http://cid.oxfordjournals.org/content/early/2016/03/10/cid.ciw082.abstract
"Population-Based Retrospective Cohort Study"
In one word it is very difficult to get a causation link with this kind of study.
0.5 vs 0.3/100 000 another time with such a magnitude of the supposed effect causation is far from probable even if the stats tell us a CI of less than 5% to be wrong.
It is is just a trend which has to be proved by a RCT.
Aucun commentaire:
Enregistrer un commentaire