New York Times gift articles

BMI is specifically not an amazing metric, for populations

It's a great measure for populations specifically because it doesn't rely on getting extra measurements. Height is easy, weight is easy. These are things we have data on for huge groups of people without doing any extra work at all. If you want BRI, you need to go out and measure people, which is expensive, or you need to rely on self-reporting, which is almost always wrong.

or individual people.

That's why I said it's a good indicator, which is not the same thing.

There are plenty of people who are .5 points short of being obese, but thanks to being very sedentary are still carrying too much fat, so it's a bad metric. But if you're .5 points short of being obese, you should take note of the fact that you're basically obese, so it's a good indicator.

Its 200 years old, based nearly entirely on white men, with no design consideration for women or POC. It also fails utterly to account for muscle, so it classifies many very fit people as obese.

This is such a stupid argument. A model that uses 2 variables does indeed not use every other variable in the universe. BRI doesn't account for lymphedema, race, amputation, pregnancy, gender, hydrocephalus, bladder obstruction, weight, congestion, muscle, cardiac health, age and a million other variables.

BRI uses exactly 2 measurements: Waist circumference and height, which is the exact same number of variables as BMI (height and weight). Which basically removes every argument except "BMI is older", which is a very strange argument to make.

Now, anticipating your reply: Using BRI to compute bodyfat and VAT requires many other inputs, like gender, race, age, etc, making it a more nuanced, better model. But that's not what's being discussed here. If you look up a "BMI based" bodyfat calculator, it will ask for your waist circumference. If you look up a "BRI based" one, it will ask for your weight. Those are completely different models doing something completely different.

And this is exactly my problem with this paper. They say "This model is much better!" and then they proceed to not actually use their model in validation, but a derived model.

Now, what I DO agree on is that BRI is very useful for one thing, even without strict validation: It removes the "BMI is bad" excuse for people, so from a public health viewpoint I fully support using BRI as the new "BMI killer" metric that i'll gladly say is amazing for everyone. If that gets people to focus on solving the obesity crisis in the west, I'll happily support it as the best thing in healthcare since indoor plumbing and germ theory. But in reality, it's a model based on just 2 variables.