Skip to content

change the reference level in plot(summary(cph-fit))

2 messages · zhu yao, David Winsemius

#
On Jan 4, 2014, at 8:37 PM, zhu yao wrote:

            
I do not think we have enough information to be sure what you mean, but I would remind you that Cox models are always estimations of the relative risk for differences in covariates. If the era for which 1999-2002 is one instance as a covariate (rather than as the time variable), then you could certainly construct differences from that era.

--

David Winsemius
Alameda, CA, USA