficult stuff to get robust data about—it’s awash in complex correlation/causation stuff, and getting hold of data is very hard. It’s also overlaid with a lot of moralising that seems pretty unhelpful.
If, for the sake of argument, we make the (unreasonable and wrong!) assumption that the variance in mortality is caused by the variance in body mass index, it looks to me more like being fat is much less dangerous than being thin. Look at the shape of the curves as you move away from the minimum mortality trough around BMI 19-26 or so (which is slightly higher than many official guidelines for ‘normal’ weight)
I would agree with you if we were looking only at that chart and making no use at all of any other knowledge or common sense. But if one adds in other knowledge and common sense, it’s reasonable to believe that there is cause and effect on the right side of the chart.
The left side is more difficult to interpret since there are many unhealthy conditions which can cause weight loss.
This is very difficult stuff to get robust data about—it’s awash in complex correlation/causation stuff,
This is true of just about any data concerning human health. But health decisions need to be made. Under the circumstances, I’m reasonably comfortable with (1) not smoking; (2) restricting calories enough so that I do not become fat; (3) eating plenty of fruits and vegetables; (4) exercising regularly. I think there’s a good chance that these things will improve my longevity and perhaps more importantly I think it’s pretty unlikely that I will be significantly worse off for having done these things.
My concern is particularly describing “thin” as healthy, or low risk for mortality. If by “thin” you mean BMI 18-25, then I’m with you, but that’s officially labelled “healthy” or “normal” weight and is not what most people mean by thin. The official “underweight” category (<18) is much riskier than the official “overweight” category (25-30). The risk profile either side of official “healthy” weight is not symmetrical—and indeed there are sound reasons to think that tending towards the top end of “healthy” and in to “overweight” as you age is the least-risk track for weight over a life course.
here are many unhealthy conditions which can cause weight loss.
Indeed—eating disorders being a particularly notable group. I am concerned that erroneous messages that “thin is good and healthy” are exacerbating those problems, causing significant avoidable mortality. Thin is not good and healthy.
(You suggested “being fat puts a lot of abnormal extra strain on your system almost all the time”; I suspect being thin does too, since it means your body will your body will struggle to find sufficient metabolic resources for things like healing processes, regeneration, the immune system and cellular repair mechanisms.)
The curves on that graph for “healthy subjects who never smoked” should exclude people with unhealthy conditions and diseases that affect their weight, and show the same pattern, albeit reduced—you have to get up in to the “obese” category (plotted at BMI ~31) to get a mortality risk as high as the “underweight” one.
One might be able to make a case that there is a particular subset of underweight people who do not experience the significantly raised mortality risk that other underweight people do, but I’ve not (yet?) seen a convincing one.
it’s reasonable to believe that there is cause and effect on the right side of the chart.
Sure—but it’s not simple and one-way. One can also reasonably interpret the data to find “being low socio-economic status” as a causal factor of both higher BMI and higher mortality risk. (And of course there are also diseases that cause weight gain and increased mortality.)
I think there’s a good chance that these things will improve my longevity and perhaps more importantly I think it’s pretty unlikely that I will be significantly worse off for having done these things.
Absolutely—that list seems a good distillation of my understanding of what the evidence supports too.
My concern is particularly describing “thin” as healthy, or low risk for mortality. If by “thin” you mean BMI 18-25, then I’m with you,
I didn’t precisely define the word “thin,” but what I had in mind was “not fat.” I suppose that BMI 18-25 is roughly what I mean by “thin.” I consider myself to be “thin” even though my BMI of 24 puts me close to the official line for “overweight.”
(You suggested “being fat puts a lot of abnormal extra strain on your system almost all the time”; I suspect being thin does too, since it means your body will your body will struggle to find sufficient metabolic resources for things like healing processes, regeneration, the immune system and cellular repair mechanisms.)
Depending on your definition of “thin,” I would agree with you. And it’s part of the reason I am not engaging in what I have defined as “severe calorie restriction.”
