Is it possible that some medications are not being approved because some people are non-responders versus the drug not being beneficial?

Yes, because a drug may be beneficial to a small proportion of people who are responders, present study designs may dilute results due to the non-responders. The authors of a recent study suggest the use of a study design called enriched enrolment with randomised withdrawal, EERW. In this study design, subjects are tried on a drug first, and only those who have a positive response are then randomised to study it further.