Abstract
BACKGROUND: Red blood cell transfusion is a well-studied intervention with numerous randomized trials to guide management, yet practice remains highly variable. OBJECTIVE: In this study, we explored hospital and patient factors influencing red cell transfusion in a large multi-center database of hospitalized patients. DESIGN: Retrospective analysis PARTICIPANTS: All adults admitted at hospitals in the GEMINI database-a multi-center inpatient database in Ontario, Canada, between December 2016 and June 2022. MAIN MEASURES: The pre-specified primary analysis compared average treatment effect (ATE) of hemoglobin value on transfusion probability in general internal medicine (GIM) patients by hospital type (teaching or non-teaching) and selected medical diagnoses. KEY RESULTS: The primary analysis cohort consisted of 525,510 GIM patients from 24 hospitals. The overall cohort had a 24-h transfusion probability of 79.2% in the < 6.0 g/dL range, 76.2% in the 6.0-6.9 g/dL range, 18.2% in the 7.0-7.9 g/dL range, and 1.1% in the > 8.0 g/dL range. In comparative primary analyses based on medical diagnosis, transfusion probability was higher in the 6.0-6.9 g/dL range for patients presenting to non-teaching hospitals with sickle cell disease (28.7% vs 11.9%, ATE 16.8%, 95% CI 12.6-21.0, p < 0.001), cardiac disease (80.5% vs 75.8%, ATE 4.7%, 95% CI 3.4-5.9, p < 0.001), and gastrointestinal bleeding (87.8% vs 86.1%, ATE 1.8%, 95% CI 0.9-2.6, p < 0.003), and similar for patients with iron deficiency (77% vs 76.5%, ATE 0.5%, 95% CI 0.6-1.6, p = 0.39). CONCLUSIONS: In this longitudinal cohort study, GIM patients with sickle cell and cardiac disease faced substantially different probabilities of red cell transfusion at teaching and non-teaching hospitals.