The curves on that graph for “healthy subjects who never smoked” should exclude people with unhealthy conditions and diseases that affect their weight
I doubt that they do so completely. Between people lying about their health and simply being unaware of latent serious health problems, I am pretty confident that the curves for “healthy people who never smoked” include a decent number of people who are actually not healthy.
One can also reasonably interpret the data to find “being low socio-economic status” as a causal factor of both higher BMI and higher mortality risk.
It would depend what data you are looking at. If you look at the entire universe of general knowledge and common sense, this seems unlikely—at least as the primary factor.
I consider myself to be “thin” even though my BMI of 24 puts me close to the official line for “overweight.”
Aha! I think we’ve found the main source of our disagreement here, and it’s purely terminology. Totally agree that maintaining a BMI around 24 is a reasonable, broadly-supported aspiration (all other factors being equal), particularly if you’re younger than middle age.
this seems unlikely—at least as the primary factor
Agreed it’s probably not the largest effect, but I do think there’s good reason to think there is an effect going that way. There seems to be a growing amount of evidence that low socio-economic status is bad for mortality, mostly indirectly (makes you more likely to do things like smoking, eating a diet with less fresh fruit and vegetables, etc) but also directly (low social status makes you die sooner), although of course separating that out of any naturalistic data is hard. (See e.g. this, and the older Whitehall studies.)
There seems to be a growing amount of evidence that low socio-economic status is bad for mortality, mostly indirectly (makes you more likely to do things like smoking, eating a diet with less fresh fruit and vegetables, etc)
This is subject to the same cause and effect issues you alluded to earlier. It’s reasonable to hypothesize that worse-than-average impulse control is likely to result in both low socio-economic status and overweight/obesity.
Anyway, for my purposes it doesn’t really matter. The evidence is strong enough that at a minimum I’m willing to stake my health on the claim that it’s better for one’s health to avoid getting fat.
That’s not how I read that chart and the many similar ones showing mortality as a function of body mass index.
If, for the sake of argument, we make the (unreasonable and wrong!) assumption that the variance in mortality is caused by the variance in body mass index, it looks to me more like being fat is much less dangerous than being thin. Look at the shape of the curves as you move away from the minimum mortality trough around BMI 19-26 or so (which is slightly higher than many official guidelines for ‘normal’ weight). Sure, mortality increases steadily as BMI increases from the minimum, but it shoots up much more steeply as BMI decreases. Indeed, for all subjects, the BMI=17.5 bucket (the only one thinner than the minimum mortality trough) has higher mortality than the BMI=42.5 bucket (significantly fatter). To put that in concrete terms, the mortality risk for a six-foot high person (1.83m) is higher if they weigh 130 lb (59 kg, 9 stone 3 lb, BMI 17.5) than if they weigh 310 lb (142 kg, 22 stone 6 lb, BMI 42.5).
There aren’t any data points in the buckets below BMI 17.5, and my guess is that’s because people tend not to have BMIs in that range—or if they do it’s not for very long, because they die. Of course, that’s often because—contra the assumption above- causality there often clearly runs the other way: people who get terminally ill often lose an awful lot of weight before they die (cachexia).
Curves like the one you link to are a common finding. Another common finding is that the curve shifts to the right as people age—i.e. the lowest-risk BMI increases over time for an individual, and being officially ‘overweight’ (BMI 25-30) has lower all-cause mortality than being officially ‘healthy weight’ (BMI 18-25). (Can’t instantly put my hand on a good example, but this study found the mortality minimum was 27.1 for people aged 70-75.)
That seems to be to be at least some evidence against the idea that to maximise your lifespan you should make your weight decrease as your life progresses.
This is very difficult stuff to get robust data about—it’s awash in complex correlation/causation stuff, and getting hold of data is very hard. It’s also overlaid with a lot of moralising that seems pretty unhelpful.
