Background: Many red blood cell transfusions (RBCT) are unnecessary or harmful according to clinical trials and guidelines. These RBCT put patients at risk and waste scarce healthcare resources. Clinical decision support (CDS) using the Epic™ computer provider order entry system has reduced unnecessary transfusion (Goodnough et al, Transfusion 2014) at one center with a post-intervention RBCT rate higher than our baseline rate. We sought to improve already low RBCT rates by adapting proven CDS interventions to our health system, reducing patient risk and healthcare costs.
Methods: We conducted a prospective, before and after study at an academic medical center with a diverse patient population. We included all inpatient adults except those with gastrointestinal bleeding or within 12 hours of a surgical procedure. The CDS intervention alerted providers if they ordered blood for a patient with a hemoglobin > 7 g/dL and required a rationale for protocol deviation (active bleeding, anticipated surgery, or Hgb <8 g/dL with brain or heart ischemia). The default dose was changed from 2 units to 1. Also, RBCT thresholds were changed for the stem cell transplant service’s ordersets. Education (meetings, ID badge cards and fliers) supported the effort. We monitored the RBCT rate for Hgb >7g/dL, the multi-unit RBCT rate, and the total transfusion rate and costs. We considered 1/14 to 9/14 our baseline period, 10/14 to 4/15 our intervention period, and 5/15 to 4/16 our post-intervention period.
Results: The rate of RBCT for Hgb >7 g/dL fell from 72.3% to 57.6% during the intervention period and to 39.9% for the post-intervention period (p < 0.0001), while the rate of multi-unit RBCT fell from 59.9% to 41.1% during the intervention period and to 20.8% post-intervention (Figure; p < 0.0001). The total RBCT rate (units/1000 inpatient days) fell from 41.2 to 40.5 during the intervention period and to 39.0 during the post-intervention period (p = 0.009). On the BMT service, transfusions for Hgb >8 g/dL dropped from 34% to 0%, and multi-unit transfusions dropped from 35% to 4%. Goodnough et al limited transfusions for Hgb >8 to 24% of transfusions; we limited them to only 8.5%, but our estimated annual savings ($64,900) were only 4% of theirs.
Conclusions: CDS can further improve low RBCT rates, but the incremental benefits and savings may be limited in facilities with low baseline rates. Our educational tools and orderset improvements can be easily implemented by other facilities.
