Impaired Platelet Recovery Is Associated with Mortality after Severe Neonatal Intraventricular Hemorrhage Brief Report

Technical Summary: Impaired Platelet Recovery Is Associated with Mortality after Severe Neonatal Intraventricular Hemorrhage

Problem Statement
Severe neonatal intraventricular hemorrhage (IVH), particularly grades 3 and 4, is a critical complication of extreme prematurity associated with high mortality, post-hemorrhagic ventricular dilatation (PHVD), and adverse neurodevelopmental outcomes. Current risk stratification relies heavily on cranial ultrasound findings and clinical course. However, severe IVH occurs within a context of systemic critical illness, inflammation, and hematologic instability. While thrombocytopenia is common in critically ill preterm infants, a single platelet count has been shown to have limited prognostic value regarding bleeding risk or recovery. This study addresses the gap in understanding whether the longitudinal trajectory of platelet recovery, rather than an isolated measurement, differs by survival status in infants with severe IVH.

Methodology
The authors conducted a retrospective secondary analysis of a single-center cohort from the University Hospital neonatal intensive care unit (NICU) between January 2010 and January 2025.

• Cohort: The parent cohort included 43 infants with grade 3–4 IVH (mean birth weight 895 g, mean gestational age 25.9 weeks), of whom 16 (37.2%) died before discharge.
• Longitudinal Subset: For the primary analysis, infants were included only if they had at least three numeric platelet measurements and known survival status. This yielded a subset of 17 infants (12 survivors, 5 non-survivors) comprising 403 platelet observations.
• Statistical Approach: Two complementary methods were employed:
  1. Observation-Level Linear Regression: Used to plot raw platelet trajectories against postmenstrual age, stratified by survival status. This was treated as a descriptive visualization.
  2. Linear Mixed-Effects Model: Used for primary inference to account for the correlation of repeated measurements within the same infant. The model included a random intercept for each infant and fixed effects for day of life, survival status, same-day platelet transfusion, delayed platelet transfusion, and interaction terms with survival status. P-values were calculated using the Satterthwaite approximation.
• Limitations in Modeling: Due to the small analytic subset, the model was not adjusted for all markers of illness severity (e.g., standardized illness scores, specific timing of sepsis or necrotizing enterocolitis, vasopressor exposure).

Key Results

• Unadjusted Trajectories: In the observation-level analysis, platelet counts diverged sharply by survival status. Survivors showed a significant increase in platelet counts over time (slope +3.47 × 10⁹/L/day), while non-survivors showed a decrease (slope −1.34 × 10⁹/L/day). The time-by-survival interaction was highly significant (p < 2 × 10⁻¹⁶).
• Mixed-Effects Model: After accounting for repeated measures and transfusion timing, the association persisted but with attenuated effect sizes.
  • Survivors demonstrated continued platelet recovery (slope +2.22 × 10⁹/L/day).
  • The time-by-death interaction was −1.46 × 10⁹/L/day (95% CI −2.48 to −0.44; p = 0.005).
  • This corresponds to an estimated slope for non-survivors of approximately +0.76 × 10⁹/L/day, indicating significantly impaired recovery compared to survivors, rather than the steep decline suggested by the unadjusted model.
• Transfusion Impact: Same-day platelet transfusion was associated with lower measured counts (consistent with transfusions occurring at low thresholds), but delayed transfusion was not significant. The authors note these terms reflect clinical indication rather than causal treatment effects.
• Specificity: Exploratory plots of white blood cell count, hemoglobin, and red blood cell count did not show a separation by survival status as pronounced as the platelet trajectory.

Significance and Claims
The paper concludes that failure of platelet recovery is associated with mortality following severe neonatal IVH. The authors emphasize that the central finding is not that a single platelet count predicts mortality, but that the failure to recover may reflect an adverse physiologic trajectory indicative of global illness severity (e.g., ongoing consumption, sepsis, coagulopathy, or marrow suppression).

The study claims that platelet trajectory may provide prognostic information beyond a single count and could complement existing ultrasound-based and clinical risk stratification. However, the authors maintain a modest stance, explicitly stating that:

1. The study is exploratory, retrospective, and limited by a small sample size and potential biases (survivorship and informative missingness).
2. The model lacks adjustment for all relevant confounders.
3. Platelet trajectory should not be used for prognostication until it is validated prospectively in a larger, multicenter cohort that adjusts for gestational age, birth weight, IVH burden, PHVD velocity, infection, and transfusion exposure.