Analysing and presenting results from systematic reviews of complex interventions

Article type
Authors
Jamtvedt G, Young J, Tove Kristoffersen D, Oxman A
Abstract
Background: Audit and feedback is commonly used to improve professional practice and in quality improvement programs. A systematic review included 85 studies involving audit and feedback, and an ongoing update of this review will include at least 25 more studies. As with many Effective Practice and Organisation of Care reviews the studies included are heterogeneous. Analysis and presentation of results in these reviews are challenging and should be improved.

Objectives: The aim of this paper is to discuss ways of analysing explanatory factors for heterogeneity and presenting the results in complex interventions.

Methods: The following explanatory factors were identified as possible sources of heterogeneity in the audit and feedback review; type of intervention, intensity of audit and feedback, complexity of targeted behaviour, seriousness of outcome (importance of the behaviour), baseline compliance and study quality. Sizes of effect were calculated as adjusted RD and RR for dichotomous outcomes and as adjusted post difference and % change relative to control after for continuous outcomes. A median effect size was calculated across multiple outcomes. Both visual analysis and meta-regression was performed. We used tables, bubble plots, regression lines and whisker plots that grouped comparisons relative to each of these variables in relation to the size of the effect. The size of the bubbles corresponded with number of health care professionals and we used colours to visualise comparisons that belonged to the same study. Meta-regression was performed with effect size as dependent variable and weighted by number of health professionals.

Results and conclusion: The results of the trials included in the current version of the review were heterogeneous and it was only possible to explain this heterogeneity to a limited extent based on the analyses that were undertaken. Analysis and results of the updated version of audit and feedback review will be presented and we will discuss the pros and cons of the analytical methods that were used relative to other options.

Reference:
Jamtvedt G, Young JM, Kristoffersen DT, Thomson O'Brien MA, Oxman AD. Audit and feedback: effects on professional practice and health care outcomes. The Cochrane Database of Systematic Reviews 2003, Issue 3.