The use of quality improvement interventions in reducing rarely appropriate echocardiograms: A systematic review and meta-analysis.

Abstract

The volume of cardiac imaging continues to increase, with many tests performed for rarely appropriate indications. Appropriate use criteria (AUC) documents were published by the American Society of Echocardiography and American College of Cardiology, with quality improvement (QI) interventions developed in various institutions. However, the effectiveness of these interventions has not been assessed in a systematic fashion. We searched Ovid MEDLINE, EMBASE, Scopus, Web of Science, Google Scholar, and EBSCO CINAHL for studies reporting association between cardiac imaging, AUC and QI. The search was not restricted to time or publication status. We selected studies assessing the effect of QI interventions on performance of rarely appropriate echocardiograms. The primary endpoint was reduction of rarely appropriate testing. Nine studies with 22,070 patients met inclusion criteria. Mean follow up was 15 months (1-60 months). QI interventions resulted in statistically significant reduction in rarely appropriate tests (OR 0.52, 95% CI: .41-.66; p < .01). The effects of QI interventions were analyzed over both the short (<3 months) and long-term (>3 months) post intervention (OR 0.62, 95% CI: .49-.79; p < .01 in the short term, and OR 0.47, 95% CI: .35-.62; p < .01 in the long term). Subgroup analysis of the type of intervention, classified as education tools or decision support tools showed both significantly reduced rarely appropriate testing (OR 0.54, 95% CI: .41-.73; p < .01; OR .47, 95% CI: .36-.61; p < .01). Adding a feedback tool did not change the effect compared to not using a feedback tool (OR 0.49 vs. 0.57, 95% CI: .36-.68 vs. 39-.84; p > .05). QI interventions are associated with a significant reduction in performance of rarely appropriate echocardiography testing, the effects of which persist over time. Both education and decision support tools were effective, while adding feedback tools did not result in further reduction of ordering rarely appropriate studies.