Yep, makes a lot of sense given the ethnic and person-to-person variation of what a healthy BMI is.
Visaria cited a new policy issued by the American Medical Association a couple of days before his presentation, as reported by Medscape Medical News, which advises that BMI "be used in conjunction with other valid measures of risk such as, but not limited to, measurements of visceral fat, body adiposity index, body composition, relative fat mass, waist circumference, and genetic/metabolic factors."
"We're at the start of the end of BMI," declared Visaria during a press briefing at the ENDO meeting.
Yep, makes a lot of sense given the ethnic and person-to-person variation of what a healthy BMI is.
While I agree that BMI is just... not a great way of measuring overall health of the individual, it's a passable way of measuring the health of a large group. It has the benefits of being...
While I agree that BMI is just... not a great way of measuring overall health of the individual, it's a passable way of measuring the health of a large group. It has the benefits of being basically free, and quickly verifiable by a professional with completely verifiable hard data point. We're not going to suddenly start giving individuals routing DEXA scans at their yearly checkup, unfortunately. I doubt we're going to start seeing wide use of calipers in doctor's offices to measure body fat (the average person would 100% hate this), waist circumference would likely be self reported from pant size, which opens it up to lying.
I sadly don't think anything is going to change any time soon. Unless there's a technological breakthrough, or digital scales that use electrical current to measure body fat while you stand on them actual improve enough to pass medical scrutiny. The article even talks about these scales, but it doesn't even mention the idea of including them in doctor's offices. It mentions them being used in homes and that somehow being used as a gauge to diagnose obesity.
To your point, I think that for the la(z)yperson, a simple ratio like the BMI is accessible and serves a purpose of being intuitively understandable. Also it’s simple enough to educate someone...
To your point, I think that for the la(z)yperson, a simple ratio like the BMI is accessible and serves a purpose of being intuitively understandable. Also it’s simple enough to educate someone that BMI is by itself not telling the whole story. But to remove it entirely in favor for more specific measures would raise the “barrier to entry” to engage in the topic, so to speak.
It is funny to me that humans need “a number” and somehow we extrapolate the state of a complex world from it (largely to satisfy emotional needs). Stock markets up? The economy is great! BMI down a point? I’m healthy!
I did have to laugh, because women's pants sizes have essentially zero direct correlation with pants size. They tell you nothing other than the general "larger size = larger measurement". But I...
I did have to laugh, because women's pants sizes have essentially zero direct correlation with pants size. They tell you nothing other than the general "larger size = larger measurement". But I always get a bit thrown with the assumption that pants size is useful in that way.
I'm very much on team "BMI is not helpful" and the way we talk about weight is usually even less helpful so I'm not sure what measures are necessarily better.
fat percentages are generally better, but the thing about health risks due to fatty tissue is that what matters much more is where that fat is accumulating. So some doctors advocate for using...
fat percentages are generally better, but the thing about health risks due to fatty tissue is that what matters much more is where that fat is accumulating. So some doctors advocate for using waist measurements rather than BMI to measure healthy weights (with the caveat that it's still a very imperfect benchmark).
Right and I did understand why waist measurement would be useful in assessing risk but also just that pants size doesn't correlate for a large portion of the population. Is somebody who has that...
Right and I did understand why waist measurement would be useful in assessing risk but also just that pants size doesn't correlate for a large portion of the population.
Is somebody who has that particular risk factor... It's very frustrating as it isn't something you can actually change. It's not my fault. My body has decided to put all the weight there. Not me!
What is BMI even telling people? I’m pretty sure a person that’s obese already knows they’re obese. Assigning a number to it doesn’t reveal anything new or change the course of action/advice....
What is BMI even telling people? I’m pretty sure a person that’s obese already knows they’re obese. Assigning a number to it doesn’t reveal anything new or change the course of action/advice.
Similarly if you look like The Rock but BMI says you’re obese, it’s pretty clear it doesn’t apply to you.
I think if you’re at the point where you’re discussing these things with your doctor, further, better tests are probably reasonable
I think you'd be surprised - a BMI of 30 is 92kg (203 lbs) for an average height man (175cm / 5'9"), and to a lot of people's eye that looks totally normal. Not actively thin or fit looking, for...
What is BMI even telling people? I’m pretty sure a person that’s obese already knows they’re obese.
