Can we trust equivalence and non-inferiority trials?