Mixed effects survival model

AprilS — Tue, 13 Feb 2018 07:39:40 GMT

The data has a 3-level clustered structure: first cluster level is A (e.g. student), second level B(e.g. class), third level C (e.g. school), and a student may have multiple events(recurrent). I wonder if any SAS proc could do a recurrent event survival analysis with 3-level clustered data?

In addition, there are more than 2 time dependent covariates (i.e. may change during the time interval from the start date to an event date), I wonder if there is an instruction or example how to deal with more than 2 time dependent covariates in a survival model?

Any help would be much appreciated!

Re: Mixed effects survival model

Thuva — Thu, 13 Dec 2018 17:11:55 GMT

Hello,

Were you able to resolve this? I am looking to create a multilevel model with 2 clusters (ID and center) with recurrent events. The PROC PHREG statement only seems to allow for one random effect in the RANDOM statment. When I write the RANDOM statement twice for the 2 variables, it doesn't appear to work. I took a look into PROC SURVEYPHREG, which allows you to have more than one RANDOM effect, but I'm not sure how to adjust of the recurrent events.

proc phreg data=final_survival_export covs(aggregate) covm;
class 	pt_id prev_enteral_feed (ref='0') /*need this to specify for random effect*/;
model  (visit0 visit1)*rs(0)= prev_enteral_feed/rl ties=efron; /* RL= 95% (default alpha level) confidence limits for the hazard ratios.*/
random pt_id; *shared frailty;
id pt_id; /*not sure if this ID step is needed*/
run;

This is what I had for the 1 random effect model that accoounts for recurrent events, however I need 2 random effects.

topic Re: Mixed effects survival model in Statistical Procedures

Mixed effects survival model

Re: Mixed effects survival model