Hello



I'm currently doing a register-based cohort study on maternal smoking during pregnancy and the risks of cryptorchidism and hypospadias (two male genital malformations).



During my study period, hospitals changes their registries on patients smoking habits from never/former/current to categorizing in number of cigarettes per day.

I created a variable for the whole period smoke_tot but I also wish to look into a dose-response pattern with the latter variable smoke_cat.

smoke_cat consists of the following categories:

0: Never smoked during pregnancy

1: Stopped smoking during pregnancy

2: Smoked <=5cigarettes per day during pregnancy

3: 6-10 cig/day

4: 11-20 cig/day

5: >=21cig/day

6: Smoked during pregnancy, amount unknown

.: Missing



I have 7 dependent variables and 4 independent variables.

My STATA syntax is the following


mi set mlong

mi register imputed smoke_tot smoke_cat gest_age birthweight parity education_year nationality marriage

mi impute chained (ologit) smoke_tot smoke_cat education_year (mlogit) nationality (regress) gest_age birthweight = birthyear_grp matage cryptorchidism hypospadias



So my problems are the following:

I wish to use multiple imputation on both categories, but smoke_cat does not cover the whole study period and it has a missing values, which needs to be imputated over all categories + a category 6, which needs to be imputated only over category 2-5.
I also need to restrict imputations for smoke_cat to birthyear>1996

I hope you can help