improved BMI?



D

Dan Connelly

Guest
From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png

I estimate mass is roughly proportional to height^2.2 (there's a lot
of variability in this estimate, it's true, but it's clearly > 2,
the BMI assumption).

This suggests a more height-invariant BMI might be estimated, using:

BMI' = mass / height^2.2

However, I suspect people have enough trouble grasping units of "kg/m^2",
trying to get them to swallow "kg/m^2.2" may perhaps be a bit much. But
there seems a strong case to be made that the BMI index is "unfair" to tall
people (assuming lower is better, usually the case among those with enough
income to read this message).

Dan
 
Dan Connelly wrote:
> From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png
>
> I estimate mass is roughly proportional to height^2.2 (there's a lot
> of variability in this estimate, it's true, but it's clearly > 2,
> the BMI assumption).
>
> This suggests a more height-invariant BMI might be estimated, using:
>
> BMI' = mass / height^2.2
>
> However, I suspect people have enough trouble grasping units of "kg/m^2",
> trying to get them to swallow "kg/m^2.2" may perhaps be a bit much. But
> there seems a strong case to be made that the BMI index is "unfair" to tall
> people (assuming lower is better, usually the case among those with enough
> income to read this message).
>
> Dan


I'm evidently heavy boned. I don't float very well.

I have 36" waist, am 5' 6" and weigh about 188 lbs. This puts me on the
obese borderline.

Even on the old health height weight charts with "heavy frame" I was
always at the very top end of the heavy frame.

With BMI scaling as Hheight^2 (^ is an easy way to represent the
exponent), it is much more realistic than the height^3 or cubic scaling
law. The height^(2.2) is an improvement which has the disadvantage of
being incomprehensible to the average medical or health worker, and
incomprehensible to the average "personal trainer" who has no training
credentials at all.

So, there is a lot wrong with the existing BMI, and it is not much
better than the old height-weight charts, except that it claims to give
a number for your "fatness".

It isn't a great number, but it is one of those small steps forward, or
small stumbles forward, that the health community takes from time to time.

Health is too important to be left to health workers.
--
................................


Keepsake gift for young girls.
Unique and personal one-of-a-kind.
Builds strong minds 12 ways.
Guaranteed satisfaction
- courteous money back
- keep bonus gifts

http://www.alicebook.com
 
jbuch wrote:

> With BMI scaling as Hheight^2 (^ is an easy way to represent the
> exponent), it is much more realistic than the height^3 or cubic scaling
> law. The height^(2.2) is an improvement which has the disadvantage of
> being incomprehensible to the average medical or health worker, and
> incomprehensible to the average "personal trainer" who has no training
> credentials at all.
>


It may not be an improvement. On another forum, it was pointed out to me a recent paper
was published on the subject, examing various populations in different regions, that the
"true" number is likely below 2, rather than above. However, I suspect shorter people have
a lower susceptibility to thinness than taller people, as they need less food, but aren't
necessarily served less (rich) or produce less (poor). Thus I suspect using elite
athletes is a more reliable guide. Nevertheless, in the end, BMI is just a crude measure
of anything (for example, the paper points out Asians, at the same body fat percentage,
tend to have a lower BMI than caucasians, and of course women and men differ), so I don't
see anyone switching from 2 anytime soon.

Dan
 
jbuch wrote:

> I'm evidently heavy boned.


Great line. (It was either that, or "Family newsgroup, buddy.")

Bill "Friday" S.
 
"Dan Connelly" <d_j_c_o_n_n_e_l@i_e_e_e.o_r_g> wrote in message
news:seVjf.26355$q%[email protected]...
> From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png
>
> I estimate mass is roughly proportional to height^2.2 (there's a lot
> of variability in this estimate, it's true, but it's clearly > 2,
> the BMI assumption).
>
> This suggests a more height-invariant BMI might be estimated, using:
>
> BMI' = mass / height^2.2
>
> However, I suspect people have enough trouble grasping units of
> "kg/m^2",
> trying to get them to swallow "kg/m^2.2" may perhaps be a bit much.
> But
> there seems a strong case to be made that the BMI index is "unfair" to
> tall
> people (assuming lower is better, usually the case among those with
> enough
> income to read this message).
>
> Dan


I get 21 using your method and 23 using yahoo health. I'm 6'2", 180lbs
with a 32" waist. Even if I input 170 (my fighting weight) I'm still at
22 with the conventional method yet this is on the ragged edge for me. I
would have to weigh 145 to be on the lower end of
"normal".........absurd.

