The performance assurance mechanisms that have been proposed over the past decade or so have many aims, some laudable, some less so. Sometimes they are designed essentially as cost containment measures; sometimes they are motivated by a genuine concern for raising the quality of healthcare. In their paper, ‘Informed consent and surgeons’ performance', Clarke and Oakley (2004) argue that data on surgeons' performance should be collected and disseminated for another reason: to provide patients with information they need for genuinely informed consent. Clarke and Oakley suggest that promoting informed consent is vital, inasmuch as so doing respects patient autonomy; a good which is so significant that its promotion trumps most other considerations. Indeed, they give only one example of a good that is important enough to restrict (though not to violate) patient autonomy – surgeon's privacy with respect to their sexual orientation – and explicitly argue that even a reduction in overall surgical utility may not be a weighty enough consideration to justify a restriction on autonomy (2004, p. 19 and p. 23).

I suggest, however, that Clarke and Oakley are mistaken in thinking that respecting autonomy requires giving it weight sufficient to trump most rival goods. Respecting autonomy does not require maximizing it; it requires taking it seriously. We respect patient autonomy by always taking it into consideration in ethical decision-making, just as we respect a person by always taking her interests into consideration, not by treating her interests as trumping all rival goods (if it were the case that respecting a person required taking her interests as overriding, it would require a miraculous harmony of interests for us to be able to simultaneously respect many people).