Sarah A. Mustillo <[email protected]> asked about the
difference between a shared frailty model and the -cluster- option when
using -stcox-:
> I am running a cox model on data that are clustered by study site.
> "One solution would be to fit a standard Cox model, adjusting the standard
> errors of the estimated hazard ratios to account for the possible
> correlation by specifying cluster(patient).
>
> Alternatively, one can model the correlation by assuming that the
> correlation is the result of a latent patient-level effect, or frailty.
> That is, rather than fitting a standard model and specifying
> cluster(patient), fit a frailty model by specifying shared(patient)."
Sarah asks:
> How does one decide which method to use?
I want to say "go with the shared frailty", because the -cluster- option
is not fitting a more sophisticated model; it's just giving you a more
generous standard error estimate. My instinct says to me that building
the correlation within study-site into the model has to be better.
BUT that is only good advice IF shared frailty is the correct model.
Thus, to compare the models, I would look at goodness-of-fit tests and
the residuals (as suggested in chapter 11 of Stata's Survival Analysis
book: http://www.stata.com/bookstore/saus.html).
Also, I found some similar questions by searching the Statlist archive
(http://www.stata.com/statalist/archive/)
Here are two repsonses on the same subject:
http://www.stata.com/statalist/archive/2002-06/msg00450.html
http://www.stata.com/statalist/archive/2003-05/msg00431.html
-- May
[email protected]
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
