I am fitting mixed models on a mids object obtained after Multivariate Imputation by Chained Equations (MICE).

Following the MICE method, I get a pooled estimate of the mixed models fitted to each data set of my mids object.
Now I would like to make post-hoc tests on this pooled estimate (namely contrasts using emmeans), as I would do after classically fitting a mixed model. Of course, this last step does not work as is (see code below) and needs some adaptation.

Any clue on how to do so? Thanks a lot!

# STEP 1: fit mixed model on mids object
fit_mm.hf = with(data = imputation_mids, exp = lme4::lmer(hf_tv ~ group*session + age_months + iq_score_norm + ses_index +  sex +
            (1|participant_id)))

# STEP 2: get pooled estimate
estimate = pool(fit_mm.hf)

# STEP 3: make contrasts
emmeans(estimate, pairwise ~ session|group)