topic Re: Survival Analysis with time-dependent covariates in Statistical Procedures

Survival Analysis with time-dependent covariates

deleted_user — Fri, 11 Dec 2009 17:19:02 GMT

Hi,
I am trying to use PHREG procedure in 9.2 SAS to analyse my data on bahaviour of crayfish, and I encountered problems with modeling effects of time-dependent covariates. Maybe somebody would be able to help me to solve my problem. In short, I observed a number of crayfish over 30 minutes. I recorded time after which they were leaving their shelter (censored = 0 if they did not leave after 30 minutes) in control and treatment groups. In addition, water in tanks with crayfish changed its quality after 15 minutes - crayfish were exposed to a time-dependent state of water (time-dependent covariate).
I used the following model:

proc phreg data=tak;
class treatment;
model time*censored(0)= treatment status;

if time <= 15 then status= 0.0;
else status= 1.0;
run;

but I do not receive statistical results for the time-dependent covariate "status" (in the result table 0 valuues are desplayed for this variable). I would be grateful for any suggestions how to solve this problem and what can be a problem. You can also contact me directly via my e-mail:
bamwandzara@yahoo.pl

best regards
Marcin, Poland

Re: Survival Analysis with time-dependent covariates

Doc_Duke — Fri, 11 Dec 2009 19:43:09 GMT

part of your posting is missing.

Re: Survival Analysis with time-dependent covariates

deleted_user — Sat, 12 Dec 2009 16:54:23 GMT

Hi, I try again to post my question. I analyse data on bahaviour of crayfish, and I encountered problems with Phreg procedure with time-dependent covariates. I observed a number of crayfish over 30 minutes, and recorded time after which they were leaving their shelter (censored = 0 if they did not leave after 30 minutes) in 2 groups. Water in tanks changed its quality after 15 minutes (time-dependent variable - 2 states of water).
I used the following model:
proc phreg data=tak;
class treatment;
model time*censored(0)= treatment status;
if time < 15 then status= 0.0;
else status= 1.0;
run;
Something goes wrong with this procedure, as I obtain 0 values for the effect of "status" variable. I should be grateful for your suggestions.
Marcin, Poland