Mode of administration does not cause bias in patient-reported outcome results: a meta-analysis