Guidelines and 'evidence-based medicine': the importance of the role distinction