Perhaps someone could help with this problem ...
I want to include left and right censored data ... along with non-censored
data in a Weibull model within the Lifereg procedure in SAS. I can see
how this is done separately for left and for right censored observations
but I am not clear on how right censored observations are dealt with when
left censored observations are included.
Here's my problem ...
I have a long length of dna (say time data) along which events can happen,
called microsatellites. These microsatellites signal the start and the
finish times, so they can be both the entry of an individual into the
trial and the death of that individual. See the line example below:
-----x-------x------------------x----------x--x------------
x is a microsatellite
The distance between microsatellites is a non-censored observation ...
To the right of the line there is a right censored observation
and also to the left of the line is a left censored observation ...
To enter left censored observations in SAS you need to create a left
and right variable containing start and finishing times. To indicate
a left censored observation you simply make the left variable missing.
For the right censored observation there is the problem of telling
SAS the starting time (absolute) and the length time relative to that
starting time when the individual was last monitored but was not known
to be decreased. The problem I have is how do I get these two pieces
of information included in the model and still inform sas I have a right
censored observation; for example, can I
create observations for left and right as say 8000 and 8200 respectively
where 8000 is the absolute time when the individual was entered into the
trial and 8200 indicates how long we know he has been in the trial and
then use an indicator variable to indicate that the status is right
censored ... that is this acceptable in SAS to say, based on the above,
model (left,right)*status(1)=explanatory vars;
where 1 is a right censored observation.
Your help will most greatly appreciated ...
Trevor McMullan
M.Sc. Biometry
Reading University
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|