Statistical methods
The mean scores of the radiological signs formed the outcomes of interest. The exposures of interest were the SHS exposure in the past 12 months and the total lifetime SHS exposure index as continuous variables. Home and work exposures were analysed separately and combined. Multiple linear regression was used to study the relations between SHS exposure and the radiological signs, adjusting for age, asbestos-exposure index, smoking status (ex-smoker vs. never smoker), potential image interpreter effect (1=RJ/TVi, 0=TA/TVe), body mass index (weight [kg]/height [m]2)17 and pack-years of active smoking for ex-smokers. Results are given as regression coefficients and 95% confidence intervals (95% CI). Due to the concern that collinearity could distort regression estimates, the variance inflation factors (VIF) were determined for each model. Multiple logistic regression with the same covariates was used for binary radiological signs.
Statistical methods
The mean scores of the radiological signs formed the outcomes of interest. The exposures of interest were the SHS exposure in the past 12 months and the total lifetime SHS exposure index as continuous variables. Home and work exposures were analysed separately and combined. Multiple linear regression was used to study the relations between SHS exposure and the radiological signs, adjusting for age, asbestos-exposure index, smoking status (ex-smoker vs. never smoker), potential image interpreter effect (1=RJ/TVi, 0=TA/TVe), body mass index (weight [kg]/height [m]2)17 and pack-years of active smoking for ex-smokers. Results are given as regression coefficients and 95% confidence intervals (95% CI). Due to the concern that collinearity could distort regression estimates, the variance inflation factors (VIF) were determined for each model. Multiple logistic regression with the same covariates was used for binary radiological signs.
การแปล กรุณารอสักครู่..
