Why BMI is B.S.

Confession: I'm overweight. Clarification: I'm overweight according to a flawed metric. 

I remember the resentment I felt during a physical evaluation for life insurance a few years ago: just before stealing what felt like a few gallons of blood from my arm (I survived, but barely), the nurse hurriedly measured my height and weight. 
"Hmm, weight, looks like 198, height, just over 5'11. Okey doke!" and that was the full extent of body composition measurements taken. 
As she wrote down the numbers on her sheet and I envisioned my monthly insurance rates ticking upwards because of how "overweight" I was, I wanted to say, "ahem, so, wanna see my abs? I'm pretty fit." ...but that might have made things a bit weird. 
See, I did not have, nor have I ever had, a high percentage of body fat (what the BMI purports to indirectly measure). Yet my BMI came in at 27.6 on that day, squarely in the "overweight" zone.

The problem is, though a higher BMI certainly can (and frequently does) correlate with higher body fat, it has the potential to give woefully inaccurate measures of actual body fatness. And yet, this cursory (at best) measurement of height and weight is par for the course. Yes, in more specialized settings there will certainly be more body composition measurements taken; but for most of you, your tailor probably knows more about your body composition and implicated health risks than your doctor or health screener does. And that's pretty bogus, given that BMI is used for things like published global, national, and local health statistics, insurance rates, general health screenings, prediction of health risks, medical advice, and more. 

So what's wrong with BMI? Why do I take issue with its widespread use? Let me explain.

1. First, a bit of history...
The formula for BMI was first introduced in the early-mid 1800s by a mathematician, statistician, astronomer, and sociologist named Adolphe Quetelet, and the term BMI (body mass index) was popularized in the mid-late 1900s by Ancel Keys, who determined BMI to be the preferred measurement for body composition in population studies (at least, when compared to another arbitrary and oversimplified measurement). Quetelet (who, I will reiterate, was not a medical professional nor a physiologist) himself warned that his method was not perfect, and that it was more of an estimation based on trends in body composition at that time and place; and Keys determined BMI to be satisfactory for population studies, noting that it should not be used to assess individuals. And yet, here we are. 

So, our most widely used body composition metric today is a 200-year-old quick and easy "eh, it works out in a pinch" formula developed by a mathematician. Seems legit.

2. It can be very, very inaccurate. 
Here are some pictures of overweight people, according to BMI: 

Lebron James / BMI: 27.4 / Overweight
Just look at that chubby face! ¹

Aneta Florczyk / BMI: 27.5 / Overweight
Go on, say it to her face, I dare you. ²

Rich Froning / BMI: 29.5 / Overweight (almost obese)
Abs, fat folds, same difference. ³

Okay, how about some obese people. They're bound to be at least a little chubby, right?

Dwayne Johnson / BMI: 30.8 / Obese
Soft and jiggly like a rock (get it?). ⁴

Brock Lesnar / BMI: 36.9 / Obese
He heard what you said. ⁵

Ronnie Coleman / BMI: 41.4 / Morbidly Obese
8x Mr. Olympia is morbidly obese. Sure. ⁶

So, in case the point is not yet abundantly clear, BMI has the potential to give very, very misleading results. Athletes, particularly strength and power athletes, tend to have higher bone density and much greater muscle mass than sedentary populations, and therefore have BMIs that suggest much greater body fatness than they actually have. 

So what, some athletes have misleading BMIs and pay a little more for life insurance and skew statistics a bit... But most of us aren't professional athletes anyways—what's the big deal?

Well, first off, this doesn't apply just to a few pro athletes. I'm about as average as they come athletically, and as mentioned above, I am classified as overweight. With the rising popularity of sports like weightlifting, powerlifting, and Crossfit in otherwise "ordinary" populations, more people are packing on extra muscle to their frame and becoming "overweight" and "obese" as they exercise more. Yes, exercise might make you "overweight" or "obese"—you heard it here first.

In addition, it's not just the more muscular crowd that's affected. BMI can be incredibly inaccurate in the other direction as well. BMI measurements are very likely to underestimate body fatness. The New York University School of Medicine conducted a cross-sectional study in which they compared BMI measurements with the results from a DEXA scan (which is arguably the most accurate method of measuring body composition) in a group of adults. The study found that while BMI classified 26% of the participants as obese, 64% of the subjects had obese levels of body fatness according to the DEXA scan. In all, 39% of subjects who were not considered obese according to BMI were found to be obese according to the DEXA scan. In addition, the BMI measurements were considerably more likely to misclassify women than men (48% vs 25% misclassification). Additional studies have shown similarly dismal results for the accuracy of BMI in correlating to actual body fatness and predicting disease. 

Now, look around the internet and see just how many population health studies and statistics are based on BMI (I'll just tell you: it's almost all of them). All of those studies and statistics could be immensely inaccurate and the implications misleading. Example: according to the CDC, just over 1/3 of the American population is classified as obese. What if that number was more like 60-70%? If the inconsistencies found in the NYU study are any indication, this may be closer to the truth.

To make matters worse, studies have shown that being "overweight" according to BMI was not associated with increased mortality, and another demonstrated that among patients with coronary heart disease, those who were underweight or normal weight had the highest risk for mortality, while those who were overweight had the lowest risk for mortality, and moderately obese patients had no increased risk for total mortality. (Please note that this does not mean we should all try to get fatter because that's healthier. These studies merely suggest that BMI alone does not accurately predict body fatness and therefore disease and mortality risks.)

