Would you have expected that people who are slightly overweight have a lower mortality than those who are of normal weight? Though BMI is imperfect, it does all

Would you have expected that people who are slightly overweight have a lower mortality than those who are of normal weight? Though BMI is imperfect, it does allow for interesting observations such as that. I would be very enthusiastic about a noninvasive, cheap, rapid, and accurate way of measuring whole-body (and especially abdominal) fat, of course!