Dear all,

I’ve used `rstanarm::stan_clogit`

to estimate the effect of a certain intervention on mortality. In short, our design is a matched risk-set case-control study with 1 case and 5 controls per risk set/stratum. I used `rms::rcs`

as suggested previously but I’m now struggling a bit with getting conditional effects of the splines. `brms`

(which has the nice `conditional_effects`

function) doesn’t support the conditional logistic likelihood, as far as I know, which is why I’m sticking with `rstanarm`

. It is possible to re-cast the conditional logistic regression as a Cox regression, but that requires defining the strata, and I can’t seem to find out how to do that in `brms`

.

I feel like I’m missing something obvious, but since I’m quite stuck, I wanted to reach out to the community and hopefully get some pointers on moving forwards with this.

So, to continue with the example from my other post:

```
library(dplyr)
library(rstanarm)
library(rms)
dat <- arrange(infert, stratum) %>%
filter(parity <= 2) # yields 1 case and 2 controls per stratum
post <- stan_clogit(case ~ rcs(age) + spontaneous + induced,
strata = stratum, data = dat, QR = TRUE)
```

As per the `posterior_predict`

documentation,

Drawing from the posterior predictive distribution at interesting values of the predictors also lets us visualize how a manipulation of a predictor affects (a function of) the outcome(s).

All good, but the Details section specifies,

For models estimated with

`stan_clogit`

, the number of successes per stratum is ostensibly fixed by the research design. Thus, when doing posterior prediction with new data, the`data.frame`

passed to the`newdata`

argument must contain an outcome variable and a stratifying factor, both with the same name as in the original`data.frame`

. Then, the posterior predictions will condition on this outcome in the new data.

My question is, then, if I wanted to see how age affects the outcome (e.g. over a grid in [18:45]) for women with `spontaneous = 0`

and `induced = 0`

, how would I go about it? That is, what `newdata`

should I give the `posterior_predict`

? Or am I completely off track and need to do something else? I’m running rstanarm v2.18.2.

Cheers,

Ben