I think you'd be surprised - a BMI of 30 is 92kg (203 lbs) for an average height man (175cm / 5'9"), and to a lot of people's eye that looks totally normal. Not actively thin or fit looking, for sure, but it appears (and is) absolutely average in many places. I actually think that's one of the biggest values that BMI does bring, despite its flaws: it's so easy to quantify that you can immediately know those "few extra pounds" are a genuine health concern, and it very quickly breaks the assumption most people have of obesity only looking like the TV depiction of the absolute extremes.
I understand all of this but the problem with BMI is that it doesn’t differentiate between muscle, bone, and fat. A relatively short amount of time in the gym skews the results pretty drastically....
I understand all of this but the problem with BMI is that it doesn’t differentiate between muscle, bone, and fat. A relatively short amount of time in the gym skews the results pretty drastically. On the flip side, there are “fat” people with very little muscle mass (especially among the elderly) who track as “healthy” on the BMI charts.
Something as simple as waist circumference is a much better indicator and isn’t difficult to obtain.
I don’t think a short time in the gym for the average western person would skew their bmi. The average western person is already obese, spending three weeks power lifting in the gym is first of...
I don’t think a short time in the gym for the average western person would skew their bmi. The average western person is already obese, spending three weeks power lifting in the gym is first of all the wrong thing to do, and second of all wouldn’t change their bmi drastically. You’d really need months of lifting to go from a bmi of around 20 to over 30.
There’s a definite underprediction issue - as the top level article is pointing out, a “healthy” BMI isn’t enough to go on at all - and yeah, you can argue that the overweight category can skew...
There’s a definite underprediction issue - as the top level article is pointing out, a “healthy” BMI isn’t enough to go on at all - and yeah, you can argue that the overweight category can skew both ways a bit, especially closer to the borderline values. But obesity is often the specific measure of concern, and if BMI says you’re obese it’s almost certainly correct.
When you get down to it, BMI is an approximate normalization of weight over height. It's intuitive that someone who's 5'6'' and weight 160 lb is in a different situation than someone who's 6'3''...
When you get down to it, BMI is an approximate normalization of weight over height. It's intuitive that someone who's 5'6'' and weight 160 lb is in a different situation than someone who's 6'3'' and weights 160 lb. BMI distills that into a easy to calculate number which is approximately correct as, despite the fact that the body isn't uniformly dense and that there's actually a cubic relationship with volume and height, human bodies usually don't vary that much in height, and varies much less in the other two dimensions.
A part that gets ignored is that BMI is a constant for a normal person. If you have a 30, you should probably lose weight and get that lower, and having a new BMI, which would also be indicated by...
A part that gets ignored is that BMI is a constant for a normal person. If you have a 30, you should probably lose weight and get that lower, and having a new BMI, which would also be indicated by lower weight, would lead to one more non-weight metric to track that uses body proportions. This is at least how I have used BMI. Exceptions are generally obvious.
Honestly, the point is pretty much statistical. BMI under 20 and over 25 have been associated with higher all-cause mortality. If you want to live as long as possible, statistically speaking, it...
Honestly, the point is pretty much statistical. BMI under 20 and over 25 have been associated with higher all-cause mortality. If you want to live as long as possible, statistically speaking, it is best to maintain a BMI between 20 and 25. Statistics don't mean you are going to die if you have a BMI of 30. Similar to how drinking alcohol doesn't mean you're going to get cancer, but it does statistically increase your risk of cancer of the mouth, throat, esophagus, larynx (voice box), liver, and breast.
The Rock is paying professionals and doctors to advise him on his health, so I honestly don't think worrying about a statistical outlier like him is really helpful for any of us. I've been lifting weights for 4 years now, I'm still not really a statistical outlier. My charts say obese now, maybe if I got a DEXA scan I'd merely be overweight? It doesn't really change the fact that I would be more healthy if I dropped some fat. Unless you're a woman squatting over her own body weight or a man squatting 2x his body weight, you're probably not a statistical outlier either.
The relative fat mass index may count as a new breakthrough. https://en.wikipedia.org/wiki/Relative_Fat_Mass Personally, I think there are opportunities for using simple machine learning...
The relative fat mass index may count as a new breakthrough.
Personally, I think there are opportunities for using simple machine learning techniques to estimate body composition from easily acquirable metrics like height, weight, waist, age, etc. that far exceed any simple linear or polynomial equation like BMI or RFM. I’m not sure why this isn’t being explored more.
