Abstract
BACKGROUND: This paper extends the findings of the Cochrane systematic review of audit and feedback on professional practice to explore the estimate of effect over time and examine whether new trials have added to knowledge regarding how optimize the effectiveness of audit and feedback. METHODS: We searched the Cochrane Central Register of Controlled Trials, MEDLINE, and EMBASE for randomized trials of audit and feedback compared to usual care, with objectively measured outcomes assessing compliance with intended professional practice. Two reviewers independently screened articles and abstracted variables related to the intervention, the context, and trial methodology. The median absolute risk difference in compliance with intended professional practice was determined for each study, and adjusted for baseline performance. The effect size across studies was recalculated as studies were added to the cumulative analysis. Meta-regressions were conducted for studies published up to 2002, 2006, and 2010 in which characteristics of the intervention, the recipients, and trial risk of bias were tested as predictors of effect size. RESULTS: Of the 140 randomized clinical trials (RCTs) included in the Cochrane review, 98 comparisons from 62 studies met the criteria for inclusion. The cumulative analysis indicated that the effect size became stable in 2003 after 51 comparisons from 30 trials. Cumulative meta-regressions suggested new trials are contributing little further information regarding the impact of common effect modifiers. Feedback appears most effective when: delivered by a supervisor or respected colleague; presented frequently; featuring both specific goals and action-plans; aiming to decrease the targeted behavior; baseline performance is lower; and recipients are non-physicians. DISCUSSION: There is substantial evidence that audit and feedback can effectively improve quality of care, but little evidence of progress in the field. There are opportunity costs for patients, providers, and health care systems when investigators test quality improvement interventions that do not build upon, or contribute toward, extant knowledge.