Earth Essences.Com

The definitive guide to BMI, and how much you should really pay attention to it

For a measure that’s taken so often, Body Mass Index (BMI) is surprisingly controversial.

The BMI – which is not far off its 200^th birthday, having been invented by the mathematician Adolphe Quetelet in the 1830s – is now a completely routine part of our lives.

It’s customary for doctors to measure their patients’ BMI and for scientists to record and analyse it in studies. That’s partly because it’s so easy to calculate: all you need to do is divide someone’s weight in kilograms by the square of their height in metres.

We’re all familiar with the charts, which split the BMI into arbitrary categories: if your BMI is below 18.5, you’re underweight; if it’s above 25 you’re overweight; and if it’s over 30, you’re obese. Official government statistics on obesity rates are based entirely on these BMI categories.

And yet, BMI is constantly under fire as a measure. There are articles providing the “top 10 reasons why the BMI is bogus”, implying that beyond these ten, there are many more.

BMI has been described as “inaccurate and misleading”. Other sources call it “flawed, crude, archaic, and [an] overrated proxy for health”.

A 2022 article by Harvard academics said BMI is “the organizing principle of a massively sprawling surveillance system”.

“BMI”, said one psychologist in a viral tweet last year, “is trash”.

It’s worth looking at some of those “top 10 reasons” for BMI’s uselessness, as they come up quite often. A few of them hinge on BMI being overly simplistic. It doesn’t, for example, include information on waist size or the proportion of fat in someone’s body. For that reason, it’s going to miscategorise people whose weight is high because they have a lot of muscle or a high bone density, saying they’re overweight and implying they’re unhealthy.

But although BMI doesn’t expressly include waist size or body fat percentage, it is very strongly correlated with those measures. That is, people who have higher BMI are much more likely to have higher values on these measures too. On the standard correlation scale where 0.0 means there’s no relation whatsoever and 1.0 means there’s a perfect positive correlation, BMI and waist circumference are often shown to be correlated at nearly 0.9; the correlation between BMI and body fat percentage is a bit lower at somewhere between 0.7 and 0.8.

The fact that these aren’t perfect correlations leaves plenty of room for outliers: body-builders with huge amounts of muscle who have a high BMI despite very low body fat, say, or people with relatively low overall BMIs but a high proportion of fat. And it means that, at least in some studies, taking into account waist circumference and body fat percentage as well as BMI can help make even better predictions of people’s health. 

But they’re called outliers for a reason. They’re exceptions to the very broad trend: these very strong correlations mean that BMI actually tells us a great deal about the average person’s waist size and body fat percentage.

The common criticism of BMI for not being a perfect correlate of other important measures is a double standard: there are all sorts of non-perfect correlations that we take for granted, and sometimes see fit to act on. The vast majority of the evidence on the relation between sleep quality and health, the relation between living in a city with high pollution levels and asthma, or the relation between eating sugar and tooth decay, to take just a few examples, comes from correlations that are an awful lot lower than that of BMI with waist size and body fat percentage, and yet there’s very little attempt to pick them apart. 

It’s all very well correlating together different bodily measures, but that doesn’t tell us what we really want to know, which is whether BMI is a good “proxy for health”. How good is BMI at predicting people’s risk of illness and their risk of early death?

It’s clear from reviews of the evidence that BMI is a commonly-found risk factor for all sorts of health conditions. Perhaps most prominent are heart-related problems: for issues including heart attack, atrial fibrillation, and sudden cardiac death, BMI can be an appreciable risk factor.

That’s not to say it’s a brilliant measure of these outcomes, or even the best measure we have in theory. In a perfect world we’d have even more measurements of body composition for every person, like fat percentage, bone density, and more. These measures are often discussed as predicting health problems “independently of BMI”. But, as noted above, that normally means that BMI itself explains a big chunk of the risk and these other factors can make the prediction even better.

But as part of a suite of different pieces of information we can have about a person, BMI plays a significant role in predicting their health. Given how much we can learn from something that’s so cheap and easy to measure, and if you’ll excuse the pun, it punches above its weight.

And it’s not just physical health: recent large-scale reviews of the scientific literature find that obesity, measured by BMI, is related to a 10 to 30 per cent increase in the risk of depression in adults, and a nearly 80 per cent higher risk of depression in children. Of course, a substantial part of this relation could be due to the stigma of being obese, but it remains possible that some of the higher levels of bodily inflammation that come from obesity might be part of the cause of mental health problems.

What about the most final outcome of all: mortality? If you imagine the graph with BMI on the x-axis and risk of early death on the y-axis, the relationship is a wide U-shape: people with very low BMIs and those with very high BMIs are much more likely to die earlier. 

But the question is where the bottom part of the U lies, and how flat it is before it starts rising. A large study from July this year found that, while those in the obese category were definitely at higher risk of early mortality (up to 108 per cent more likely to die in the nine years after their BMI was measured), there wasn’t much of an extra risk in the “overweight” category. Of course, being overweight has been associated with increasing risk of other, non-fatal health conditions, but it’s a useful reminder that not all risks rise with increasing weight in a simple and straightforward line.

Of course, an even bigger criticism here is that all these correlations and associations aren’t causal. Yes, BMI is related to these health outcomes, and yes, it seems to remain related to them after you take into account other factors like education and income. But we know from many different cases that even this could be confounded by some other factor that we haven’t measured in our purely correlational studies. 

This is where the new weight-loss drugs can help us: we can now run experiments where we randomise people to a treatment which we know will substantially reduce their BMI (that is, taking a drug like semaglutide), and then wait to see what effects this has on their health. We now have evidence from studies of patients with and without diabetes showing that lowering BMI did indeed cause a lower likelihood of major cardiovascular events like heart attacks (and there’s similar evidence from patients who’ve had bariatric surgery).

A blinkered focus on any one single measure is always a bad idea: whether it’s your BMI, your resting heart rate, your cholesterol levels, your exam grades, or your IQ score, every measurement inherently includes errors and will never be perfectly related to whatever it is you really want to know, like your overall health and fitness, your risk of illness, your academic potential, your happiness, or anything else.

And bear in mind that the studies are all focusing on overall relationships across samples of people: as we saw in our discussion of outliers, there will always be people who buck the trend, and that’s just by definition when we have imperfect correlations between variables (another way of putting it is that your BMI doesn’t “necessarily” tell you about your own health levels).

To focus too much on the trend-bucking individuals is to be like the people who refer to their great-grandmother who survived to age 105 while smoking 10 packs of cigarettes a day. Yes, it’s possible, but your great-grandmother isn’t representative of the average person and shouldn’t be part of the advice we give to the average person on whether to start smoking.

Many of the criticisms of BMI are effectively arguments that we – individuals, doctors, scientists – should only use measurements when they’re perfect, rather than when they’re useful. If that’s your standard, we can rule out effectively all the ways of measuring things in the world, which will always produce outliers, and will always come with their own limitations. 

To the extent to which the reasoned criticisms of BMI have led us to consider those limitations, they’ve been a very good thing: everyone who uses any measure should be aware of its pros and cons.

But the criticisms of BMI have been grossly overstated. Whereas it’s fair to say that BMI is far from a perfect measure, and can sometimes be misleading, it’s also silly to call it “bogus” or “trash” and encourage people to disregard it entirely.

If we want to understand our societal obesity problem, BMI is just one tool among many – but it’s a disproportionately helpful one.