BMI not a mortality risk in itself
August 31, 2022 7:33 AM Subscribe
"When fasting insulin and the natural logarithm of c-reactive protein were included in the model, an inverse association between BMI and mortality was present" Risk is overestimated for fat people, some of whom don't have the real risk factors and underestimated for thin people, some of whom do have the risk factors.
Can someone who is good at reading medical studies provide a summary of what this one actually says? There's a lot of things mediated and parameterized in this article and I'm not sure I actually know what it all means. I get the top level summary is basically 'fat doesn't make you die, insulin levels and inflammation make you die'. And then maybe it says 'insulin levels and inflammation make you fat, but being fat doesn't necessarily change your insulin levels and inflammation' but I am unsure about whether that interpretation is correct.
posted by jacquilynne at 8:27 AM on August 31, 2022 [4 favorites]
posted by jacquilynne at 8:27 AM on August 31, 2022 [4 favorites]
I am endeavoring not to reply with snark, not because I think this study is wrong, but because I think this study is absolutely right, just as educated medical professionals and fat activists have been saying for YEARS. Nancy Lebovitz, thank you very much for making this post; I'm glad to see this study. I'm frustrated it's taking so long to chip away at the fatphobia in medicine (and society at large), but I hope this will help.
posted by epj at 8:54 AM on August 31, 2022 [20 favorites]
posted by epj at 8:54 AM on August 31, 2022 [20 favorites]
It's good news for individuals with high BMIs but who have low fasting insulin and CRP levels, but that's a relatively small group:
That doesn't necessarily mean that having a lower BMI would necessarily reduce those risk factors, although my understanding is that they tend to go hand in hand, at least for insulin issues (e.g. the GLP-1 agonist drugs both reduce weight and improve insulin sensitivity).
Overall it seems to be evidence that there should be more focus on the proximate causes of disease and mortality rather than weight as-such, the causal role of which is complicated at best, especially given the difficulty of lasting weight loss. As the authors put it:
posted by jedicus at 8:57 AM on August 31, 2022 [27 favorites]
the proportion of those in the top 25th percentile of fasting insulin (≥87 pmol//L) increased with increasing BMI (from 1.3% to 81.3%) ... the proportion of those in the top 25th percentile of CRP (≥4.4 mg/L) increased with increasing BMI (from 6.0% to 74.5%)So whichever direction the causality arrow points between weight and fasting insulin and CRP, it remains the case that heavier people tend to also have these risk factors. This study suggests that health at any size is possible (at least where health is measured by all-cause mortality*), but it does not appear to be especially common, with the substantial majority of people with higher BMIs also having these risk factors.
That doesn't necessarily mean that having a lower BMI would necessarily reduce those risk factors, although my understanding is that they tend to go hand in hand, at least for insulin issues (e.g. the GLP-1 agonist drugs both reduce weight and improve insulin sensitivity).
Overall it seems to be evidence that there should be more focus on the proximate causes of disease and mortality rather than weight as-such, the causal role of which is complicated at best, especially given the difficulty of lasting weight loss. As the authors put it:
Current public health strategies have tended to highlight the importance of weight reduction. At the population level, these strategies have increased weight stigma, weight cycling and weight gain —all of which associate with increased inflammation and hyperinsulinemia. Our findings suggest the need for renewed focus on alternative interventions that reduce levels of insulin, inflammation, and perhaps the link between physiological stress, hyperinsulinemia, and inflammation.* This was an acknowledged limitation of the study, which "could not explore the temporal associations between BMI, hyperinsulinemia, chronic inflammation, and the development of noncommunicable chronic disease such as hypertension, cancer, diabetes, or coronary disease, as mortality is the only longitudinal outcome available in this dataset."
posted by jedicus at 8:57 AM on August 31, 2022 [27 favorites]
I have no doubt there are better measurements of health than BMI. It is used because it is old and cheap, not because it is especially accurate.
I do wonder about BMI being a confounding factor for cancer mortality (which is the second cause of death in the US). Cancer treatments can lead to a lot of weight loss and when the patient dies, they may not be in the same BMI category as when they started. As the study says, it cannot factor in if BMI is associated with getting any of these diseases, but is only measured at the time of death.
posted by meowzilla at 9:29 AM on August 31, 2022 [2 favorites]
I do wonder about BMI being a confounding factor for cancer mortality (which is the second cause of death in the US). Cancer treatments can lead to a lot of weight loss and when the patient dies, they may not be in the same BMI category as when they started. As the study says, it cannot factor in if BMI is associated with getting any of these diseases, but is only measured at the time of death.
posted by meowzilla at 9:29 AM on August 31, 2022 [2 favorites]
Also what's with the stark difference in hazard ratios between males and females? One is a flat line, the other is U-shaped.
