Blood transfusion has been associated with poor outcomes in many disciplines, yet transfusion practices and related outcomes in esophagectomy are unknown. We analyzed the Society of Thoracic Surgeons General Thoracic Database to determine patient factors associated with transfusion after esophagectomy, risk-adjusted variation in transfusion practice among institutions, and the association of transfusion practice with mortality.
We performed a retrospective review of patients undergoing esophagectomy for cancer from October 2008 to December 31, 2014. Patient comorbidities and procedure variables were used to construct a risk model for transfusion. Using this model, each institution was assigned an observed to expected (O:E) transfusion rate. We examined institutional factors associated with variation in O:E transfusion rate. Finally, O:E transfusion rate was compared to risk-adjusted mortality to determine if there was an association of transfusion practice and survival.
Seven thousand one hundred thirty-seven patients underwent esophagectomy at 182 institutions during the study period. The median unadjusted transfusion rate was 23.1%. The risk model for transfusion demonstrated patients who received transfusions were more likely to be older, female, and have low preoperative hemoglobin and other comorbidities, such as CAD, COPD, and low creatinine clearance. Patients who received a minimally invasive procedure were less likely to have received a transfusion.
After adjusting for the characteristics above, 13 centers (7.1%) were classified as having lower than average O:E transfusion rate and 16 centers (8.7%) were classified as higher than average O:E transfusion rate.
Institutions with lower than expected transfusion rates also had lower risk-adjusted perioperative mortality than institutions with higher than expected transfusion rates (median [IQR] = 0.90 [0.77–0.94] vs. 0.99 [0.94–1.06], P = 0.028).
Age, female sex, CAD, COPD, renal insufficiency, and open technique are associated with transfusion after esophagectomy, while tumor stage and preoperative chemoradiation are not. There is wide variation in transfusion practice. Centers with lower than expected transfusion rate also had lower than expected perioperative mortality. At an institutional level, lower transfusion rates are associated with improved outcomes.