Background: Best practice alerts (BPAs) and computerized provider order entry (CPOE) are electronic health record (EHR) tools used to promote evidence-based practice and adherence to clinical practice guidelines. Many institutions have implemented and published on the use of red blood cell transfusion (RBC) BPA/CPOEs to promote evidence-based practice since transfusion is the most common inpatient procedure, and the clinical practice guidelines are informed by high-quality clinical trial data. However, the most widely followed transfusion clinical practice guidelines from the AABB contain multiple recommendations informed by complex data from multiple trials, making it difficult to accurately program the guidelines or trial data into a BPA/CPOE. If BPA/CPOEs are either incomplete or inaccurate because they cannot be programmed to reflect nuanced practice guideline recommendations or the underlying trial data, then they may undermine the purpose of the BPA/CPOE and potentially negatively impact patient outcomes. As a result, we conducted a systematic review of studies describing the implementation of an RBC transfusion BPA/CPOE to assess accuracy and adherence to recommended guidelines and evidence-based practice.
Methods: We identified 32 pre-post observational studies that explored the effects of the implementation of an RBC transfusion BPA/CPOE. We included studies involving adult patients, pediatric patients, or patients admitted to specialized hospital units. Accuracy and adherence of the BPA/CPOE to recommended guidelines were assessed through a review of the articles using a grading rubric developed by the study team. The rubric assessed the content of each BPA/CPOE for accuracy related to the AABB RBC transfusion guidelines or cited trial data, the BPA/CPOE implementation within a health system, and the patient outcomes attributed to the BPA/CPOE.
Results: Of the 32 studies included, 28 (88%) cited practice guidelines or trial data to justify the use of their institutions’ BPA/CPOE. Of these, 27 (96%) deployed a BPA/CPOE in the EHR that did not match the recommendations from published AABB guidelines or was not completely representative of the underlying trial data. The majority (56%) of BPA/CPOEs were deployed to hospitalists based only on a set hemoglobin/hematocrit value trigger and did not take clinical context or preexisting conditions into account as recommended by the AABB guidelines. The majority of studies (69%) did not mention a plan for updating the BPA/CPOE as new evidence emerges. Calculating reductions in red blood cell orders was the most frequently studied outcome (97% of articles). The majority of articles (56%) estimated cost savings for their institution due to the reduction in red blood cell usage, citing this as a benefit of adherence to restrictive transfusion promoted by the use of the BPA/CPOE.
Conclusions: These findings suggest that there is variation by institution in the accuracy and deployment of transfusion BPA/CPOE compared to published practice guidelines and the underlying trial data informing those guidelines. This is problematic because an inaccurate BPA/CPOE may bias the decision-making process of the hospitalist, ultimately harming patients. Future work should focus on implementation science approaches to improving the accuracy and implementation of transfusion BPA/CPOE.
