Multivariate survival analysis with 3 variables analyzed as cubic splines in a competing risk setting

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

Multivariate survival analysis with 3 variables analyzed as cubic splines in a competing risk setting

KatjaN
This post has NOT been accepted by the mailing list yet.
I'm doing a survival analysis with cancer-specific mortality (DSM) calculated as cumulative incidence and death of other diseases as competing risk. I'mm interested in finding prognostic factors for DSM. I have 1065 patients and 361 events of interest. Based on the litterature I have chosen 6 variables (all independent) to be included in a multivariate analysis. Among other variables I have age, duration of symptoms, and tumorsize as continuos variables. I would like to include these in the multivariate analysis using cubic spline regression. There is no problem making the spline variables using either -mkspline- or -spbase-.  I would like to make a graphical illustration of the relation between each of the continous varialbes and their corresponding HR (after adjusting for all the other variables). In univariate analysis this can easily be done by using -predict- specified as HR after the cox regression followed by at -twoway- graph, however this seems not to work after a multivariate regression?
Furthermore I would like to compare HR for specific values of the continuous variables, say e.g. age 15, 30, 45, 60, 75 and 90 with age 15 as the reference. Is this possible?