Those aren't mutually exclusive. The first thing you learn in any machine learning class is the closed form solution for linear regression. I'm pretty sure the BMI scale is derived from machine...
simple machine learning techniques to estimate body composition from easily acquirable metrics like height, weight, waist, age, etc. that far exceed any simple linear or polynomial equation like BMI or RFM. I’m not sure why this isn’t being explored more.
Those aren't mutually exclusive. The first thing you learn in any machine learning class is the closed form solution for linear regression. I'm pretty sure the BMI scale is derived from machine learning, in that technicality. In fact, most of classical machine learning is about linear models - ridge regression, linear regression, LASSO, SVMs, and so forth.
I'm doubtful of the ability for more complicated models to be more accurate. For one, you can't claw information from nowhere just because the model is more complicated. In low dimensional spaces, you are at much higher risk of overfitting.
RFM can be more accurate, but the issue is that people don't currently measure their waist as one of their primary body measurements, and that measuring waists is significantly more variable due to people not following instructions (and not understanding where their waist is), unlike height and weight which are fairly foolproof.
Yea I'm not sure a complicated model is required. From the article, it does sound like there is some room for increasing the dimensionality of BMI but I could be misunderstanding:
Yea I'm not sure a complicated model is required. From the article, it does sound like there is some room for increasing the dimensionality of BMI but I could be misunderstanding:
Further analysis showed that when Visaria added waist circumference to BMI to enlarge the diagnostic net for obesity it cut the percentage of adults missed as having obesity by BMI alone nearly in half.
That is an interesting way of measuring things. I plugged my numbers into a calculator I found here and it calculated my RFM ~3 points lower than BMI calculations do. I would kind of expect this...
That is an interesting way of measuring things. I plugged my numbers into a calculator I found here and it calculated my RFM ~3 points lower than BMI calculations do. I would kind of expect this in my case since I weight lift recreationally. That said, I'm still in the 'obese' category myself, so it doesn't really change anything for me, it just does away with me seeing a number like BMI 30 and saying "BMI is flawed, I don't need to loose weight". If this gets people to actually listen when their doctor says "obese" and not try to justify that the numbers "are flawed" than it is a huge win, so there is that benefit.
All of this said, coupling something like BMI with a baseline strength measurement would probably be just as effective for the health of our population, if not even better. Identifying something like not being able to lift 30 lbs overhead as a health issue could be even better.
I think this is at the heart if the issue-- I think many people are in denial about their obesity because "I'm muscular" and "BMI isn't very accurate." But the reality is BMI underestimates...
If this gets people to actually listen when their doctor says "obese" and not try to justify that the numbers "are flawed" than it is a huge win, so there is that benefit.
I think this is at the heart if the issue-- I think many people are in denial about their obesity because "I'm muscular" and "BMI isn't very accurate." But the reality is BMI underestimates obesity much more often than it overestimates it, and people should be aware of this.
Of course, this won't solve the obesity problem considering how hard losing weight is when you know you're obese, but regardless I'm a big believer that people should be well informed about their health and that if you have risk factors it's better to know than to not know.
There are even comments here in this thread regarding bmi being flawed. It can be, especially if you’re close to the healthy range, but over it because you actually are a professional athlete...
There are even comments here in this thread regarding bmi being flawed. It can be, especially if you’re close to the healthy range, but over it because you actually are a professional athlete level of fitness.
The higher (or lower) your bmi is, the more likely it is that your weight is unhealthy. BMI is really only a flawed measure if you’re an outlier, which most people aren’t.
I've been using relative fat mass to help people who are trying to lose weight since it's first came out. The exciting thing it does is provide a good proxy for an individual's body fat percentage...
I've been using relative fat mass to help people who are trying to lose weight since it's first came out.
The exciting thing it does is provide a good proxy for an individual's body fat percentage using just two inputs. That's quite an improvement for trying to tell an individual's health screen based on their height and weight. RFM does a better job at proxying bodyfat percentage than those body impedance analysis scales that are all the rage these days.
Because relative fat mass uses the waist circumference and the subjects height, it is more 3 dimensional than BMI. The inability to appreciate body fat percentage within BMI has always been it's weakness.
Strange but I can't read the article without logging in, other commenters don't seem to have this issue. Here's the archived link: https://archive.vn/hXaH4 If I understand it correctly, what in...
