Tony Lachenbruch found a problem with -sampsi_sccs- a program that
calculates sample sizes for the self-controlled case series design.
He is correct, his inputs are not "dopey" but part of my code is.
Briefly, I believe the code for methods "binomial" and "srl" are OK. Both
rely on Musonda's equation (3) to calculate the number in the exposed group
(once the overall total has been calculated).
I also believe that the *overall total number* calculated for the
age-adjusted method is correct (or at least in so far as Musonda's method
and formulae are correct and the assumptions therein are maintained).
The problem is with the *number in the exposed group* for the age adjusted
method. This should of course be less than or equal to the overall total.
Although it is not made completely clear in Musonda's paper, I think my
application of his Equation (3) to the overall total to get the number in
the exposed is inappropriate in the age-adjusted context. Unfortunately,
after a closer read of the Appendix of his paper and optimistically writing
some new code, the problem with negative numbers perists :-( .
I will communicate with Patrick Musonda and get back to the List. In the
meantime I will send a new version of -sampsi_sccs- to Kit for putting up on
SSC. It will simply omit reporting the number in the exposed for the
age-adjusted method.
Thanks to Tony for alerting me to the problem.
Phil
Philip Ryan
University of Adelaide
Australia
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Lachenbruch,
Peter
Sent: Tuesday, 21 April 2009 1:35 AM
To: [email protected]
Subject: st: RE: new program sampsi_sccs
I downloaded this program and played with it a bit. Here is one bit of
output. At the end you may note that the number of events required in
unexposed subjects if -6. I may have chosen dopey inputs, but I don't think
I can envision any group with -6 events. Phil - or someone - can you help?
I tried another combination and got -6 events needed...
. sampsi_sccs, rho(3) method(age)
Signed root likelihood method controlling for age effects
input number of age groups (e.g. 5)
. 5
input age specific incidence [relative to age group 1] in age group 2 . 3
input age specific incidence [relative to age group 1] in age group 3 . 5
input age specific incidence [relative to age group 1] in age group 4 . 7
input age specific incidence [relative to age group 1] in age group 5 . 7
input length of observation period for age group 1
. 10
input length of observation period for age group 2
. 10
input length of observation period for age group 3
. 10
input length of observation period for age group 4
. 10
input length of observation period for age group 5
. 10
input post exposure risk period [assumed the same for each group] . 10
input proportion of subjects exposed during obs period in age group 1 . .2
input proportion of subjects exposed during obs period in age group 2 . .2
input proportion of subjects exposed during obs period in age group 3 . .3
input proportion of subjects exposed during obs period in age group 4 . .3
input proportion of subjects exposed during obs period in age group 5 . .3
==================================================================
========== sample size for SCCS Design with age effects ==========
==================================================================
2-tailed alpha is: .05
power is: .8
relative incidence associated with exposure is: 3
common post-exposure risk period is: 10
total number of events required in exposed subjects is: 32
total number of events required in unexposed subjects is: -6
Total number of events required is: 26
==================================================================
.
Tony
Peter A. Lachenbruch
Department of Public Health
Oregon State University
Corvallis, OR 97330
Phone: 541-737-3832
FAX: 541-737-4001
-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Philip Ryan
Sent: Sunday, April 19, 2009 4:31 PM
To: [email protected]
Subject: st: new program sampsi_sccs
Listers
Thanks again to Kit Baum, a new program -sampsi_sccs- is now available on
the SSC server.
-sampsi_sccs- calculates sample sizes for self-controlled case series study
designs. Full refs to a tutorial on this type of study and to Musonda et
al's paper on sample size estimation are included in the accompanying help
file.
Requires Stata v10
To check it out:
ssc describe sampsi_sccs
Philip Ryan
University of Adelaide
South Australia
*
* 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/
*
* 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/
No virus found in this incoming message.
Checked by AVG - www.avg.com
Version: 8.0.238 / Virus Database: 270.12.1/2070 - Release Date: 04/20/09
17:56:00
*
* 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/
