[R] Kaplan Meier analysis: 95% CI wider in R than in SAS

[Frank Harrell]

I used log-log in my book too until Terry Therneau alerted me to the
significant problems this creates.  In the 2nd edition it will use log S(t).
Frank

Paul Miller wrote
> 
> Hello Drs. Colosimo and Harrell,
>  
> Thank you for your replies to my question. From Dr. Colosimo, I was able
> to determine that the SAS results can be replicated by adding the
> option conf.type="log-log" to my code as in :
>  
> survobj <- survfit(survfrm, conf.type="log-log", data=Survival)
>  
> Originally, it looked like the SAS results could be replicated using
> conf.type="plain". Applying this option to my actual data revealed that
> this was not the case, however. 
>  
>>From Dr. Harrell, I learned that using conf.type="log-log" may not be such
a good idea. Interestingly though, I've seen at least one instance where
experts in the R community use this option in their book. The book is about
10 years old. So maybe opinion about the use of this option has shifted
since then. 
>  
> Thanks,
>  
> Paul  
>  
> 	[[alternative HTML version deleted]]
> 
> 
> ______________________________________________
> R-help@ mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide
> http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
> 


-----
Frank Harrell
Department of Biostatistics, Vanderbilt University
--
View this message in context: http://r.789695.n4.nabble.com/Kaplan-Meier-analysis-95-CI-wider-in-R-than-in-SAS-tp4554559p4557695.html
Sent from the R help mailing list archive at Nabble.com.