Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Competing risks with time-varying covariates/episode-splitting


From   Steve Samuels <sjsamuels@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Competing risks with time-varying covariates/episode-splitting
Date   Thu, 12 Aug 2010 12:26:35 -0400

The 2004 Statalist message was written by May Boggess of StataCorp,
not by Enzo Coviello. I suspect you confused it with:
http://www.stata.com/statalist/archive/2004-05/msg00461.html, which
gives the original reference for the data expansion technique.

The Statalist FAQ request that you show us what you typed and what
Stata did. Do this with a toy data set of a few ids ID, and one
time-dependent covariate. However I suspect that you have a
bookkeeping problem. Each line in the expanded data set should be
uniquely identified by (in May's terminology): type, id, and _t.
Rather than repair your code, I suggest that you prepare a separate
data set for each failure type, treating the other types as censored;
then append the data sets.

Note that the the effects of covariates estimated by the techniques in
the Boggess message might not predict effects on cause-specific
cumulative incidence curves. (See, the example of a treatment that
does _not_ affect the hazard of one failure type but affects the
cumulative incidence curve of that type in
http://www.stata-journal.com/sjpdf.html?articlenum=st0059.) For
competing risks analysis, I recommend -stcompet- by Coviello,
available from -ssc- and -stcrreg- in Stata 11.

Steve


On 8/11/10, Bettina Heiss <heiss@usc.edu> wrote:
> Dear group,
>
> I have used Enzo Coviello's helpful suggestions
> (http://www.stata.com/statalist/archive/2004-05/msg00506.html) about
> how to expand a dataset in order to estimate a competing-risks model
> with two different outcomes. However, duplicating the entries
> requires, as I found out experientially and by perusing Mario Cleves's
> article (http://www.stata.com/support/faqs/stat/stmfail.html), that
> the ID variable has to be left unspecified when it comes to stsetting
> the data.
> As my data are split into yearly episodes to accommodate a variety of
> time-varying covariates, I depend on the ID variable to group the
> spells by subject. Leaving it in, however, confuses Stata as it is
> being confronted with duplicate rows for each subject.
>
> It would be great if any of you had some suggestions on how it might
> be possible to run a competing-risks model while episodes are split.
>
> Your help is much appreciated!
> B. Heiss
> *
> * For searches and help try:
> * http://www.stata.com/help.cgi?search
> * http://www.stata.com/support/statalist/faq
> * http://www.ats.ucla.edu/stat/stata/
>


-- 
Steven Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax: 206-202-4783
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index