|
The program was updated on October 30, 1998 (version 2.02b).
The manual was updated on August 9, 2000.
Download the program package (including
the manual) as a self-extracting ZIP file. (~820Kb)
Download the latest version of the
manual in PDF format. (~530Kb)
The PDF file can be viewed and printed using Adobe Acrobat
Reader.
The program is freeware, although we request users register
so we can justify continued funding for the project.
Version 1.6
of the program is also available.
A Windows version
(SURV3) is currently being developed. A trial version can be
downloaded.
About the relative survival
analysis program
The relative survival analysis program is a DOS program designed
for the analysis of patient survival based on data collected
by population-based cancer registries. The program uses two
input data files, containing the patient data and general population
(all-cause) mortality rates. Life tables based on either monthly
or annual intervals are produced, containing estimates of, among
other quantities, observed survival rates, expected survival
rates (using a choice of three methods), and relative survival
rates. The program can be run without the general population
mortality input file, in which case only observed survival rates
can be estimated. High-resolution graphs of the survival rates
can be displayed and tests carried out for differences in survival
rates between patient groups. The program also outputs a data
file containing life table information, which can be input to
a commercial statistics package (e.g. GLIM or SAS) to fit a
life table regression model for the relative survival in the
framework of generalised linear models.
Technical requirements
Required
- IBM compatible micro-computer
- DOS, runs also in DOS boxes under Windows 3.1, Windows 95,
and Windows NT 4.0
- a mathematical co-processor (80x87) is not required
- a minimum of 1.5Mb hard disk space
- a minimum of 500Kb free memory (RAM)
- VGA grey scale monitor, 16 scale
Highly recommended:
- VGA/SVGA colour monitor
- mouse
Installation
The program package (including the manual) is distributed as
a self-extracting ZIP file. A self-extracting ZIP file is a
Windows executable file which contains the program files in
compressed format. Running the executable file results in all
program files being extracted to a user-specified directory,
after which the program is ready to run without requiring further
setup. By default, the program is installed in C:\SURV2, although
an alternative directory can be specified during the file extraction
process. It is recommended that the main program directory (C:\SURV2
by default) be added to the path (by editing AUTOEXEC.BAT).
If a previous version of the program has been installed, care
must be taken not to overwrite important parameter files previously
altered by the user (TABU.PAR for example) and it may be best
to first extract the program files to a temporary directory.
All program files are stored in the main program directory or
its subdirectories. The program can therefore be uninstalled
by deleting the main program directory and its subdirectories.
The self-extracting ZIP file is a Windows executable and will
not run under DOS. If you are not running Windows, please contact
the authors and we will send you a version which runs under
DOS.
Updating the
installed version
When the program is updated, the complete program package is
redistributed as a self-extracting ZIP file which can be installed
in the same manner as the original installation. By default,
the installation program overwrites files in the target directory.
As such, the default installation procedure should not be followed
if parameter files in the main program directory (TABU.PAR for
example) have been updated by the user. In this situation, it
may be best to first extract the files to a temporary directory
and then manually copy the files which have been updated (generally
SURV2.EXE, SURV2.OVR, MANUAL.PDF and the contents of the UTILITY
directory) to the main program directory.
Main
differences between versions 1 and 2
The major differences between version 1 and version 2 are in
terms of the user interface. The program now consists of a single
executable file, although the same parameter files are used
to define the input data files and the desired analyses. The
program is now more user-friendly in terms of error-checking
and ease of constructing parameter files. All parameter files
can be edited within the program using templates. A `check-up'
function provides the user with feedback on how the program
has interpreted the input files. High resolution graphics features
have been added, which produce survival curves from the life
table output and graphs of mortality from the population mortality
file (popmort.dat).
A serious limitation in the current version is that the number
of life tables which can be processed in a single run is limited
by the amount of available memory. Depending on the amount of
memory available, the maximum number of tables which can be
produced in a single run varies between 80 and 300. Version
1 is not subject to this limitation and has been used successfully
to produce more than 25,000 life tables in a single run. We
are currently working on removing this limitation from version
2.
Program history
The first version of this program, `A Computer Program Package
for Cancer Survival Studies', was prepared in 1987-88 by Timo
Hakulinen, Robert Gibberd, Kamal Abeywickrama, and Bengt Söderman
in a co-operative effort between the Finnish Cancer Registry
and the University of Newcastle, Australia. The program was
written in FORTRAN and ran under DOS and VAX/VMS. Due to the
memory limitation of personal computers at the time the program
was written, the package consisted of three separate executable
files (TABULATN, PRINTNG, and GROUPTST). The latest update of
version 1 (version 1.6) is available from the Finnish Cancer
Registry. The DOS version can be downloaded from the Finnish
Cancer Registry WWW site.
The new version (version 2) is written in Turbo Pascal and is
available for DOS only. The package now runs via a single executable
file. The main theory and formulae remain the same, but the
user interface has been completely revamped. The current version
(version 2) was developed at the Finnish Cancer Registry during
1996--98 by Esko Voutilainen, Paul Dickman, and Timo Hakulinen.
Parameter files written for version 1.x can be read and interpreted
by version 2. We are currently developing a Windows version
(version 3 of the package) written in Delphi and hope to release
a beta version during March 2000. The authors welcome any suggestions
for how the program might be updated or improved.
Contacting the authors
All correspondence concerning the program should be addressed
to Paul Dickman:
Dr. Paul Dickman
Department of Medical Epidemiology
Karolinska Institutet
PO Box 281
171 77 Stockholm
Sweden
E-mail: paul.dickman@mep.ki.se
Fax: +46 8 314 975
Web: http://www.pauldickman.com/
Related files and
links
Version 1.6
of the program for DOS is also available.
A SAS program for estimating
expectation of life from annual probabilities of death and
creating a 'popmort' file.
ch11.glm is a GLIM4 command file containing
the data and GLIM directives to reproduce the analyses shown
in Chapter 11 of the software manual using different settings
for the scale parameter.
tadek.glm is a GLIM4 command file containing
the data and GLIM directives to fit the model shown in Chapter
11 of the software manual and then use the Dyba and Hakulinen
macro to estimate the model-based cumulative relative survival
rates and 95% confidence intervals reported in Table 11.2. Note
that the original Dyba and Hakulinen macro was written for GLIM
3.77. It has been modified slightly to run under GLIM4 with
this specific example. There is no guarantee that the macro
will run under GLIM4 with other examples, although Tadek Dyba
plans to update the macro to GLIM4 some time in the future.
dyba_mac.txt
is the GLIM 3.77 macro for producing model-based estimates of
cumulative relative survival rates with 95% confidence intervals
from the paper by Dyba and Hakulinen in GLIM Newsletter 23.
(This macro will not run under GLIM4)
PROC
GENMOD FAQ from the SAS technical support pages.
A generalized
linear models page, including links to software for fitting
GLMs and links to books on GLMs.
Page last updated by Paul
Dickman, 3 October, 2000
|
