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

Re: st: Failure time in stset.


From   Ronán Conroy <rconroy@rcsi.ie>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Failure time in stset.
Date   Fri, 17 Sep 2004 12:41:17 +0100

adiallo5@worldbank.org wrote:

Hi,
I have a trouble knowing what exactely is the time variable
in stset command.
I want to compute a weibull model on the determinants of
child survival and I have an age variable along with
household and child characteristics.
Is it ok to set age as the analysis time, with correcting for
those children who are dead?
I do not have a origin and exit dates, only the age of child and the
interview date.

su agechild
replace agechild = agedeath if dead==1

stset agechild, failure(dead) id(childid)

This is the most logical way to treat age, and is the subject of an excellent paper by Ed Korn which points out that we often treat time prior to coming under observation (age) as a risk factor but time after coming under observation as the time variable.

Korn EL, Graubard BI, Midthune D. Time-to-event analysis of longitudinal follow-up of a survey: choice of the time-scale. Am J Epidemiol 1997; 145(1):72-80.

In your example, remember that a child could die before coming under observation. So your origin is age zero, but children enter at the age when they were first observed and exit at either age of death or age of last observation.


Is it normal that stsum produces missings for 25, 50 and 75 percentiles?

25% of children have to die before there can be a 25%ile. This probably hasn't happened. Since percentiles are a matter of fact, they -stsum- isn't prepared to infer them. Your Weibull model could be used to infer them, but it would be going beyond that data, which is never a wonderful idea, especially when making gloomy statements about the deaths of children (there are many conditions characterised by a significant early death rate, followed by a flattening out, as those who are going to recover do so.)


Best regards.
Amadou.

*
* 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/




--

Ronan M Conroy (rconroy@rcsi.ie) Senior Lecturer in Biostatistics Royal College of Surgeons Dublin 2, Ireland +353 1 402 2431 (fax 2764) -------------------- Just say no to drug reps http://www.nofreelunch.org/

*
* 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