You would gain some power if you switched to a parametric model, but the question would be whether the assumed distribution would fit reasonably well, something the Cox PH approach sidesteps (to some degree anyway) by making the hazard nonparametric.
I forget the name of the Stata command for parametric lifetime regression but if you use help on the Cox procedure it should show up (or look in the menu). The log-logistic distribution is often nice.
-----Original Message-----
From: "moleps islon" <[email protected]>
To: [email protected]
Sent: 3/29/2009 11:26 AM
Subject: st: cox regression alternatives
I'm struggling with a dataset of 90 patients with only 10 events and
several covariates I'd like to investigate. Seeing the recommended
number of events for a cox regression to be at least 10-what are my
other options in investigating this rare tumor except from detailing
the descriptives and producing a Kaplan Meier plot?
Regards,
Moleps
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
