How much can we trust medical literature?