One of the most common sources of discrimination is the assumption that the way people tend to be – i.e., the statistical majority – is the way any one person ought to be.
Sometimes the majority is right, of course. Most people are kind to children, and you should be too. But the majority is wrong sometimes, and sometimes it's just a majority – there aren't any "shoulds" or "shouldn'ts" to infer. For instance, the majority of American adults identify as heterosexual (86.7% in a 2021 Gallup poll), but that tells you nothing about how an individual should identify, nor does it justify discrimination against the LGBTQ community.
In this newsletter I'll be talking about a well-known statistic called the Body Mass Index, or BMI. The BMI was invented in the 1830s to describe something about how populations tend to be, but it's widely used to describe how individuals ought to be.
Although the BMI is useful in some contexts, experts recognize that applying it to individuals causes substantial harm. Last week, for example, MedPage Today published an article by Harvard professors Bryn Austin and Tracy Richmond entitled "It's time to retire BMI as a clinical metric." What struck me about this article is that many of the concerns raised are the same ones experts have been pointing out for decades, and yet the BMI continues to be widely used in medical settings, in the health and fitness industry, and in research. So, I want to address four questions here: What is the BMI? In what sense is the BMI flawed? Why are its flaws harmful? And, what are the alternatives?
1. What is the BMI?
The BMI is a way of describing the relationship between a person's height and weight.
The plain English interpretation of BMI varies from expert to expert. Some refer to it as an estimate of body fat. Others present it as an indicator of whether or not a person's weight is healthy. Regardless of how it's characterized, the underlying math is the same. Specifically, in the metric system (height in meters, weight in kilograms), the standard formula is:
In other words, BMI equals weight divided by height squared. But don't reach for a calculator just yet. In the English system (height in inches, weight in pounds), a conversion factor must be added:
Thus, if you're 5 feet 4 inches tall, and you weigh 140 pounds, your BMI would be: 140/(64 x 64) x 703 = 24.03.
What do numbers like this tell us? Typically, clinicians and even researchers only use BMI scores for one thing: To assign individuals to weight categories. For adults in the U.S., and in many countries, the following categories are used:
Underweight: BMI less than 18.5.
Normal: BMI ranging from 18.5 to 24.9.
Overweight: BMI ranging from 25 to 29.9.
Obese: BMI 30 or more.
Different categories may be used for certain populations, and organizations like the WHO add further distinctions, but the breakdown you see above is the industry standard. It's quite an "industry". As Drs. Austin and Richmond note in their MedPage Today article,
"BMI is assessed at nearly every touchpoint in primary and specialty care. BMIs are often included in electronic health record work flows even when not relevant. Payors and clinic or hospital quality-improvement policies incentivize clinicians to assess, categorize, and intervene upon BMI at most encounters."
Apart from medical use, people are encouraged to track their BMI by major health organizations, including the WHO, the CDC, and the NIH, as well as by gyms, wellness programs, websites, and other players in the health and fitness industry. Some employers even incentivize their employees to meet BMI goals – Whole Foods, for instance, offers employee discounts for BMI reductions. Meanwhile, you can purchase BMI calculators on Amazon, or just go to any one of the countless web pages that calculate your BMI for you.
2. In what sense is the BMI flawed?
The BMI was created in the early 1830s by a Belgian polymath named Adolphe Quetelet. Quetelet wished to discover the characteristics of the "average man" (a phrase he invented), because he believed that when large groups of people are measured, the average values capture what's most essential about our species.
Quetelet baked moral evaluation into his statistics. He treated the average value of a characteristic as a sort of ideal. Any substantial deviation from that value, unless it stems from measurement error, indicates that the individual is "abnormal" (in a negative sense).
Here we see, in the realm of statistics, the origin of the idea that the way people tend to be is the way an individual ought to be. Nowadays most of us would disagree with this idea, at least as a general rule.
Quetelet studied height and weight separately, but he also noticed that when calculating what we now call the BMI, the averages are consistent across large groups of people. (Quetelet invented the formula, but it wasn't called the "body mass index” until that term was used in an influential 1972 paper.)
Although Quetelet considered an average BMI to be superior to values above or below average, he didn't indicate anything about how much deviation in either direction would be problematic. He said nothing about individuals, because they weren't his focus. Rather, he gathered population-level statistics and wrote about Man. It was only in the 1970s, in the hopes of identifying weight problems more accurately, that the BMI began to be used for evaluating individuals. (The researchers whose 1972 paper launched the BMI "industry" cautioned that it's a population statistic and not suited for individual use, but their warning was largely ignored.)
One of the reasons the BMI doesn't work well on the individual level is that it's a very imprecise measurement of body fat. Here's why:
(a) The BMI doesn't distinguish between fat and muscle.
Muscle is roughly 18% denser than fat. One consequence of this difference is that the BMI misclassifies relatively muscular people as overweight or obese. For similar reasons, the BMI underestimates body fat among people with relatively low muscle and bone density.
