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

Re: st: stcox problem

From   "Svend Juul" <[email protected]>
To   <[email protected]>
Subject   Re: st: stcox problem
Date   Fri, 18 Jan 2008 15:03:46 +0100

Simon Bach wrote:
New to Stata so this may be very simple.

Running stcox: have generated a model to determine factors 
associated with recurrence of bowel cancer following surgery.
I isolated 3 tumour variables that are independently 
associated with recurrence (i.e. size, depth invasion, 
presence lymphatic invasion).
Two different types of surgery were performed (A and B). 
A is associated with much higher recurrence rates. This 
further improves the model.
In stcurve I can fix values for these variables and estimate 
the proportion of subjects who remain disease free at 60 
months (eg 1cm diameter tumour, superficial invasion, no 
lymphatic invasion, excised using method A).
I can then generate estimates for other flavours of tumour 
(eg 2cm, 3cm and so on).
BUT - I am struggling to obtain the corresponding raw figures 
for the proportion of patients who remain disease free at 
60 months (using the different variable values)?
While I can read from the graph this is obviously clumsy/inaccurate.
Also stcurve does not appreciate stratified models, not a huge 
deal but predictably enough my model does fit better when stratified.
Simon Bach
(surgeon not statistician - unfortunately)
(Well, I would prefer to be operated on by a surgeon rather
than a statistician)
I guess that -sts list- is what you need, like this:
     sysuse cancer
     stset studytime, failure(died==1)
     sts list if age>55 & drug==1
Hope this helps
Svend Juul
Institut for Folkesundhed, Afdeling for Epidemiologi
(Institute of Public Health, Department of Epidemiology)
Vennelyst Boulevard 6 
DK-8000 Aarhus C,  Denmark 
Phone, work:    +45 8942 6090 
Phone, mobile:  +45 2634 7796 
Fax:            +45 8613 1580 
E-mail:         [email protected] 

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index