Update of Cochrane Systematic Review of audit and feedback demonstrates stability in findings. Will future trials be helpful?

Article type
Authors
Ivers N1, Jamtvedt G2, Flottorp S2, Young J3, French S4, O’Brien M5, Johansen M2, Odgaard-Jensen J2, Oxman A2, Grimshaw J6
1Department of Family Medicine, Women’s College Hospital, Canada
2Norwegian Knowledge Centre for the Health Services, Oslo, Norway
3Surgical Outcomes Research Centre (SOuRCe), Central Sydney Area Health Service, Camperdown, Australia
4Primary Care Research Unit, University of Melbourne, Carlton, Australia
5School of Rehabilitation Science, McMaster University, Hamilton, Canada
6Clinical Epidemiology Program, Ottawa Health Research Institute, Ottawa, Canada
Abstract
Background/Objective: This review updates a previous Cochrane review aiming to assess the effects of audit and feedback (AF) on healthcare provider professional practice, which found a median 5% improvement in compliance with desired practice across a range of dichotomous outcomes.

Methods: We searched MEDLINE, EMBASE, CINAHL and the Cochrane Effective Practice and Organization of Care (EPOC) database up to December 10, 2010 for randomized trials of AF. We defined AF as a summary of clinical performance given to health care providers. For this update, trials were included only when AF was judged to be the core, essential feature of the quality improvement intervention; thus, some previously included trials were excluded. Data were double-abstracted independently. For each comparison, we calculated the risk difference (RD) of compliance with desired practice, adjusted for baseline compliance.

Results: 144 trials were included in the review; 91 were from the previous review. The interventions were heterogeneous in their design and addressed a variety of clinical problems. Considering only trials with dichotomous outcomes, when AF alone was compared to usual care (41 comparisons), there was a 4.1% increase in compliance with desired practice. When AF was considered a core, essential aspect of a multifaceted intervention (56 comparisons), the median increase in compliance was 5.3%. In trials comparing combined educational outreach plus AF to AF alone (16 comparisons), the median increase in compliance was 4.4%.

Conclusions: AF has a small but important positive effect on clinician behaviour. This effect may be increased by concurrent interventions, especially educational outreach. This update led to minimal changes in the estimate of effect despite a substantial increase in the number of studies. The stability in effect estimate suggests that future trials comparing AF to usual care may not provide valuable new knowledge, unless carefully crafted to assess specific aspects of intervention design.