posted by meowzilla at 9:32 AM on August 31, 2022
posted by meowzilla at 9:32 AM on August 31, 2022
Just imagine how much more dramatic these results would be if they were corrected for the medical neglect of people with high BMI!
posted by heatherlogan at 9:35 AM on August 31, 2022 [38 favorites]
posted by heatherlogan at 9:35 AM on August 31, 2022 [38 favorites]
I’m so looking forward to reading this! Based on my last review of the literature (I’m a lapsed registered dietitian with a masters in nutrition) all we know is that overweight and insulin resistance are correlated. What we don’t know is which direction the causality lies. Most people assume overweight causes insulin resistance, but from my personal exposure to the research I’m convinced it goes the other way, ie. insulin resistance a cause of overweight.
Since insulin triggers a cascade of hormones designed to alter your behavior to make you want to eat less and move more to clear that sugar out of your blood, if your body is ignoring insulin you just stay hungry and sluggish despite being fully fed. So of course it’s going to lead to weight gain in most people.
The really interesting question is, what’s causing the rise in insulin resistance?
posted by antinomia at 9:37 AM on August 31, 2022 [11 favorites]
Since insulin triggers a cascade of hormones designed to alter your behavior to make you want to eat less and move more to clear that sugar out of your blood, if your body is ignoring insulin you just stay hungry and sluggish despite being fully fed. So of course it’s going to lead to weight gain in most people.
The really interesting question is, what’s causing the rise in insulin resistance?
posted by antinomia at 9:37 AM on August 31, 2022 [11 favorites]
Trying to tread lightly here, but as a fat person I find the preoccupation in some corners with establishing that fatness is healthy (whether or not that ultimately turns out to be the case, I couldn't say) to be... kind of ableist? My body is fine and valid and is the way it is, whether or not that state can be classified as optimal health. Hopefully studies like this can help reduce the medical fatphobia of people being given ignorant and dismissive healthcare, but, healthiness isn't any more moral or legitimate than thinness.
posted by dusty potato at 9:53 AM on August 31, 2022 [32 favorites]
posted by dusty potato at 9:53 AM on August 31, 2022 [32 favorites]
It's good news for individuals with high BMIs but who have low fasting insulin and CRP levels, but that's a relatively small group:
That's a very good point about the arrow of causality, and I think there is an interpretation of pointing the arrow in the unconventional direction, from high levels of fasting insulin and CRP toward high BMI, that doesn’t get the attention it deserves.
What if getting fat helps to protect a person from the deleterious effects of the underlying conditions which are associated with high fasting insulin and CRP?
Somewhat like the way high fevers help keep people from dying from infections. If running a temperature had the level of stigma attached to it that being fat does, would we have a segment of the health profession running around saying 'oh my goodness, people with high fevers have an elevated risk of death and many of them also seem to have these bad infections — we must do everything we can to get these fevers down, and presumably that will help with the infections, too!'?
I’d like to see a double blinded study of a group of people with high BMI who all embark on the same weight loss program and have their fasting insulin and CRP levels measured in advance and as the program goes forward, but without disclosing the results to them or their doctors until the end of the program.
If it were to turn out that the people who had lower levels of fasting insulin and CRP at the beginning and as the study progressed also were more successful at losing weight and keeping it off, that would be revelatory.
posted by jamjam at 10:35 AM on August 31, 2022 [1 favorite]
the proportion of those in the top 25th percentile of fasting insulin (≥87 pmol//L) increased with increasing BMI (from 1.3% to 81.3%) ... the proportion of those in the top 25th percentile of CRP (≥4.4 mg/L) increased with increasing BMI (from 6.0% to 74.5%)So whichever direction the causality arrow points between weight and fasting insulin and CRP …
That's a very good point about the arrow of causality, and I think there is an interpretation of pointing the arrow in the unconventional direction, from high levels of fasting insulin and CRP toward high BMI, that doesn’t get the attention it deserves.
What if getting fat helps to protect a person from the deleterious effects of the underlying conditions which are associated with high fasting insulin and CRP?
Somewhat like the way high fevers help keep people from dying from infections. If running a temperature had the level of stigma attached to it that being fat does, would we have a segment of the health profession running around saying 'oh my goodness, people with high fevers have an elevated risk of death and many of them also seem to have these bad infections — we must do everything we can to get these fevers down, and presumably that will help with the infections, too!'?
I’d like to see a double blinded study of a group of people with high BMI who all embark on the same weight loss program and have their fasting insulin and CRP levels measured in advance and as the program goes forward, but without disclosing the results to them or their doctors until the end of the program.
