Assume we use the following example from the mice package:

imp <- mice(nhanes, m=5, print = FALSE, seed = 55152)

I want to obtain the means of nhanes$chl grouped by nhanes$hyp. Without using multiple inputation, one would do:

nhanes %>% 
    group_by(hyp) %>% 
    summarise(chl = mean(chl, na.rm = T))

How can I obtain such pooled estimates of means of chl per group of hyp after multiple imputation with mice?