KM survival for death and NOT complication

[Rose Matthews]

Hi I create a survival KM graph for those surgeries that had a complication. However, I now want to plot a KM graph which only has death as the failure
Am I plotting this correctly?

I have a variable with survivaltime - survivalt-, this takes:

1. Date of complication (if patient had a complication)
OR
2. Date of death (if patient died)
OR
3. Last date i.e end of follow up date if patient did not die and did not sustain a complication

///This is the normal survvial km graph is patient had a complication
stset survivalt [pw=attweight], failure(complication==1)
sts graph, by(procedure)

////What about death
st survivalt [pw=attweight], failure(died==1)

//KM curves to compare death
sts graph, by(procedure)

I suppose if the patient had a complication and then died a few years late - the variable -survivalt- should take the date of death rather than the date of the complication, as death is what I am interested in, is this assumtpion correct?

[Clyde Schechter]

I suppose if the patient had a complication and then died a few years late - the variable -survivalt- should take the date of death rather than the date of the complication, as death is what I am interested in, is this assumtpion correct?

Yes. It should be the date of death for those who died, and the last date of follow-up for those who did not die. The failure option in -stset- should be -failure(died == 1)-.