Phil H
 
On Fri, 02 Dec 2005 10:21:12 +0000, Dan Connelly wrote:

> From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png
>
> I estimate mass is roughly proportional to height^2.2 (there's a lot
> of variability in this estimate, it's true, but it's clearly > 2,
> the BMI assumption).


If you scale up a body proportionally, then the mass would be proportional
to height^3. Since the limb-strength (and ability to support the mass)
only grows as the cross-sectional area, it would be proportional to
height^2 if it were merely scaled. So, reasonably, taller people would
require proportionally more robust bones, so the proper scaling of
mass should exceed height^3.

I have no idea why BMI uses mass proportional to height^2.

--

David L. Johnson

__o | Some people used to claim that, if enough monkeys sat in front
_`\(,_ | of enough typewriters and typed long enough, eventually one of
(_)/ (_) | them would reproduce the collected works of Shakespeare. The
internet has proven this not to be the case.
 
On Fri, 02 Dec 2005 06:54:31 -0600, jbuch wrote:

> With BMI scaling as Hheight^2 (^ is an easy way to represent the
> exponent), it is much more realistic than the height^3 or cubic scaling
> law.


why would BMI be more realistic?

The height^(2.2) is an improvement which has the disadvantage of
> being incomprehensible to the average medical or health worker, and
> incomprehensible to the average "personal trainer" who has no training
> credentials at all.


Units are irrelevant.

--

David L. Johnson

__o | Let's not escape into mathematics. Let's stay with reality. --
_`\(,_ | Michael Crichton
(_)/ (_) |
 
David L. Johnson <[email protected]> wrote:

> If you scale up a body proportionally, then the mass would be proportional
> to height^3. Since the limb-strength (and ability to support the mass)
> only grows as the cross-sectional area, it would be proportional to
> height^2 if it were merely scaled. So, reasonably, taller people would
> require proportionally more robust bones, so the proper scaling of
> mass should exceed height^3.


I don't remember the rationale, but a doctor in a local newspaper argued
that the formula should actually be mass/height^2.6 . Closer to cube
than square, but neither are really close. I guess this might've been
based on a statistical study on height, weight and fat %, which should
give you a reasonable constant value for the formula.

-as
 
On Fri, 02 Dec 2005 22:41:39 -0500, David L. Johnson wrote:

> On Fri, 02 Dec 2005 10:21:12 +0000, Dan Connelly wrote:
>
>> From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png
>>
>> I estimate mass is roughly proportional to height^2.2 (there's a lot
>> of variability in this estimate, it's true, but it's clearly > 2,
>> the BMI assumption).

>
> If you scale up a body proportionally, then the mass would be proportional
> to height^3. Since the limb-strength (and ability to support the mass)
> only grows as the cross-sectional area, it would be proportional to
> height^2 if it were merely scaled. So, reasonably, taller people would
> require proportionally more robust bones, so the proper scaling of
> mass should exceed height^3.
>
> I have no idea why BMI uses mass proportional to height^2.


I've always wondered about why anyone cares about the BMI thing. A quick
dip in the pool will tell you a lot more than any formula.

Breath out fully, if you float, you are in real trouble with regard to
body fat. If you sink slowly, consider adding 5 miles per day to your
riding. If you can sit cross legged on the bottom of the pool and
meditate, then I hear that the Dalai Lama is hiring.

Breath in to full capacity, do you float a little, or float a lot. If you
float a lot while full, but sink when empty, you're not in danger, but
could probably stand one less serving on Christmas and Thanksgiving day.
If you are neutrally buoyant with full lungs, or perhaps even sink slowly,
your body fat percentage is very low regardless of what the BMI might say.
 
rwwff wrote:
> On Fri, 02 Dec 2005 22:41:39 -0500, David L. Johnson wrote:
>
>> On Fri, 02 Dec 2005 10:21:12 +0000, Dan Connelly wrote:
>>
>>> From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png
>>>
>>> I estimate mass is roughly proportional to height^2.2 (there's a lot
>>> of variability in this estimate, it's true, but it's clearly > 2,
>>> the BMI assumption).

>>
>> If you scale up a body proportionally, then the mass would be
>> proportional to height^3. Since the limb-strength (and ability to
>> support the mass) only grows as the cross-sectional area, it would
>> be proportional to height^2 if it were merely scaled. So,
>> reasonably, taller people would require proportionally more robust
>> bones, so the proper scaling of
>> mass should exceed height^3.
>>
>> I have no idea why BMI uses mass proportional to height^2.

