Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

st: stcascoh.ado - CASE-cohort


From   "Grote, Veit Dr." <Veit.Grote@kk-i.med.uni-muenchen.de>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: stcascoh.ado - CASE-cohort
Date   Thu, 17 Apr 2003 15:35:30 +0200

Hi, 
I am using the stcascoh.ado to model a case-cohort dataset. To figure out how things work and if things go along properly, I tried to reproduce the example of BArlow (Barlow WE, Ichicawa L, Rosner D, and Izumi S: Analysis of Case-Cohort Designs; Journal Clinical Epidemiology 1999; 52: 1165-1172) from the Breslow and Day nickel dataset. 
The dataset provided on the net (http://lib.stat.cmu.edu/general/robphreg) used by Barlow is a the full dataset, however their results are based on an specific subcohort-sample. - stcascoh - samples a subcohort to prepare the dataset for analysis. Thus I expect this dataset always to be slightly different to the one of Barlow, however, results should be quite similar.
My problem is that I am not getting there and cannot reproduce the results (I tried it in SAS and it worked).  
At examining the dataset after - stcascoh - I had the impression that weights were wrongly assigned in the programm in lines 179 and 184. But I am not very familar with the programming language. However, also unweighted results do not work out properly. I would be very grateful for some hint.

Here is what I do in Stata 7:

/***********************************************
*                                              *
*   INPUT THE DATASET AS PROVIDED ON THE NET   *
*                                              *
***********************************************/
input id icdcode exposure dob age_emp age_strt age_stop subco20
3 0 5 1889.0192 17.4808 45.2273 92.9808 0
4 162 5 1885.978  23.1864 48.2684 63.2712 0
...
0 0 0 1895.5 27.6753 38.7465 39.7219 0
end

/*************************************************
*                                                *
*   LABELING, CODING AND RECODING AS PROVIDED    *
*   IN THE SAS PROGRAM ON THE NET                *
*                                                *
*************************************************/
replace id=990 if id==0 & icdcode==490
replace  id=991   if id==0 & icdcode==177 
replace id=992   if id==0 & icdcode==430 
replace id=993  if id==0 & icdcode==0   

lab var id		"identification number"
lab var icdcode		"160=death due to nasal sinus cancer"
lab var exposure	"exposure level"
lab var dob	"Date of birth"
lab var   age_emp 		"Age at first employment"
lab var age_strt  "Age at start of study"
lab var age_stop	"Age at death or end of study, whichever is earlier"
lab var subco20 "1=included in 20% subcohort, 0=not included"
gen yfe10=( dob+age_emp-1915)/10
gen yfe100=(( age_emp-1915)^2)/100
gen logafe=log(age_emp-10)
gen logexp=log(exposure+1)

/***********************************************
*                                              *
*   PREPARING THE DATASET FOR ANALYSIS         *
*   USING TIME SINCE EMPLOYMENT AS TIME AXIS   *
*   AND THE 20% SAMPLE FRACTION                *
*   AS IN BARLOW                               *
*                                              *
***********************************************/

stset  age_stop,failure(icd==160) id(id) enter( age_strt) exit( age_stop) origin( age_emp)

stcascoh, alpha(20)

/*******************************************************************
*                                                                  *
*   LASTLY I MODEL AS SUGESTED IN THE HELP FILE                    *
*   OF - STCASCOH -                                                *
*                                                                  *
*   1- Prentice: stcox varlist, robust                             *
*   2- Self and Prentice: stcox varlist, offset(_wSelPre) robust   *
*   3- Barlow: stcox varlist, offset(_wBarlow) robust              *
*                                                                  *
*******************************************************************/

stcox  yfe10 yfe100 logafe logexp, robust nohr 
stcox  yfe10 yfe100 logafe logexp, robust offset(_wSelPre) nohr
stcox  yfe10 yfe100 logafe logexp, robust offset(_wBarlow)nohr

What is wrong?
Thanks a lot.
Veit Grote
 
.
----------------------------------------
Dr. Veit Grote, MSc
Klinikum der Universität München
Dr. v. Haunersches Kinderspital
Lindwurmstr. 4
D-80337 München
Tel.: +49 (89) 5160-7798
Fax: +49 (89) 5160-2951
veit.grote@med.uni-muenchen.de
----------------------------------------



*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   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   |   What's new   |   Site index