[R] independent censoring
Therneau, Terry M., Ph.D.
therneau at mayo.edu
Tue Nov 29 14:20:38 CET 2016
On 11/29/2016 05:00 AM, r-help-request at r-project.org wrote:
> Independent censoring is one of the fundamental assumptions in the survival analysis. However, I cannot find any test for it or any paper which discusses how real that assumption is.
> I would be grateful if anybody could point me to some useful references. I have found the following paper as an interesting reference but it is not freely available.
> Leung, Kwan-Moon, Robert M. Elashoff, and Abdelmonem A. Afifi. "Censoring issues in survival analysis." Annual review of public health 18.1 (1997): 83-104.
This is because there is no test for independent censoring. Say I am following a cohort
of older gentlemen (65 years old) who were diagnoses with condition "x", and after 8 years
there a a dozen who no longer answer my letters. Why not?
a. Perhaps because they are in a nursing home, with dementia.
b. Perhaps because they have moved to another city to be interact and be near to
In case a, those lost to follow-up are much sicker than the average subject, and in case b
they are most likely the most healthy and active of the group. In a) the KM will
over-estimate survival and in b it will underestimate.
The main point is that there is absolutely no way to know, other than actually tracking
the subjects down. Any study which has a substantial fraction with incomplete follow-up
is making a guess. The more accurate phrase would be "a blind hope for independent
censoring" than "assume". There are cases where simple reasoning or experience tells me
that this hope is futile, but mostly we just hope. The alternative is proactive
follow-up, i.e., devote enough staff and resources to actively contact all of the study
subjects on a regular schedule. Even then you will lose a few. (In one study several
years ago, long term follow-up of cancer, there was a new Mrs Smith who refused to
acknowledge the existence of the prior wife, even to forward letters.)
More information about the R-help