Hacker News Reader

Comment by germinalphrase | Hacker News Reader

Isn’t BMI an intrinsically flawed measure?

Edit: the change in population argument makes sense. I’ve always thought it was flawed as a good friend of mine is short, very stocky, and fit - but is technically “obese”.

Edit 2: I was curious how this trend looks going back further than the 80’s. “The evolution of BMI values of US adults: 1882-1986” https://voxeu.org/article/100-years-us-obesity

View on HN

Not really. It's inaccurate for some very athletic people, but it is accurate for a vast majority of the population.

It's accurate when measuring populations, where the missing variables average out. It isn't meant to be applied to individuals but is good enough for expedient's sake.

That is untrue. There's no research to support the idea that BMI is a particularly useful tool (compared to, say, neck vs waist circumference) or that the thresholds it sets are meaningful (e.g., that an increase from 24.7 to 24.9 is a lesser or greater increase of risk than an increase from 24.8 to 25.0).

How do we measure how accurate it is or not?

It's not like there is some underlying ground truth on what 'overweight' is, is there?

Measured overtime for a population - it seems to be very good given how easy it is to measure.

I'd love it if there was technology at some point accessible to everyone that would allow for more informative measurements (w/ privacy considerations).

It's not like most of those Americans with high BMIs are bodybuilders… or is it? Only 15% of Americans belong to a gym.

Not everyone into fitness belongs to a gym.

This comment makes it sound like while Americans don't belong to gyms, tons of them are into fitness. This is absolutely not the case and seems rather disingenuous to suggest.

It is for individuals but it's reasonable to use as a population measure. Waist-to-height and waist-to-hip ratios are also good measures for population-level health and those work a little better for individuals.

Underlining this point. As imprecise as it may be for an individual, it is remarkably good directionally.

More, for many individuals, it is more accurate than folks give credit.

This chart from the CDC shows obesity prevalence from 1999 to 2018. It's a very good indication of where we are going as a nation bmi-wise.

https://www.cdc.gov/nchs/images/databriefs/351-400/db360-fig...

It's especially reasonable when we're talking about changes in BMI instead of levels regarding a population. That can only be driven by the population getting fatter.

BMI is pretty accurate for most people. Having said that you don't need a measurement to see if someone is fat!

> BMI is pretty accurate

What is 'accurate' in this context? Accurate against what? What are you comparing it against to get a measure of accuracy?

It indicates against 'normal weight' - but if the population's weight changes then the 'normal weight' changes with it doesn't it?

Admittedly accurate is the wrong term, but it is a reasonable inductor of health, as the heavier someone is for a given weight, the shorter their expected lifespan will be.

The normal weight may change, but the ideal healthy weight doesn't. The idea of a healthy weight is misleading, as there is more of a healthy body fat percentage. Even that's is an oversimplification, as visceral fat has far more health implications that subcutaneous fat.

As a benchmark, 5'6" and 185lbs is considered obese.

Is your friend built like a running back?

https://www.gssiweb.org/sports-science-exchange/article/sse-...

No - I guess he would be “overweight” not obese.

BMI is as flawed as people's x/y varies, where x is e.g. shoulder width and y is height.

That isn't very much, really. Some people are far from the norm, but generally the broad-shouldered people aren't the short ones.

Maybe, but negativity towards BMI is a very good predictor for BMI :-)

Reply on news.ycombinator.com