# [R] Testing strata by covariate interactions in coxph

From: Pyy-Martikainen Marjo <Marjo.Pyy-Martikainen_at_stat.fi>
Date: Thu 06 Oct 2005 - 18:33:00 EST

Dear list members,

I am working with a Cox ph model for the duration of unemployment. The event of interest
in my analysis is getting employed. I have various background variables explaining this
event: age, sex, education etc. I have multiple unemployment spells per person. I use a model with person-specific frailty terms in order to take into account the correlation of spells by the same person.

The persons can be divided into 3 groups, say A, B and C. I am interested to see whether there are differences between estimated covariate effects between the groups. Therefore
I specify a model with strata by covariate interactions. I would like to conduct a Wald test
for the null hypothesis "no differences between any covariate effects in the 3 groups".
This is similar to the example by Therneau & Grambsch in their book "Modeling survival data. Extending the Cox model", p. 47, except that I have interactions with many covariates that I would like to test jointly (and a frailty term). In S-plus there seems to be a function waldtest that does the job. Is there anything similar in R that I could use?

Here is the code for my model. The covariates are a subset of all the covariates that I use.

fit=coxph(Surv(tykesto,
tyoll)~nainen*strata(group)+I(ika-mean(ika))*strata(group)
+I(ika2-mean(ika2))*strata(group)+keski*strata(group)+korkea*strata(group)
+frailty(hnro),data=coxdata)

tyoll2 =1, if event "getting employed" has occurred, 0 otherwise nainen = 1, if female 0 otherwise
ika = age in years at the start of the spell ika2 = age squared
keski =1, if secondary education, 0 otherwise korkea=1, if higher education, 0 otherwise hnro = person identifier
group = group identifier

Thank you in advance for any help.

Marjo Pyy-Martikainen

R-help@stat.math.ethz.ch mailing list