(b) The BMI formula is biased for extremes of height.
Quetelet used an exponent of 2 in the denominator of his formula, because that enabled the best fit with the data. His formula remains the standard one. And yet, given scaling differences between changes in height and weight, the BMI formula tends to overestimate body fat for especially tall people and underestimate it for especially short ones.
These two sources of imprecision are nicely illustrated by a pair famous athletes, former bodybuilder Arnold Schwarzenegger and contemporary sprinter Sha'Carri Richardson.
As with many other athletes, Schwarzenegger's body fat would've been overestimated owing to his muscular density. At the peak of his career, he was 6' 2" and 235 pounds, which yields a BMI of 30.2. In other words, he'd be classified as "Obese". This illustrates how wrong the BMI can be, because, as a professional bodybuilder, his body fat would've actually been extremely low, perhaps to the point of jeopardizing his health.
At the other end of the spectrum is the elite sprinter Sha'Carri Richardson. Sprinters have low body fat as a consequence of training, but not as low as that of bodybuilders, and their BMIs usually fall within the "Normal" range. Richardson is a bit of an exception, because she's relatively short. At 5' 1" and 99 pounds, her BMI is 18.7, which is borderline Underweight. And yet, as with Schwarzenegger, this reflects a problem with the BMI rather than with her body.
Here are two additional limitations of the BMI:
(c) The BMI doesn't take into account how body fat is distributed.
A person with a BMI in the normal range may have abdominal obesity – a relatively large amount of "belly fat" – which increases the risk of health problems such as cardiovascular disease.
(d) BMI categories may not generalize well.
Since Quetelet, the BMI has been measured mostly in white males. The exclusion of women and people of color until recently means that the standardized cut-offs for under- and over-weight may not be suitable for all populations. (Conventional guidelines, used by the WHO, the CDC, the NIH, most clinicians, and most researchers, date back to the 1990s.) Alternatives are being proposed, but the formula hasn't changed in most settings.
3. Why are the BMI's flaws harmful?
The BMI is useful in epidemiological research, because extreme values reliably predict negative health outcomes. In short, it can be a helpful tool for making generalizations about populations.
At the individual level, the BMI might be useful as a quick screening tool, so long as it's understood that there will be many false positives (classifying people as "underweight", "overweight" or "obese" when they're not) as well as false negatives (classifying people as "normal" when their weight is unhealthy).
Much of the harm created or potentially created by the BMI can be traced to those misclassifications. Here are some examples:
(a) Diminished quality of health care.
A 2016 study of over 40,000 adults revealed that 31% of those with a Normal BMI were metabolically unhealthy, according to measurements of their blood pressure, cholesterol, glucose, insulin resistance, and other key variables. At the same time, roughly half of people in the Overweight category, and 16% in the Obese category were metabolically healthy. This is one of many studies indicating that the BMI misclassifies a substantial number of people.
If health care professionals assume that someone with a Normal BMI is healthy, they might miss medical conditions such as excess abdominal fat (an oversight that's more likely given the rise of telemedicine).
At the same time, assuming that a person with a BMI in the Overweight or Obese range needs to reduce their weight could cause health care professionals to focus on weight reduction and overlook other physical problems.
Diminished quality of care may differentially impact people of color. In particular, experts have noted that Black Americans have higher BMIs on average than white Americans do, even though higher BMIs among Black individuals do not necessarily predict worse health outcomes. The concern here is that Black patients may be more likely to encounter health care providers who overemphasize BMI and weight reduction, when these individuals are in fact either healthy or have an undiagnosed medical condition.
(b) Susceptibility to misrepresentation.
Online BMI calculators are often accompanied by misleading statements about it being a reliable indicator of body fat or healthy weight. Even when the web page cautions you that the BMI isn't perfect, the advice provided may be unwise. For instance, the Harvard Medical School web page that contains a BMI calculator tells you simply that "If your BMI is above 25, losing weight is a smart idea."
Actually, no. If your BMI is 28 and you're pretty muscular, losing weight might be a bad idea. Regardless of your musculature, a BMI over 25 should prompt you to connect with a medical professional who can give you a more comprehensive evaluation of your health. What's not smart is to treat the BMI Normal range as a simple, universal goal. Once again, it's where people tend to be, not where they necessarily ought to be.
(b) Occupational discrimination.
Weight discrimination is common in the workplace, but American job applicants and employees are only partially protected by the law. For instance, the Americans with Disabilities Act only protects those whose high BMI scores result from a specific underlying medical condition such as diabetes. The Equal Employment Opportunity Commission prohibits weight discrimination, although it allows weight to be considered in hiring and retention "if the employer can demonstrate how the need is related to the job". There are many examples of employers taking advantage of that qualification. As one expert noted earlier this year, "With the exception of Washington and Michigan and a few cities, it is legal to discriminate on the basis of weight."
