Monet

Monitoring Outcome in Neonatal Thrombocytopenia

Rationale

Approximately 10% of preterm neonates admitted to a neonatal intensive care unit develop major hemorrhage during their admission. Some baseline models have been developed to predict bleeding risk, but these do not take into account clinical variables that change over time, such as sepsis or whether a neonate is mechanically ventilated. Thrombocytopenia is considered a risk factor for neonatal bleeding, but the relationship between platelet count and bleeding risk is unclear. Moreover, the practice of correcting low platelet counts in order to prevent bleeding is not evidence-based.

Objective

To develop a clinically applicable dynamic prediction model to facilitate the prediction of an individual neonate’s probability of developing a major bleed.

Study design

Multicenter observational cohort study

Population

Neonates with a gestational age of <34 weeks at birth, born in one of 7 participating hospitals: University Medical Center Groningen (UMCG), Isala Klinieken Zwolle (Isala), University Medical Center Utrecht (UMCU), Academic Medical Center Amsterdam (AMC), Leiden University Medical Center (LUMC), Maxima Medical Center Veldhoven (MMC) and Erasmus University Medical Center Rotterdam (Erasmus MC). >Neonates born between januari 1 2010 and december 31 2014 were included.

Inclusion criteria

  1. Neonates with a gestational age at birth of < 34 weeks 
  2. Admitted to one of seven participating neonatal intensive care units (NICU) and 3) with a platelet count of less than 50x109/L

Exclusion criteria 

  1. Significant congenital malformations 
  2. High suspicion of spurious platelet count (e.g. clots in the sample, or very rapid ‘recovery’ to previous non-severely thrombocytopenic levels)
  3. Prior admission to another NICU or readmission, 4) platelet count <50x109/L exclusively in context of exchange transfusion, 5) major bleeding prior to the start of severe thrombocytopenia

Main study endpoint 

The primary outcome measure is major bleeding, which is defined as intraventricular hemorrhage (IVH) grade 3, IVH of any grade with parenchymal involvement, other types of intracranial hemorrhage visible on ultrasound scans, pulmonary hemorrhage or any other type of bleeding requiring immediate interventions such as erythrocyte transfusions, inotropic support and volume boluses.

Secondary study endpoint 

Mortality

Clinical prognostic variables 

Prognostic variables will be identified through a review of literature in combination with expert advice. Both baseline and time-dependent variables will be collected.

Sample size 

The expected thrombocytopenia rate in this population is approximately 10%. Few studies record the bleeding rate in this specific population, the rate was estimated to be at least 10%. It is unclear what proportion of the major bleeds develop prior to the occurrence of thrombocytopenia. We expect to be able to include 1 variable per 100 neonates included in the study.

Assessments

Only data generated through standard care will be collected. This includes platelet counts, cerebral ultrasound scan reports, information about bleeding and transfusions, and multiple clinical variables. All variables will be recorded including date and time.

Data collection

Data will be collected using digital extracts of baseline data and platelet counts from the national neonatal research database and the local hospital computer systems. Additional data will be retrieved and recorded manually using digital and paper patient records. Data will be entered into a GCP certified online database.

Statistical analyses 

A statistical analysis plan will be drafted and signed prior to any data analyses. Variables to be included in the main model will be identified based on literature and expert opinion, prior to analyses of the data, and recorded in the statistical analysis plan. The model will be developed using dynamic prediction modeling using landmarking.

Guidelines 

The study will be performed according to the recommendations of the PROGRESS framework, and results will be presented according to the TRIPOD guidelines.

People

Suzanne Gunnink
Johanna (Anske) van der Bom
Camila Caram Deelder
Isabelle Ree
Karin Fijnvandraat
Enrico Lopriore (LUMC)
Hein Putter (LUMC)

Many neonatologists from hospitals throughout The Netherlands.