Body mass index (BMI) as a measure of obesity and health: a critical appraisal

i-8540bc6ecce1d00a067cb4c76466a07c-obesity measure.jpg

If you go to your physician's office and inquire about your weight status, he or she will measure your height and weight to derive your BMI (weight in kg divided by height in m squared). Then they will compare your BMI to that of established criteria to decide whether you are underweight (<18.5 kg/m2), normal weight (18.5-24.9 kg/m2), overweight (25-29.9 kg/m2), or obese (>30 kg/m2) . Often times, this measure alone determines whether or not you receive lifestyle treatment. But how useful is this measure anyways? What does it tell you about your health? And finally, how helpful is it to measure when assessing the effect of a lifestyle (diet/exercise) intervention?

For quite some time I have been meaning to discuss some of the issues of solely relying on BMI as a measure of obesity and health, and a nice nudge from our friend ERV was just the motivation  I needed to finally get to work.

Before I get into the various limitations of BMI, I must point out that the measure is quite useful across large populations, as it is well correlated with the degree of adiposity, and of course it is extremely simple to measure in clinical practice.

Nevertheless, here are some of the key issues with BMI, particularly when used on an individual basis.

1. BMI does not differentiate between the Michelin Man and The Terminator

Ok, we might as well just get this abundantly obvious problem out of the way. I have heard countless times how one buff celebrity or another (e.g. Tom Cruise, Arnold Schwarzenegger, The Rock etc.) would be classified as overweight or obese according to their BMI due to their excess amount of muscle. Yes, this is absolutely true. BMI is a measure of relative weight; fat mass and muscle mass are not distinguished. Here's what is equally true: the large majority of the general population with a BMI in the overweight or obese range does not look like Jerry Maguire or the Terminator. Also, if you seek advice from your physician about your "excess weight", in case you have body dysmorphia and cannot yourself decide, they will quickly be able to assess whether your excess weight is due to your bulging muscles or your rolls of adipose tissue. So while this is an obvious problem, I would argue not the main issue.

2. BMI does not differentiate between apples and pears

For over 60 years, we have known that independent of how heavy a person is, the distribution of their body weight, or more generally the shape of their body is a key predictor of health risk. It is now well established that individuals who deposit much of their body weight around their midsection, the so called apple-shaped, are at much greater risk of disease and early mortality in contrast to the so called pear-shaped, who carry their weight more peripherally, particularly in the lower body. Thus, two individuals with a BMI of 32 kg/m2 could have drastically different body shapes, and thus varying risk of disease and early mortality.

Fortunately, a very simple measure allows you or your physician to decide whether your elevated BMI is of the apple or pear variety: waist circumference. Current thresholds suggest that a waist circumference above 88 cm in women and 102cm in men denotes abdominal obesity. Interestingly, for the same BMI level, those individuals with an elevated waist circumference have a greater risk of diabetes, cardiovascular disease, mortality, and numerous other health outcomes. Thus, as studies from our laboratory have consistently suggested, waist circumference may be a more important measure of obesity and health risk than BMI. Currently, most researchers would agree that waist circumference should be measured along with BMI to adequately classify obesity-related health risk.

You can measure your own waist circumference by using a tape measure and wrapping it around your abdomen, at the level of the top of your hip bones. Make sure you measure at the end of exhalation, without sucking in your gut - you're only fooling yourself!

3. BMI does not always budge in response to lifestyle change

Given the number of papers my supervisor, Dr. Ross, and I have published on the topic, I would argue this is the biggest drawback of using BMI: it doesn't always change even though you may be getting healthier. This is particularly so if you adopt a physically active lifestyle, along with a balanced diet, but are not necessarily cutting a whole lot of calories. This lack of change in BMI or body weight is all too often interpreted as a failure, resulting in the disappointed individual resuming their inactive lifestyle and unhealthy eating patterns.

However, as we have argued most recently in a paper in the Canadian Journal of Cardiology, several lines of evidence suggest that weight loss or changes in BMI are not absolutely necessary to observe substantial health benefit from a healthy lifestyle. Thus, an apparent resistance to weight-loss should never be a reason for stopping your healthy behaviours.

First, it is well established that increasing physical activity and associated improvement in cardiorespiratory fitness are associated with profound reductions in coronary heart disease and related mortality independent of weight or BMI. Second, exercise (even a single session) is associated with substantial reduction in several cardiometabolic risk factors (such as blood pressure, glucose tolerance, blood lipids, etc.) despite minimal or no change in body weight. Third, waist circumference and abdominal fat (arguably, the most dangerous fat) can be substantively reduced (10-20%) in response to exercise with minimal or no weight loss. In fact, significant reductions in fat mass often occur concurrent with equal increases in muscle mass in response to physical activity - equal but opposite (and beneficial!) changes which are not detected by alterations in body weight on the bathroom scale, and thus BMI.

So in the end, while BMI surely has its strengths in ease of use and pretty good reliability in large populations, on an individual basis, the greater focus should be on healthy behaviors: physical activity and a healthy diet. And if you must measure something, check your waist circumference.

