Tuesday, July 29, 2014

The BMI ('Body Mass Index') - As Ridiculous as It Is Unscientific

No automatic alt text available.
In one of the Sunday magazines, we behold a brief article (thankfully):” ‘What’s Your BMI and Why Should You Care?  In the lead paragraph ‘the Doctors’ write:

The BMI (body mass index) is a good indicator of how much body fat you have. Health professionals use it to screen for weight problems in adults.

They do add that “it doesn’t paint a full picture of your health” and that’s an understatement. As noted in a Penn & Teller ‘Bullshit’ episode,  lampooning BMI and the whole “obesity is an epidemic” baloney, both Michael Jordan and Brad Pitt would be overweight, and Russell Crowe and George Clooney would be “obese” on the BMI index scale.

Apart from such whacked out nonsense, as one Univ. of Virginia prof quoted in the segment observed:

Another problem with the government using BMI is that it says everyone needs to be a certain weight within a certain height range in order to be healthy.”

But this disdains the range of variations for most humans pertaining to a host of attributes. It mandates that only a certain human height-weight body profile is acceptable while labeling the outliers “unhealthy” or “obese” or “overweight”. Using this bogus index we’ve actually come to believe “one third of Americans are obese” – based on having a BMI of 30 or higher.

But this is nonsense!

As Penn & Teller observed it was a Belgian polymath,  Adolphe Quetelet who devised the  BMI formula in 1832 in his quest to define the "normal man" in terms of everything from his average arm strength to the age at which he marries.  Obviously and clearly, his numerical basis would be irrelevant to today given the “normal man” ca. 1830s Belgium would not be in an way comparable to the normal man today - especially in the US of A. His diet would be more frugal, less protein for one thing as well as fewer nutrients,  and hence he’d naturally bear more a resemblance to reed-thin Stan Laurel than George Clooney, or Russell Crowe.

So his project had nothing to do with obesity-related diseases, nor even with obesity itself. Rather, Quetelet used the formula to attempt to describe the standard proportions of the human build—the ratio of weight to height in the average adult- in that reduced nutrition era. Using data collected from several hundred countrymen, he found that weight varied not in direct proportion to height (such that, say, people 10 percent taller than average were 10 percent heavier, too) but in proportion to the square of height. (People 10 percent taller than average tended to be about 21 percent heavier.)

The new formula had little impact among the medical community until long after Quetelet's death. While doctors had suspected the ill effects of obesity as far back as the 18th century, their evidence was purely anecdotal. The first large-scale studies of obesity and health were conducted in the early 20th century, when insurance companies began using comparisons of height and weight among their policyholders to show that "overweight" people died earlier than those of "ideal" weight. Subsequent actuarial and medical studies found that obese people were also were more likely to get diabetes, hypertension, and heart disease. (Of course, this  later allowed the medical insurers to either invoke "pre-existing conditions" to bar people from coverage or, more often, have an excuse to increase their premiums.)

By the early 1900s, it was fairly well-established that these ailments were the result of having too much adipose tissue—so the studies used functions of height and weight as little more than a proxy for determining how much excess body fat people had.  The problem with proxies, of course, is that they are not direct quantifiers or indicators and are only so good as the physical basis really allows.

It would actually have been more accurate for the actuaries to compare longevity data with more direct assessments of body fat—such as caliper-measured skinfold thickness or hydrostatic weighing. But these data were much harder for them to obtain than standard information on height, weight, and sex.  So they punted!

Medical researchers , meanwhile, needed a standard measure of fatness, so they could look at the health outcomes of varying degrees of obesity across an entire population. For decades doctors couldn't agree on the best formula for combining height and weight into a single number—some used weight divided by height; others used weight divided by height cubed. It arrived in 1972, when physiology professor and obesity researcher Ancel Keys published his "Indices of Relative Weight and Obesity," a statistical study of more than 7,400 men in five countries. Keys examined which of the height-weight formulas matched up best with each subject's body-fat percentage, as measured more directly. He concluded that the best predictor came from Quetelet’s BMI: weight divided by height squared. Keys renamed this number the body mass index.

But this was decidedly premature.

A recent critique (in PDF) of the body mass index in the journal Circulation suggests that BMI's imprecision and publicity-friendly cutoffs distort even the large epidemiological studies. (For example, there's no definitive count of how many people are misclassified by BMI, but several studies have suggested that the error rate is significant for people of certain ages and ethnicities. That old natural variation bugbear again!) It's impossible to know which studies have been affected and in what direction they might have been skewed.

Further, the BMI is actually a solid example of the “proofiness” that Charles Seife referenced in his book,
Proofiness: How You're Being Fooled by the Numbers.

Seife decries the tactic of using numbers not just to lie but to baffle the susceptible with bullshit.  He refers to a common failing of most people unversed in math to be hoodwinked merely because some form of math or numbers are interjected into arguments.  Not just using numbers to bolster one's argument. In his words, to use fake numbers to prove falsehoods and to seek to prove something is true - even when it's not- is one of the most egregious forms of  intellectual  fraud.

