Multiple linear regression analysis with FEV1 (% predicted) as the dependent variable, and age, sex, exposure to kerosene heater emissions, and agricultural employment for ⩾10 years as the independent variables
| β4-150 | 95% C I | p value | |
|---|---|---|---|
| All study participants | |||
| Age (years) | −0.7 | −0.9 to −0.5 | <0.001 |
| Sex (effect of male sex) | −14.9 | −21.5 to −8.3 | <0.001 |
| Using kerosene heater | −16.2 | −29.0 to −3.4 | <0.05 |
| ⩾10 years in agricultural employment | −18.0 | −30.0 to −6.0 | <0.01 |
| Men | |||
| Age (years) | −1.0 | −1.2 to −0.8 | <0.001 |
| Using kerosene heater | −5.8 | −11.4 to 23.0 | NS |
| ⩾10 years in agricultural employment | −14.6 | −32.2 to 3.0 | NS |
| Women | |||
| Age (years) | −0.4 | −0.6 to −0.2 | <0.01 |
| Using kerosene heater | −28.4 | −36.6 to −10.2 | <0.01 |
| ⩾10 years in agricultural employment | −21.1 | −36.9 to −5.3 | <0.01 |
↵4-150 FEV1 = Intercept + β4-150age + β4-150sex + β4-150kerosene + β4-150agriculture. The variance of FEV1 (% predicted) explained by the variables (R2) was 35% in the series as a whole, 45% in the men and 26% in the women.