Strange but I can't read the article without logging in, other commenters don't seem to have this issue. Here's the archived link:
If I understand it correctly, what in this article is mentioned is what is called "skinny-fat", which isn't a new notion, but good to see things backed up by some numbers. Shows the usefulness of at least a minimum of strength training, for all ages and genders.
Thanks for the link! I think my question here is there value to requantifying this? I am not sure I am seeing that in the article. Is skinny fat a major predictor of health consequences for heart...
Thanks for the link!
I think my question here is there value to requantifying this? I am not sure I am seeing that in the article. Is skinny fat a major predictor of health consequences for heart disease and other conditions? How significantly different is it as a predictor? Or is it just a different way of measuring fat content and redefining fat?
Womens body typically do contain more fat vs muscle for variety of reasons compared to male counterparts but is there a reason I need to decrease my fat stores? Because to me they've been more helpful than hurtful as I am nourishing a child.
Curious to understand if there's value in casting a wider net on defining obesity from a treatment perspective. Did I miss something in the article?
I mean, the other side of it is if you have highly below average body fat percentage, and if you're absolutely shredded at 10% bodyfat, then I don't think you need any health measurements to check...
I mean, the other side of it is if you have highly below average body fat percentage, and if you're absolutely shredded at 10% bodyfat, then I don't think you need any health measurements to check whether or not you're healthy weight wise, and it should be obvious to you and everyone with eyes.
You don’t have to be a bodybuilder to have fairly useless BMI stats. I have pretty average body fat, but I’m fairly overweight according to BMI and just a normal person by dexa scan and RFM.
You don’t have to be a bodybuilder to have fairly useless BMI stats. I have pretty average body fat, but I’m fairly overweight according to BMI and just a normal person by dexa scan and RFM.
Perhaps technically true on occasion, but that's exactly why studies like this are so valuable - they let us quantify how applicable that is across the board. Going by the most in depth one I'm...
Perhaps technically true on occasion, but that's exactly why studies like this are so valuable - they let us quantify how applicable that is across the board. Going by the most in depth one I'm aware of, the "overweight" band could go either way and is, at best, a nudge to take a look at your habits, discuss with a doctor, or go get a proper DEXA scan; if, on the other hand, BMI says you're obese (≥ 30 kg/m2) it's almost definitely correct (95% accurate in men and 99% in women).
Both studies agree that BMI is likely to underpredict obesity by a factor of two. So sure, a person might be less fat than their BMI number suggests, but the statistics say it's a lot more likely to go the other way.
Haven’t we always known that BMI alone doesn’t say much at the individual level? I’m thinking of how a super fit person will have a high BMI even as they have a low body fat percentage, but the...
His findings highlight that "BMI should be supplemented with other measures of obesity" for the management of individual patients
Haven’t we always known that BMI alone doesn’t say much at the individual level? I’m thinking of how a super fit person will have a high BMI even as they have a low body fat percentage, but the reverse would be true too. Waist circumference, height, gender, age, if extra weight is carried in the mid-section all affect a person’s BMI, muscle mass all affect BMI, and there are probably other things that do too.
Do you think if there are exceptions to a general rule that the general rule is bunk? BMI is fine for looking at populations of people, but of course, because humans are diverse, there will be...
Do you think if there are exceptions to a general rule that the general rule is bunk? BMI is fine for looking at populations of people, but of course, because humans are diverse, there will be exceptions. Doesn't mean BMI is an inherently bad measurement.
Every discussion of BMI on the internet is an interesting look at the general populace and their relationship with statistics. People tend to take both weight and statistics very personally,...
Every discussion of BMI on the internet is an interesting look at the general populace and their relationship with statistics. People tend to take both weight and statistics very personally, comparing their personal situation with the reported average, which makes me think that the general understanding of the concept of distributions is not well internalized by many people. Looking at the typical histogram of a normal distribution, the bars are clustered around the average, but very few people are actually in that center bar, so the vast majority can truthfully say that their personal situation is different from the average. Same thing happens with inflation when the CPI data comes out and people start talking about how an average for the entire nation doesn't match what they see in the grocery store.
