Curiosity and Caution: Interpreting Standardized Maternal Health Data

Abstract

In the United States, pregnant people face a higher risk of morbidity and mortality than in any other high-income country, and the U.S. is the only one of these countries with rising maternal mortality. Profoundly, we know from maternal mortality reviews that occur in every U.S. state that 80% of these deaths are preventable. Nationwide efforts are underway to measure and understand causes of adverse outcomes – including the need for robust and standardized data to better understand the clinical and psychosocial drivers of outcomes for this population.Premier’s statistically deidentified PINC AI™ Healthcare Database (PHD), is a hospital-based, all-payer discharge database for geographically diverse inpatient and outpatient visits. The PHD also contains linked maternal and infant records which enable insights within the linked populations. International Classification of Diseases, Tenth Revision (ICD-10) coding is used to define maternal clinical conditions for analysis. The database is robust, standardized, and validated; but even such datasets come with their challenges. ICD-10 codes allow for standardized measurement, but when hospital coding practices are nonstandard, more questions need to be asked to understand the full story the data is telling.During this session, attendees will learn about how Premier collects disparate data sources, reconciles, validates, maps, and conducts quality assurance to create a robust, standardized database. They will then learn how even standardized data needs to be approached with caution and curiosity, to explore the data and create meaningful insights to better understand morbidity and mortality among the maternal population.