Peter Janiszewski

Ross R, & Janiszewski PM (2008). Is weight loss the optimal target for obesity-related cardiovascular disease risk reduction? The Canadian journal of cardiology, 24 Suppl D PMID: 18787733

More like this

something is off with your blog, the font is tiny, even compared to other blogs on scienceblogs.com

I would just like to say that extreme measures to lose weight -- like bariatric surgery -- should only be taken when the weight truly is extreme. Extreme enough to render someone unable to walk or bend over to retrieve a dropped item, for example.

And there is some truth to the "it's muscle, not fat" statements that overweight people make. I had bariatric surgery 20+ years ago. Now I weigh just 10lbs less than when I had it, but my body is much larger. Before, I was able to move the refrigerator to clean behind and underneath it. Within a few years, I couldn't even budge it. Muscle loss is part of the weight loss along with the fat.

I'm not saying being fat is healthy, rather I'm saying that it's healthier than going to extremes to be not fat.

As for metabolism, mine has changed tremendously. My body temp is most often (I won't say normally) between 96.5 and 97. I suffer "spells" when it drops as low as 94. These are painful, involving uncontrolled shivering and they are sometimes relieved only by putting a pair of sweats and a blanket in the dryer to warm them and going to bed. It's a little scary and I've considered going to the ER, but the shivers also sometimes go away as suddenly as they appear.

As for eating, you betcha bariatric surgery changes that. At first, even drinking too much water can induce projectile vomiting. Later, as inflammation reduces and the pouch swells, patients learn what they can eat and how much. By this time, one's esophagus may be scarred. And, while I don't know that this is normal for all bariatric patients, for me everything I eat makes me nauseous. Not the "too full" nausea, but just plain nausea where you sort of wish you could throw up so it would go away. I put off eating anything until the gnawing hunger is worse than the nausea I know I'll get.

For a bariatric surgery patient, I'm really pretty lucky. While it has completely screwed up my digestive system, I'm not suffering the most severe side effects. Some would even say mine is successful because I still don't have the high cholesterol, type II diabetes, or atherosclerosis every one of my doctors thinks I should have because of my weight.

When I wanted to join a fitness center a few years ago, I had to get permission from my doctor. My PCM was afraid to sign off on that without a stress test. So I had two and failed both of them. While the cardiologists involved suspected that breast size was interfering with the measurements, the only way to make sure (according to them) was a cardiac cath. Everyone seemed so surprised that my arteries were so unobstructed. (I was 56 at the time.)

My only purpose in posting is that applying what is normal for a large population to an individual is not necessarily helpful. I think this applies to those who are deemed as needing to gain weight also.

For those who are deemed underweight, it strikes me as ludicrous that they should be instructed to eat what are deemed "empty calories" for those overweight simply to gain weight. There's a disconnect here that I don't understand.

How much does height factor into this as well? Because BMI for my height (6'6") seems off to me. I would think a healthy weight would be around 220 to 230 while BMI suggests around 180 - 215. I know I am above average in height and that could also play a role in things.

Welcome to the SB! Thank you for the BMI article. This is a really useful blog, I appreciate your taking the time to share your knowledge.

I have a question about body fat measurement. Covert Bailey in his "Fit or Fat" book(latest version) recommended a body fat formula using waist, hip, forearm and wrist sizes to determine body fat percentage(for Men over 30). Women and different age groups had different formulas. It seemed reasonable as I trained for a marathon, reducing from 26% to 17%.

Is there good evidence that this type formula is accurate for body fat measurement? Is it possibly correct within 2-3 %?

"Fit or Fat" got me off my ass in college. "Exercise as fast as your heart tells you" got me to run for more than 10 minutes.

Tedd

Welcome! I look forward to reading your blog. I am blogging on food, farming and genetics and will send my readers your way.

BMI (weight in kg divided by height in m squared)

Really? (consults teh google)
Ah. I was going to point out that this is stupid, since weight should scale to the cube, not the square, of a linear measure like height, but I see this has been pointed out and, apparently, generally ignored (*shrug*):

For a given height, BMI is proportional to weight. However, for a given weight, BMI is inversely proportional to the square of the height. So, if all body dimensions double, and weight scales naturally with the cube of the height, then BMI doubles instead of remaining the same. This results in taller people having a reported BMI that is uncharacteristically high compared to their actual body fat levels. In comparison, the Ponderal index is based on this natural scaling of weight with the third power of the height. However, many taller people are not just "scaled up" short people, but tend to have narrower frames in proportion to their height. It has been suggested that instead of squaring the body height (as the BMI does) or cubing the body height (as the Ponderal index does), it would be more appropriate to use an exponent of between 2.3 and 2.7

--('kipedia)

By Sven DIMilo (not verified) on 10 Mar 2010 #permalink

"weight should scale to the cube, not the square, of a linear measure like height"

Only if we were cubes :) Generally speaking, humans shouldn't grow outwards in direct proportion to growing upwards.

BMI is an approximate measure of health - but the point is that it's easy. If you made it more accurate (eg. more measurements, more complicated calculations), it would be less useful in that respect.

By theshortearedowl (not verified) on 10 Mar 2010 #permalink

Two issues that worry me - height loss due to injury and aging that could have incorrect interpretation of a person's BMI being added to medical records, which might affect insurance costs. Second measuring waist standing up does not account for posture problems. Supine is a much more precise measurement. Indeed sagittal abdominal supine height would be even better but that doesn't translate into the simplistic BMI most doctors want on the records.

