We are interested in looking at survival model in which the multiple outcomes are considered jointly. Naturally, there is more than one approach. For a parametric model, I might consider multiple likelihoods, where we have global and local parameters for each likelihood, and partial pooling between those parameters (I haven’t worked it out completely).
EDIT: multiple likelihoods? gibberish, we can write our own likelihood and check to see it’s a PDF
Can someone please point me to pre-existing packages and related papers that do survival accounting for multiple outcomes? I’ve heard the term “ordinal ranks” thrown around
we describe a multivariate random effects model here with implementation in SAS: cqo paper The outcomes in that case were (time to) mortality, emergeny department visits and hospitalisations. I referenced a paper i really like from 1990 that analysed nonmelanoma skin cancers (SCC, BCC) ie Abu-Libdeh et al.
edit: in this paper i mentioned R packages that are available for estimating the model: stat mod
I don’t have experience with this myself, but I believe the Merlin package in Stata and R can handle recurrent events, competing risks, joint longitudinal and time to event and various other things: https://arxiv.org/pdf/1806.01615.pdf