R help - May 2010 - Problem with plotting survival predictions from cph model

Dear R-helpers,

I am working with 'cph' models from 'rms' library. When I build
simple
survival models, based on 'Surv(time, event)', everything is fine and I 
can make nice plots using plot(Predict(f, time=3)).

However, recently I tried to be more specific and used 'Surv(start, 
stop, event)' type model. Using this model 'plot(Predict(f))' works
OK,
but 'plot(Predict(f, time=3))' throws an error:

"Error in summary.survfit(g, print.it = FALSE, times = times) :
   object 'n.enter' not found"

Actually, using 'survest(f, time=3)' has the same result.

Has anybody encountered this kind of error? Is there a workaround?

Best regards,

-- 
Michal J. Figurski, PhD
HUP, Pathology & Laboratory Medicine
Biomarker Research Laboratory
3400 Spruce St. 7 Maloney
Philadelphia, PA 19104
tel. (215) 662-3413

On 05/26/2010 10:55 AM, Michal Figurski wrote:
> Dear R-helpers,
>
> I am working with 'cph' models from 'rms' library. When I
build simple
> survival models, based on 'Surv(time, event)', everything is fine
and I
> can make nice plots using plot(Predict(f, time=3)).
>
> However, recently I tried to be more specific and used 'Surv(start,
> stop, event)' type model. Using this model 'plot(Predict(f))'
works OK,
> but 'plot(Predict(f, time=3))' throws an error:
>
> "Error in summary.survfit(g, print.it = FALSE, times = times) :
> object 'n.enter' not found"
>
> Actually, using 'survest(f, time=3)' has the same result.
>
> Has anybody encountered this kind of error? Is there a workaround?
>
> Best regards,
>
Please provide a tiny self-contained reproducible example, e.g., using 
simulated data.

Frank

-- 
Frank E Harrell Jr   Professor and Chairman        School of Medicine
                      Department of Biostatistics   Vanderbilt University

> but 'plot(Predict(f, time=3))' throws an error:
> "Error in summary.survfit(g, print.it = FALSE, times = times) :
>    object 'n.enter' not found"
Congratulations!  You just discovered a new, never-before-encountered
bug in the survival package.  I should have a repair posted to Rforge by
the end of the day.

Terry Therneau (author of the package)

Josh,

Good point. And it actually accidentally solved my problem.

Previously I attached .Rdata files. After your email I exported this 
dataset into a csv file and then re-loaded it to see if the problem 
persists, and... the problem disappeared.

Nevertheless, if I load the original .Rdata file, the problem is back 
again. This makes no sense to me - there must be a bug in R, or in the 
survival package.

Regards to all,

--
Michal J. Figurski, PhD
HUP, Pathology & Laboratory Medicine
Biomarker Research Laboratory
3400 Spruce St. 7 Maloney
Philadelphia, PA 19104
tel. (215) 662-3413

On 2010-05-27 13:39, Joshua Wiley wrote:
> On Thu, May 27, 2010 at 10:32 AM, Michal Figurski
> <figurski at mail.med.upenn.edu>  wrote:
>>
>> Actually, I have another problem with the same data - this time with
plotting simple KM lines. The dataset is attached.
>
> Michal,
>
> Just as a heads up, only certain types of files can be attached,
> others get scrubbed.  At least I did not see any data attached from
> your last two emails.
>
> Josh
>
> --
> Joshua Wiley
> Senior in Psychology
> University of California, Riverside
> http://www.joshuawiley.com/