0
$\begingroup$

I am hoping to use joint modelling (shared random effects models using JM) to test whether change in a longitudinal marker variable contributes to prediction of a survival event, incremental to current levels of that same marker variable.

In other words, if two participants have the same score on a marker variable, do they have the same risk of experiencing the survival event, if the score represents a decrease for one and an increase for the other?

In a Cox regression, I would do this by entering both current score and change score (e.g., difference between marker variable at assessment time and prior two weeks) as time-varying predictors to see if change remains a significant predictor after accounting for current score, but I am not sure how to adapt this to a joint model. Would it make sense to use a change score as my longitudinal marker variable, with current score as a covariate of that(change~time+current)? Would the reverse make more sense (current~time+change)? Do neither make sense and it's a stupid question because prior scores are explicitly factored into joint models, thus embedding change?

Any help is appreciated!

$\endgroup$
0
$\begingroup$

The parameterization option for jointmodel allows users to specify if they'd like to include slopes of a marker variable, current levels, or both to determine risk for the survival event. To test whether slopes incrementally predict the survival event, I would set the argument to parameterization="both"

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.