Assessment of education and computerized decision support interventions for improving transfusion practice.

BACKGROUND: Overuse of blood products is common, but prior efforts to improve transfusion decisions have met with limited success. STUDY DESIGN AND METHODS: This study examines transfusion practices before and after a conventional educational intervention followed by a randomized controlled trial of a decision support (DS) intervention with computerized physician order entry (CPOE) for red blood cell, platelet, and fresh-frozen plasma orders. The study was conducted in an academic medical center between April 2003 and June 2004. Orders originating from units not using CPOE with DS (e.g., the emergency department) were excluded. Junior housestaff were randomly assigned into a control group and an intervention group who received DS for transfusion orders. Transfusion orders were initially classified according to guideline rules as DS-agree or DS-disagree. Chart reviews assessed inappropriateness for all DS-disagree orders and a sample of DS-agree orders. The total of inappropriate transfusion orders included chart review confirmed DS-disagree orders and DS-agree orders reclassified as inappropriate. RESULTS: The percentages of inappropriate nonemergent transfusion orders during the baseline phase for the entire staff and randomly assigned junior housestaff were 72.6 percent (2154/2967) and 71.9 percent (1259/1752) and improved after conventional education to 63.8 percent (1699/2663; p < 0.0001) and 63.3 percent (1263/1996; p < 0.0001), respectively. The percentage of inappropriate orders in the DS intervention group continued to improve (59.6%, 804/1350; p < 0.0001). Physicians accepted 14 percent (133/939) of new DS-recommended orders, especially recommendations to increase transfusion doses (73%). CONCLUSIONS: Education and computerized DS both decreased the percentage of inappropriate transfusions, although the residual amount of inappropriate transfusions remained high.