>
> I've always wondered about why anyone cares about the BMI thing. A
> quick dip in the pool will tell you a lot more than any formula.
>
> Breath out fully, if you float, you are in real trouble with regard to
> body fat. If you sink slowly, consider adding 5 miles per day to
> your riding. If you can sit cross legged on the bottom of the pool
> and meditate, then I hear that the Dalai Lama is hiring.
>
> Breath in to full capacity, do you float a little, or float a lot.
> If you float a lot while full, but sink when empty, you're not in
> danger, but could probably stand one less serving on Christmas and
> Thanksgiving day. If you are neutrally buoyant with full lungs, or
> perhaps even sink slowly, your body fat percentage is very low
> regardless of what the BMI might say.


I'm a sinker for sure. I've always been scared since I was young that I'm
going to drown one day because I'll be stranded in the water from a boat
sinking. Therefore, I will only board a boat with a life jacket, a ring
buoy, flare gun, GPS transponder, and water wings on.
--
Phil, Squid-in-Training
 
On Fri, 02 Dec 2005 10:21:12 GMT, Dan Connelly
<d_j_c_o_n_n_e_l@i_e_e_e.o_r_g> wrote:

>However, I suspect people have enough trouble grasping units of "kg/m^2",
>trying to get them to swallow "kg/m^2.2" may perhaps be a bit much. But
>there seems a strong case to be made that the BMI index is "unfair" to tall
>people (assuming lower is better, usually the case among those with enough
>income to read this message).


Personally, I don't need a fricking calculator to tell me I'm overweight.
I'd be much better off at two thirds or a bit oover half my current
weight, but according to Quetelet/BMI I'd still be overweight then, which
would clearly not be the case.

Jasper
 
On Fri, 02 Dec 2005 22:43:11 -0500, "David L. Johnson"
<[email protected]> wrote:
>On Fri, 02 Dec 2005 06:54:31 -0600, jbuch wrote:
>
>> With BMI scaling as Hheight^2 (^ is an easy way to represent the
>> exponent), it is much more realistic than the height^3 or cubic scaling
>> law.

>
>why would BMI be more realistic?


Because height^3 leads to massive aberrancies even within the normal human
healthy range of between a bit under 5 feet and 6.5 feet, more so than
than ^2 does. Empirically determined. Basically it's because smaller or
taller doesn't mean everything on your body scales in depth and width
equally. Look at genetic dwarves, or the vertically challenged, or
whatever the current PC term is: they're still very nearly as wide and
deep as the average human, just shorter (which makes them look wildly out
of proportion). Similarly, NBA players aren't wider or deeper in
proportion to their excessive height, which also makes them look wildly
out of proportion.

Jasper
 
On Sun, 4 Dec 2005 00:46:14 -0500, "Phil, Squid-in-Training"
<[email protected]> wrote:

>I'm a sinker for sure. I've always been scared since I was young that I'm
>going to drown one day because I'll be stranded in the water from a boat
>sinking. Therefore, I will only board a boat with a life jacket, a ring
>buoy, flare gun, GPS transponder, and water wings on.


Don't you look silly in the on-board casino wearing that?

Jasper
 
Jasper Janssen wrote:
> On Sun, 4 Dec 2005 00:46:14 -0500, "Phil, Squid-in-Training"
> <[email protected]> wrote:
>
>> I'm a sinker for sure. I've always been scared since I was young
>> that I'm going to drown one day because I'll be stranded in the
>> water from a boat sinking. Therefore, I will only board a boat with
>> a life jacket, a ring buoy, flare gun, GPS transponder, and water
>> wings on.

>
> Don't you look silly in the on-board casino wearing that?


I forgot to mention the flippers, the mask, and snorkel. I keep the knife
and harpoon gun on the side for those pesky sharks. It only gets in the way
when I pull the slot machine levers.

--
Phil, Squid-in-Training
 
Jasper Janssen wrote:
>
> Because height^3 leads to massive aberrancies even within the normal human
> healthy range of between a bit under 5 feet and 6.5 feet, more so than
> than ^2 does. Empirically determined. Basically it's because smaller or
> taller doesn't mean everything on your body scales in depth and width
> equally. Look at genetic dwarves, or the vertically challenged, or
> whatever the current PC term is: they're still very nearly as wide and
> deep as the average human, just shorter (which makes them look wildly out
> of proportion). Similarly, NBA players aren't wider or deeper in
> proportion to their excessive height, which also makes them look wildly
> out of proportion.


Er, at 6'10", I'm slightly taller than the NBA average (6'7.25"), I
don't think I look "wildly out of proportion". I'm about 17% taller than
the average male of European descent, not that big a deal. BMI seems to
scale pretty accurately for me. There's a lot more variation in
proportion from fat -- way more than 17%.
 
