Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down on April 23, and its replacement, is already up and running.

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: Median survival times via Poisson model, using -margins- ?

To   Statalist <>
Subject   st: Median survival times via Poisson model, using -margins- ?
Date   Tue, 26 Mar 2013 10:55:47 +0800

Dear 'Listers,

I have found multiple references to calculating median survival times
for sub-populations from Poisson survival models (eg. Kelsey JK,
Whittemore AS, Evans AS, Thompson WD. Methods in observational
epidemiology. 2nd ed. New York: Oxford University Press, 1996, pgs.
31-3). A general scheme reported is :

We computed mu as the unweighted average of these 18 death rates [Per
categorical from Poisson model]. The risk of death at time t, R(t),
was computed as R(t) = 1 - e ^-(mu*t), where t was in years. Median
survival was computed by finding the value of t satisfying the
relation, R(t) = 1 - e ^-(mu * t) = 0.5.

I suspect *anything* is possible using -margins- but am stymied on how
to get there with confidence intervals for the median times.

A (not very interesting or well-modelled) example:

***** Start example *****
use, clear

gen survdays = dox - doe
gen survyear = survdays/365.25
stset survyear, failure(chd == 1) id(id)

stsplit, at(fail)

collapse (sum) survyear chd, by(hieng job)
sum surv, detail

nbreg chd hieng i.job, exposure(survyear) irr

// margins

***** End example *****

Any thoughts on how to predict median survival times with 95% CIs for
the three levels of 'job' at say, hieng ==1?

Andrew Lover

Centre for Infectious Disease Epidemiology Research (CIDER)
Saw Swee Hock School of Public Health
National University of Singapore
*   For searches and help try:

© Copyright 1996–2016 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index