I've only been reading your blog for a few months, so I don't know to what extent you've discussed "fat denialism" -- the rationales people come up with to explain why they are really not overweight (when they are), or why they can't seem to lose weight. Common rationales of course include "it's genetic" -- and after a recent media report, I expect "it's the gut bacteria" to join the repertoire shortly.

And I can't tell you how many people I have heard explain that their BMI may be elevated but they aren't overweight because it's really all muscle -- including some people with suspiciously large waistlines. It certainly may be true in some cases, but it also gets used as a denial tactic.

I very much appreciate your sensible, level-headed and clear approach to these things.

(I'm overweight, pear-shaped and not active enough, but trying to work on it....)

Excellent point and you did a great job explaining the known limitations of BMI being used as a diagnosis criteria, but the situation in which it is being used should be more defined.

In a clinical setting a BMI measure alone is not adequate for a health screening by the person's PCP. All of the hospitals I frequented required several other clinical labs for an RD referral and an MD intervention. Each hospital has different cut offs but normally it involves total cholesterol, albumin and/or electrolyte values outside of what they have decided is normal.

There is also a "confidence interval" for lack of a better term within the BMI calculation and the extremes either of very short or very tall people tend to get skewed results.

You mention that waist circumference is quite important, and I agree but was under the impression it isn't JUST waist circumference but waist-to-hip ratio that is normally calculated. A person could have a relatively small waist measurement and still be considered apple shaped depending on their hip measurements. It depends on the particular risk factors you are interested in, which I assume in your case is cardiovascular in nature. I agree with your statements on risk factors associated with increased waist circumference, and have read many of the same sources.

I think it's great that you point out that it is used because it is easy to calculate. One of my former coworkers asked why every facility doesn't do DXA or BodPod measurements to determine adiposity and it's a matter of cost, space and time commitment. Another factor is where you are getting these measurements from and how they are collected. In one of your previous posts you mentioned the public getting advice from the "experts" who work in the gym, the same is true when getting body fat measurements from an untrained person.. this is why the use of anthropometric caliper measurements have gone the way of the dodo in nearly every place I've visited.

If you would like to read about some of the absolute failures brought on by BMI, look into Arkansas' use of the BMI Initiative, where they placed a child's BMI on their report card.. Failed horribly as you can imagine and I was still living there when everything hit the fan.

Great idea to bring some of the limitations of this into the light, since most people don't even seem to understand what the numbers mean!

..PS.. I forgot to mention I used to do pediatric obesity related research at one of the USDA sites, so we might know some of the same people :)

weight should scale to the cube, not the square, of a linear measure like height

The closest I can come to explaining this (although I am not fully satisfied with the answer) is that the capacity of your support systems (bone, muscle, and blood vessel capacity) scales as the square of height. This is the reason why scaling up an insect to human size, as is done in so many science fiction B movies, doesn't work in real life: the giant cockroaches would collapse under their own weight. It also explains why having too low a BMI could be problematic. It fails, by itself, to explain why too high a BMI is a problem--this is the muscle vs. fat issue discussed in the post.

Using the cube of the height also has problems. Exceptionally tall people (230 cm or more) are disproportionately likely to suffer from heart and circulation problems simply because it is so difficult to pump blood over that height, not because they have a high BMI. ISTM that the right exponent is somewhere in between, but it would take some research to find what the best exponent is. Of course BMI and adiposity are correlated, because fatter people tend to have higher BMIs, but I don't know whether someone has established that they should be linearly correlated.

By Eric Lund (not verified) on 10 Mar 2010 #permalink

As I understand it, the original data came from Metropolitan Life Insurance. The idea was to see if height and weight were predictors of longevity. To collect the data, they had height and weight on application forms. When the data proved useless, the company released it to anyone who wanted it.

Note that they accepted the applicants' answers uncritically, and asked only once, at the time of the application, which used to be usually upon turning 21. There is no regard for what occupation the person would have, how many children a woman will bear, or anyone's lifestyle. Among the customers were nurses who smoked to avoid gaining weight. Looking back, nobody would expect any utility in height and weight upon turning adult.

The first attempts at recommended weight for height used a linear model, as if weight scaled linearly with height, allowing base weight plus so many pounds for every inch over the base height, but that never worked.

Next somebody went to a square model, as if weight scales to the square of height, and the BMI is one of these. Arbitrarily, any cutoff is exactly a whole number or something-point-five, proof that somebody pulled numbers out of the air (or worse).

This is what happens when somebody tries to understand physiology without first studying scaling laws. Body mass scales roughly to the cube of length, but not tightly.

If I change the calculation from 2 to 2.5, I come out with a normal BMI of 20.05 roughly (6'6" 250). Yet, I am pretty sure I am overweight. I will go with my gut and try and lose a few pounds, get back down to around 225.

RE: weight loss

There is this funny thing called the First Law of Thermodynamics, which posits that energy can be changed in form, but cannot be created or destroyed.

This makes weight loss terribly easy: eat fewer and expend more calories.

Gut bacteria and genetics don't trump this law. If you eat 2000 calories/day and expend 2200 calories/day, you will lose weight.

I've never understood why people can't see something so obvious.

Good post!