Jasper Janssen wrote:
>
> Because height^3 leads to massive aberrancies even within the normal human
> healthy range of between a bit under 5 feet and 6.5 feet, more so than
> than ^2 does. Empirically determined. Basically it's because smaller or
> taller doesn't mean everything on your body scales in depth and width
> equally. Look at genetic dwarves, or the vertically challenged, or
> whatever the current PC term is: they're still very nearly as wide and
> deep as the average human, just shorter (which makes them look wildly out
> of proportion). Similarly, NBA players aren't wider or deeper in
> proportion to their excessive height, which also makes them look wildly
> out of proportion.


Er, at 6'10", I'm slightly taller than the NBA average (6'7.25"), I
don't think I look "wildly out of proportion". I'm about 17% taller than
the average male of European descent, not that big a deal. BMI seems to
scale pretty accurately for me. There's a lot more variation in
proportion from fat -- way more than 17%.
 
Jasper Janssen wrote:
>
> "David L. Johnson"
> >
> >why would BMI be more realistic?

>
> Because height^3 leads to massive aberrancies even within the normal human
> healthy range of between a bit under 5 feet and 6.5 feet, more so than
> than ^2 does. Empirically determined. Basically it's because smaller or
> taller doesn't mean everything on your body scales in depth and width
> equally.


The problem with BMI *is* geometric, desipte what you assert, but it's
also statistical.

It derives generalities from the fat part of the bell curve and applies
it to to the ends. That's what it's designed to do, and that's
fundamentally mistaken. There is no single way to be at the 99th
percentile for whatever characteristic. I used to have a neighbor who
at 6'10" or 6"11 was noticeably taller than me, but weighed probably of
60% what I did-- and at that time I was relatively lean. His waist was
about the same girth as one of my thighs.

You use the example of achondroplastic dwarves to defend the geometric
flaw contained in BMI, but they are a perfect example of why BMI
doesn't work. Someone 4-1/2 feet tall has to be an achondroplastic
dwarf to have a normal BMI at a normal body composition, while e.g. a
4-1/2 foot tall Filipina of typical build would have to be very high in
adiposity to register a normal BMI.

Peter Cole at 6'10" weighs some 230 lbs., but at that weight I look
like a terminal drug addict-- and I'm 2 inches shorter than he is.

The statistically flawed nature of BMI is in common with height/weight
tables and other such contrivances, but its geometrical flaw insures
that the further you deviate from the statistical mean, the more
inaccurate it becomes as an indicator of health or body composition.

If it were only the province of the quacks who created it, then it
would not be a problem. The trouble comes when insurers and medical
professionals treat as if it had any scientific value.

Chalo Colina
 
David L. Johnson wrote:
> On Fri, 02 Dec 2005 06:54:31 -0600, jbuch wrote:
>
>
>>With BMI scaling as Hheight^2 (^ is an easy way to represent the
>>exponent), it is much more realistic than the height^3 or cubic scaling
>>law.

>
>
> why would BMI be more realistic?
>
>



You can find some information on scaling laws for animals here....
1) http://smccd.net/accounts/goth/other/Life_On_The_Scales.pdf
2) http://www.primidi.com/2005/02/21.html
and it is a tad technical.


Take a 7 foot person and a 5 foot person.

Would you expect the 5 foot person of "normal build" to have a waist
which is 5/7 of that for a 7 foot tall person?

This linear relationship is what happens if the height^3 (cubed) scaling
law is assumed..... But, if you understood the topic, you would have
known that.

Taking the 34 inch waist for the 7 foot person as "fit physique", would
you expect the waist of the 6 foot person to be 24.3 inches (5/7 of 34
is 24.3 inches) for a "fit physique"?

This is absurd.

Short people of "normal physique" don't have such absurdly small waist
sizes.

If you use the height^2 scaling law, you would expect the waist size
would scale as (5/7)^0.667 (the two thirds power) or you would suspect
that maybe 27.2 inches would be a reasonable waist size for a 5 foot
tall person, rather than 24.3 inches.

The discussion is leading to the suggestion that maybe the waist size
for a "normal physique" 5 foot person would be a little larger than 27.2
inches.....

And if you go into a clothing store that deals in short people's
clothes, you would find a range of waist sizes that you could use to
better calibrate the scaling law that actually fits the body
configurations that humans have.


One suggestion is to use the 3/4 power scaling law (in the second
reference above), which would then give (5/7)^.75 and then a 5 foot
person of "nominal physique" ( with a 7 foot 34 inch waist as "nominal
physique for reference) would have a waist size of 28.0 inches.

