Sensitivity analysis
Sensitivity to the covariates included and how they are modelled | ||||||||
Model 1: covariate adjustment equivalent to model 5, table 2 | Model 2: all non-significant covariates in table 1 added to model 1 | Model 3: all continuous covariates in model 1 dichotomised at median split | Model 4: removal of 2 plausible mediators—media parenting and friend smoking—from model 1 | |||||
n=1546 | n=1532 | n=1546 | n=1554 | |||||
OR | 95% CI | AOR | 95% CI | AOR | 95% CI | AOR | 95% CI | |
Sensitivity to how film-smoking exposure is modelled | ||||||||
Exposure to film smoking | ||||||||
Modelled as a continuous variable | ||||||||
Scaled so lowest is 0 and highest is 1 | ||||||||
All values >95th percentile set at 95 percentile | 2.08 | (1.22 to 3.55) | 2.09 | (1.21 to 3.61) | 2.63 | (1.66 to 4.18) | ||
No outlier trimming | 2.95 | (1.37 to 6.36) | ||||||
Scaled so that median value is 1 | ||||||||
All values >95th percentile set at 95 percentile | 1.29 | (1.07 to 1.55) | ||||||
No outlier trimming | 1.25 | (1.07 to 1.47) | ||||||
Modelled as dummy variables | ||||||||
Quartiles of exposure | ||||||||
Quartile 1 | Ref | Ref | ||||||
Quartile 2 | 0.91 | (0.60 to 1.37) | 1.10 | (0.78 to 1.58) | ||||
Quartile 3 | 1.25 | (0.83 to 1.88) | 1.45 | (1.02 to 1.06) | ||||
Quartile 4 | 1.49 | (1.00 to 2.22) | 1.91 | (1.35 to 2.70) | ||||
Median split | ||||||||
Below median | Ref | |||||||
Median or above | 1.44 | (1.08 to 1.92) | 1.54 | (1.18 to 2.01) |