[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
Re: st: using stcox with clustered data
on 1/10/02 7:30 PM, Seidel, Kristy at email@example.com wrote:
> I have been asked to come up with a plan for analyzing an unusual clinical
> data set that consists of paired time-to-event observations. The setting is
> an infant intensive care unit (IICU). The research question is: which of
> two topical medications is better at shortening the duration of diaper rash
> in IICU patients. Because there is so much variability in the types of
> problems that infants are admitted to the IICU with, the investigators wish
> to use a paired design in which the nursing staff will use medication A on
> one side of the infant's rear and medication B on the other side.
Maybe you will be in a situation where the time to recovery will be known in
all cases. It seems likely that they will continue treatment until the baby
gets better. In this case, you can compute the difference in time for each
baby and use that as your dependent variable.
Using Cox regression will give you a hazard ratio but this is hard to
interpret as a measure of effect size. Ideally, what you want to say is
'[insert product name here] shortens nappy rash by X days on average'. I say
try and ensure you have no censored data (or use -intreg-, which is a very
handy tool!) and estimate the mean difference.
Ronan M Conroy (firstname.lastname@example.org)
Lecturer in Biostatistics
Royal College of Surgeons
Dublin 2, Ireland
+353 1 402 2431 (fax 2329)
Wisdom, Kai Lung reflected, will make you a better cyclist;
It will not, alas, make your bike new again.
* For searches and help try: