At least on the internets, a lot is made about how body mass index (BMI) is a poor measure of fatness. While this seems true in an academic sense, in a practical sense, I’d be shocked of most of the people with BMIs of 30+ felt happy with their physical fitness. The amount of lean muscle mass necessary to nudge oneself over to the obese category is not a trivial matter. These masters of rhetoric make it seem like if you’re a moderately active person that there’s a real risk that your BMI will poorly reflect your physical makeup. To get a sense of how skewed a measurement BMI is for measuring physical composition, I analyzed the height and weight measurements of the athletes on the rosters of the four major sports leagues in 2010 (NFL, NBA, MLB, NHL). The general results do not entirely support a dissociation of BMI from an intuitive grasp of physical makeup.

League Average BMI
NFL 31.6[1]
NBA 25.1
MLB 26.7
NHL 26.6

While it is true that the average BMI for the average player in every professional sports league was in the overweight category, a closer inspection of the data supports our intuitions. The NFL’s average is in the obese category, but there is significant variation in the BMIs of each position. If you follow the sport, you would guess the linemen would likely have much higher BMIs than say the kickers. On the offensive and defensive line, extra weight is advantageous, not detrimental. To be sure, these players need to maintain a minimum level of quickness. Elite offensive tackles have 40-yard dash times in the low to mid 5 second range. To put this in perspective, the extrapolated 100 meter time of a 5.25s 40 would be 14.18s (and yes, I realize this is not physiologically valid). This isn’t too shabby for a 300+ pound guy. This is exactly what the data shows.

Likewise, we should expect the positions with lower BMIs to be those that place a premium on speed or lack an incentive for being large. The positions with the lowest BMIs are: wide receiver, cornerback, defensive back, kick returner, free safety, punter, safety, kicker, quarterback, and strong safety. Seven “speed” positions, the quarterback and two kickers.

It shouldn’t be a surprise that Kris Jenkins, 6’4″ and 360 pounds, and Chad Ochocinco, 6’1″ and 192lbs, look very different. While Kris Jenkins is very accomplished at his position, his body isn’t aesthetically pleasing. It is a body constructed to fit into a mold that requires both massive amounts of strength and sheer size. Using online calculators, we can estimate he needs to take in over 5,000 calories per day purely to maintain his current size. His size isn’t an accident. He has worked very hard to consume the necessary calories to add mass (both muscular and fat) while working out extremely hard.

For the NBA, the story is similar. Centers have the highest average BMIs and guards are in the lowest group, though there is less variability in basketball due to the necessary cardiovascular endurance and minimum levels of speed that are generally required. The player level data also isn’t surprising. Shaquille O’Neal (7’1″, 325lbs and a BMI of 31.6) has the highest BMI; is a player infamous for his struggles with his weight. Whereas, Corey Brewer (6’9″, 188lbs, BMI of 20.144) is the leanest.

It is noteworthy that only five (1%) NBA players have BMIs that would be classified as obese: Shaquille O’Neal, Dexter Pittman, Garret Siler, Glen Davis, DeJuan Blair vs. 56% of the NFL[2] vs. 7% of the MLB and 1% of the NHL. The notion that there is a vast reserve of “obese” athletes in the professional ranks is simply false. Yes, many of the athletes at the pro level have BMIs that would be classified as overweight, but even the staunchest supporters of BMI would be comfortable conceding this.

To the critic of BMI, this is all well and good but not particularly impressive. Sure, they’d say, but we still claim that BMI doesn’t mean anything. It was created by a Belgian mathematician over 150 years ago. Our bodies have changed. Our understanding of what causes disease has changed. Why are we clinging to such an antiquated measure?

A study[3] done by Abel Romero-Corral and colleagues used data collected in the NHANES III (National Health and Nutrition Examination Survey) to determine the ability of BMI to accurately label individuals as being overweight or obese. To achieve this, they used a subset of the total sample who submitted to bioelectrical impedance analysis, which is a more accurate method of measuring body fat[4]. BMI had a high specificity, which means that a high percentage of the individuals labeled as obese by BMI were also labeled obese by the body fat measure. 96% of men and 99% of women who were labeled obese by BMI were also considered obese by their body fat percentage.

Unfortunately, BMI also had low sensitivity, which means that a decent percentage of the individuals that were not labeled obese by their BMI were labeled obese by their body fat percentage. Only 36% of men and 49% of women who had obese body fat percentages were labeled as such by their BMI. This finding indicates that if your BMI is =30, then you could stand to lose a few pounds. But if your BMI is <30 then get too cocky and think you’re in the clear[5].

As with all measures, BMI is merely a tool. It can do some things well and imperfectly do others. BMI is a quick and dirty method for determining the health of an individual. Sure, it does so imperfectly. Sure, using BMI when there are far better methods available ought to receive our ire. But it still has value.

As an aside, an astute observer will notice that the correlations are weaker for men than women. That is because in men, BMI’s variance was more strongly due to lean mass than it was with body fat percentage (r2s of 0.53 and 0.44 respectively). In women, body fat percentage fared much better with it explaining 71% of the variance vs. lean mass’s 55%. This is likely due to a combination of it being much more socially acceptable for men to add muscle mass and it being physiologically easier for men to do so, plus women’s body types seem to have a higher variance in the amount of fat that their body type requires.

[1]As a technical note, working with height entered in as Feet’Inches” is a pain. Joshua Levine’s InchCalc Excel add in was a life saver, though I cannot vouch that it hasn’t infected my machine.

[2]Dexter Pittman’s BMI can be considered a success story since he has lost almost 100 lbs since graduating high school. He would not have been an NBA prospect if he weighed in in the mid to upper 300s. Glen Davis himself has lost nearly 20 pounds since turning pro.

[3]Romero-Corral, A., Somers, V.K., Sierra-Johnson, J., Thomas, R.J., Bailey, K.R., et al. (2008). Accuracy of body mass index to diagnose obesity in the US adult population. International Journal of Obesity, 32(6), 959-966.

[4]It isn’t nearly as accurate as an expertly administered skinfold test or hydrostatic analysis, but even critics of BMI would agree that it ought to be more accurate than BMI.

[5]This is the sort of finding I could see warped into a misleading ad campaign for a gym since it could be interpreted as saying that regardless of your BMI, you should be in the gym.