How to conduct an LPA with missing or imputed data?

[Eleanor Braithwaite]

Hi there, 

I'm hoping to run a latent profile analysis (LPA) using three variables, however we have issues with missing data for these data. I have already used mice imputation in R to impute multiple datasets for other analyses and had originally planned to use these when running the LPA too. However, I am unsure if this is possible/ how to do so using the tidyLPA package in R - is this possible and if so, how can this be implemented please?

In my reading around this I also came across suggestions to impute within the LPA analyses, which seems to be possible when using MPlus? However, I am also not sure if it is possible/ how to do this using MPlusAutomation in R, or if this can only be done in MPlus directly. If so, I am not entirely against using only MPlus, however this would be my first experience of that software and I am much more familiar using R, so that would be my preference if possible.

Below is the code I would use if I were doing the LPA with only one dataset (i.e. with no imputation) - hopefully this looks okay. I can adapt this code to look through the multiple (imputed) datasets, however I understand it would be difficult to pool the outputs from these in a meaningful way.

lpa <- data[]%>%
        dplyr::select(var1, var2, var3)%>%
        estimate_profiles(1:6,
                    package='MPlusAutomation',
                    ANALYSIS = "estimator = mlr;
                    type = mixture;
                    starts = 200 50;
                    processors = 10;
                    lrtstarts = 2 1 50 10;",
                    OUTPUT= "sampstat residual tech 1 tech11 tech 14;",
                    models = 1)

Apologies if this is not the best format or if I missed any information, but I'd be really grateful for any help/ pointers for how I might proceed.

Thanks!