In this regard, one of the surest signs of proofiness is the failure to provide attached uncertainties to the measurements - any measurements! Since BMI is always recorded as an absolute single number, say 29, and never as 29 + 2 or whatever, then it is inherently proofy - a bogus quantity. Seife emphasizes there can NEVER be a 100 percent accurate number if based on physical measurements, and he's right. Maybe the scale used is off by a pound or two, and maybe the height isn't evaluated for the associated probable error - based on the instrument used to measure it.  OR......maybe, just maybe the presumed cutoffs along the BMI chart indices have been majorly distorted by earlier misclassifications in large epidemiological studies. 

The BMI also takes this to new level because the combination of the 2 quantifiers make no sense. I mean the ratio of  weight in pounds to height in inches squared? And then multiplying by 703? That’s pure baloney and in no way even comparable to say obtaining metric mass by dividing the weight (in newtons) by the acceleration of gravity in N/kg.
Where does 703 come from anyway? Well it’s the correction factor introduced if one used Imperial units (foot, pounds) in stead of metric system. In the metric system the BMI is simply:

Mass (kg)/ [height (m)]2

Again, this is bollocks, since the result  (mass per unit area) yields no conceptually consistent physical quantity as applied to human biology!  It’s fully an example of more proofiness: In this case putting two unrelated units together in a ratio and making people believe the result (in kg/m2) has some innate core physical meaning. It doesn’t. It’s bullshit. (As Penn and Teller also pointed out in their show on “Obesity”.)

The medical -industrial -insurance- PhrMA whackos will try to tell you the ratio is valid because height and weight "are related", but this is a presumption unwarranted by the total constellation of data- especially applied to distinct ethnic groups. Also, if one investigates the fundamental units of physics that comprise it, s/he will find no such equivalent anywhere. (Which can also be deduced by using the basic SI units in various combinations.)

The closest one can come is the combination of units:

kg m -3    Which is mass divided by the length cubed or M L -3

This yields what we call “density”.   And at least the use of density would make some physical sense, but the ratio for BMI makes zero sense, because no comparable physical quantity in terms of mass per unit area exists for human bodies. It make no difference how many idiotic trials were used to attempt to validate it in the health sense. I could as well take the Martian mass in kg, and divide it by its assorted  Earth opposition values in meters to show that UFO sightings increase whenever the ratio approaches a certain value (say 1013 kg m-1.)  It is pure nonsense, and any “findings” add up to little more than lucky coincidence.

The use of this dumb obesity quantifier is even more enraging given there’s at least a more rational alternative. It turns out that the circumference around a person's waist provides a much more accurate reading of his or her abdominal fat and risk for disease than BMI. One unit, no hocus pocus. Simple. Besides, wrapping a tape measure around your belly is no more expensive than hopping on a scale and standing in front of a ruler. That's why the American Society for Nutrition, the American Diabetes Association, and other prominent medical groups have lately promoted waist circumference as a replacement for, the body mass index. (Some have indicated as a “supplement” but why waste time with proofy contrived numbers at all?)

Alas, few doctors  - including our own - have made the switch. This is probably because waist measurements require slightly more time and training to interpret  than it takes to record a BMI reading and use some fake out chart, which doesn’t come with any “official cutoffs”. (Right now, my BMI is 29.5 but I laugh when anyone says I am “over weight” for the reasons given above, especially the proofiness of the index and nonsensical units.)  The sensitivity of doctors to these slight inconveniences signals just how difficult it will be to unseat Quetelet's antiquated and irrational, proofy formula. See, the body mass index is cheap and easy to get (never mind the absence of uncertainty), and it has the incumbent advantage in that the Lords from On High in Health Central have conferred their benediction – along with the political-Pharma –lobby enclave – so who’s going to argue with them? Well, I am!

Sadly, just like tea leaves, natal horoscopes and palm reading, BMI is here to stay—despite its flaws – the chief of which is that it’s irrational and has no bearing to any real physical quantity (as the examination of its units discloses)

But that doesn’t mean I have to treat it any more seriously than other monkey fool bollocks, including horoscopes, palm reading and tarot cards.

 As for rationality in terms of obesity these are the key obesity- BMI  myths you need to know, summarized from Prof. Paul Campos in his book, 'The Obesity Myth':
1)  Weight is a good proxy for health ("97 percent false" according to Campos)
2) Health improvement comes via transition from being fat to thin. (Hardly ever, for most people - especially the elderly who are more at risk if they become frail.)
3) We know how to produce long term weight loss.
In respect of the last, Prof. Campos makes it clear that despite the bloviations of the medical-industrial-insurance complex and the government health brigade as well as the health diet faddists, no one really has a clue how to sustain long term weight loss. Yes, they say, "balance intake of food with exercise" -but if a lot of weight gain is traced to gut bacteria this is  a non-starter. 

Then there is the leptin factor, which hormone levels decrease when people don't get enough sleep - and which causes them to eat inordinately. As explained on one medical site (WebMD):

"When you don't get enough sleep, it drives leptin levels down, which means you don't feel as satisfied after you eat. Lack of sleep also causes ghrelin levels to rise, which means your appetite is stimulated, so you want more food,"

 The sad fact is too many overworked and  tech over- connected Americans are in this latter category. Dieting won't help them but getting on a regular, decent sleep schedule might!
Stay tuned, and in the meantime don't get hysterical over your BMI!

No comments: