[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: repeat: predict after -streg-

From   Steven Samuels <>
Subject   Re: st: repeat: predict after -streg-
Date   Thu, 12 Jul 2007 10:32:41 -0400

Jeph, -streg- is fitting censored parametric distributions (one for each combination of predictors) . Most of the "true" unobserved failure times are >30 days, so the medians of the fitted distributions will be >30 days. In rough terms, you are estimating entire distributions without data beyond the 22nd percentile.


On Jul 12, 2007, at 9:10 AM, Jeph Herrin wrote

I posted this a few days ago, but got no replies;
thought I'd try again, shorter

My data are time to readmission (failure) after
hospitalization, with censoring on death:

-> stset ftime, failure(radm)

failure event: radm != 0 & radm < .
obs. time interval: (0, ftime]
exit on or before: failure

---------------------------------------------------------------------- --------
424787 total obs.
0 exclusions
---------------------------------------------------------------------- --------
424787 obs. remaining, representing
95529 failures in single record/single failure data
1.07e+07 total analysis time at risk, at risk from t = 0
earliest observed entry t = 0
last observed exit t = 30

My model is

. streg `mycovars' , d(lnormal)

And my problem is:

. predict median, median time
. count if median<=30

So, though the data have 95,000+ failures in less than 30 days, my
model predicts only 30 such failures.

I get the same results with other parameterizations, though lognormal
seems the best fit. Either I don't understand what the model is doing,
or I'm not calculating the correct predicted value. Should I expect

Thanks in advance for any insights.


* For searches and help try:
Steven JH Samuels
18 Cantine's Island
Saugerties, NY 12477
EFax: 208-498-7441

*   For searches and help try:

© Copyright 1996–2015 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index