What is so bad about reverse epidemiology anyway?

What is so bad about reverse epidemiology anyway?