Missing variables were imputed by multiple imputation (n = 5) using the mi command in Stata.
Five imputed data sets were considered sufficient 29 and were constructed using the mi command in Stata, based on the following baseline variables: smoking behaviour, HSI, age, ethnicity, qualifications, employment, IMD score and partner smoking status.
For each scenario the missing X data were imputed conditional on Y using MVNI, which was implemented in Stata release 12 using the mi impute mvn command [ 18].
Multiple imputation by chained equation was used to impute missing data using the "mi impute chained" command in Stata version 12.0 (StataCorp, College Station, TX).
For this reason we could assume that these data were missing at random and we could carry out multiple imputation using the "mi impute chained" command in STATA version 12.1.
Multiple imputation was performed using the mi impute chained (pmm) command in Stata 12 (StataCorp, 2011), which uses the iterated chained equations approach (van Buuren et al., 1999) with predictive mean matching (Little, 1988; White et al., 2011).
We used multiple imputation [ 22] methods to deal with the missing data, using the mi impute chained and mi estimate commands for chained equations and subsequent regression model estimation.
We used the mi impute chained command in Stata to perform multiple imputation using chained equations to generate 42 imputed data sets, based on the rule of thumb suggested by White et al (2011).
The following imputation methods were used with m = 20 imputations performed for each procedure: – Linear regression imputation (applied using the Stata command: mi impute regress) with no post-imputation rounding.
For each of the two outcomes (breast cancer-specific mortality and overall mortality), 50 imputed datasets were generated and results were combined using the PROC MI and PROC MIANALYZE commands in SAS (version 9.3) with appropriate correction for variance.
