Participants | CD14/-260 genotype | p-value | ||||||

All | Full confounder information | Full confounders and genotype | CC | CT | TT | Heterogeneity of effect | Trend of interaction^{¶} | |

Participants n | 902 | 814 | 527 | 110 | 271 | 146 | ||

FEV_{1} | ||||||||

Unadjusted | -0.01 (-0.16–0.15) | 0.01 (-0.15–0.18) | 0.07 (-0.11–0.26) | 0.51 (0.11–0.92) | 0.04 (-0.21–0.30) | -0.24 (-0.61–0.11) | 0.020 | 0.003 |

Adjusted^{+} | 0.04 (-0.11–0.19) | 0.08 (-0.11–0.27) | 0.62^{§} (0.21–1.04) | 0.01 (-0.25–0.27) | -0.22 (-0.59–0.13) | 0.008 | 0.005 | |

FVC | ||||||||

Unadjusted | -0.09 (-0.24–0.06) | -0.06 (-0.22–0.09) | -0.02 (-0.22–0.16) | 0.31 (-0.14–0.77) | -0.03 (-0.30–0.23) | -0.33 (-0.68–0.02) | 0.008 | 0.059 |

Adjusted^{+} | -0.05 (-0.22–0.10) | 0.00002 (-0.20–0.20) | 0.49 (0.03–0.96) | -0.04 (-0.32–0.23) | -0.29 (-0.65–0.06) | 0.029 | 0.013 | |

FEV_{1}/FVC | ||||||||

Unadjusted | 0.15 (0.02–0.29) | 0.15 (0.01–0.29) | 0.14 (-0.03–0.31) | 0.24 (-0.16–0.65) | 0.09 (-0.14–0.33) | 0.20 (-0.11–0.52) | 0.775 | 0.976 |

Adjusted^{+} | 0.13 (-0.01–0.27) | 0.13 (-0.005–0.28) | 0.18 (-0.24–0.60) | 0.05 (-0.18–0.29) | 0.27 (-0.03–0.59) | 0.538 | 0.809 |

Data are presented with 95% confidence intervals.

^{#}: a male of age 50 yrs and height of 180 cm is predicted to have a FEV_{1}of 3.8 L using the equation of Quanjer*et al.*[17]. If his observed FEV_{1}is 4 L, then the difference between observed and predicted FEV_{1}is 0.2 L. Following standardisation (again using the standard deviation provided by Quanjer*et al.*[17]), his standardised residual is 0.38. This means the difference between his measured and predicted FEV_{1}is 0.38 standard deviations greater than the mean of all of these values.^{¶}: assuming additive model.^{+}: confounders were smoking status in pack-yrs and household variables (keeping a cat or dog, age of the mattress, presence of mould in bedroom, presence of damp in bedroom, household density and age of home).^{§}: a coefficient of 0.62 in this table means that for a 10-fold increase in endotoxin levels (it is 10-fold because endotoxin levels have been logarithmically transformed using base 10) there is a change of 0.62 standard deviations in the difference between measured and predicted FEV_{1}.