7

I have a dataset with survival data and a few missing covariates. I've successfully applied the mice-package to imputate m-numbers of datasets using the mice() function, created an imputationList object and applied a Cox PH model on each m-dataset. Subsequently I'ved pooled the results using the MIcombine() function. This leads to my question:

How can I get a p-value for the pooled estimates for each covariate? Are they hidden somewhere within the MIcombine object?

I understand that p-values isn't everything, but reporting estimates and confidence intervals without corresponding p-values seems weird to me. I'm able to calculate an aprox. p-value from the confidence intervals using e.g. the formula provided by Altman, but this seems overly complicated. I've searched around for an answer, but I can't find anyone even mentioning this problem. Am I overlooking something obvious?

E.g.:

library(survival)
library(mice)
library(mitools)
test1 <- as.data.frame(list(time=c(4,3,1,1,2,2,3,5,2,4,5,1), 
          status=c(1,1,1,0,1,1,0,0,1,1,0,0), 
          x=c(0,2,1,1,NA,NA,0,1,1,2,0,1), 
          sex=c(0,0,0,0,1,1,1,1,NA,1,0,0)))

dat <- mice(test1,m=10)

mit <- imputationList(lapply(1:10,complete,x=dat))

models <- with(mit,coxph(Surv(time, status) ~ x + strata(sex)))

summary(MIcombine(models))

I've tried to sort through the structure of the MIcombine object, but as of yet no luck in finding a p-value.

slamballais
  • 3,161
  • 3
  • 18
  • 29
Kjetil Loland
  • 83
  • 1
  • 4
  • Have you tried `str` to examine your object? Perhaps you could see `str(summary(your.object))` to see if there's a p-value in there. In any case, could you post a minimal reproducible example that shows how the result looks like (for example of function `MIcombine`)? http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example – Roman Luštrik Jan 03 '13 at 14:14
  • Yes, I tried sorting through the structure without any luck. Thanks for the tip of adding an example! – Kjetil Loland Jan 04 '13 at 07:48

1 Answers1

1
models <- with(dat,coxph(Surv(time, status) ~ x + strata(sex)))
summary(pool(models))
gofr1
  • 15,741
  • 11
  • 42
  • 52
Andreu FG
  • 11
  • 1