The research, led by Bailey B. Zeiler and Robert D. Christensen of the University of Utah, together with colleagues at Intermountain Health, ARUP Laboratories, and Boston Children’s Hospital, set out to answer a deceptively simple question: what is actually normal for a newborn’s platelet count? Rather than relying on the conventional adult-derived figure, the team turned to the international hematologic consensus definition of cytopenia, which classifies a blood count as abnormally low only when it falls below the 2.5th percentile of an appropriate comparison population. Applying that standard to neonates required accurate reference intervals for newborn platelet counts, which historically have been difficult to establish.

The key methodological innovation was the refineR algorithm, a statistical tool developed to estimate reference intervals from large volumes of routine clinical laboratory data. Traditional approaches to defining reference intervals require recruiting large numbers of demonstrably healthy individuals and measuring them under standardized conditions, an approach that is logistically daunting in neonatology, where ethically obtaining blood samples from healthy newborns is nearly impossible. Indirect methods such as refineR sidestep this problem by mining the enormous datasets that hospitals already generate, mathematically separating the distribution of results from healthy individuals from the skewed contributions of sick patients whose abnormal values pull the tails of the distribution.

RefineR works by modeling the mixture of distributions present in real-world laboratory data, identifying the dominant, presumably healthy subpopulation and estimating its central tendency and spread without requiring manual exclusion of pathological results. The algorithm has been validated for a growing list of laboratory parameters and has been recommended in guidance from the International Federation of Clinical Chemistry and Laboratory Medicine on indirect methods for reference interval determination. The Utah and Boston team had previously used refineR to establish neonatal reference intervals for other hematologic measures, including nucleated red blood cell counts and red cell parameters, and platelets were the natural next target.

The scale of the analysis is what gives the new intervals their statistical power. Drawing on large multihospital clinical databases, the researchers assembled 155,062 neonatal platelet counts. From this dataset they generated reference intervals for the day of birth, stratified by gestational age, and for the first 30 days after birth across all gestational ages. The results were striking. For neonates born at 34 weeks of gestation or later, the lower reference limit, corresponding to the 2.5th percentile, was 119,600 platelets per microliter, with a 95 percent confidence interval of 117,700 to 135,300. For neonates born before 34 weeks, the lower limit was even lower, at 103,000 platelets per microliter, with a confidence interval of 98,700 to 105,600.

Those numbers carry an immediate clinical implication that the authors state plainly: platelet counts in the range of 120,000 to 150,000 per microliter at birth are normal for many newborns. Under the traditional definition, a term infant with a platelet count of 135,000 per microliter would be labeled thrombocytopenic, potentially triggering additional testing, parental anxiety, and in some cases interventions. Under the new evidence-based intervals, that same count sits comfortably within the expected range. The gap between the old and new definitions is not a marginal statistical quibble; it encompasses a substantial band of counts that clinicians encounter every day in nurseries and neonatal intensive care units.

The stakes extend well beyond labeling. Platelet transfusions are among the most common blood products administered to preterm infants, and the thresholds at which transfusion is given have been the subject of intense debate. The landmark PlaNeT-2 randomized trial, published in the New England Journal of Medicine in 2019, found that prophylactic platelet transfusion at a higher threshold of 50,000 per microliter did not reduce major bleeding compared with a lower threshold of 25,000 per microliter in neonates with thrombocytopenia, and follow-up studies have raised concerns about long-term outcomes associated with more liberal transfusion practices. Subsequent work has documented the challenges of implementing restrictive transfusion guidelines in neonatal units, and international clinical practice guidelines from the AABB and the International Collaboration for Transfusion Medicine Guidelines have endorsed more conservative approaches.

Accurate definitions of thrombocytopenia are the foundation on which such transfusion decisions rest. If the diagnostic threshold itself is miscalibrated, then every downstream decision, from the ordering of infectious disease workups to the administration of transfusions, inherits that error. Overdiagnosis of thrombocytopenia can lead to unnecessary transfusions, which carry known risks including transfusion-associated circulatory overload, infection transmission, and, according to some recent research, possible contributions to conditions such as bronchopulmonary dysplasia. Underdiagnosis, conversely, could delay evaluation of infants who genuinely have pathological platelet counts caused by sepsis, necrotizing enterocolitis, alloimmune processes, or congenital syndromes.

The new reference intervals also refine the picture of how platelet counts naturally vary in the earliest days of life. By modeling counts across the first 30 days, the study provides a dynamic view of the expected trajectory, acknowledging that the normal range in a three-day-old preterm infant differs from that of a newborn on the day of birth. Previous efforts, including a well-known 2009 analysis of more than 47,000 patients from the same Intermountain healthcare system, established that neonatal platelet counts vary with gestational age and postnatal age, but the refineR approach allows continuous, statistically robust estimation from routine data rather than relying on manually curated healthy cohorts.

The authors emphasize that the study was a retrospective review of de-identified records, approved by the Intermountain Health Institutional Review Board with a waiver of informed consent, and conducted in accordance with federal human subjects protections. The team reports no competing interests, and the statistical code is available upon request. As neonatal units and clinical laboratories consider adopting the new intervals, the practical effect will likely be a recalibration of what triggers concern: a count that once read as mildly low may now read as normal, reserving the label of thrombocytopenia, and the interventions that follow it, for infants whose platelet counts genuinely fall below the physiological range for their gestational age and postnatal day. In a field where the smallest patients receive the most cautious care, that shift in the denominator of normal could spare thousands of newborns each year from unnecessary labels and the cascade of interventions that labels invite.

Subject of Research: Establishment of evidence-based neonatal platelet reference intervals using the refineR algorithm to redefine thrombocytopenia in newborns

Article Title: Redefining neonatal thrombocytopenia: an evidence-based approach, improved by the refineR algorithm

Article References: Zeiler, B. B., Bahr, T. M., Doyle, K., Henry, E., Carlson, L. C., Davenport, P., Sola-Visner, M. C., Ohls, R. K., & Christensen, R. D. (2026). Redefining neonatal thrombocytopenia: an evidence-based approach, improved by the refineR algorithm. Journal of Perinatology. https://doi.org/10.1038/s41372-026-02888-3

Image Credits: AI Generated

DOI: 10.1038/s41372-026-02888-3

Keywords: neonatal thrombocytopenia, platelet count, refineR algorithm, reference intervals, newborns, preterm infants, platelet transfusion, neonatology, hematology, Journal of Perinatology, gestational age, clinical laboratory data