If it were to turn out that the people who had lower levels of fasting insulin and CRP at the beginning and as the study progressed also were more successful at losing weight and keeping it off, that would be revelatory.
posted by jamjam at 10:35 AM on August 31, 2022 [1 favorite]
I've been arguing this point on the Blue for years--given the choice of being in good shape (regular exercise, good diet, which often leads to lower fasting insulin and c-reactive protien) and being skinny, you are better off being in good shape. (Being skinny sometimes follows, but doesn't have to).
Lots of doctors are 'BMI stupid' and don't bother looking at any other aspect of their patients health.
posted by eye of newt at 10:45 AM on August 31, 2022 [9 favorites]
Lots of doctors are 'BMI stupid' and don't bother looking at any other aspect of their patients health.
posted by eye of newt at 10:45 AM on August 31, 2022 [9 favorites]
It's not an amazing coincidence that "Overweight" starts at 25.0 BMI and "Obese" starts at 30.0 BMI. They're just whole numbers. That should tell you everything you need to know about how precisely useful the measure is.
posted by Etrigan at 11:16 AM on August 31, 2022 [10 favorites]
posted by Etrigan at 11:16 AM on August 31, 2022 [10 favorites]
Michael Hobbes & Aubrey Gordon are going to be alll ooooover this.
posted by wenestvedt at 11:28 AM on August 31, 2022 [12 favorites]
posted by wenestvedt at 11:28 AM on August 31, 2022 [12 favorites]
Like I.Q., BMI has an origin story that makes it impossible for me to consider that number anything even resembling "scientific."
posted by MrJM at 12:04 PM on August 31, 2022 [6 favorites]
posted by MrJM at 12:04 PM on August 31, 2022 [6 favorites]
I am not very sold.
First, the main result is that patients with high BMI but without the biochemical changes that strongly correlate with obesity have lower mortality. That sounds like a story about the limitations of BMI as a measure of obesity. The biggest factor for cadiovascular health is exercise, which only weakly reduces weight for most people.
Second, it has long been recognized that "overweight" is the lowest risk group for mortality. A lot of that is due to confounding (lower BMI is associated with smoking, chronic disease, cancer, and poor nutrition), which they try to adjust for. However, it doesn't take a lot of mis-measurement to create associations of this strength.
Third, the message that things like insulin resistance and inflammation come before obesity is not very convincing. Interventions like sustained diet and exercise that reduce BMI strongly decrease the hormonal changes that they point to as the "real causes." Diabetes and hypertension which are strongly correlated with obesity have substantial morbidity and mortality, and again modest weight loss results in improvements in both of those.
posted by a robot made out of meat at 12:21 PM on August 31, 2022 [9 favorites]
First, the main result is that patients with high BMI but without the biochemical changes that strongly correlate with obesity have lower mortality. That sounds like a story about the limitations of BMI as a measure of obesity. The biggest factor for cadiovascular health is exercise, which only weakly reduces weight for most people.
Second, it has long been recognized that "overweight" is the lowest risk group for mortality. A lot of that is due to confounding (lower BMI is associated with smoking, chronic disease, cancer, and poor nutrition), which they try to adjust for. However, it doesn't take a lot of mis-measurement to create associations of this strength.
Third, the message that things like insulin resistance and inflammation come before obesity is not very convincing. Interventions like sustained diet and exercise that reduce BMI strongly decrease the hormonal changes that they point to as the "real causes." Diabetes and hypertension which are strongly correlated with obesity have substantial morbidity and mortality, and again modest weight loss results in improvements in both of those.
posted by a robot made out of meat at 12:21 PM on August 31, 2022 [9 favorites]
The excellent Maintenance Phase podcast is a great source for information. If you haven't heard it before, I recommend starting with these two episodes:
Is Being Fat Bad for You?
The Body Mass Index
posted by AlSweigart at 12:50 PM on August 31, 2022 [5 favorites]
Is Being Fat Bad for You?
The Body Mass Index
posted by AlSweigart at 12:50 PM on August 31, 2022 [5 favorites]
I agree with a robot made out of meat. We'll have to wait and see if this statistical analysis stands up to scrutiny. This kind of analysis is very difficult to do: if a single study shows a wildly unexpected result, it's quite possible the statistics is wrong rather than everything we thought we knew.
So the conventional story is that obesity causes insulin resistance and inflammation, which cause multiple conditions which reduce lifespan. e.g., e.g., e.g.
This study is saying if we control for insulin resistance and inflammation, obesity is associated with longer lifespans.
It's a bit as if previously we thought being shot with a gun caused blood loss and tissue damage which reduce lifespan. But instead if you carefully process the statistical correlations, if you control for blood loss and tissue damage, being being shot with a gun is associated with longer lifespans.
The thing is that there's a ton of experimental evidence that if you over-feed animals from zebrafish to roundwords to mice to fruit flies they get the same kind of obesity-related conditions as over-fed humans.
