Declining use of red blood cell transfusions for gastrointestinal cancer surgery: A population-based analysis.

802 Background: Perioperative anemia is common in gastrointestinal (GI) cancer surgery patients and is often treated with red blood cell transfusion (RBCT), which carries risks for inferior oncologic outcomes. Despite level-1 evidence for restrictive transfusion strategies, RBCT use is often not consistent with guidelines leading to a high rate of unnecessary transfusions. Understanding of RBCT use at the population-level is necessary to develop system-level efforts to minimize perioperative RBCT for cancer. We sought to evaluate the secular trends of transfusion in a large North American population. Methods: We conducted a population-based retrospective cohort study of patients undergoing GI cancer resection between 2007-2018 using linked administrative health datasets in Ontario, Canada. Primary outcome was administration of any RBCT during the hospitalization. Temporal RBCT trends were analyzed with Cochran-Armittage tests for trend. Modified Poisson regression assessed trends while controlling for potential confounders. Results: Of 79,764 patients undergoing GI cancer resection, median age was 69 (IQR: 60-78) years old and 55.5% were male. The most frequent cancer site was colorectal cancer (n = 63,243), followed by esophago-gastric (n = 7,307), hepato-pancreato-biliary (n = 6,510), and small bowel (n = 2,704). 30% of patients received RBCT. The proportion of patients transfused decreased from 26.5% in 2007 to 18.9% in 2018 (p < 0.001). This trend remained consistent when stratified by sex, age, cancer type, operative approach, procedure setting, and institution teaching status. After adjusting for patient and institution factors, the time period was associated with receipt of RBCT with a relative risk of 0.94 (95% CI 0.91-0.96) for 2011-14 and 0.75 (95% CI 0.73-0.78) for 2015-2018 compared to the period of 2007-10. Conclusions: Over the 11-year study period, we observed a decrease in RBCT for GI cancer resection. These findings may reflect the dissemination of clinical guidelines and implementation of patient blood management programs. An evaluation of institutional variation and the relationship with outcomes is warranted to identify opportunities for further improvement.