Variation in Transfusion Practices and the Effect on Outcomes After Noncardiac Surgery

Abstract

Objectives: To identify the patient-level effects of blood transfusion on postoperative outcomes and to estimate the effects of different transfusion practices on hospital-level risk-adjusted outcomes.

Background: Postoperative transfusion practices and their effects on short-term outcomes in patients undergoing noncardiac surgery are not well understood.

Methods: Demographic, operative, and outcomes data for 48,720 patients undergoing general or vascular surgery at 52 hospitals between July 2012 and April 2014 were obtained. The main exposure variable was receipt of any blood transfusion within 72 hours after surgery. Thirty-day mortality, any morbidity, infectious complications, and postoperative myocardial infarction were the outcomes of interest. Propensity score matching was used to minimize confounding by indication. Hospitals were categorized as having a restrictive, average, or liberal transfusion practice based on average trigger hemoglobin values.

Results: A total of 2243 (4.6%) patients received a postoperative blood transfusion. After propensity matching, a postoperative transfusion was associated with increased 30-day mortality (3.6% excess absolute risk), any morbidity (4.4% excess absolute risk), and infectious morbidity (1.0% excess absolute risk). However, a transfusion was associated with 3.5% absolute risk reduction in postoperative myocardial infarction. At the hospital level, there was a wide variation in transfusion practices. Hospitals with liberal practices were twice as likely to transfuse patients and had higher risk-adjusted mortality rates than restrictive hospitals (3.1% vs 2.2%; P = 0.002).

Conclusions and relevance: Postoperative transfusions after noncardiac surgery are associated with increased adverse postoperative outcomes, with the exception of postoperative myocardial infarction. Hospitals that are liberal in their transfusion practices have higher 30-day mortality rates, suggesting potential interventions for quality improvement.