Multimediate package with multiply imputed mediators and outcome

[kellysaunders17]

I am reaching out on behalf of my project team who is looking to run a causal mediation analysis using multiple mediators. In researching different programming options, we came across your R package multimediate.

One additional aspect of our analysis is that we are dealing with missing data on our mediators and our outcome that we’d like to account for using multiple imputation (using the mice package) and then pooling the estimates from multimediate, using Rubin’s rule. Given the mulitmediate function does not produce standard errors, we are looking at alternative ways to pool the estimates across multiply imputed datasets. With the use of the Monte Carlo draws, an alternative seems to entail using the bootstrapped results to calculate the variance for each imputed dataset, as recommended by Stef van Buuren in this exchange: confidence interval - How can I pool bootstrapped p-values across multiply imputed data sets? - Cross Validated (stackexchange.com).

1. Is there a way to extract the p-values from each of the Monte Carlo draws from the multimediate output?
2. Do you have any other recommendations on how we can use the multimediate package with multiple imputation?

Thank you very much for your help with our analysis. If I can provide additional details to help answer these questions, please let me know.