Journal of Clinical Epidemiology 55 (2002) 717–722 0895-4356/02/$ – see front matter © 2002 Elsevier Science Inc. All rights reserved. PII: S0895-4356(02)00403-1 Attributes of clinical recommendations that influence change in practice following audit and feedback Robbie Foy*, Graeme MacLennan, Jeremy Grimshaw, Gillian Penney, Marion Campbell, Richard Grol The University of Edinburgh, Department of Obstetrics & Gynaecology, 37 Chalmers Street, Edinburgh EH3 9ET, UK Received 28 February 2001; received in revised form 28 January 2002; accepted 2 February 2002 Abstract The object of this study was to determine which attributes of clinical practice recommendations influence changes in clinical practice following audit and feedback. This was an observational study using multilevel modeling to examine the relationship between attributes of clinical practice rec- ommendations and compliance with the recommendations before and after audit and feedback. Sixteen hospital gynecology units in Scotland partic- ipated in a national audit project. Clinical practice recommendations covering selected gynecological topics were developed and data collected to as- sess baseline (preintervention) compliance. Summaries of performance were fed back to consultant gynecologists in each hospital and follow-up (postintervention) data were collected. Trained audit assistants used standardized forms to abstract data from case notes. Compliance data were available at baseline and follow-up for a total of 42 clinical practice recommendations. Altogether, 4,664 case notes contributed to baseline data and 4,382 to follow-up data. Thirteen attributes describing clinical practice recommendations were developed, based upon previous work, and pretested. A panel of seven consultant gynecologists rated the extent to which each of the 42 recommendations possessed each of the 13 attributes. The main outcome measures were the association of each attribute with compliance and with changes in clinical practice. Recommendations compatible with clinician values and not requiring changes to fixed routines were independently associated with greater compliance at baseline and follow-up. How- ever, recommendations incompatible with clinician values were independently associated with greater change in practice following audit and feed- back. Attributes of recommendations may influence the effectiveness of audit and feedback in secondary care. Recommendations seen as incompat- ible with clinician values are associated with lower compliance but greater behavioral change following audit and feedback. © 2002 Elsevier Science Inc. All rights reserved. Keywords: Audit and feedback; Clinical practice guidelines; Professional practice; Behavior 1. Introduction The implementation of valid clinical guidelines can im- prove the quality of health care [1]. Passive dissemination of a guideline is unlikely to lead to changes in clinical prac- tice [2]. Combining more active interventions, such as re- minders or interactive education, with guideline dissemi- nation and implementation is more likely to change professional and organizational practice. Various factors, or effect modifiers, can influence the ef- fectiveness of such interventions [3]. Until recently, most research has focused on characteristics of clinicians or health care organizations, such as local attitudes or pre- paredness to change. However, the characteristics of clini- cal practice recommendations themselves may also influ- ence their rate of adoption [4]. Grilli and Lomas first assessed the association between such characteristics and compliance with clinical guideline recommendations [5]. They reviewed published studies re- porting compliance rates with 143 different recommendations developed or endorsed by official organizations. Compliance was higher for recommendations displaying “trialability” (which could be tried out temporarily and discarded if found wanting) and lower for complex recommendations. The “ob- servability” of recommendations (how readily their benefits could be seen to be achieved) had no impact. Grol et al. [6] assessed the extent to which Dutch general practitioners’ compliance with 47 guideline recommenda- tions was influenced by 12 characteristics (or attributes) of the recommendations. The guidelines were disseminated via journals and continuing medical education programs. Com- pliance was lower if recommendations were vaguely worded, incompatible with clinician norms and values, and disruptive to routine practice. * Corresponding author. Tel.: +44 131 229 2575; fax: +44 0131 229 2408. E-mail address: R.Foy@ed.ac.uk (R. Foy).