Replication is hard and expensive, particularly in the medical field, and does not get the researcher any real science points.

Besides, it's more fun to do a meta-analysis on a flock of sorta-vaguely-similar studies.