R help - Feb 2012 - Survival analysis and comparing survival curves

Hei,

I have a one simple question which does not seem to be that simple as I
cannot find any solution/answer:

Is it possible to compare multiple survival curves in R with
survdiff-function when there is interaction term involved in predictor
variables (and this interaction is significant)?

Example:

survdiff(Surv(death,status)~treatment*gapsize)

R is making "problems" with it ie.e. it does not want to perform the
test.
And all the examples I have found so far, may involve multiple predictor
variables but in additive format (e.g. treatment+gapsize).

If survdiff is not the way to go, are there any other solutions in order to
compare statistically the curves i.e. if they differ significantly from each
other?

I would really appreciate if someone can answer!

Cheers, Minna

--
View this message in context:
http://r.789695.n4.nabble.com/Survival-analysis-and-comparing-survival-curves-tp4414316p4414316.html
Sent from the R help mailing list archive at Nabble.com.

DEAR R-girl:

On Thu, Feb 23, 2012 at 8:12 AM, R-girl <jhovinen at gmail.com>
wrote:
> Hei,
>
> I have a one simple question which does not seem to be that simple as I
> cannot find any solution/answer:
>
> Is it possible to compare multiple survival curves in R with
> survdiff-function when there is interaction term involved in predictor
> variables (and this interaction is significant)?
I have one simple answer. Get local statistical help. You do not
understand the meaning of an interaction.

-- Bert
>
> Example:
>
> survdiff(Surv(death,status)~treatment*gapsize)
>
> R is making "problems" with it ie.e. it does not want to perform
the test.
> And all the examples I have found so far, may involve multiple predictor
> variables but in additive format (e.g. treatment+gapsize).
>
> If survdiff is not the way to go, are there any other solutions in order to
> compare statistically the curves i.e. if they differ significantly from
each
> other?
>
> I would really appreciate if someone can answer!
>
> Cheers, Minna
>
> --
> View this message in context:
http://r.789695.n4.nabble.com/Survival-analysis-and-comparing-survival-curves-tp4414316p4414316.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org 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.


-- 

Bert Gunter
Genentech Nonclinical Biostatistics

Internal Contact Info:
Phone: 467-7374
Website:
http://pharmadevelopment.roche.com/index/pdb/pdb-functional-groups/pdb-biostatistics/pdb-ncb-home.htm

On Feb 23, 2012, at 11:12 AM, R-girl wrote:
> Hei,
>
> I have a one simple question which does not seem to be that simple  
> as I
> cannot find any solution/answer:
>
> Is it possible to compare multiple survival curves in R with
> survdiff-function when there is interaction term involved in predictor
> variables (and this interaction is significant)?
I've never had a problem with survival failing to construct an  
interaction term.
>
> Example:
>
> survdiff(Surv(death,status)~treatment*gapsize)
>
> R is making "problems" with it ie.e. it does not want to perform
the
> test.
Is 'gapsize' a numeric vector rather than a factor?
> And all the examples I have found so far, may involve multiple  
> predictor
> variables but in additive format (e.g. treatment+gapsize).
>
> If survdiff is not the way to go, are there any other solutions in  
> order to
> compare statistically the curves i.e. if they differ significantly  
> from each
> other?
Without a better description of the data layout (say using hte str()  
function) and the code you have used so far, the words "the curves",  
the indefinite pronoun "they" and the phrase "each other"
will remain
far too nebulous for further comment.
>
> I would really appreciate if someone can answer!
Only if "someone" can provide the information requested in the Posting
Guide.
>
> PLEASE do read the posting guide
http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
David Winsemius, MD
West Hartford, CT