In response to BeamStalk:

Height is an important factor. Often in our genetic studies, when we see that a genetic variant associates with a measure of obesity (e.g., BMI, waist circumference, waist-hip ratio), there is also an association with height in the shorter direction. That is, a reduction in height. Typically this is 1-2 cm and for a person of 1.8 m height (5 foot, 11 inches) weighing 90 kg (198 pounds) this is a change in BMI of 0.5 unit.

You noted the muscle mass/fat mass distiction, but what about bone mass? For example, comparing one of my brothers to myself, his wrists are bigger in circumference than my ankles; the difference being in the size of our respective bones.
I'd also ask about the issue of limb length. Height in torso adds more mass than height in the legs, right? Given two people of the same height, but one with longer legs and the other with a longer torso, is it reasonable to suggest that both should have the same weigth, as BMI does?

@2 Tedd

When I decided to get into better shape a few years ago (6'2" 196 lbs, doesn't sound too bad but I'm a very small build naturally), I bought one of the high-end electronic scales with the impulse/body fat measurement systems. It gives BMI as well (if you input your height), but I've found the body fat % much more indicative of fitness progress.

A dozen or so times, I "tested" the measurement using a technique (similar to yours) apparently implemented in the US Military and, except for one time when I was under 10% body fat (after a surgery), the numbers matched up to within a few tenths of a percent.

So, I'm inclined to trust either measurement as a decent estimate, with the one caveat that the impulse measurement seems to be somewhat affected by hydration. The scale's just easier to do every morning.

Thanks to all for the fantastic comments! We're not quite accustomed to the volume and the length of the comments (not to mention the quality), so please bare with us. I am amazed by the caliber of insights everyone is bringing to the discussion - so rewarding for us to learn from our readers.

@BeamStalk - I just did the calculations, and you're absolutely correct - 215lbs would be the upper range of a normal weight. @Larry Parnell and @Science Bear's points may help explain why the normal weight cutoff at the extremes of height may be an issue. I wonder though, what's your waist circumference? If you're below 102cm, I'd tend to think I wouldn't give BMI too much consideration.

@Tedd - I've definitely heard of the book, though I must admit I have yet to read it. There has been no shortage of equations developed in an attempt to quantify body fat. Unfortunately, most of these have significant error. I have not come across a specific formula using circumferences only (at least not one that has been accepted in the literature or in clinical practice), but the use of skinfolds as measured using calipers has been often used along with age, and gender specific formulas (this remains popular in gyms, but is likely on the way out). Even though skinfolds would obviously be more accurate than mere circumferences, there is still great error - no idea how much visceral fat one may have, for example. In truth, short of DEXA, CT, or MRI - there is no cheap, accurate, and reliable method for measuring body composition. Also, despite what many claim to be true about bioimpedance analysis (like the handheld gadgets, or the scales that measure body fat) - these are rarely accurate and again depend on various predictive equations, which have substantial error, and as has been pointed out is affected by hydration among other factors.

Thanks to all for bringing up the issue of cubing rather than squaring height - I had never heard that before - very interesting.

@Kevin C. - Interesting point regarding bone mass. From my basic understanding the skeleton of an adult contributes only something like 10% to total body mass (don't quote me on that number:) - if you've ever held a bone, you'll note how light they actually are. Thus, despite significantly larger skeletal structure, I'm not sure the weight of one's bones would have that much of an impact. In other words, while your suggestion may be absolutely true, I'd argue there are much bigger issues with BMI.

@Pam Ronald - Thanks for the welcome - looks like our blogs will nicely complement one another. Looking forward to reading more of your posts!

Excellent point about skin fold measures. Most people who are the "trained" person at gyms do them incorrectly.

Even with DXA and BodPod, you have people in research settings doing them incorrectly. I recently saw some footage of a group of football players having their body fat percentage checked and the method in which they did it would have completely compromised their results. I believe they had on baggy shorts, a cloth on their head and a cut off shirt... not gonna work guys :-)

@ Rob #12

Follow Obesity Panacea posts more closely. The problem is not with knowing one has to spend 2200 calories/day, it is with measuring how many calories are actually being spent.

Not everyone is spending same number of calories when they do the same activities. There is a thing called metabolism, you know.

Your points at BMI issue #3 are a huge relief to me - my cardiorespiratory fitness and physical activity levels are very good, but BMI says I'm at the low end of overweight. I can (and should) lose weight by further increasing physical activity (which I would much prefer to dieting), but it will be a slow process. I'm below the abdominal obesity measurement for women that you mentioned as well, but I'd like to be further below it, because I'm middle-aged now, and a number of non-obese (but probably overweight) relatives have developed type II diabetes later in life.

I'd much rather pay a personal trainer to push me to work out harder at the gym, than try to severely restrict calories in my well-balanced, low fat (but tasty and undeniably high carb) diet. Almost every time I see a cardiologist colleague at work, he harangues me about my weight, even though he knows perfectly well that my cardiorespiratory fitness and metabolic indicators are healthy. I put a stent in someone your age yesterday! Ugh! I swim a mile or two each week, walk at least 2 miles each day (including some stairs on a lunchtime walk with an older friend at work), do weights and cardio at the gym at least three times a week, and practice an equestrian sport (or do dressage exercises in the arena, if the field's too muddy) once or twice a week. Physical activity has always been an important part of my life, but lately I've felt pretty frustrated by the BMI emphasis. Good to read the critiques from an exercise physiologist.