"Nominal Physique"

7 foot tall, 34 inch waist
5 foot tall, 28 inch waist (3/4 power scaling law)

5 foot tall, 24.3 inches (linear scaling law, BMI =K * Height^3)

I go with the experts who have studied scaling laws in animals.

BMI as implimented simply sucks and fails to meet what has long been
known about how dimensions of animals scale with size.










The height^(2.2) is an improvement which has the disadvantage of
>
>>being incomprehensible to the average medical or health worker, and
>>incomprehensible to the average "personal trainer" who has no training
>>credentials at all.

>
>
> Units are irrelevant.
>



Anything with matematics is pretty incomprehensible to 95% of the
population. We may not be illiterate, but we have been called
innumerate.... meaning we really screw up stuff based on arithmetical
operations more complex than add, subtract and multiply. Nobody expects
modern youth to actually be able to do long division anymore.... or
count change.



--
................................


Keepsake gift for young girls.
Unique and personal one-of-a-kind.
Builds strong minds 12 ways.
Guaranteed satisfaction
- courteous money back
- keep bonus gifts

http://www.alicebook.com
 
Phil, Squid-in-Training wrote:
> rwwff wrote:
>
>>On Fri, 02 Dec 2005 22:41:39 -0500, David L. Johnson wrote:
>>
>>
>>>On Fri, 02 Dec 2005 10:21:12 +0000, Dan Connelly wrote:
>>>
>>>
>>>> From http://anonymous.coward.free.fr/rbr/nba-tdf-bmi.png
>>>>
>>>>I estimate mass is roughly proportional to height^2.2 (there's a lot
>>>>of variability in this estimate, it's true, but it's clearly > 2,
>>>>the BMI assumption).
>>>
>>>If you scale up a body proportionally, then the mass would be
>>>proportional to height^3. Since the limb-strength (and ability to
>>>support the mass) only grows as the cross-sectional area, it would
>>>be proportional to height^2 if it were merely scaled. So,
>>>reasonably, taller people would require proportionally more robust
>>>bones, so the proper scaling of
>>>mass should exceed height^3.
>>>
>>>I have no idea why BMI uses mass proportional to height^2.

>>
>>I've always wondered about why anyone cares about the BMI thing. A
>>quick dip in the pool will tell you a lot more than any formula.
>>
>>Breath out fully, if you float, you are in real trouble with regard to
>>body fat. If you sink slowly, consider adding 5 miles per day to
>>your riding. If you can sit cross legged on the bottom of the pool
>>and meditate, then I hear that the Dalai Lama is hiring.
>>
>>Breath in to full capacity, do you float a little, or float a lot.
>>If you float a lot while full, but sink when empty, you're not in
>>danger, but could probably stand one less serving on Christmas and
>>Thanksgiving day. If you are neutrally buoyant with full lungs, or
>>perhaps even sink slowly, your body fat percentage is very low
>>regardless of what the BMI might say.

>
>
> I'm a sinker for sure. I've always been scared since I was young that I'm
> going to drown one day because I'll be stranded in the water from a boat
> sinking. Therefore, I will only board a boat with a life jacket, a ring
> buoy, flare gun, GPS transponder, and water wings on.



I too am a sinker, when I got my weight down from being moderately obese
(42+ inch waist on a 5' 6" body).

I always had a problem floating in water.

Bone, muscle and connective tissue are denser than water, and fat floats.

When I was fat enough, I floated nicely.





--
................................


Keepsake gift for young girls.
Unique and personal one-of-a-kind.
Builds strong minds 12 ways.
Guaranteed satisfaction
- courteous money back
- keep bonus gifts

http://www.alicebook.com
 
On Sun, 04 Dec 2005 18:12:31 -0600, jbuch wrote:

> You can find some information on scaling laws for animals here....
> 1) http://smccd.net/accounts/goth/other/Life_On_The_Scales.pdf
> 2) http://www.primidi.com/2005/02/21.html
> and it is a tad technical.


"Technical" is an interesting term, here. From the first site, it
suggests that all animals have basically the same number of heartbeats in
a lifetime; that metabolic rate is proportional to mass. But humans live
longer than elephants. Oops.

It also says, and I quote:

"If one animal is, say, twice as big as another animal in each linear
dimension, then its total volume, or mass, is 23 times as large, but its
skin surface is only 22 times as large."

Umm, no. If one animal is proportionally twice as big as another in each
linear dimension (presuming 3-dimensional animals, here), it would have 8
times the volume, and 4 times the skin surface.

Such "experts" are open to question.

--

David L. Johnson

__o | Become MicroSoft-free forever. Ask me how.
_`\(,_ |
(_)/ (_) |
 

Similar threads

C
Replies
1
Views
642
L