[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: Survival time analysis/non-proportional hazard/stpm by Royston
I want to run a survival data analysis with Stata.
Applying 'stcox' and controlling for non-PH with 'estat phtest' suggest that some of my treatment variables as well as several covariates violate the PH-assumption clearly.
Now, I'm came across the ado-file 'stpm' implemented by Royston (Flexible parametric alternatives to the Cox model, and more; P Royston; The Stata Journal; Number 1, pp. 1–28 2001).
Not beeing able to make head or tail of the specifications in his Paper I can't figure out, if I have to put all variables that violate the PH assumtion into the 'stratify'-option or only the relevant treatments that doing so.
Depending on which of both alternatives I apply, the resulting hazard rate curves (conditioned on treatment=1) differ considerably. With my sample of about 550.000 obs. and over 50 variables it is time-intensive and, thus, difficult to apply 'trial-and-error'. Has anybody any ideas?
* For searches and help try: