Table 2

Relation of ECRHS “slope” to body mass index or weight, with three methods of adjustment for baseline FEV1, estimated by random effects meta-analysis of regression coefficients across 34 centres. All analyses adjusted for independent variables as described in text

Men (n=5899)Women (n=5354)
Adjustment for baseline FEV1Meta-analysis regression coefficient95% CIp value for regression coefficientp value for heterogeneityMeta-analysis regression coefficient95% CIp value for regression coefficientp value for heterogeneityp value for interaction of sex and BMI
*Internally standardised difference between FEV1 and value predicted for sex, age and height, divided by residual standard deviation.
†Regression coefficient on weight with adjustment for FEV1, height, and height2.
Standardised difference*−0.027−0044 to −0.0100.0020.10−0.014−0.033 to 0.0050.140.0110.41
FEV1/height2−0.027−0044 to −0.0100.0020.10−0.005−0.033 to 0.0040.130.0110.46
Regression method†−0.009−0.014 to −0.0030.0030.09−0.002−0.013 to 0.0020.130.0050.60