It's hard to reconcile this result with the body of human and animal studies which say causality works the other way. I think there's something wrong with the statistics here.
posted by TheophileEscargot at 10:52 PM on August 31, 2022 [3 favorites]
So the conventional story is that obesity causes insulin resistance and inflammation, which cause multiple conditions which reduce lifespan. e.g., e.g., e.g.
This study is saying if we control for insulin resistance and inflammation, obesity is associated with longer lifespans.
It's a bit as if previously we thought being shot with a gun caused blood loss and tissue damage which reduce lifespan. But instead if you carefully process the statistical correlations, if you control for blood loss and tissue damage, being being shot with a gun is associated with longer lifespans.
The thing is that there's a ton of experimental evidence that if you over-feed animals from zebrafish to roundwords to mice to fruit flies they get the same kind of obesity-related conditions as over-fed humans.
It's hard to reconcile this result with the body of human and animal studies which say causality works the other way. I think there's something wrong with the statistics here.
posted by TheophileEscargot at 10:52 PM on August 31, 2022 [3 favorites]
TheophileEscargot, there might be more that one sort of obesity.
posted by Nancy Lebovitz at 11:35 PM on August 31, 2022 [1 favorite]
posted by Nancy Lebovitz at 11:35 PM on August 31, 2022 [1 favorite]
So the conventional story is that obesity causes insulin resistance and inflammation, which cause multiple conditions which reduce lifespan
As a result of this conventional story, fat people are exhorted to lose wait. This is, at a population level certainly and probably at an individual level, not effective at reducing obesity. In fact, if anything obesity levels have risen.
The mechanisms that cause diabetes, and other health problems clearly correlated with obesity are not as well understood as they could be. It isn't quite as simple as "obesity causes diabetes and that's it" because, for example, Type I diabetes is also correlated with obesity to some extent, but for those individuals the Type I diabetes came first.
There is a compelling need to find a new story or a nuanced story that we can better build public health interventions around. It may or may not be the case that this is based on a different understanding of the underlying mechanisms involved. My instinct is that at the very least, if you want to prevent diabetes you should focus on improving blood sugar control directly and let the weight do what it does.
posted by plonkee at 1:51 AM on September 1, 2022 [2 favorites]
As a result of this conventional story, fat people are exhorted to lose wait. This is, at a population level certainly and probably at an individual level, not effective at reducing obesity. In fact, if anything obesity levels have risen.
The mechanisms that cause diabetes, and other health problems clearly correlated with obesity are not as well understood as they could be. It isn't quite as simple as "obesity causes diabetes and that's it" because, for example, Type I diabetes is also correlated with obesity to some extent, but for those individuals the Type I diabetes came first.
There is a compelling need to find a new story or a nuanced story that we can better build public health interventions around. It may or may not be the case that this is based on a different understanding of the underlying mechanisms involved. My instinct is that at the very least, if you want to prevent diabetes you should focus on improving blood sugar control directly and let the weight do what it does.
posted by plonkee at 1:51 AM on September 1, 2022 [2 favorites]
Part of what's interesting is that people default to a wide range of BMIs and body compositions in the same environment. It's more complicated than you'll learn from overfeeding animals.
posted by Nancy Lebovitz at 5:34 AM on September 1, 2022 [1 favorite]
posted by Nancy Lebovitz at 5:34 AM on September 1, 2022 [1 favorite]
There is a compelling need to find a new story or a nuanced story that we can better build public health interventions around.
I despair a little over the idea of a new story. It seems clear that there is one, something along the lines of, it's the fat you CAN'T see that causes the problems--i.e., turning the attention more to visceral fat and its effect on our organs and hormones, rather than the bits on display. But we're so entangled with our just-so story: "Look, by an amazing coincidence, the reason we're looking down on your because of your appearance, is the same reason we think you're unhealthy (and lazy!)", and how on earth do you pry people out of that, when even doctors believe it wholeheartedly?
posted by mittens at 8:57 AM on September 1, 2022 [4 favorites]
I despair a little over the idea of a new story. It seems clear that there is one, something along the lines of, it's the fat you CAN'T see that causes the problems--i.e., turning the attention more to visceral fat and its effect on our organs and hormones, rather than the bits on display. But we're so entangled with our just-so story: "Look, by an amazing coincidence, the reason we're looking down on your because of your appearance, is the same reason we think you're unhealthy (and lazy!)", and how on earth do you pry people out of that, when even doctors believe it wholeheartedly?
posted by mittens at 8:57 AM on September 1, 2022 [4 favorites]
« Older I think a part of me always knew it would never... | “This is not a soccer podcast; this is a euphoria... Newer »
This thread has been archived and is closed to new comments
posted by rogerroger at 7:58 AM on August 31, 2022 [2 favorites]