Hi There,
I have a cancer dataset with 170 observations, I calculated the length of progression(in days) as the difference between date of surgery and date of cancer progression. Now that the date for cancer progression (1's) are only available to 20 patients only and rest 150 patients do not have a date as the cancer is not progressed( 0's) - these 0's are the censored cases. Now the problem is the length of progression can only be calculated for patients whose cancer is progressed (just 20 patients ). Can I go ahead and proceed with KM plot and multivariate Cox model? Any suggestions would be very thankful.
You seem to need a tutorial on time to event analysis. In this type of problem, the relevant data for one given patient are either the date of the event or the last date known where the event has not took place (date of last news). In practice, your relevant data are te couple (date of last news, state at last news). Google "survival analysis tutorial" for a $#!+load of useful pages.
BTW : why SAS ? R has excellent survival analysis packages, to begin with Terry Thernau's "survival" package, which is standard in any R installation. Unless you have masochistic urges to fulfill or a large IT budget to spend SAS licenses are *expensive*), wasting your time with PROC PHREG seems a bit unhelpful...
You can even use RCommander with the survival Plugin, if you are not familiar with the R commands.
A comprehensive lecture notes with examples are presented by prof. David Madigan (Columbia University, New York, NY) at
http://www.stat.columbia.edu/~madigan/W2025/notes/survival.pdf
- Badges
- Science topic
- Similar topics
- Mathematics
- Statistics