Hi
Context: I am working with a SEER data set, with ~ 36000 records with each record having information on age at diagnosis, gender and various socio-economic factors at baseline. I have a very long study period from 2000 to 2013 for inclusion and study cut-off is dec2018. At the end of the study, 85% pop is dead. I am focusing for now on comparing survival outcomes between 2 races. I was asked by my mentor and lead biostatistician to run Propensity Score Matching analysis to complement the COX PH regression I already ran.
After much research reading Stata literature (stteffects intro, advanced, rebalance, teoverlap), it seems that I need to use stteffects command (and not psmatch2 or effects psmatch which seem to have been designed for binary outcomes and other types but not survival outcomes).
Is my understanding correct?
If yes, is IPW estimator a reasonable approach?
Data sample below:
This is the syntax I am currently using:
stteffects ipw (raceb aged sex geogr, probit) (aged sex geogr), atet
rebalance overid
teoverlap
I would like to insert here the output but have not found way in the Stata FAQ to share output.
Many thanks
Arnaud
Context: I am working with a SEER data set, with ~ 36000 records with each record having information on age at diagnosis, gender and various socio-economic factors at baseline. I have a very long study period from 2000 to 2013 for inclusion and study cut-off is dec2018. At the end of the study, 85% pop is dead. I am focusing for now on comparing survival outcomes between 2 races. I was asked by my mentor and lead biostatistician to run Propensity Score Matching analysis to complement the COX PH regression I already ran.
After much research reading Stata literature (stteffects intro, advanced, rebalance, teoverlap), it seems that I need to use stteffects command (and not psmatch2 or effects psmatch which seem to have been designed for binary outcomes and other types but not survival outcomes).
Is my understanding correct?
If yes, is IPW estimator a reasonable approach?
Data sample below:
Code:
* Example generated by -dataex-. For more info, type help dataex clear input float raceb byte sex float(geogr event) int survmonth 0 2 1 1 65 1 1 1 0 74 0 1 1 1 13 0 2 1 1 136 1 1 1 1 3 0 2 1 0 66 0 1 1 1 71 0 2 1 1 30 0 2 1 1 5 0 1 1 1 1 end label values sex sex1 label def sex1 1 "Male", modify label def sex1 2 "Female", modify label values geogr geogr1 label def geogr1 1 "Big Metro", modify label values event event1 label def event1 0 "Alive", modify label def event1 1 "Dead", modify
stteffects ipw (raceb aged sex geogr, probit) (aged sex geogr), atet
rebalance overid
teoverlap
I would like to insert here the output but have not found way in the Stata FAQ to share output.
Many thanks
Arnaud