Meanwhile, some companies offer incentives for employees to participate in wellness programs and to meet objective targets such as reductions in BMI, a practice occasionally portrayed as discriminatory (on the grounds that it creates implicit pressure for some employees to participate, while excluding those with Normal BMIs from access to the incentives).
At the other end of the spectrum, countries such as France, Italy, Spain, and Israel have laws banning models whose BMIs place them in the Underweight category. I agree with the need for these laws, given traditional overemphasis on thinness in the fashion industry, but I wonder too whether some models with healthy levels of body fat have been discriminated against because their bone and muscle density is relatively low.
(c) Unrealistic expectations.
Part of the BMI's appeal is that it feels "sciency". The calculation is just complicated enough to make people feel like they're doing something technical – and therefore more accurate than the crude estimate that it is. Informational resources, even when they include caveats, inadvertently promote this impression. For instance, here's what you see on a National Institutes of Health (NIH) web page:
"A good way to decide if your weight is healthy for your height is to figure out your body mass index (BMI). You and your health care provider can use your BMI to estimate how much body fat you have."
I struggle with this statement, because the BMI is not really "a good way" to decide whether your weight is healthy. Also, saying that you and your health care provider can use the BMI to estimate your body fat sounds more complicated than what actually happens. What health care providers do is to discuss your BMI. They probably won't convert it to an estimate of body fat, because that estimate would be very imprecise and still need to be interpreted.
A key problem with expectations around BMI categories is that people sometimes take the Normal range as their goal, whether or not this would be the healthiest and most comfortable weight for them. Medical professionals often comment on this tendency. Again, what's "normal" here is just what's close to average for the population. It's not automatically what's best for the individual.
(Note for stats people: From Quetelet on, experts have assumed that BMI is roughly normally distributed in the population, and so the mean, median, and mode are expected to be roughly the same. Thus, Normal BMI represents a narrow range around the population average, it contains the midpoint in values for the population, and more people are observed in this range than in the Underweight and Overweight/Obese categories.)
(d) Promotion of stigmas.
Stigmas about weight undermine a person’s emotional well-being. Even doctors turn out to have a so-called "anti-fat" bias which may translate into lower-quality care for some patients. In their MedPage Today article, Drs. Austin and Richmond describe some related concerns:
"[A] growing body of research on weight stigma in medicine has identified routine BMI assessments as a key barrier to care for people living in larger bodies and for others experiencing weight-based shame. Studies and patient stories tell us that anticipating being weighed in medical settings leads many to delay or avoid medical care altogether, resulting in missed preventive care or worse. When patients do arrive to care, a focus on BMI can cause more problems than it resolves. Clinicians' focus on BMI can lead to unproductive weight-related conversations that fracture the doctor-patient relationship and may introduce mistrust. This can lead to patients opting not to follow physician advice, even when that advice is not weight-focused, and not pursuing follow-up care due to faltering trust -- a vital element of effective doctor-patient relationships."
4. What are the alternatives to the BMI?
(a) Mathematical tweaks to the BMI formula have been proposed (e.g., slightly increasing the exponent in the denominator), but these tweaks only address specific problems, like the overestimation of body fat among tall people. They don't help with the most fundamental sources of imprecision in the BMI or the challenges of interpreting individual scores.
(b) Other simple measures have been proposed (e.g., waist circumference, or waist divided by height). These measures sometimes perform as well as, or better than the BMI at predicting health outcomes, but that's not saying much. None of them come close to what can be learned from direct assessments of metabolic health.
(c) Body fat can be directly measured with expensive, highly technical procedures (hydrostatic weighing, dual-energy x-ray absorptiometry, etc.). But even if a patient has access to these procedures – and can afford them – it would be hard to evaluate the data. Because the procedures are difficult to standardize, current research doesn't offer consistent guidance on translating test results into body fat estimates.
(d) Body fat can also be measured with calipers (available on Amazon, naturally). What's actually measured is density, which is a more accurate proxy for fat than weight/height ratios. Unfortunately, caliper measurements rely on complex measurement procedure. (You're not just pinching your belly a couple of times.) Even then, accuracy is limited because the caliper readings must be entered into a formula that incorporates age but not fitness. The fitter you are, the more calipers overestimate your body fat.
In short, no measure of body fat works particularly well as an individual-level measure. Perhaps this is inevitable.
(d) Arguably the best alternative to the BMI is an inclusive approach that looks at various health indicators (blood pressure, cholesterol, triglycerides, etc.) as well as overall well-being and body satisfaction. By means of an inclusive approach, medical professionals can improve health and quality of life without excessive or harmful emphasis on weight. Weight loss (or gain) might be recommended in an inclusive approach, but it's not the sole focus.
Conclusion
Live well, be healthy, love your body. Seek out medical professionals who take an inclusive approach to your physical and mental well-being. Calculate your BMI, set goals if that makes you feel good, but don't take the specific numbers too seriously.
Thanks for reading!