Proportional hazards valid even if curves cross

classic Classic list List threaded Threaded
1 message Options
Reply | Threaded
Open this post in threaded view

Proportional hazards valid even if curves cross

This post has NOT been accepted by the mailing list yet.
Hi Statalisters,

I wondered if there was a situation wherein survival curves cross and
yet the proportional hazards assumption still holds (or at least does
not require any sort of correction).

I am conducting a retrospective analysis wherein my treatment and
unexposed groups both have fairly shallow survival curves that cross
about half-way along the x-axis.

As it happens, analytic tests for the proportional hazards assumption
(i.e., including a time-varying covariate, Schoenfeld residuals using
-estat phtest-, etc) do not yield a significant result.

Are the initial results (i.e. without adjustment for the curves
crossing) reportable as they are?