Substantial Variability in Platelet Transfusion Practice in a Patient-Level Audit of 56,204 Transfusions across 22 Hospitals

Introduction: Platelet transfusions are the second most commonly transfused blood product after red blood cells, with over 2 million units transfused annually in the United States. Despite randomized trials and evidence-based guidelines, recent audits have found high rates of unnecessary transfusion, ranging from 22% to 42%, driven primarily by prophylactic transfusion in non-bleeding patients at a threshold over 10,000/uL. Our multicenter, retrospective observational study sought to characterize transfused patients, assess time trends, and describe variability in pre-transfusion platelet counts across 22 hospital sites. Methods: We conducted a retrospective, multicenter observation study of general medicine wards, subspecialty wards (including hematology-oncology), and critical care areas at 32 hospitals from January 1, 2017 to June 30, 2022 participating in the GEMINI data platform (https://geminimedicine.ca/). Ten sites were excluded for invalid platelet transfusion data. A platelet unit was defined as any platelet type, including apheresis units and pooled whole blood derived platelets. Any platelet units issued within 60 minutes of one another without repeat platelet count between units was considered part of one transfusion event. A transfusion event was tied to the closest platelet count that occurred within the preceding 24 hours. Our primary analysis examined variability in pre-transfusion platelet count across clinical diagnosis, patient subgroups, hospital sites, and clinician characteristics. Results: Across 804,067 admissions over 22 hospitals, 17,777 (2.2%) involved at least one platelet transfusion. The analysis included 56,204 platelet transfusion events. The most common primary diagnoses of transfused patients were malignancy (25.1%), cardiovascular disease (19%), gastrointestinal disease (10.2%) and traumatic injury (8.9%). In an analysis of platelet units per transfusion event, 93.6% were given as 1 unit, 5.2% as 2 units and 1.2% as >2 units. No pre-transfusion platelet count was performed in the preceding 24 hours for 7.2% of transfusion events. Most platelet transfusions were prescribed by Hematology-Oncology (43.2%) and Internal Medicine (20.6%) with median pre-transfusion platelet count being 9,000/uL [IQR 7,000-17,000] and 18,000/uL [9,000-38,000] respectively. Specialties transfusing at the highest median thresholds were Cardiothoracic Surgery (108,000/uL [66,000-174,000]), other surgical specialties (65,000/uL [40,000-100,000]) and Cardiology (53,000/uL [27,000-127,000]). After adjusting for prescriber subspecialty, transfusion threshold was significantly higher with increasing years out of practice, but there was no statistically significant association with prescriber sex. The proportion of platelet transfusion events with pre-transfusion platelet count above 50,000/uL ranged widely from 1.6% to 55.4% across hospital sites. Patients with a diagnosis of hematologic malignancy or other malignancy were transfused at a lower threshold (10,000/uL [7,000-18,000] and 16,000/uL [8,000-34,000] respectively) than patients without known malignancy (42,000/uL [20,000-84,000]). Pre-transfusion platelet count thresholds were stable across the 5-year study period. Conclusion: Our study highlights substantial variability in pre-transfusion platelet counts across different hospital sites, patient diagnoses, and prescriber characteristics. Highlighting practice variation may allow for targeted change interventions to promote guideline adherence and reduce unnecessary transfusion and associated harms.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.