In response to my request for recomended software to perform a
non-parametric survival curve estimate on left & right censored data, I
received 8 replies. Thanks to all of you. The results were
However one person did claim specifically that S-plus could not perform the
desired analysis. Incidentally, I posted the original question because
neither SPSS nor R can do it.
I've produced a digest of the responses below.
hat think that STATA version 5 (or later) will help you.
> I'd be grateful if anyone could recommend some software that can produce a
> non-parametric estimate of a survival curve and do log-rank tests, for data
> which has left as well as right censoring. Some references would be very
> useful also.
Lecturer in Medical Statistics
Department of Epidemiology and Public Health
University of Leicester
EGRET, STATA, SAS and S-plus allow for right censoring and left-truncation.
We have been working with elderly populations where the survival data have
the above characteristics. We have a paper about this issue.
'Left-truncated data with age as time scale: an alternative for survival
analysis in the elderly population' Lamarca et al Journal of Gerontology Med
sci 1998 vol 53A(5):
M337_M343. In the appendix you'll find the commands that are used in this
case in the aforementioned statistical packages. I the bibliography there
are included the main papers about this under my criteria.
Health Services Research Unit
Institut Municipal d'Investigació Mèdica (IMIM)
C/ Dr. Aiguader, 80. E-08003 Barcelona, SPAIN.
Stata has the possibility for subjects to enter at arbitrary times and
exit and re-enter if necessary.
This not be of much use to you but for our survival
analysis unit we used S-Plus to perform the analysis for
our assignement, this may be or not some use to you.
>try STATA, information canbe found on the web
I can recommend Stata (Statacorp 1999), which does all the above and loads
of other things in survival analysis (and many other areas of stats). Their
web site is http://www.stata.com/.
I hope this is useful.
All the best
Statacorp. Stata Statistical Software: Release 6.0. 1999; College Station,
Texas, USA: Stata Corporation.
Lecturer in Medical Statistics
Department of Public Health Medicine
Guy's, King's and St Thomas' School of Medicine
Try the ICE macro in SAS (I think in the IML package), which uses
Turnbull's (1976 JASA I think) method for dealing with interval censored
data, as well as left, right and uncensored data. To my knowledge, there is
nothing in Splus which does this (please let me know if you find out
The PROC LIFETEST procedure of SAS does that, according to the methods
described in Kalfleisch and Prentice (...). If you want to use PH-models,
then the procedure LIFEREG would be more appropriate.
The code for LIFETEST is as easy as the following:
proc lifetest data=tmp3 outsurv=tmp4;
The TIME statement specifies that SECTION is the time variable and that the
value 1 stands for a censored observation. The STRATA statement specifies
that each value of the variable CV defines a group of observations and that
the survival curves of the different groups are to be compared.