False positives on the one hand, false negatives on the other hand, false implications about health risks in both hands—regardless of how you look at it, BMI doesn't inspire much confidence. 

3. It ignores most of the factors that affect body weight and composition.
BMI assumes a relatively equal body type for equal heights and weights. But look around next time you're in public and note the massive variability in body type. Two people who are built similarly don't necessarily weigh the same, and two people who weigh the same aren't necessarily built the same. When looking at how BMI is measured, you have to ask: What about people with higher or lower bone mass (bone is about twice as dense as fat)? How about those with higher or lower muscle mass (muscle is about 5/4ths as dense as fat)? What about variations in frame size? Body type differences by gender? Inherent variations in healthy levels of body fat by gender? Variations in body shape and proportions that come with age? Waist size? What about amputees? BMI ignores all of these factors. 

4. It's arbitrary.
Weight (kg) divided by height (m) squared. That's the formula for BMI. What's the magic of weight divided by the square of height? Well, nothing. Quetelet looked at trends in body weight/height and body composition and constructed a formula that very roughly correlated. In addition, the rigidity and the arbitrary nature of this formula (because squaring the height really has no real-world basis) means that BMI tends to overestimate the body fatness of taller individuals and underestimate the body fatness of shorter individuals. Nick Trefethen, an Oxford mathematics professor, has suggested a formula of 1.3*weight (kg) divided by height (m) raised to the 2.5th power as a more accurate measure than the traditional BMI formula, though he recognizes the absurdity of doggedly sticking to overly-simplistic metrics for complex things like body composition, and proposes it only as a somewhat better alternative to an inherently flawed metric. 

Most arbitrary of all is the rigid categorization of "underweight," "normal weight," "overweight," and "obese." What makes someone suddenly obese (and subject to all the risks of obesity) when the decimal point of their BMI ticks from 29.9 to 30.0? In addition, these numbers have varied considerably over the years as different governing bodies have changed what they think qualifies as "underweight," "normal weight," etc. The absurdity of this all should be evident by now. But apparently the CDC, insurance companies, and much of the medical community don't give a damn about that. We sent a man to the moon almost 50 years ago, and yet we're still cool with using a 200-year-old, over-simplified, and arbitrary metric as our primary method for measuring body composition and predicting health? 

5. It rests on assumptions about body type.
Put simply, BMI assumes a particular body type for everyone (specifically, an unfit and sedentary body type). Not only does body type vary immensely from person-to-person, but it also varies with geographical location and over time. To give credit where it's due, some national and international organizations do try to adjust for this by setting their own cut-off points; however, these distinctions are often arbitrary themselves, and have the downside of muddling the results of long-term longitudinal studies or global comparisons. 

6. It's a statistical, population-based measurement that we consistently use to assess individuals.
Need I say more?

7. There are a number of more accurate measures of body composition and predictors of health complications.
At this point, there really is no reason to use BMI. Yes, it's convenient and easy and you can make the kids in a high school health class memorize the formula. But we have countless other methods that do their job far better than BMI. Of course we have precise metrics such as DEXA scans and lipid profiles that can give accurate and detailed assessments of numerous health factors. However, to get an accurate assessments of body composition and health, we don't always have to lean on high-tech, high-cost procedures. I won't get into all of the alternative methods here (there are plenty), but there's one that stands out in particular. Remember how I mentioned your tailor probably knowing more about your body composition and health risks than your health screener or doctor? That's because numerous studies have demonstrated a much higher accuracy of waist circumference, waist-to-height ratio, and waist-to-hip ratio in the prediction of all-cause mortality, disease risk, etc. These findings are significant considering how easy it is to measure waist and/or hip circumference. Using waist-to-height ratio or a similar measurement in health screenings, medical check-ups, and statistical analyses requires virtually no additional cost, training, or time, and yet it would vastly improve the accuracy of our population health studies and individual health assessments. 



It's time to say bye-bye BMI, and I'm certainly not the only one to be saying this. Countless doctors, health professionals, health organizations, and reasonable people are realizing the massive shortcomings of using BMI as our standard of body composition and health risk assessment. Perhaps someday the global health organizations, standard-setters, and governing bodies will catch up with the blog-osphere and change their ways. 

---

Be sure to subscribe, share, and comment.
Follow me on Twitter @PSprimont and on Instagram @PSprimont

Strong of heart, strong of mind, strong of swole.

---

1 By Steve Jurvetson (Flickr: LeBron James) [CC BY 2.0 (http://creativecommons.org/licenses/by/2.0)], via Wikimedia Commons^↩
2 By Artur Andrzej (Own work) [CC BY-SA 3.0 (http://creativecommons.org/licenses/by-sa/3.0) or GFDL (http://www.gnu.org/copyleft/fdl.html)], via Wikimedia Commons^↩
3 By Lance Cpl. Derrick K. Irions [Public domain], via Wikimedia Commons^↩
4 (Wikipedia:Contact us/Photo submission) [CC BY-SA 3.0 (http://creativecommons.org/licenses/by-sa/3.0)], via Wikimedia Commons^↩
5 By Miguel Discart [CC BY-SA 2.0 (http://creativecommons.org/licenses/by-sa/2.0)], via Wikimedia Commons^↩
6 www.localfitness.com.au. [Attribution, CC BY-SA 3.0 (http://creativecommons.org/licenses/by-sa/3.0) or GFDL (http://www.gnu.org/copyleft/fdl.html)], via Wikimedia Commons^↩