Personalized Prediction of Platelet Transfusion Outcomes in Preterm Infants with Severe Thrombocytopenia

Highlight

• Prophylactic platelet transfusion in preterm infants with severe thrombocytopenia shows significant individual variability in preventing major bleeding or mortality.
• A novel dynamic prediction model was developed and externally validated to estimate 3-day risk of adverse outcomes under two transfusion strategies: prophylaxis versus no prophylaxis.
• The model incorporates time-varying clinical factors and transfusion timing to guide individualized transfusion decisions dynamically during the first week after thrombocytopenia onset.
• Model validation demonstrated good discrimination and calibration, supporting its potential utility in clinical decision-making for this vulnerable population.

Background

Severe thrombocytopenia (platelet counts below 5099/L) is common in preterm infants, especially those born before 34 weeks of gestation. Such platelet deficiency predisposes to life-threatening bleeding complications, which remain a leading cause of mortality and morbidity in neonatal intensive care units (NICUs). Consequently, platelet transfusions are frequently administered prophylactically to prevent bleeding. However, the benefits of platelet transfusions remain uncertain, and emerging data suggest potential harm from unnecessary transfusions, including infection, alloimmunization, and increased mortality risk. Hence, there is an unmet clinical need for individualized strategies to optimize platelet transfusion in this vulnerable group to balance risks and benefits appropriately.

Study Design

This study developed and validated a dynamic prediction model in two multicenter cohorts of preterm infants with severe thrombocytopenia. The development cohort included 1,042 infants from 14 NICUs across the Netherlands, Sweden, and Germany from 2017 to 2021. The external validation cohort comprised 637 infants from 7 Dutch NICUs over 2010 to 2014. Inclusion criteria were infants born before 34 weeks gestation diagnosed with platelet counts less than 5099/L.

Two transfusion strategies were compared dynamically at multiple prediction time points throughout the first week after thrombocytopenia onset: (1) receiving a platelet transfusion within 6 hours (prophylaxis), and (2) no platelet transfusion for the subsequent three days (no prophylaxis). The primary outcome was development of major bleeding or mortality within three days after each prediction point. Key predictors captured at each landmark time included gestational and postnatal age, small-for-gestational age status, presence of necrotizing enterocolitis or sepsis, respiratory support requiring mechanical ventilation, usage of vasoactive agents, platelet count, and prior transfusion history.

Sophisticated statistical methods combining landmarking and the clone-censor-weight approach allowed rigorous adjustment for time-varying confounding and enabled comparison of dynamic risks under each transfusion strategy.

Key Findings

The median gestational age in both cohorts was approximately 28 weeks, with median birthweight around 900 grams. Approximately 59% of infants were male. Major bleeding or death occurred in 23% of infants in the development cohort and 21% in the validation cohort.

Model performance in the validation dataset showed satisfactory discrimination with a time-dependent area under the receiver operating characteristic curve of 0.69 (95% CI, 0.60-0.76) for the prophylaxis strategy and 0.85 (95% CI, 0.76-0.92) for the no prophylaxis strategy. Calibration plots confirmed good agreement between observed and predicted risks.

Importantly, predicted 3-day risks of major bleeding or death varied substantially among individual infants depending on their evolving clinical status and transfusion strategy. For example, infants exhibiting severe clinical conditions (e.g., ongoing sepsis, mechanical ventilation) had higher baseline risks, and prophylactic transfusion demonstrated differential impact across these risk strata. This underscores the limitation of uniform transfusion thresholds and highlights the potential benefit of personalized decision-making guided by dynamic risk prediction.

Expert Commentary

This study represents a significant advance in neonatal hematology by providing a validated tool to individualize platelet transfusion timing in preterm infants with severe thrombocytopenia. Previous clinical guidelines predominantly recommend transfusions based on static platelet count thresholds, without accounting for the complex interplay of clinical factors affecting bleeding risk.

The model’s incorporation of time-updated patient conditions and its rigorous methodology to account for treatment-confounder feedback is a methodological strength, addressing a common challenge in observational data analysis. Such dynamic modeling aligns with personalized medicine principles and could potentially reduce unnecessary transfusions and their associated risks.

Limitations include the moderate discrimination for the prophylaxis strategy, reflecting perhaps the multifactorial and unpredictable nature of bleeding complications. External validation was performed within Dutch centers, and broader geographic validation would strengthen generalizability. Future prospective validation and impact studies are warranted to assess clinical utility and integration into bedside decision support.

Conclusion

In summary, this international multicenter study developed and externally validated a dynamic prediction model that estimates three-day risk of major bleeding or death in preterm infants with severe thrombocytopenia under alternative platelet transfusion strategies. The model highlights considerable individual variability in potential benefits or harms from prophylactic transfusions based on current clinical parameters. Implementation of this model could enable tailored, evidence-based transfusion decisions, minimizing overtreatment and optimizing clinical outcomes in this fragile population. Ongoing research should focus on prospective validation and user-friendly integration into neonatal care pathways to realize personalized platelet transfusion in routine practice.

Funding and Clinical Trials

The study was supported by collaborative funding from international neonatal research consortia. The clinical trial registration and funding details can be found in van der Staaij et al., JAMA 2025.

Reference

van der Staaij H, Prosepe I, Caram-Deelder C, Keogh RH, Deschmann E, Dame C, Onland W, Prins SA, Cassel F, d’Haens EJ, van Westering-Kroon E, Andriessen P, Vrancken SL, Hulzebos CV, Vijlbrief DC, Fustolo-Gunnink SF, Fijnvandraat K, Lopriore E, van der Bom JG, van Geloven N. Individualized Prediction of Platelet Transfusion Outcomes in Preterm Infants With Severe Thrombocytopenia. JAMA. 2025 Oct 14;334(14):1267-1277. doi: 10.1001/jama.2025.14194. PMID: 40952748; PMCID: PMC12439188.