Something is weird about Body Mass Index

[tedly]

Something about the Body Mass Index puzzles me. To start where I went in, the units are wrong. BMI is your mass in kg divided by your height in metres squared. So the units are units of pressure, where it seems they should be units of density (mass/length cubed).

I googled for BMI, and the first references I found go back to the mid '90s from the NIH. I haven't, so far, got a reference to the defining paper. I get the hunch that health records were mined, and the first correlation with a high explanatory power was selected. (If you torture data enough it will confess to anything.)

But let's take a people example. I am told that Raquel Welch is quite short. (Oops - google gives her 167 cm which is 3-4 over american average.) OK, not to let facts interfere with a good story, let's take a woman who is 60 inches tall, and relatively fit, give her a BMI of 24. That means she weighs 123 lbs. If we scale her up to 72 inches, she is 1.2 times as tall, 1.2 times as wide, and 1.2 times as thick. She should weigh 1.728 times as much. She would look exactly the same, in the absence of any article to give a length scale. But a 6' woman, weighing 212 lbs has a BMI of 29, which is close to obese.

If we go with a constant BMI The 6' woman should weigh 177 lbs. She would be 20% taller, but only 10% wider and 10% thicker. I think she would look considerably leaner than the shorter woman of the same BMI.

So why are taller people more prone to health effects of overweight? ( If you take the BMI as valid over a reasonable range of human height.) Or, why isn't BMI based on height cubed?

 

[El Greco]

Body mass index is hardly useful because it doesn't take into account the bodyfat level. It has been used mainly by insurance companies.

 

[Walter Wayne]

So why are taller people more prone to health effects of overweight? ( If you take the BMI as valid over a reasonable range of human height.) Or, why isn't BMI based on height cubed?

It's based on height squared because short people aren't usually just a scaled version of tall people. Take a look at pictures of short people and tall people. Even without other objects around, one can often tell they are short or tall by their proportions, and they still look like healthy people.

Mass is not proportional to height cubed. If I double in height, I don't need to have my internal organ mass increased by a factor of eight (my heart, liver etc do not to need to be scaled). As such my chest cavity and general struture doesn't have to be twice as wide and twice as deep.

Walt

 

[BTox]

BMI is a crock. I am overweight according to that calculation, yet I have a greek-god like body with very low body fat. If I add another 20 lbs of muscle I'll be considered obese!

 

[TeaBag420]

Re: Re: Something is weird about Body Mass Index

It's based on height squared because short people aren't usually just a scaled version of tall people. Take a look at pictures of short people and tall people. Even without other objects around, one can often tell they are short or tall by their proportions, and they still look like healthy people.

Have you ever actually tried that experiment? Are you sure you're not talking about "little people"?

 

[tedly]

Re: Re: Re: Something is weird about Body Mass Index

Have you ever actually tried that experiment? Are you sure you're not talking about "little people"?

I know nothing about GIMP or PhotoShop, but ...

could you take a photo and expand it vertically by 1.2 and horizontally by 1.095, and compare it to a photo expanded by 1.2 in both directions? This would be (by my lights) what would happen for two similar people, holding BMI constant, of differing height.

 

[WildCat]

Re: Re: Re: Re: Something is weird about Body Mass Index

I know nothing about GIMP or PhotoShop, but ...

could you take a photo and expand it vertically by 1.2 and horizontally by 1.095, and compare it to a photo expanded by 1.2 in both directions? This would be (by my lights) what would happen for two similar people, holding BMI constant, of differing height.

OK, here ya go.
Original photo:

Increased vertically by 1.2, horizontally by 1.095:

Increased 1.2 in both directions:

So there ya go!

 

[tedly]

Wildcat:

Thank you, I hope that was as easy as you make it look, but if I was a Pain ITA I apologise.

I can't tell from the expanded photo if it is a tall man at the same distance or a short man close up. The tall thin one looks skinnier.

The cube relationship holds across different animals, so it should hold in something with as small a range as human height. If the internal organs are smaller in proportion that cavity is still full of something, all of which will have a specific gravity close to 1.

I'm still curious.

 

[Martin]

So why are taller people more prone to health effects of overweight?

Caveat: I know sod-all about physiology. But the standard explanation would be that volume and therefore mass vary with the cube of scale, while the cross-sectional area of the legs that have to carry that mass vary with the square. So the mass per unit are increases linearly with scale, which will cause problems beyond a certain point.

 

[WildCat]

Only takes a few seconds in Photoshop, tedly. How about an embossed Bush?

 

[Stimpson J. Cat]

I was denied health insurance because of the BMI nonsense.

Those who saw me at the TAM2 know that I am significantly overweight, but by no means to a debilitating extent. My blood pressure is fine, and I have absolutely no problem doing intensive martial arts training twice a week with people who are in very good shape.

