PRedicting OutcomeS in Preterm nEonates With thromboCyTopenia (PROSPECT) — PROSPECT  

Intraventricular Hemorrhage Clinical Trial

Official title: Dynamic Prediction of Bleeding in Thrombocytopenic Preterm Neonates

Status Recruiting

Clinical Trial Summary

Rationale: Preterm neonates with low platelet counts receive prophylactic platelet transfusions with the aim to prevent bleeding. However, it is not clear in which cases platelet transfusions reduce the risk of bleeding or whether they do more harm than good. A large, randomized trial showed that the higher platelet count threshold for transfusion was associated with a higher rate of death and major bleeding, which suggests that platelet transfusions caused harm in neonates. To gain insight into the risk/benefits of platelet transfusions, the investigators will validate a recently developed dynamic prediction model for major bleeding in multiple NICUs in Europe and investigate the effects of prophylactic platelet transfusions on the risks of bleeding and potential transfusion-associated adverse events. This model could then be used in future studies to define enhanced indications for transfusion, with the ultimate goal to prevent transfusion-associated harm in this vulnerable population. Objectives: 1. Validation of the existing dynamic prediction model in an international cohort of preterm neonates with severe thrombocytopenia (platelet count <50x10^9/L) admitted to a NICU. 2. Model amendment to enable prediction of bleeding risks under various hypothetical platelet transfusion strategies in preterm neonates with severe thrombocytopenia. 3. To examine whether prophylactic platelet transfusions are causally associated with the occurrence of bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), proven sepsis, retinopathy of prematurity (ROP), major bleeding, and mortality. Study design: Multicenter international retrospective cohort study. Study population: Neonates with a gestational age <34 weeks and a platelet count <50x10^9/L, admitted to a NICU between January 1st 2017 and January 1st 2022. Main study endpoints: Major bleeding, BPD, NEC, proven sepsis, ROP and mortality. Nature and extent of the burden and risks associated with participation, benefit, and group relatedness: Not applicable, as this is a retrospective study.

