I understand the objections to BMI comparisons at an individual level (though I believe people think themselves a bit too exceptional too often), but as part of a larger study, I would think it'd make a good additional data point, no?
It most definitely would, but measuring body fat percentage would be better.
But then you couldn't easily extrapolate to the population. Well, you could, but you'd require the population to have an understanding of their body fat percentages. BMI isn't ideal, but it's quick and dirty, easy to collect and it should be relatively reliable for the population.