Not to be nasty, but I didn’t ask for WAGs, but for informed answers.
No, they did not use a cluster analysis and no such clustering exists. Mortality in the US is decreased for the “overweight” compared to the “normal” and more so compared to “obese” and “underweight”. There is an apparent linear increase in some morbidities (incidence of Type 2 DM, hypertension, for examples) from a BMI of 21 on up … but again fewer deaths in the “overweight” group. For some morbidities there is a U-shaped distribution, with best outcomes clustering in the 24 to 27 range.
And while the CDC may intend that clinicians use BMI cautiously, BMI has become the definition of overweight and obesity in all guidelines and nearly all studies done.
Yes, it is a very poor measure on its own, but it is easy to use and reproducible. Skin folds are difficult and give different results by different measurers. Underwater weighing is not available in most doctors offices. Clinically eyeballing for body type is too subjective. So on.
Do we do a service to the public by telling the majority of them that they are fat and setting targets for goal weights that they are near certain to fail to achieve? Especially in the situation when they are less likely to die at a less stringent target? Don’t we instead dilute the importance of weight control and fitness as a message to the truely obese?
Does anyone actually know if there was anything other than an arbitrary “Hey 25 is an easy number for those stupid clinicians to remember” consideration to the choosing of these standards?
BTW, I’m a pediatrician and it is even worse for the pediatric BMI’s … we are now supposed to calculate and chart BMI and BMI percentile for age for kids. The problem is that the BMI percentiles are wrong and were wrong by intent when they were created. If I tell you that your child is at the 85%ile for BMI for age, you’d expect that to mean that in the US today, 85% of kids your childs age are leaner (or at least have a smaller BMI) than your child, right? But you’d be wrong. Your kid is probably nearer the middle of the curve, much nearer. The data was based on pooled studies from the 60’s to 80’s and the most recent of those studies, NHANES III, had its data on kids over 6 yo excluded because it would have produced percentiles that showed fewer kids to be “fat” by BMI percentiles.
I’m not saying that we should not adress the problem of obesity in this country, and particularly the very real and significant problem of pediatric obesity. But creating a generation of scale watchers, telling kids in a very alternate dimension Lake Woebegone way that they are all above average, is counterproductive.