this is my first post so please correct me if I am doing something wrong or if there is a better place to post my question.
Package: survival
I currently am analyzing survival data in R that includes nested clusters of interval censored data. There are two covariates I am primarily interested in (we'll call them CV1 and CV2, each can take two possible values). The data therefore generally looks like the following (but with ~1200 individuals, 5 groups under Group 1, and 100 groups under Group 2):
I would like estimates for the following terms:
- CV1 (fixed)
- CV2 (fixed)
- CV1:CV2 (fixed)
- Group 1 (random)
- Group 2 (random)
- CV2:Group 2 (random)
This is currently how I have set up my model:
survreg(survival.object ~ CV1*CV2 + cluster(Group 1) + cluster(Group 2) + CV2:cluster(Group 1), dist = "weibull")
Everything works fine with the three fixed effects. However, there are no outputs when I include the three random effects terms, and I cannot find a way to access that information. Using cluster() also does not allow for an interaction between CV2 and Group 2.
I have already tried using frailty() instead of cluster(), but it does not let me add more than one random effects term to the model. Furthermore, I have tried the standard notation for linear models in R (1|Group 1), but the estimates for the random effects terms all read 0.000 with missing z and p values. I would use a cox model, but there are few that allow for interval censored data and none I have seen that allow for mixed effects models.
I have spent the past few weeks trying every source I can find online. I have only a moderate background in both R and survival analysis (currently learning these techniques as a grad student). I am not looking for someone to spoon-feed me answers, but I have reached the extent of what I think I can do alone. Any feedback or recommendations to proceed from here would be highly appreciated. I plan to contact the package maintainer, Terry Therneau, as well.
Aucun commentaire:
Enregistrer un commentaire