Kind of wonder the same thing about BMI and water composition. Your body weight can vary by up to 10 pounds depending on how much water you're retaining. I've never been able to find information about the composition assumed or measured when coming up with BMI numbers.
Granted I don't think the ultimate effect is huge, and you can eliminate it by weighing yourself daily and taking an average. But most people don't do that, and a spot-check at a doctor's office certainly can't do that.
> I've never been able to find information about the composition assumed or measured when coming up with BMI numbers.
BMI is just weight divided by height squared. No distinction is made between type of mass. Muscle mass, fat, bone, water? BMI couldn't care less. It sums all that stuff up into a single value.
Think of it as a number that roughly correlates to disease. There will always be false positives and false negatives. False positives are acceptable. We want to minimize the number of false negatives.
There are nearly ten billion humans on Earth. It is not possible to fully evaluate every single one of them. Gotta run a SELECT statement. Filter them based on some criteria, and fully evaluate those that match. BMI isn't perfect, but it takes less than one minute to measure the variables and compute it. The equipment required is cheap and easy to use. Speed, efficiency, cheapness and ease are extremely important factors when you're applying this at national scales.
Patient might turn out to be a physically fit 100 kg 1.7 m 34.6 kg/m2 body builder. That's alright. Our objective is to make sure the obese and the malnourished can't escape the sieve.
Granted I don't think the ultimate effect is huge, and you can eliminate it by weighing yourself daily and taking an average. But most people don't do that, and a spot-check at a doctor's office certainly can't do that.