Instrumental variable estimates of the causal effect of BMI on current asthma, non-atopic asthma, and atopic asthma in children at age 7½ and 9 y using two-stage GMM estimator.

aChi-squared test comparing estimate effects for non-atopic and atopic asthma.bControls were children with no current asthma.‡Test for the null hypothesis that the RR in males is the same as the RR in females (likelihood ratio test).*Based on residuals from regression models of mass measure (fat/lean mass) on height, height squared, and gender. Fat mass was divided by two so that its standard deviation was similar to that of BMI and lean mass.