Good morning,
I'm beginning to work on a problem for which I believe survival analysis of
some kind is appropriate. Not being very familiar with survival analysis, ,
however, I need some advice as well as how to implement it in S (on Windows XP
using S+ 7.0).
The situation is that a client has a medication for which physicians write
prescriptions with dosage on an as-needed basis with multiple refills possible.
Whenever the patients fill the onset of the illness, they take the prescribed
dosage. The refills depend on the frequency of the reoccurrence of the
illness, an unknown. The longer the time that goes by without a refill, the
more likely the patient either recovered or died.
The client wants to know the average (or median) time between refills
(basically, the half-life of a prescription), where "death" is the
prescription being completely used and needing a refill (except that death can
occur many times, i..e many refills). Possible data available include dosage,
number of refills possible, date of initial prescription, refill dates, and
patient demographics.
Any suggestions, especially what to do in S? Since the refills are a function
of time (time until the next occurrence of the illness), this sounds like
survival to me. I'm thinking of a parametric survival model to include the
demographics.
Thanks for any help,
Walt Paczkowski
________________________
Walter R. Paczkowski, Ph.D.
Data Analytics Corp.
44 Hamilton Lane
Plainsboro, NJ 08536
(V) 609-936-8999
(F) 609-936-3733
|