A minor quibble - dried/bleached bone specimens in an anatomy lab will almost certainly weigh less than bone full of cells, blood vessels, and nerves (and surrounded by periosteum) in a living person. How much less, I don't know off the top of my head, but surely someone has measured this.

@Rob: First, a lot of people refuse to admit they are overweight in the first place. They are merely "big boned" or "naturally large". Second, for many people, overweight carries a huge amount of shame with it; they have been told over and over that overweight means you are a bad or worthless person. So a lot of people are in denial about their physical shape, and in many cases, their primary coping tactic is to come up with reasons why whatever shape they are in is "not my fault." Blaming it on "genetics" is one such rationalization.

I do hear a lot of people claim that they do indeed only take in X calories per day and still don't lose weight. Here again there are numerous ways people can fool themselves. They can -- usually without realizing it -- lower their activity level and burn fewer calories than they think they are burning. They can under-estimate (sometimes by a factor of 2 or 3) how much food they are actually eating, due to a lack of education about what a "portion" actually looks like or where "hidden" calories are likely to lurk. They can "forget" to write down every bite that enters their mouth, including snacks. All of these can be perfectly unconscious, but they explain why someone can claim that they "can't lose weight" and so cannot be blamed for the physical shape they are.

The Obesity Panacea guys have said a lot about how people can counteract some of the self-defeating tricks that lead them to think they are more physically active than they really are. Counteracting the self-defeating tricks that lead people to think they are eating less than they really are is no easier.

Actually there is a substantial literature on the phenomenon of people saying they eat like a bird and still lose no weight. When you study them and evaluate their intake and expenditure you see a very natural bias in their reporting. Basically, people will always overestimate their exercise, and underestimate their eating. It's not surprising, and a well documented phenomenon.

Human bodies are machines that are governed by the laws of physics and that is the great fail of the obesity cranks that try to suggest they either have bodies that violate laws of thermodynamics or argue their incredibly unhealthy lifestyle is not, in fact, unhealthy. Weight loss = greater energy output than input. People do not have metabolic rates that allow them to obtain more calories than exist in food. In fact, when studied metabolically obese people are actually less efficient than skinny people - they actually burn more calories per pound.

I appreciate the criticisms of BMI too, but people who attack it always have some bizarre motive. BMI is just a tool, as a physician I use it to evaluate people along with, for instance, my freaking eyes and common sense. No one makes the mistake of offering Arnold Schwarzenegger bariatric surgery just because he's a BMI outlier. That's not the point. It's a measure that allows us to study health of populations, and to some degree make recommendations that generally apply to individuals who fall within a few standard deviations of the mean. You will of course hear from a lot of people who are the exceptions, but that's beside the point. It's a blunt tool, we know it, get over it.

Welcome to the sb, even if don't post a lot I keep up, and appreciate the good work. We need someone who specializes in just taking on the obesity BS. It's an augean task, good luck.

I appreciate the criticisms of BMI too, but people who attack it always have some bizarre motive.

This is not true. Most people who attack BMI don't want to be viewed as "fat" - being viewed as "fat" comes with substantial social costs. That's not a "bizarre" motive at all; it's a perfectly normal facet of human psychology. Also - most of them don't want the health consequences associated with obesity, and if they can fool themselves into thinking they are not obese, they feel safer. This is another perfectly normal aspect of human psychology, and not "bizarre" at all. Finally, there are a few people who attack BMI because it has some significant (but hardly disastrous) weaknesses, as pointed out in this article. This is also not "bizarre"; it's normal for professionals and scientists to look for better ways to measure things.

One thing I've been wondering about for a while is how much variation is there really in adult human metabolism. I realize that human bodies have to obey physics, but there's no reason my body has to burn the same number of calories as someone else when I'm sleeping, sitting, or even when I'm running. There's obviously a minimum amount of energy needed, but how much variation is there above that?

I've been in shape a few times in my life, but it's always seemed to take a lot more effort on my part than on the part of the people around me. Even if I eat the same amount of the same things and exercise with another person, it always seems like I lose weight much more slowly, so I've always wondered.

@Peter - I haven't measured. I wear a size 34 to 36 waist in jeans but I have a gut above that. That is why I think I am overweight. I play basketball from time to time but I need more regular exercise, plus basketball is high impact. I already have jumper's knee and plantar fasciitis both in my left leg, which I jump off of the most.

I think the frustration with BMI tends to be more what it's used for. I've had an unhealthily low BMI for most of my life and not a single doctor has made a deal of it, told me to do more weight training or increase my calorie count. My overweight friends certainly can't say the same. Seeing as being underweight as opposed to overweight (though not obese) has been correlated with much poorer health, you'd think this oversight would've been taken care of.

I find it difficult to believe that this isn't due to some sort of cultural bias, and that it doesn't speak to the medical professionals' lack of understanding about their own shortcomings in judgment.

Cube, sphere, cylinder, human body--it doesn't matter. Any shape's volume scales to the cube of any linear measurement if the shape stays the same and only size is changed. Weight will scale directly with volume if composition is constant.
Now of course, human body proportions change some with growth and size, so the correct "neutral" scaling exponent is unlikely to be exactly 3. There are probably small changes in composition as well. But the "correct" exponent is very unlikely to be 2, which is implicit in the conventional BMI calculation.

It's the cross-sectional area of support structures (legs/bones) that scales as the square of height, not their volume.

