I think its important to be clear as to why we use levels of evidence. They
are good in guidelines to give an idea of how sound the evidence is as a
very coarse measure. Lots of As is better than none. But individual
studies have to be appraised carefully; an RCT is useless if the outcomes
are surrogates, if the population is highly selected etc. Similarly, there
must be times when an observational study can trump or quastion an RCT.
e.g. A tablet for giving up smoking looks good in an RCT but the subjects
are highly selected. Then an observational study shows that very few real
patients have quit.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|