Expectation eaximization with repeated measures data

I use SPSS 21.

I have repeated measures data with multiple time points. After the first time point every other time point (two follow-up time points) has significant participant attrition (close to 6oo participants at baseline and around 400 at each follow-up time point). I would like to use expectation maximization to address this missing data so that when I run my analyses I have complete data from every participant, from baseline forward (complete data at all time points for all 600 participants).

First, can expectation maximization in SPSS handle this (missing value analyses --> EM estimation)? Second, how would I actually run these EM analyses? Should I run EM on the scales for all time points simultaneously? What I mean by this is, suppose I have scale X (comprised of 3 continuous variables) measured at 3 time points. Should I first run expectation maximization for scale X at each time point simultaneously (i.e. have 9 items in the "quantitative variables" box)?

If EM can't work, what's a good alternative? I appreciate the help.