We do this shit all the time in biology, calculate "condition indices", a comparison of an individual's weight with the predicted weight from its length. You can use data to generate an empirical scaling relationship (typical exponents are 2.5-3) and calculate residuals from that, or divide weight by length raised to the empirical exponent, or in absence of an empirical exponent, use 3, which is usually close enough. Using 2 would be wrong every time.

I don't see how dividing by the cube of height rather than the square makes it a more unwieldy and less useful index.

By Sven DiMilo (not verified) on 11 Mar 2010 #permalink

Forgive me if I am wrong as this is in no way my area of expertise, but I remember reading a study that showed that women with PCOS had to keep within a rather low amount of calories compared to normal women in order to to maintain a normal weight. Something like 10% of women have this. Thinking of cases like this is why I am not fond of the whole "2nd law of thermodynamics means you're lying, fatty" attitude that seems so common. I understand that you have to try to be healthy no matter what crappy illnesses life dealt you, but at what point should doctors look for innate reasons instead of just assumig the patient is lying?

Re Sven, we often dose drugs as a relation to body surface area instead of expected body volume. i.e m^2 instead of m^3. This tends to give a target plasma concentration closer to what we are going for with tricky medication. However, you would certainly expect blood volume to increase in proportion to a three dimensional shape, go figure. I would imagine the accuracy of BMI is similar.

Re Chiral, metabolism usually has a pretty tight distribution. Yes, it can certainly differ though in order to do so you would typically see changes in body temperature. One common variable is the amount of activity we do. Just cleaning the house burns a lot more calories than sitting at the computer. One perk, is that exercising is good even if you are not loosing weight so try not to be discouraged.

Clinical judgement is of-course important in determining someone's health status. As I pointed out on the ERV blog one key uses of "BMI" is that it's neutral terminology that allows for a conversation about health/lifestyle without using more of the words such as "fat" "obese" "glutton" etc... It really is amazing how many derogatory words for obesity that we have. Anyway, it makes a much better communications tool than a diagnostic one. As mentioned, measuring the waist circumference is more accurate and just as easy. It's just that a lot of people feel devalued when their doctors pull out a tape measure.

I'm a fan of using waist circumference as a measure of overweight or obesity.

Waist circumference does a very good job of distinguishing between the Michelin Man and the Terminator. Even if the Terminator has been eating heartily recently and isn't ready to show off in the movies, his waist circumference will be within the normal range. When the Terminator goes on a diet to show off his muscles, the waist circumference will drop a couple of inches. When the Michelin Man goes on a diet, his waist circumference will drop several inches.

In addition, it's a very simple and inexpensive measure that can be done rapidly.

By Ruth Sponsler (not verified) on 11 Mar 2010 #permalink

I remember reading that it's not only girth per se, but whether you have subcutaneous fat or fat around the organs. The jiggly fat on your belly that you can pinch is less harmful to your health than the tight-as-a-drum beer belly?

How much of a risk is a BMI that puts you in the overweight range if your other health indicators (blood sugar, cholesterol levels, blood pressure, stress test) are normal?

Sven: It would be possible to devote several blog posts to the inappropriate uses of ratio and scaling in biological data. Maximal oxygen uptake divided by body weight is one of my favorite examples.

I might be completely wrong on BMI but I think that part of the popularity is that the number is believed to be a proxy for percent body fat. I say that because I have overheard several people state something along those lines while explaining BMI.

Wow, 12 comments in on the new blog -- on a topic other than weight loss, even -- and already the de rigeur thermodynamics comment. Which is usually just some normal weight or mildly overweight dude who feels the need to let the world know that he thinks he's smarter than the fatties. I applaud your good self esteem, thermodynamics guy, but I'm pretty sure very little in the way of useful therapeutic approaches have ever come out of the hypothesis "the patient is just dumb". And when you have a problem that is very broadly distributed over a majority of the population, it seems pretty likely that "dumb" is not the cause.

BMI is just a measure of weight adjusted for height. So it measures fatness about as well as your weight does, which is not terrible but not perfect either.

If weight increases linearly with height, then the exponent of height in the index is a function of the slope and intercept of the linear regression and 2 is about right. It doesn't really have anything to do with saying that weight is a function of height squared. It's actually expressing a linear relation in a roundabout fashion. yes, there are scientific articles that have been written about this, although you have to go back to the 1960s and 1970s.

Re MS3

Well my body temperature has always been a bit low, it's usually 97.2 when I'm not feeling ill. My heart also beats slow enough that I've got sinus bradycardia written in my charts somewhere. Maybe that's what's going on.

102cm waist! Here in Japan I get a bad mark on my annual health check for reaching the heady heights of 85cm. Given my 196cm height and 82kg weight, you can easily guess I am not overweight (or even fat-waisted) in any conventional sense. Of course, neither do I have the vital statistics of a conventional Japanese man. But hey, the numbers don't lie :-)

I don't see how dividing by the cube of height rather than the square makes it a more unwieldy and less useful index.

Useful for what? If you're trying to predict health outcomes with a simple metric, you have to include the adverse effects of increasing height as well as obesity. For that purpose, an exponent of 2 may well be better than an exponent of 3; I don't doubt that Shaq has a shorter overall life expectancy than someone closer to the median height and a BMI of 21 or so.

