*advocates applying the results of population studies over
   mechanistic reasoning * in diagnosis, prognosis and therapy."

   in short, it has to do with population generalization and the
   overuse of it. It *is* overused. Things with nice neat lines
   like physics are easy to use statistics with. Messy things like
   biology don't conform as well to statistics, medicine even less
   so. Case studies are appropriate generalizations - and they're
   not even generalizations; they're case studies - archetypes
   that, when considered in an overlapping fashion, can lead to
   some similarities to a current case.

   They work.

   Statistics are alluring and tempting but they're not appropriate
   as the MAIN basis for diagnosis, only as a supporting role.