2

I am trying to evaluate the proportional hazard assumption of a Cox PH model that includes both an interaction term and a frailty term. When I call cox.zph, I get this error:

Error in imatr[kk, kk] : subscript out of bounds

I don't get the error when I remove the frailty term from the model, and, in the reduced model, all of my fixed factors meet the proportional hazards assumption. Do I need to run cox.zph on my full model as well? If so, how can I get around the error message I'm getting?

This code should recreate the error message I'm getting:

library(survival)

set.seed(80)

fake.data <- data.frame(event = c(rbinom(75, 1, 0.7), rbinom(75, 1, 0.4)),
                        time = runif(150, min=0, max=3),
                        time.period = rep(c("pre", "peri", "post")),
                        condition = rep(c("fed", "not fed")),
                        momID = floor(runif(150, min=1, max=35)))



phmodel <- coxph(Surv(time, event) ~ time.period*condition + frailty(momID) , data = fake.data )

assump.check <- cox.zph(phmodel)

Thanks!

momce
  • 21
  • 1

0 Answers0