topic Re: Joint modeling of longitudinal and survival data in Statistical Procedures

Joint modeling of longitudinal and survival data

kc — Sun, 16 Oct 2022 01:39:33 GMT

I am looking for any papers/references along with the code if possible (proc nlmixed) for modeling longitudinal continuous outcome (range 0-100) and survival data. The intent is to calculate the mean difference of the longitudinal outcome between 2 treatment groups at various timepoints. It would be great to understand how the equations required in the proc nlmixed are arrived at for accomplishing the said task. Any pointers in this regard would be extremely helpful.

Re: Joint modeling of longitudinal and survival data

StatDave — Sun, 16 Oct 2022 15:33:51 GMT

You should take a look at the joint modeling capabilities in PROC GLIMMIX using DIST=BYOBS to see if it can handle your situation. See the joint modeling example in the Examples section of the GLIMMIX documentation.

Re: Joint modeling of longitudinal and survival data

Ksharp — Mon, 17 Oct 2022 12:34:51 GMT

StatDave ,
PROC GLIMMIX could handle censored data ?
As far as I know COX regression is semi-parameter method ,unlike other GLM assuming Y conform to a distribution.
COX regression using K-M estimator for Y variable.

Maybe @SteveDenham @lvm know the MIXED model for censored data ?

Re: Joint modeling of longitudinal and survival data

SteveDenham — Mon, 17 Oct 2022 15:22:28 GMT

I would not expect GLIMMIX to do a good job with censored data, as the missing values are not at random. However, with complete data the BYOBS method for joint distributions that @StatDave brought up is a good idea. The procs FMM and HPFMM allow for truncated normal and Poisson distributions. Censored data can be done with PHREG as you mention, but tobit models (PROC QLIM or ENTROPY in SAS/ETS, LIFEREG in SAS/STAT) may provide better estimates (stress on the word MAY).

SteveDenham