As you say, exceptions don't invalidate the data, and correlations can help us make better policy decisions than if we didn't have this data. So it's not true that "BMI is useless," summarizing some comments here and in previous threads. Adding a waist circumference measurement is stated to be imperfect form the get-go in the OP article, but adding it apparently would make our data more accurate. I'm sure people would still complain about the inaccuracy of "body mass circumference index" (BMCI?) as well. I always wonder what we can do from a science communication standpoint to improve this situation, since it's always the same comments over and over in these kinds of threads. This tells me that people are not understanding the usefulness of these measures.
Hah, sure. Measurements like BMI are great for aggregating large groups and saying something about trends. Probably it was developed with that in mind? So I wouldn’t argue it’s a bad or worthless...
Hah, sure. Measurements like BMI are great for aggregating large groups and saying something about trends. Probably it was developed with that in mind? So I wouldn’t argue it’s a bad or worthless measurement. But I would argue that this BMI score = obese without considering all of the other factors of that person’s physical situation is unhelpful to that individual.
Except for the fact that over 90% of those with a bmi over thirty are actually obese. It’s a good, simple, measure to quickly check one’s status. The few people who aren’t obese with a high bmi...
Except for the fact that over 90% of those with a bmi over thirty are actually obese. It’s a good, simple, measure to quickly check one’s status. The few people who aren’t obese with a high bmi aren’t that hard to suss out through other measures like the mentioned dexa scans.
But I mean at a certain point, other factors are irrelevant, right? It doesn't matter how many other things you consider, if you're 500 lbs, you're obese and at serious risk of health hazards....
But I would argue that this BMI score = obese without considering all of the other factors of that person’s physical situation is unhelpful to that individual.
But I mean at a certain point, other factors are irrelevant, right? It doesn't matter how many other things you consider, if you're 500 lbs, you're obese and at serious risk of health hazards.
Splitting hairs over 28 vs. 29.5 for example is where I would with you.
The fact is also that epidemiological studies looking in to the health of a population consistently see worse health outcomes in BMI bins over 30 - regardless of cause for the high BMI. Those who...
The fact is also that epidemiological studies looking in to the health of a population consistently see worse health outcomes in BMI bins over 30 - regardless of cause for the high BMI.
Those who have a BMI over thirty, yet are as healthy as an average person in a BMI bin below 30 are a true outlier, meaning there are few such individuals around. This in turn simply means that BMI is a good yardstick for healthiness.
I really don't know what to say, other than everyone truly needs to eat better, eat less, and be more active. We've known this for years and we can't get people to do this. This article relays the...
I really don't know what to say, other than everyone truly needs to eat better, eat less, and be more active. We've known this for years and we can't get people to do this.
This article relays the two ways we measure obesity, DEXA scans and bmi as one being accurate but rarely obtained as one as obtainable but not accurate, and while this may be true seems like the idea is that they should use this electrical pulse attached to bath scales to measure BMI. We'll see if that technology appears, regardless people need to just eat less,and be more active.
At this point there needs to be political action due to the severity of the problem at least in the US since ita become such a security issue. Nobody will do anything because weight is a 3rd rail.
Yep, makes a lot of sense given the ethnic and person-to-person variation of what a healthy BMI is.
While I agree that BMI is just... not a great way of measuring overall health of the individual, it's a passable way of measuring the health of a large group. It has the benefits of being basically free, and quickly verifiable by a professional with completely verifiable hard data point. We're not going to suddenly start giving individuals routing DEXA scans at their yearly checkup, unfortunately. I doubt we're going to start seeing wide use of calipers in doctor's offices to measure body fat (the average person would 100% hate this), waist circumference would likely be self reported from pant size, which opens it up to lying.
I sadly don't think anything is going to change any time soon. Unless there's a technological breakthrough, or digital scales that use electrical current to measure body fat while you stand on them actual improve enough to pass medical scrutiny. The article even talks about these scales, but it doesn't even mention the idea of including them in doctor's offices. It mentions them being used in homes and that somehow being used as a gauge to diagnose obesity.
To your point, I think that for the la(z)yperson, a simple ratio like the BMI is accessible and serves a purpose of being intuitively understandable. Also it’s simple enough to educate someone that BMI is by itself not telling the whole story. But to remove it entirely in favor for more specific measures would raise the “barrier to entry” to engage in the topic, so to speak.
It is funny to me that humans need “a number” and somehow we extrapolate the state of a complex world from it (largely to satisfy emotional needs). Stock markets up? The economy is great! BMI down a point? I’m healthy!
