[R] coxph with time-dependent covariates

[David Winsemius]

On Jan 13, 2014, at 10:50 AM, Phil Stevenson wrote:

> I am using the coxph() function (from the 'survival' package) for a data
> set that requires one of the covariates to be treated as time-dependent.
> The issue is with this data set is there are some covariate values that are
> missing at the time point when the event occurs.
> 
> For example:
> 
> Person A:
> 
> Day      time-dependent covariate         deathday          status
> 
> 1          34                                                    6
>               0
> 
> 2          47                                                    6
>               0
> 
> 3          78                                                    6
>               0
> 
> 4          29                                                    6
>               0
> 
> 5          NA                                                  6
>             0
> 
> 6          NA                                                  6
>             1
> 
> 
> 
> R throws out this event due to the missing time-dependent covariate at day
> 6. Is there a way I can keep this event without filling in a covariate
> value at day 6?
> 

What is your expectation concerning what _should_ happen?  

> 	[[alternative HTML version deleted]]
-- 

David Winsemius
Alameda, CA, USA