Association between red blood cell transfusion volume and infection risk: a dose-response analysis of a nationwide...

Infection is a major complication after trauma and is associated with worse clinical outcomes. Although red blood cell (RBC) transfusion is an essential component of resuscitation in trauma patients, its relationship with infection risk remains controversial. Furthermore, the dose-response relationship has not been fully elucidated.

We conducted a retrospective observational study using a nationwide trauma registry in Japan. Trauma patients registered between 2019 and 2021 were included. Patients were categorized into four groups according to RBC transfusion volume within the first 24 h after admission: None (0 units), Low (1-4 units), Moderate (5-9 units), and High (≥ 10 units). The primary outcome was in-hospital infection. Multivariable modified Poisson regression analyses were performed to estimate adjusted risk ratios (RRs) with 95% confidence intervals (CIs). Furthermore, we performed multivariable-adjusted restricted cubic spline analyses with four knots to assess the adjusted dose-response relationship between RBC transfusion volume and the risk of in-hospital infection.

A total of 55,807 patients were analyzed. The incidence of in-hospital infection increased across transfusion categories: 4,399/48,308 (9.1%) in the None group, 618/3,724 (16.6%) in the Low group, 396/1,733 (22.8%) in the Moderate group, and 548/2,042 (26.8%) in the High group. In multivariable analysis, compared with the None group, Low (adjusted RR 1.85, 95% CI 1.62-2.10), Moderate (adjusted RR 1.95, 95% CI 1.69-2.26) and High (adjusted RR 2.07, 95% CI 1.76-2.43) transfusion volumes were associated with significantly increased risks of infection, respectively. Restricted cubic spline analyses showed an increase in infection risk at lower transfusion volumes, followed by a plateau at higher transfusion volumes.

RBC transfusion volume within the first 24 h after trauma was associated with in-hospital infection risk. The association was generally linear at lower transfusion volumes, whereas a plateau was observed at higher transfusion volumes. Given the observational nature of this study, these findings should be interpreted cautiously.