Left truncated and right censored data
On Fri, 11 Nov 2005, Martina Gandini wrote:
I have a question concerning a non parametric estimator for left truncated and right censored data. I see that the function survfit(Surv(time1, time2, status) ~ group, data=mydata) (where time1= left truncation time time2= follow=up time status= 0:censored 1:not censored) do this, but what's the method used? Is it a Lynden-Bell (Woodroofe) estimator or something else?
I don't know what names it is called, but it is the cumulative product of
1 - number of events/number at risk
over time points. This is the nonparametric MLE, and reduces to the
Kaplan-Meier estimator in the absence of truncation.
-thomas