They have a "Role of BMI" section which strengthens your point.
> BMI accounted for a substantial proportion
In the "Role of BMI" they also say that BMI isn't perfect to detect adiposity, suggesting even further that being fat is what you want to avoid.
> Finally, given the observational nature of this study, it is possible that there is still unmeasured confounding, residual confounding, and reverse causality.
> BMI accounted for a substantial proportion
In the "Role of BMI" they also say that BMI isn't perfect to detect adiposity, suggesting even further that being fat is what you want to avoid.
> Finally, given the observational nature of this study, it is possible that there is still unmeasured confounding, residual confounding, and reverse causality.
Studying human diets is hard.