Hi Christine,
Perhaps Gang can correct me, but 3dMEMA might do that to some degree. Subjects with fewer events would probably explain less variance, and therefore be weighted somewhat less than subjects with more events when computing the group beta.
- rick