Clinical Trial Description

1. INTRODUCTION AND RATIONALE * Which severely thrombocytopenic neonates do and which do not benefit from prophylactic platelet transfusions? Most platelet transfusions are given prophylactically to non-bleeding neonates with severe thrombocytopenia (i.e., platelet count <50x10^9/L) with the aim to prevent bleeding. The underlying assumption is that correction of thrombocytopenia reduces the risk of bleeding, as platelets play a crucial role in clot formation during hemostasis. However, it is unclear below which platelet count threshold impaired primary hemostasis results in a higher bleeding risk, and if this threshold is similar for all preterm neonates. Furthermore, it is not clear in which cases platelet transfusions reduce the risk of bleeding, or whether platelet transfusions do more harm than good. An international randomized controlled trial (PlaNeT-2/MATISSE), which compared high (50x10^9/L) versus low (25x10^9/L) prophylactic platelet transfusion thresholds in preterm neonates, revealed an unexpected increase in major bleeding or death in the high threshold group (26% versus 19%, OR 1.57; 95% CI, 1.06 to 2.32). These findings suggest that platelet transfusions in preterm neonates may cause harm. This highlights the need for research examining which neonates benefit from platelet transfusions. In the current study, the investigators aim to i) validate and update a dynamic prediction model for major bleeding, and ii) examine whether prophylactic platelet transfusions may be causally associated with potential transfusion-associated adverse events, such as bronchopulmonary dysplasia (BPD), necrotizing enterocolitis (NEC), proven sepsis, retinopathy of prematurity (ROP), major bleeding and mortality * Dynamic prediction model for major bleeding in thrombocytopenic neonates The current platelet transfusion guidelines for prophylactic transfusion decisions are based on platelet count thresholds. However, serious questions are raised regarding the value of platelet counts as a trigger for prophylactic platelet transfusion decisions. Several studies found little correlation between the degree of thrombocytopenia and the incidence of bleeding in preterm neonates, suggesting that factors other than platelet count might be important determinants of bleeding risk in thrombocytopenic neonates. Two neonates with similar platelet counts but different clinical conditions may have distinct risks of bleeding, and may benefit differently from platelet transfusions. Their risks of bleeding can be predicted with a recently developed dynamic model that includes multiple clinical variables in addition to platelet count. As opposed to other prediction models for neonatal bleeding, the advantage of this dynamic model is that clinical variables that change over time (e.g. platelet count and mechanical ventilation) are also taken into account. In this way, the clinical course of neonates can be incorporated into the model and bleeding risk can be estimated at any time point during the first week after the onset of severe thrombocytopenia. * Model validation on new patients is necessary before use in clinical practice Prediction models that perform well in the development cohort often perform worse in other cohorts, because they were developed to fit the original dataset. Before considering their use in clinical practice, they need to be validated in another cohort that includes patients representative of those in whom clinicians would want to use the prediction model. Therefore, the investigators will validate the dynamic model in a new cohort of preterm infants who are admitted to a neonatal intensive care unit (NICU) in Europe. * Prediction of bleeding risks under hypothetical platelet transfusion strategies Although prediction and etiology are typically two distinct research domains that differ in their aim, use, and statistical approach, methods from both prediction research and causal inference research are required when the goal of the model is to inform decision-making. The investigators will amend the existing dynamic model to enable prediction of bleeding risks under different hypothetical platelet transfusion strategies, so the model can be used to predict bleeding risks if no prophylactic platelet transfusion would be provided ('untreated risk') and if a prophylactic platelet transfusion would be provided ('treated risk'). When the model shows a good predictive performance, it could be used to define indications for transfusion, and a randomized controlled impact study could be designed to compare the effect of model-based transfusion decisions with that of platelet count-based transfusion decisions. Ultimately, estimates of individualized treatment effects could be used for the development of individualized platelet transfusion guidelines to optimize transfusion strategies for preterm neonates. 2. SAMPLE SIZE CALCULATION In 2019, 3089 neonates with a gestational age <34 weeks were born in the Netherlands, of which 2176 (70%) neonates were admitted to one of the NICUs in the Netherlands. Data on the incidence of bleeding outcomes in severely thrombocytopenic preterm neonates admitted to a Dutch NICU were available from the development cohort of the dynamic prediction model [6]. During an inclusion period of 5 years (2010-2014) in 7 participating NICUs, 640 (6.9%) severely thrombocytopenic neonates were included out of 9333 neonates with a gestational <34 weeks. In this population, 63 out of 640 (10%) neonates developed major bleeding, 73 out of 640 neonates died (11%), and 132 out of 640 neonates either developed major bleeding or died (21%) during the 10-day follow-up period. To determine the minimum sample size needed for external validation, the investigators used a formula proposed by Riley and colleagues to target precise estimation of the observed/expected (O/E) ratio, calibration slope and c-statistic. The researcher's calculations suggest that at least 1200 participants (120 events of major bleeding) are required to precisely estimate the calibration and discrimination measures, with this number driven by the calibration slope. Over a 5-year period, the investigators expect to include approximately 900 neonates (circa 90 major bleeding events) from all NICUs of the Netherlands and 300 neonates (circa 30 major bleeding events) from NICUs of other European countries. 3. STATISTICAL ANALYSIS The investigators will write a full statistical analysis plan prior to the start of data analyses. Missing data will be handled using simple imputation or multiple imputations where appropriate. 4. ETHICAL CONSIDERATIONS This study will be conducted according to the principles of the Declaration of Helsinki (64th WMA General Assembly, October 2013) and the General Data Protection Regulation (GDPR). In addition, the study will be reported according to the Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis (TRIPOD) guidelines. The Medical Research Involving Human Subjects Act (WMO) does not apply. The investigators will submit the study protocol to the institutional review boards of the coordinating and participating centers for ethical (non-WMO) approval. 5. HANDLING AND STORAGE OF DATA A data sharing agreement (DSA) will be signed between the initiating and participating centers about the intended use, confidentiality, security, data sharing, and potential financial costs. All required data for this study will be collected on electronic standardized Case Report Forms (eCRFs) in a Castor database by study personnel, which may include research nurses, data managers, medical students and PhD students, under supervision of the local and coordinating investigator. Only approved members of the research team (e.g. the coordinating investigator, principal investigators, dedicated data managers and statisticians) will have access to the database, requiring user ID and password access. The dataset is encoded and the patient identification log will be stored separately from the data. The subject identification logs will be kept locally per site. The investigators will make the database as little identifiable as possible, requiring substantial effort to trace data back to individuals. 6. MONITORING AND QUALITY ASSURANCE Given the neglectable risk associated with this observational study, an official monitoring plan is not required. 7. PUBLIC DISCLOSURE AND PUBLICATION POLICY Planned publication in open-access peer-reviewed international scientific journals. Furthermore, results will be published in a PhD thesis. ;

Related Conditions & MeSH terms

• Hemorrhage
• Intraventricular Hemorrhage
• Neonatal Thrombocytopenia
• Platelet Transfusion
• Thrombocytopenia

• NCT number: NCT06043050
• Study type: Observational
• Source: Leiden University Medical Center
• Contact: Hilde van der Staaij, MD
• Phone: +31 (0)71 5262965
• Email: h.van_der_staaij@lumc.nl
• Status: Recruiting
• Start date: September 26, 2022
• Completion date: December 31, 2023