A Web-Based Registry Framework Linking Transfusion Practice to Long-Term Patient Outcomes: Design and Single-Center Feasibility Study

Abstract

Abstract Blood transfusions are among the most common medical interventions performed worldwide. Several comprehensive databases link transfusions to routinely collected health data but do not include the indication and the decision-making basis for a particular administration, nor patient-reported long-term outcomes. We developed a web-based registry for the standardized assessment of transfusion practice and patient outcomes, designed for an observation period of up to 30 years. The framework combines clinical data captured at the point of care with patient-reported outcome measures collected through automated digital follow-up, ensures privacy by design through physical separation of identifying and medical data, and codes diagnoses and procedures in SNOMED CT. We evaluated feasibility in a single-center pilot at Goethe University Hospital Frankfurt (11 September 2024–15 January 2026), with patient-reported follow-up collected through 13 August 2026. A total of 408 patients were enrolled, yielding 1,298 administrations across nine treatment categories in a predominantly cardiac surgical cohort. The pilot demonstrated that the registry integrates into routine clinical workflows, captures the parameters required for guideline-adherence monitoring, and supports sex-disaggregated, descriptive analyses of transfusion practice. Follow-up was completed by 12.5% of patients at 3 months and 15.2% of patients at 1 year; paired baseline–follow-up assessments of physical capacity and frailty demonstrate the data structure for capturing functional recovery. Risk-adjusted benchmarking, network-wide analyses, and donor outcomes are designed capabilities of the framework that were not evaluated in this pilot. The platform is available in 16 languages and participation is open to further institutions.

The authors' abstract, as published at the source. Journal of Medical Systems, 2026 · DOI ↗