[R] Obtaining value of median survival for survfit function to use in calculation

[David Winsemius]

The survfit.object help page says:

"The print.survfit method does more computation than is typical for a  
print method and is documented on a separate page."

It takes a bit of digging, but after first trying:

getAnywhere(print.survfit)   # and then following code and trying
getAnywhere(survmean)  # survmean is the function which does the work  
using pfun

... I think I finally understand exactly what the print.survfit help  
page means when it refers to a "side-effect".

So there is no durable element in the survfit object that is the  
median. The automation you request would involve duplicating the code  
of survmean but with assignment of the "out" matrix to something that  
does not get discarded.

-- 
David Winsemius

On Sep 8, 2009, at 6:42 PM, Polwart Calum (County Durham and  
Darlington NHS Foundation Trust) wrote:

> Hi,
>
> I'm sure this should be simple but I can't figure it out!  I want to  
> get the median survival calculated by the survfit function and use  
> the value rather than just be able to print it.  Something like this:
>
> library(survival)
> data(lung)
> lung.byPS = survfit(Surv (time, status) ~ ph.ecog, data=lung)
> # lung.byPS
>  Call: survfit(formula = Surv(time, status) ~ ph.ecog, data = lung)
>  1 observation deleted due to missingness
>              n events median 0.95LCL 0.95UCL
>  ph.ecog=0  63     37    394     348     574
>  ph.ecog=1 113     82    306     268     429
>  ph.ecog=2  50     44    199     156     288
>  ph.ecog=3   1      1    118     Inf     Inf
>
> What I want is to be able to call the median using something like:
>
> lung.byPS$median[ph.ecog=0]
>
> so that I can add it to a plot like this:
>                plot (lung.byPS,
>                                conf.int=F,
>                                lty=1:4,
>                )
>                abline(h=0.5, lty=5)
>                abline(v=lung.byPS$median[ph.ecog=1])
>                abline(v=lung.byPS$median[ph.ecog=2])
>
> Anyone got any easy solutions?  Its fairly normal to plot across and  
> down to show medians on survival curves... so I'm sure it must be  
> possible to automate.
>

David Winsemius, MD
Heritage Laboratories
West Hartford, CT