(B-252) Analyzing Neonatal Mortality: A Systematic Review of Observational Study Approaches

Background: Neonatal mortality, the death of a newborn within the first 28 days of life, is a critical global health challenge. Observational studies using healthcare databases are crucial for evaluating maternal medication use and its impact on pregnancy and neonatal outcomes. However, the methods for identifying neonatal mortality in these databases are unclear.

Objectives: To systematically review methods and algorithms used in large healthcare databases to identify neonatal death and describe the name of database, time assessed, and types of death.

Methods: We conducted a systematic review and followed the PRISMA 2021 guidelines for reporting review (PROSPERO-CRD42023494955). We searched PubMed between January 2010 and July 2024. We included English-language peer-reviewed studies that examined neonatal death using routinely collected healthcare data, excluding those relying on manual reviews or adjudication. Using a pre-defined data collection worksheet, we extracted study characteristics, including country, data source, study period, neonatal death definition, and prevalence. Screening and data extraction were performed independently by two authors, with disagreements resolved by a third author.

Results: We identified 884 studies through PubMed search, with 73 selected for full-text review. An additional 18 studies from reference reviews brought the total to 75. Of these, 42 were from the U.S. and 33 from other countries, with Brazil contributing the highest number (n=5). These studies used more than 50 unique databases. In the U.S., the National Center for Health Statistics linked birth-death records was the most common data source, alongside state vital statistics. Across all studies, no claims-based algorithms for identifying neonatal death were found; linkage to mortality records was often required. Most studies (63, 84%) defined neonatal death as occurring within 28 days, with 32 of the 63 categorizing neonatal death as 0-28 days and the remainder (n=31) categorizing neonatal death as 0-27 days. Six (8%) studies had alternative definitions to operationalize neonatal death, such as in-hospital deaths of a neonate. Neonatal mortality rates (NMR, deaths per 1000 live births) vary by period, region, and population characteristics. NMR ranged from 0.04 (2010-2014) to 19 (1995 -2004) in the US and from 0.52 (Scotland, 1985-2004) to 30 (China, 2008) in non-U.S. studies.

Conclusions: This review highlights that data sources used to study neonatal mortality vary and emphasizes the need for consistent outcome identification. Standardizing neonatal mortality measures across pharmacoepidemiologic studies will improve comparability and advance public health research on medication safety in pregnancy.