I would agree with you if we were looking only at that chart and making no use at all of any other knowledge or common sense. But if one adds in other knowledge and common sense, it’s reasonable to believe that there is cause and effect on the right side of the chart.
The left side is more difficult to interpret since there are many unhealthy conditions which can cause weight loss.
This is true of just about any data concerning human health. But health decisions need to be made. Under the circumstances, I’m reasonably comfortable with (1) not smoking; (2) restricting calories enough so that I do not become fat; (3) eating plenty of fruits and vegetables; (4) exercising regularly. I think there’s a good chance that these things will improve my longevity and perhaps more importantly I think it’s pretty unlikely that I will be significantly worse off for having done these things.
My concern is particularly describing “thin” as healthy, or low risk for mortality. If by “thin” you mean BMI 18-25, then I’m with you, but that’s officially labelled “healthy” or “normal” weight and is not what most people mean by thin. The official “underweight” category (<18) is much riskier than the official “overweight” category (25-30). The risk profile either side of official “healthy” weight is not symmetrical—and indeed there are sound reasons to think that tending towards the top end of “healthy” and in to “overweight” as you age is the least-risk track for weight over a life course.
Indeed—eating disorders being a particularly notable group. I am concerned that erroneous messages that “thin is good and healthy” are exacerbating those problems, causing significant avoidable mortality. Thin is not good and healthy.
(You suggested “being fat puts a lot of abnormal extra strain on your system almost all the time”; I suspect being thin does too, since it means your body will your body will struggle to find sufficient metabolic resources for things like healing processes, regeneration, the immune system and cellular repair mechanisms.)
The curves on that graph for “healthy subjects who never smoked” should exclude people with unhealthy conditions and diseases that affect their weight, and show the same pattern, albeit reduced—you have to get up in to the “obese” category (plotted at BMI ~31) to get a mortality risk as high as the “underweight” one.
One might be able to make a case that there is a particular subset of underweight people who do not experience the significantly raised mortality risk that other underweight people do, but I’ve not (yet?) seen a convincing one.
Sure—but it’s not simple and one-way. One can also reasonably interpret the data to find “being low socio-economic status” as a causal factor of both higher BMI and higher mortality risk. (And of course there are also diseases that cause weight gain and increased mortality.)
Absolutely—that list seems a good distillation of my understanding of what the evidence supports too.
I didn’t precisely define the word “thin,” but what I had in mind was “not fat.” I suppose that BMI 18-25 is roughly what I mean by “thin.” I consider myself to be “thin” even though my BMI of 24 puts me close to the official line for “overweight.”
Depending on your definition of “thin,” I would agree with you. And it’s part of the reason I am not engaging in what I have defined as “severe calorie restriction.”
I doubt that they do so completely. Between people lying about their health and simply being unaware of latent serious health problems, I am pretty confident that the curves for “healthy people who never smoked” include a decent number of people who are actually not healthy.
It would depend what data you are looking at. If you look at the entire universe of general knowledge and common sense, this seems unlikely—at least as the primary factor.
Aha! I think we’ve found the main source of our disagreement here, and it’s purely terminology. Totally agree that maintaining a BMI around 24 is a reasonable, broadly-supported aspiration (all other factors being equal), particularly if you’re younger than middle age.
Agreed it’s probably not the largest effect, but I do think there’s good reason to think there is an effect going that way. There seems to be a growing amount of evidence that low socio-economic status is bad for mortality, mostly indirectly (makes you more likely to do things like smoking, eating a diet with less fresh fruit and vegetables, etc) but also directly (low social status makes you die sooner), although of course separating that out of any naturalistic data is hard. (See e.g. this, and the older Whitehall studies.)
This is subject to the same cause and effect issues you alluded to earlier. It’s reasonable to hypothesize that worse-than-average impulse control is likely to result in both low socio-economic status and overweight/obesity.
Anyway, for my purposes it doesn’t really matter. The evidence is strong enough that at a minimum I’m willing to stake my health on the claim that it’s better for one’s health to avoid getting fat.