The trouble is that although "you'd be healthier if you were shorter" may be true, it doesn't imply that "you should work on being shorter" is good advice.

As others have pointed out, BMI may be a useful population metric but that doesn't make it a good individual one.

By D. C. Sessions (not verified) on 12 Mar 2010 #permalink

"The trouble is that although "you'd be healthier if you were shorter" may be true, it doesn't imply that "you should work on being shorter" is good advice."
D.C. you're totally just a tall person making excuses ;-)

When conceptualizing the scaling problem, one has to look at the body as a 3 dimensional object. What has to looked at is how the other 2 dimensions scale with height. Height is the easy one and is directly related to the distance from the top of the skull to the distal end of the heel.
What is the proper measurement of width: at the shoulder, the waist, the hip? And how do each of these scale with the height?
What is the proper measure for thickness: at the sternum, at the umbilicus, at the maxilla? And how do these scale with the height?
It would be sensible to use the cube of the height if these scaled the same as height does, but that won't be true. Or if two of the dimensions scaled proportionally with the other dimension constant, then the square of the height would be close to the accurate exponent. However, that won't be true either.

One argument for using the square of the height as the exponent is the simplicity of calculating the result. If you only have a cheap calculator, you will have x2, but not x3. If the true exponent is 2.4, you have to have a log function, and who wants to mess with that?

And to add to the things that can significantly effect the calories out, think thyroid.

By natural cynic (not verified) on 12 Mar 2010 #permalink

"Current thresholds suggest that a waist circumference above 88 cm in women and 102cm in men denotes abdominal obesity."
Are these measurements based on average body length??

By rijkswaanvijand (not verified) on 12 Mar 2010 #permalink

It would be possible to devote several blog posts to the inappropriate uses of ratio and scaling in biological data. Maximal oxygen uptake divided by body weight is one of my favorite examples.

Preaching to the choir, baby. Mass-specific metabolic rates are my #1 pet peeve.

If weight increases linearly with height, then the exponent of height in the index is a function of the slope and intercept of the linear regression and 2 is about right. It doesn't really have anything to do with saying that weight is a function of height squared. It's actually expressing a linear relation in a roundabout fashion.

You are confused, I think. Weight never increases linearly with height. I mean, you can fit a linear regression to the relationship, but it won't fit well (if there is a decent range of heights involved). And even if the linear equation had a slope of 2, that would not justify dividing by the square of height--linearity is direct proportionality.
No, the linear relationship is between log-transformed height and weight. And the empirical slope of that relationship is between 2 and 3--and usually closer to 3--for pretty much all animals.

As for drug-dosing, perhaps the idea of using calculated surface area (instead of weight) is a ham-handed attempt to account for metabolic rate (which scales similarly to surface area as weight ^(0.7) or so?

By Sven DiMilo (not verified) on 12 Mar 2010 #permalink

The trouble is that although "you'd be healthier if you were shorter" may be true, it doesn't imply that "you should work on being shorter" is good advice.

More relevantly, it doesn't imply "you should work on losing weight" is good advice either. I'm perhaps likely to be less healthy than a shorter guy with the same mass/height^3 as me, but it doesn't follow I should try and slim down to his BMI.

Anyway, it ought be fairly easy to do a study to determine which exponent a leaves mass/height^a the best predictor of body fat percentage. Has nobody done so?

By Andreas Johansson (not verified) on 14 Mar 2010 #permalink

The trouble is that although "you'd be healthier if you were shorter" may be true, it doesn't imply that "you should work on being shorter" is good advice.

More relevantly, it doesn't imply "you should work on losing weight" is good advice either. I'm perhaps likely to be less healthy than a shorter guy with the same mass/height^3 as me, but it doesn't follow I should try and slim down to his BMI.

Anyway, it ought be fairly easy to do a study to determine which exponent a leaves mass/height^a the best predictor of body fat percentage. Has nobody done so?

By Andreas Johansson (not verified) on 14 Mar 2010 #permalink

Well, at some point recently I had to obtain a reasonable weight for someone 8' tall, and solved the problem by going to the biological literature. The most accurate exponent for the allometry of height vs. weight in bipeds was not 2, nor was it 3; I think it was around 2.2 or 2.3. Just FWIW. (If I can recalculate what I got, I'll post it later.)

This makes weight loss terribly easy: eat fewer and expend more calories.

Well, like most things in life, it's actually a little more complicated than that. Clearly if it were that easy, there would be no fat people. There are two problems with your reasoning:

1. When you decrease your food intake, your body doesn't always behave nicely by burning fat. While I don't really buy into that "lowered metabolism" thing, it is easy to become tired and weak if you don't eat enough. If you're too tired to exercise or move around, your energy output will decrease to match your input. I went on a diet and lost a lot of weight fairly easily, but when I was 10 pounds away from a healthy BMI, I suddenly stopped losing weight and I became too weak to manage anything other than daily functioning. I nearly fainted several times, which I have never done before in my entire life. I'm sure there are ways around this problem, like maybe the proportion of different macronutrients, but simply reducing my calorie intake further was not an option at that time.

2. People get hungry. Sure, plenty of people eat when they're not actually hungry. But there's a limit to how much people can restrict their food intake before they just go crazy from hunger. It's not really about willpower, but something that we all feel every day. Telling people to just ignore their hunger is about as effective as telling a kid to just not scratch his chicken pox.