I did have to laugh, because women's pants sizes have essentially zero direct correlation with pants size. They tell you nothing other than the general "larger size = larger measurement". But I always get a bit thrown with the assumption that pants size is useful in that way.
I'm very much on team "BMI is not helpful" and the way we talk about weight is usually even less helpful so I'm not sure what measures are necessarily better.
fat percentages are generally better, but the thing about health risks due to fatty tissue is that what matters much more is where that fat is accumulating. So some doctors advocate for using waist measurements rather than BMI to measure healthy weights (with the caveat that it's still a very imperfect benchmark).
Right and I did understand why waist measurement would be useful in assessing risk but also just that pants size doesn't correlate for a large portion of the population.
Is somebody who has that particular risk factor... It's very frustrating as it isn't something you can actually change. It's not my fault. My body has decided to put all the weight there. Not me!
What is BMI even telling people? I’m pretty sure a person that’s obese already knows they’re obese. Assigning a number to it doesn’t reveal anything new or change the course of action/advice.
Similarly if you look like The Rock but BMI says you’re obese, it’s pretty clear it doesn’t apply to you.
I think if you’re at the point where you’re discussing these things with your doctor, further, better tests are probably reasonable
I think you'd be surprised - a BMI of 30 is 92kg (203 lbs) for an average height man (175cm / 5'9"), and to a lot of people's eye that looks totally normal. Not actively thin or fit looking, for sure, but it appears (and is) absolutely average in many places. I actually think that's one of the biggest values that BMI does bring, despite its flaws: it's so easy to quantify that you can immediately know those "few extra pounds" are a genuine health concern, and it very quickly breaks the assumption most people have of obesity only looking like the TV depiction of the absolute extremes.
I understand all of this but the problem with BMI is that it doesn’t differentiate between muscle, bone, and fat. A relatively short amount of time in the gym skews the results pretty drastically. On the flip side, there are “fat” people with very little muscle mass (especially among the elderly) who track as “healthy” on the BMI charts.
Something as simple as waist circumference is a much better indicator and isn’t difficult to obtain.
There have been studies on this and depending on what health indicators you’re looking at, BMI can be very misleading. FiveThirtyEigjht did a good write up on it a few years ago: https://fivethirtyeight.com/features/bmi-is-a-terrible-measure-of-health/amp/
I don’t think a short time in the gym for the average western person would skew their bmi. The average western person is already obese, spending three weeks power lifting in the gym is first of all the wrong thing to do, and second of all wouldn’t change their bmi drastically. You’d really need months of lifting to go from a bmi of around 20 to over 30.
There’s a definite underprediction issue - as the top level article is pointing out, a “healthy” BMI isn’t enough to go on at all - and yeah, you can argue that the overweight category can skew both ways a bit, especially closer to the borderline values. But obesity is often the specific measure of concern, and if BMI says you’re obese it’s almost certainly correct.
When you get down to it, BMI is an approximate normalization of weight over height. It's intuitive that someone who's 5'6'' and weight 160 lb is in a different situation than someone who's 6'3'' and weights 160 lb. BMI distills that into a easy to calculate number which is approximately correct as, despite the fact that the body isn't uniformly dense and that there's actually a cubic relationship with volume and height, human bodies usually don't vary that much in height, and varies much less in the other two dimensions.
A part that gets ignored is that BMI is a constant for a normal person. If you have a 30, you should probably lose weight and get that lower, and having a new BMI, which would also be indicated by lower weight, would lead to one more non-weight metric to track that uses body proportions. This is at least how I have used BMI. Exceptions are generally obvious.
Honestly, the point is pretty much statistical. BMI under 20 and over 25 have been associated with higher all-cause mortality. If you want to live as long as possible, statistically speaking, it is best to maintain a BMI between 20 and 25. Statistics don't mean you are going to die if you have a BMI of 30. Similar to how drinking alcohol doesn't mean you're going to get cancer, but it does statistically increase your risk of cancer of the mouth, throat, esophagus, larynx (voice box), liver, and breast.
The Rock is paying professionals and doctors to advise him on his health, so I honestly don't think worrying about a statistical outlier like him is really helpful for any of us. I've been lifting weights for 4 years now, I'm still not really a statistical outlier. My charts say obese now, maybe if I got a DEXA scan I'd merely be overweight? It doesn't really change the fact that I would be more healthy if I dropped some fat. Unless you're a woman squatting over her own body weight or a man squatting 2x his body weight, you're probably not a statistical outlier either.