I weigh 265 lbs, which is about 120 kg. I am 6 ft tall, which is 180 cm. That gives me a BMI of about 37, which according to there silly charts, is pretty bad. What got me, though, is that when I was 18, I was the same height, and 240 lbs, which gives a BMI of about 34. According to the insurance company, that would have been fine. The thing is, at that time, I was in terrible shape. I was much fatter than now, and had very little muscle tone. Go figure.

If you look at medical sites on the BMI, you will find that they acknowledge difficulties like those mentioned here. People who are extremely muscular, like athletes and body builders, will show artificially high BMI. Likewise, unusual builds can have a significant effect. For example, the last time I bought a suit, the pants that came with the jacket that fit me were size 50! Now I only where size 42. My shoulders are just far broader than would be normal for someone 6ft tall. I just have abnormally short legs. My leg length is about typical for somebody around 5'8" or so. In fact, my dad's legs are a good two inches longer than mine, and he's only 5'10. My torso, however, as well as my arms, are longer than those of many people I know who are 6'4 or taller.

Obviously no single number is going to be able to give a good indication of a persons health risks, without taking such things into account. The good news is that doctors understand this. I asked my doctor whether he though I should be concerned about my high BMI, and he just told me that it would be meaningless to try to apply it to me. He was more concerned with the risk to my knees than anything else. The bad news is that insurance companies use this to make decisions that not only affect how much you have to pay, but even whether they will cover you at all.


Dr. Stupid

 

[T'ai Chi]

I googled for BMI, and the first references I found go back to the mid '90s from the NIH. I haven't, so far, got a reference to the defining paper. I get the hunch that health records were mined, and the first correlation with a high explanatory power was selected. (If you torture data enough it will confess to anything.)

I don't think you'll find the definining paper, because the BMI (ie. Quetelet's Index) was created in the 1800's by the Belgian statistician and astronomer Adolphe Quetelet (at least in its first form).

He did a lot of great things with social data, but, in my opinion, while the idea was good, the actual index was not thought out too well. On the other hand, of the many anthropometric measures of body mass, the BMI has the highest correlation with skinfold thickness or body density.

 

[El Greco]

If you are very interested in BMI (and frankly, I couldn't understand why), look at this:

Bibliography of BMI references

 

[The Don]

Yes BMI is a very blunt tool and yes, there are a number of elite athletes who would be classified as overweight or obese by "normal" standards but the great majority who have high BMIs do so because they are carrying rather a lot of lard around with them.

An accurate bodyfat measurement is a far better diagnostic tool BUT not may people can calculate it reliably (and a measuring it by resistance gives aritficially low readings for de-hydrated people). I laugh when I hear colleagues who do no exercise claim to have added "15 lbs of muscle" in the last few weeks on their latest fad diet.

BMI can however provide an initial flag of a potential problem. It wasn't until I calculated my BMI for the first time that I realised that I was significantly overweight (I was 220 lbs and I'm only 6' 3"). Knowing what kind of person I am, I would probably have tried to convince myself that I was over-muscled rather than a little too lardy.

I recently read that a pretty good shorthand way of determining whether you're overweight (for men in the 5' - 6'6" range) is waistline.

< 36 " - OK
36" - 38" - Chubby
> 38" - Lardy

 

[Matabiri]

Something about the Body Mass Index puzzles me. To start where I went in, the units are wrong. BMI is your mass in kg divided by your height in metres squared. So the units are units of pressure, where it seems they should be units of density (mass/length cubed).

They're not units of pressure (force/unit area, not mass/unit area). They're effectively units of pressure/acceleration, which (as far as I can tell) has no particular physical significance.

I agree with you, just not about the units...

 

[tedly]

Thanks

Thanks T'ai Chi and El Greco, I didn't find those references, and I doubt that I'd ever have found Quetelet's Index. But I'm still curious as to why BMI would correlate better to body density than would an index whose units are right.

And I really missed my guess on the development of the index. I've worked with my wife (an epidemiologist) on health statistics, and I just guessed that it was somebody with a computer and a stats package who was just doing multivariate analysis and grabbing terms as the relationship got bettter. "M/h^2 works better than M/h - let's publish!"

And, to the Don, careful with that 'only 6'3" and 220lbs'. I'm only infinitesimally taller and slightly lighter, but the 90th percentile for American males is 6' or so. That's why you always have to adjust the rear view mirror to the limit of its travel every time you get the car back.

And your BMI of 27.5 isn't far off the American average of 26.5.


Edited to add to Matabiri.
OK, but not to hide the fly-poop in the pepper, in a gravitational field we measure mass by the force it exerts against a spring. So you pick up your pressure gauge and find it's calibrated in Kg/cm^2 And since I was being picky it's only right and proper that you hold me exactly to the rules.

 

[Dr. Imago]

BMI is an "eye-ball" type of thing. It's not a hard and fast determinant. We use it prognostically to predict long-term morbidity/mortality outcomes in people with obesity. It won't, for example, account for weightlifters, etc. who have a high lean-to-fat ratio. It's more appropriate for sedentary types.

Okay... so that wasn't the issue. Sorry, I'm swamped. Be back again in a few months... just miss posting here.

-TT