A chart showing BMI vs altitude could help cut through individual bases differences. If there is a trend downward it would show a dose-response effect. I didn't find such a chart in the paper. Instead they have a chart showing how many data points were available by altitude.
