Help for COX regression

[Paolo Costa]

Hello everyone, I'm a new STATA user.
I'm analysing this database. I've performed a multiple imputation in order to calculate COX multiple regression on outcome Overalldeath (MO2). Can I choose which variables I've to imputed for multiple imputation? I mean, there are more missing variables, but among those I've imputed only the ones I'm going to use further.

mi impute chained (logit) bmi_c (logit) NYHA_dic2 (logit) smok2 (logit) MAP2 (logit) HR2 (logit) anemia2 (logit) dyskalemia2 (logit) gfr_cat2 (mlogit) age_cat2 = i.IHD2 i.hypertension2 i.diabetes2 i.valdis2 i.strokeT2 i.dementia2 i.ace i.BB i.mra i.ccb i.AC i.statin i.nitrate i.SGL i.PPI , add(10) rseed(43421) force dots

Once performed the imputation, the covariate I need to input in the regression model are the same I've imputed for imputation itself?

mi stset sos_timetodeath, failure(OM2 ==1) scale(365.25) exit(time 5000)
mi estimate, eform post: stcox i.bmi_c i.NYHA_dic2 i.smok2 i.MAP2 i.HR2 i.anemia2 i.dyskalemia2 i.gfr_cat2 i.age_cat2 i.IHD2 i.hypertension2 i.diabetes2 i.valdis2 i.strokeT2 i.dementia2 i.ace i.BB i.mra i.ccb i.AC i.statin i.nitrate i.SGL i.PPI cancer_acearbarni


thanks everyone

[Carlo Lazzaro]

Paolo:
welcome to this forum.
Yes, they are.
See example on heart attack in -Intro — Introduction to mi- entry, Stata .pdf manual.

[Paolo Costa]

I sincerely thank you. I'm new to STATA and I've just started studying statistic, so I'm stumbling across some issues on the way

[Carlo Lazzaro]

Paolo:
I also stumble across statistical/Stata issues, from time to time.
Experience reduces the frequency of that, but cannot eliminate it.

[Paolo Costa]

and how about this instead:
I'm trying to create a cumulative incidence graph of an event (X) by different groups. I've chosen time to event and exit time (express both in days).
Nonetheless, I guess I'm doing something wrong 'cause I couldn't replicate this graph. The concept I want to achieve is the same, and after that, I'll perform a stcox regression by the same groups and write the HR of them on outcome below the grap.

Code used:
stset timetoevent, failure (X) scale(365.25), exit (1000)
clonevar primary_outcome = X
stset, clear
Sts graph, by(groups)

this below the graph I want to achieve (except from colours and label stuff), why it doesn't display me cumulative incidence though?

[ATTACH=CONFIG]n1727662[/ATTACH]