The relative fat mass index may count as a new breakthrough.
https://en.wikipedia.org/wiki/Relative_Fat_Mass
Personally, I think there are opportunities for using simple machine learning techniques to estimate body composition from easily acquirable metrics like height, weight, waist, age, etc. that far exceed any simple linear or polynomial equation like BMI or RFM. I’m not sure why this isn’t being explored more.
Those aren't mutually exclusive. The first thing you learn in any machine learning class is the closed form solution for linear regression. I'm pretty sure the BMI scale is derived from machine learning, in that technicality. In fact, most of classical machine learning is about linear models - ridge regression, linear regression, LASSO, SVMs, and so forth.
I'm doubtful of the ability for more complicated models to be more accurate. For one, you can't claw information from nowhere just because the model is more complicated. In low dimensional spaces, you are at much higher risk of overfitting.
RFM can be more accurate, but the issue is that people don't currently measure their waist as one of their primary body measurements, and that measuring waists is significantly more variable due to people not following instructions (and not understanding where their waist is), unlike height and weight which are fairly foolproof.
Yea I'm not sure a complicated model is required. From the article, it does sound like there is some room for increasing the dimensionality of BMI but I could be misunderstanding:
That is an interesting way of measuring things. I plugged my numbers into a calculator I found here and it calculated my RFM ~3 points lower than BMI calculations do. I would kind of expect this in my case since I weight lift recreationally. That said, I'm still in the 'obese' category myself, so it doesn't really change anything for me, it just does away with me seeing a number like BMI 30 and saying "BMI is flawed, I don't need to loose weight". If this gets people to actually listen when their doctor says "obese" and not try to justify that the numbers "are flawed" than it is a huge win, so there is that benefit.
All of this said, coupling something like BMI with a baseline strength measurement would probably be just as effective for the health of our population, if not even better. Identifying something like not being able to lift 30 lbs overhead as a health issue could be even better.
I think this is at the heart if the issue-- I think many people are in denial about their obesity because "I'm muscular" and "BMI isn't very accurate." But the reality is BMI underestimates obesity much more often than it overestimates it, and people should be aware of this.
Of course, this won't solve the obesity problem considering how hard losing weight is when you know you're obese, but regardless I'm a big believer that people should be well informed about their health and that if you have risk factors it's better to know than to not know.
There are even comments here in this thread regarding bmi being flawed. It can be, especially if you’re close to the healthy range, but over it because you actually are a professional athlete level of fitness.
The higher (or lower) your bmi is, the more likely it is that your weight is unhealthy. BMI is really only a flawed measure if you’re an outlier, which most people aren’t.
I've been using relative fat mass to help people who are trying to lose weight since it's first came out.
The exciting thing it does is provide a good proxy for an individual's body fat percentage using just two inputs. That's quite an improvement for trying to tell an individual's health screen based on their height and weight. RFM does a better job at proxying bodyfat percentage than those body impedance analysis scales that are all the rage these days.
Because relative fat mass uses the waist circumference and the subjects height, it is more 3 dimensional than BMI. The inability to appreciate body fat percentage within BMI has always been it's weakness.
Strange but I can't read the article without logging in, other commenters don't seem to have this issue. Here's the archived link:
https://archive.vn/hXaH4
If I understand it correctly, what in this article is mentioned is what is called "skinny-fat", which isn't a new notion, but good to see things backed up by some numbers. Shows the usefulness of at least a minimum of strength training, for all ages and genders.
Thanks for the link!
I think my question here is there value to requantifying this? I am not sure I am seeing that in the article. Is skinny fat a major predictor of health consequences for heart disease and other conditions? How significantly different is it as a predictor? Or is it just a different way of measuring fat content and redefining fat?
Womens body typically do contain more fat vs muscle for variety of reasons compared to male counterparts but is there a reason I need to decrease my fat stores? Because to me they've been more helpful than hurtful as I am nourishing a child.
Curious to understand if there's value in casting a wider net on defining obesity from a treatment perspective. Did I miss something in the article?
Great. So I am actually fatter than I appear on paper? Today is a wonderful day, haha.
Or maybe you’re much less fat, it’s so hard to say with BMI.
