Dear R-users, I am checking the proportional hazard assumption of a cox model for a given covariate, let say Z1, after adjusting for other relavent covariates in the model. To this end, I fitted cox model stratified on the discrete values of Z1 and try to get beslow estimator for the baseline cumulative hazard function (H(t)) in each stratum. As far as i know, if the proportionality assumption holds, the plot of ln[H(t)] of each stratum versus time should be approximately parallel. i.e fit<-coxph(Surv(start,end,status)~sx+rated+AGLEVEL+strata(Z1),data=ALLDPinfectionandbronchitis) ss<-survfit(fit) plot(ss,fun="cumhaz") My question is on whether the cumulative hazard given by the above command is actually a breslow estimator for baseline cumulative hazard ,i.e, estimator=sum( number of death/ (sum(risk score in risk set)) or a nelson-Aalen estimator. if the above command does not give me breslow estimator, please advise on how I can get it. Thanks for any help given. kind regards, sing yee [[alternative HTML version deleted]]
Christos Hatzis
2006-Mar-07 15:12 UTC
[R] breslow estimator for cumulative hazard function
The function 'basehaz' computes the predicted survivor function for a
Cox
prop hazards model centered (at the average values of the covariates) or
non-centered (at zero level of covariates).
As far as the estimator used, basehaz defaults to the Nelson-Aalen estimate
of the cumulative hazard with a Breslow-type estimate of survival.
Alternatively, a K-M estimator for the cumulative hazard can be selected.
See ?basehaz for details.
In your example, you could do something like this:
bh <- basehaz(fit, centered=TRUE)
require("lattice")
xyplot(hazard ~ time | strata, data=ALLDPinfectionandbronchitis)
-Christos
Christos Hatzis, Ph.D.
Vice President, Technology
Nuvera Biosciences, Inc.
400 West Cummings Park
Suite 5350
Woburn, MA 01801
Tel: 781-938-3830
www.nuverabio.com
-----Original Message-----
From: r-help-bounces at stat.math.ethz.ch
[mailto:r-help-bounces at stat.math.ethz.ch] On Behalf Of singyee ling
Sent: Tuesday, March 07, 2006 9:18 AM
To: r-help at stat.math.ethz.ch
Subject: [R] breslow estimator for cumulative hazard function
Dear R-users,
I am checking the proportional hazard assumption of a cox model for a
given covariate, let say Z1, after adjusting for other relavent covariates
in the model. To this end, I fitted cox model stratified on the discrete
values of Z1 and try to get beslow estimator for the baseline cumulative
hazard function (H(t)) in each stratum. As far as i know, if the
proportionality assumption holds, the plot of ln[H(t)] of each stratum
versus time should be approximately parallel.
i.e
fit<-coxph(Surv(start,end,status)~sx+rated+AGLEVEL+strata(Z1),data=ALLDPinfe
ctionandbronchitis)
ss<-survfit(fit)
plot(ss,fun="cumhaz")
My question is on whether the cumulative hazard given by the above command
is actually a breslow estimator for baseline cumulative hazard ,i.e,
estimator=sum( number of death/ (sum(risk score in risk set)) or a
nelson-Aalen estimator. if the above command does not give me breslow
estimator, please advise on how I can get it.
Thanks for any help given.
kind regards,
sing yee
[[alternative HTML version deleted]]
______________________________________________
R-help at stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide!
http://www.R-project.org/posting-guide.html