If an answer by a recognized MD, maybe even specialized in the field the question is asked in, would not be accepted as an answer, then one should ask himself: Why would I go do a doctor when he doesn't cite the sources the communitywiki described for every sentence he says?
We should downvote if there are no sources, unless it is widely known (for example, the statement "If you get too cold you die" is a little unnecessary to backup. The actual temperature (35 C and less) should be linked.
To such a specialized MD, the statement "widely known" will have another meaning than to a normal user. And also note that it's this group of people who writes
the articles you'd like to have as citations.
So my answer to your question would be:
If there is a possibility to identify a user as a specialist in the field the question is asked in, an answer can be accepted without citations. If not, citations will be required for backing up the statements.
Please also note that decision-making in clinical environments is also not only not entirely based on evidence (as many mechanisms of treatments that evidentially help are simply unkown), but also rely on experience which is only in latter stages eventually transformed into citable articles. This is always a matter for discussion amongst professionals, but should it therefore be completely left out of this site? In that case, chances are high that not the best answer to the question, but the most citable one will end up as most upvoted (or least downvoted).