I mean, the other side of it is if you have highly below average body fat percentage, and if you're absolutely shredded at 10% bodyfat, then I don't think you need any health measurements to check whether or not you're healthy weight wise, and it should be obvious to you and everyone with eyes.
You don’t have to be a bodybuilder to have fairly useless BMI stats. I have pretty average body fat, but I’m fairly overweight according to BMI and just a normal person by dexa scan and RFM.
Perhaps technically true on occasion, but that's exactly why studies like this are so valuable - they let us quantify how applicable that is across the board. Going by the most in depth one I'm aware of, the "overweight" band could go either way and is, at best, a nudge to take a look at your habits, discuss with a doctor, or go get a proper DEXA scan; if, on the other hand, BMI says you're obese (≥ 30 kg/m2) it's almost definitely correct (95% accurate in men and 99% in women).
Both studies agree that BMI is likely to underpredict obesity by a factor of two. So sure, a person might be less fat than their BMI number suggests, but the statistics say it's a lot more likely to go the other way.
Haven’t we always known that BMI alone doesn’t say much at the individual level? I’m thinking of how a super fit person will have a high BMI even as they have a low body fat percentage, but the reverse would be true too. Waist circumference, height, gender, age, if extra weight is carried in the mid-section all affect a person’s BMI, muscle mass all affect BMI, and there are probably other things that do too.
Do you think if there are exceptions to a general rule that the general rule is bunk? BMI is fine for looking at populations of people, but of course, because humans are diverse, there will be exceptions. Doesn't mean BMI is an inherently bad measurement.
Every discussion of BMI on the internet is an interesting look at the general populace and their relationship with statistics. People tend to take both weight and statistics very personally, comparing their personal situation with the reported average, which makes me think that the general understanding of the concept of distributions is not well internalized by many people. Looking at the typical histogram of a normal distribution, the bars are clustered around the average, but very few people are actually in that center bar, so the vast majority can truthfully say that their personal situation is different from the average. Same thing happens with inflation when the CPI data comes out and people start talking about how an average for the entire nation doesn't match what they see in the grocery store.
As you say, exceptions don't invalidate the data, and correlations can help us make better policy decisions than if we didn't have this data. So it's not true that "BMI is useless," summarizing some comments here and in previous threads. Adding a waist circumference measurement is stated to be imperfect form the get-go in the OP article, but adding it apparently would make our data more accurate. I'm sure people would still complain about the inaccuracy of "body mass circumference index" (BMCI?) as well. I always wonder what we can do from a science communication standpoint to improve this situation, since it's always the same comments over and over in these kinds of threads. This tells me that people are not understanding the usefulness of these measures.
Hah, sure. Measurements like BMI are great for aggregating large groups and saying something about trends. Probably it was developed with that in mind? So I wouldn’t argue it’s a bad or worthless measurement. But I would argue that this BMI score = obese without considering all of the other factors of that person’s physical situation is unhelpful to that individual.
Except for the fact that over 90% of those with a bmi over thirty are actually obese. It’s a good, simple, measure to quickly check one’s status. The few people who aren’t obese with a high bmi aren’t that hard to suss out through other measures like the mentioned dexa scans.
But I mean at a certain point, other factors are irrelevant, right? It doesn't matter how many other things you consider, if you're 500 lbs, you're obese and at serious risk of health hazards.
Splitting hairs over 28 vs. 29.5 for example is where I would with you.
The fact is also that epidemiological studies looking in to the health of a population consistently see worse health outcomes in BMI bins over 30 - regardless of cause for the high BMI.
Those who have a BMI over thirty, yet are as healthy as an average person in a BMI bin below 30 are a true outlier, meaning there are few such individuals around. This in turn simply means that BMI is a good yardstick for healthiness.
Yes, I’m agreeing with you. BMI is a useful metric, especially in the aggregate. It’s just that it can fail the individual on occasion.
I really don't know what to say, other than everyone truly needs to eat better, eat less, and be more active. We've known this for years and we can't get people to do this.
This article relays the two ways we measure obesity, DEXA scans and bmi as one being accurate but rarely obtained as one as obtainable but not accurate, and while this may be true seems like the idea is that they should use this electrical pulse attached to bath scales to measure BMI. We'll see if that technology appears, regardless people need to just eat less,and be more active.
At this point there needs to be political action due to the severity of the problem at least in the US since ita become such a security issue. Nobody will do anything because weight is a 3rd rail.