By bananacat (not verified) on 16 Mar 2010 #permalink

Science is extremely successful at solving society's "mechanical problems" and yet it is stymied by social "problems" of this nature. We "know" more about the human body, nutrition, psychology, etc than ever in history and we "know" more about the "causes" of obesity and yet it's interesting that we are still moving in the same direction overall. No matter how advanced we become in understanding anything "about" the world through science and all the research that could possibly be performed, the whole idea of "cause and effect" is a tremendous limitation within human consciousness when applied to solving social "problems," and until it is transcended, we will continue to bang our heads against the wall, digging and digging to "know more about" or "find" the myriad of answers when its simplicity is standing right before us. Society loves to declare "war" on it's problems (the "war on obesity," the "war on drugs," etc) when time and time again they always seem to multiply. All deserve respect that have the authentic intention of helping society make strides in a positive direction, however it is the CONTEXT of our intention(s) that will determine the POWER (not force) available to transform that which we perceive as a "problem" in the world.

Hi friends,
Your style of presentation is very impressive. The meaningful contribution of your mind reflects on those people who are looking for Body Mass Index (BMI). I would like to tweet on it and keep spying at every moment you blogging.

We never use the BMI in our gym, we like to track weight and body fat percentage. It gives us a better indication of how much fat a client will lose and how much muscle they put on.

While I acknowledge that many Americans, myself included, are overweight or obese, I am becoming increasingly alarmed in how prevalent the overly simplistic and flawed BMI is being used as an undisputed health measurement for all Americans. Please consider the following:

On June 17 1998, CNN reported "Millions of Americans became "fat" Wednesday -- even if they didn't gain a pound -- as the federal government adopted a controversial method for determining who is considered overweight." Basically, by adopting the BMI, the federal government lowered the acceptable standards for being overweight by 10 pounds per height category. This move inflated the number of overweight people and strengthened the governmentâs case to start the war on obesity.

The BMI does not work for all people. I am a 5'10" non-athletic male and, according to the BMI, my ideal body weight is between 128 and 173 pounds. Everyone who is 5â10â has to be in this 45 pound box regardless of age, gender, muscle mass, bone structure... and I will never fit into it. My ideal body weight of about 215 pounds is considered obese per the BMI. I refuse to be defined by this flawed metric.

My biggest concern with the BMI is that it is now being used in legislation. On July 15 2010, CNS News reported âNew federal regulations issued this week stipulate that the electronic health records--that all Americans are supposed to have by 2014 under the terms of the stimulus law that President Barack Obama signed last year--must record not only the traditional measures of height and weight, but also the Body Mass Index: a measure of obesity.â Additionally, the Healthy Choices Act, HR 5209, proposes using the BMI as a tool to gather data, track anti-obesity efforts and counsel non-compliant individuals.

The logical next step to the BMI being legislated is penalties for non compliance. What happens if individuals cannot or chose not to comply? Are they going to be charged higher health insurance premiums, denied care or forced to make dietary and physical activity changes? Since I cannot fit inside the 45 box even at my ideal body weight, I am worried that I will be penalized for not meeting the BMI standards.

This level of government intrusion in our lives is unacceptable and must be opposed, especially because it is based on faulty science and assumptions.

Summit County Government has followed Whole Foods and other employers penalizing workers based on BMI. Summit County, Colorado, requires county employees to achieve BMI 24.9; BP 120/80; LDL 140 in order to qualify for up to $500 in co-payment for health care costs.

I ride my bicycle 16 miles a day to commute for work, about 2,400 miles a year. On my days off work, I hike at least 10 miles per week. This summer, I have already ascended 24 Fourteeners, hiking about 20 miles with vertical of 10,000 feet per week.

My VO2 Max is in the superior range. However, like many Olympic medalists and other athletes, I am borderline or fail the faulty "health-based criteria" established by my employer for the wellness program award.

A study of 250,000 patients by Mayo Clinic found that BMI is a poor predictor of health. An evaluation of more than 40,000 GM workers by Chang, et. al., found that health care costs to not begin to increase until BMI is over 27 or under 19. Evidence does not persuade my employer to end this discriminatory and irrational program.

By Kim Fenske (not verified) on 07 Sep 2010 #permalink

Science is extremely successful at solving society's "mechanical problems" and yet it is stymied by social "problems" of this nature. We "know" more about the human body, nutrition, psychology, etc than ever in history and we "know" more about the "causes" of obesity and yet it's interesting that we are still moving in the same direction overall. No matter how advanced we become in understanding anything "about" the world through science and all the research that could possibly be performed, the whole idea of "cause and effect" is a tremendous limitation within human consciousness when applied to solving social "problems," and until it is transcended, we will continue to bang our heads against the wall, digging and digging to "know more about" or "find" the myriad of answers when its simplicity is standing right before us. Society loves to declare "war" on it's problems (the "war on obesity," the "war on drugs," etc) when time and time again they always seem to multiply. All deserve respect that have the authentic intention of helping society make strides in a positive direction, however it is the CONTEXT of our intention(s) that will determine the POWER (not force) available to transform that which we perceive as a "problem" in the world.

The BMI is nonsense. The human body is 3-d not 2-d like BMI.
If the BMI is calculated for 3-d only then it would be right.