With the advent of electronic medical records (EMR), hospitalists now have access to large data sets and can harness Big Data techniques to study practice variation and optimize clinical behavior.
To use innovative methods of data analysis and visual representation to determine practice variation around transfusion behavior and identify targets to reduce unnecessary transfusions in our medical center, as suggested by the ABIM Foundation’s Choosing Wisely campaign.
We enlisted an analyst to query our Epic‐based electronic medical record (EMR) for all transfusion orders placed at our medical center during the previous year. For each transfusion order, we obtained additional clinical information including the hemoglobin value of the patient immediately prior to transfusion, and the attending physician and department associated with the order.
For the 10,248 inpatient transfusions across our institution in a year, the average hemoglobin threshold was 8.19 g/dL (Standard deviation (SD) 1.3 g/dL), with more than 50% being administered above a threshold of 8 g/dL. Due to differences in departmental practice patterns, nearly 1,500 of the 5,000 transfusions above a threshold hemoglobin of 8 g/dL were from the malignant hematology service, whereas fewer than 400 units were from the hospitalist group. We used cumulative frequency analysis (Figure 1) to identify departments with suboptimal transfusion thresholds. Within each department, we used a treemap analysis (Figure 2) to visually identify attending physicians who liberally transfuse large numbers of units of packed red blood cells. From this analysis, we created a targeted education and feedback campaign for hospitalists, who largely used restrictive transfusions strategy, and we partnered with other departments whose transfusions patterns could be optimized.
Hospitalists are called on to work on a variety of quality and value improvement projects. Using this specific example of transfusions, we demonstrate how “big data” tools, applied to EMR‐derived data and creative data visualization can help hospitalists study and design appropriate targeted interventions for quality improvement and high‐value care.
To cite this abstract:Rajkomar A, Moriates C, Mourad M, Wachter R. Innovations in Data Visualization to Drive Down Unnecessary Transfusions. Abstract published at Hospital Medicine 2014, March 24-27, Las Vegas, Nev. Abstract 710. Journal of Hospital Medicine. 2014; 9 (suppl 2). https://www.shmabstracts.com/abstract/innovations-in-data-visualization-to-drive-down-unnecessary-transfusions/. Accessed March 31, 2020.