Risk factors and predictive modeling for blood transfusion in extremely preterm infants: The role of perinatal factors and clinical outcomes

Background: Extremely preterm infants (EPIs) are at high risk for severe complications, contributing to neonatal mortality. Blood transfusion is crucial in their management, but its relationship withcomplications remains debated. This study aimed to identify transfusion risk factors in EPIs and develop a predictive model. Methods: We analyzed data from the Dryad database, focusing on EPIs with a gestational age (GA) of <33 weeks. We compared the clinical data between transfused and non-transfused groups, and developed a predictive model for blood transfusion. Results: A total of 578 EPIs were included, with an overall transfusion rate of 20.93%. The transfused group had lower GA, birth weight (BW), hematocrit at 2 hours (Hct2h), and Apgar scores at 1 and 5 minutes than in the non-transfused group ( p < .001). The transfused group also showed higher incidences of intubation, cardiac compression, chronic lung disease, death, length of stay, severe retinopathy of prematurity, necrotizing enterocolitis, any intraventricular hemorrhage (IVH), and severe IVH ( p < 0.01). GA (OR = 0.670, 95% CI: 0.548–0.819, p < 0.001), BW (OR = 0.998, 95% CI: 0.997–1.000, p = 0.007), and Hct2h (OR = 0.888, 95% CI: 0.847–0.930, p < 0.001) were independent risk factors for transfusion in EPIs. The combination of these factors predictedtransfusion needs with an AUC of 0.9145. Discussion: Blood transfusion in EPIs is associated with several complications. BW, GA, and Hct2h are independent risk factors for transfusion, and their combination can effectively predicttransfusion need in this population.