Effect of fresh-frozen plasma transfusion on intraventricular hemorrhage risk in preterm infants with prolonged APTT: a target trial emulation with causal forest analysis.

Background: Coagulopathy is a common risk factor in the occurrence of intraventricular hemorrhage (IVH) and subsequent poor neurological outcomes in preterm infants. The causal effect of fresh-frozen plasma (FFP) transfusion and IVH needed to be identified urgently. Methods: Within the framework of target trial emulation (TTE), we mimic the process of a randomized controlled trial based on retrospective neonatal clinical data collected at Daping Hospital from 2015 to 2025, ultimately including 748 preterm infants with gestational age ≤34 weeks and prolonged activated partial thromboplastin time ≥70 s. Multiple causal inference models were conducted to minimize confounding and covariate imbalance, including inverse probability weighting with generalized boosted models (GBM-IPW) method, GBM-doubly robust (GBM-DR) estimation, covariate balancing propensity score (CBPS) model and other sensitivity analyses. Causal forests algorithm further triangulated the main results and explore individual-level heterogeneity and key determinants about FFP. Results: All causal models consistently demonstrated that FFP transfusion exerted no overall estimated protective effect against IVH (all odds ratio (OR) ≈ 1; Conclusion: Routine FFP transfusion is not recommended for IVH prevention in preterm infants. FFP significantly increases IVH risk in more premature infants, while the adverse association of FFP attenuated with increasing GA.