The interaction between dysmorphology and epidemiology: Methodologic issues of lumping and splitting