(231) Characterizing inpatient electronic health records (EHR) from a large hospital network for use in neonatal pharmacoepidemiology studies in the United States

Background: The Sentinel System—U.S. Food and Drug Administration’s active surveillance system—includes inpatient electronic health record (EHR) data from HCA HealthCare (HCA). In 2020, pediatric data were made available to Sentinel. We characterized data elements that may support pharmacoepidemiologic studies in hospitalized neonates.

Objectives: To describe neonates included within a large source of inpatient EHR data, and characterize the availability of data typically required in neonatal pharmacoepidemiology studies.

Methods: We identified live births within inpatient EHR data from 144 HCA hospitals from January 2020 - October 2023. The data included coded diagnoses and procedures, inpatient pharmacy administrations, and some vital signs data. We described high-risk conditions, treatments, length of stay, and other care provided to all neonates and those with a neonatal intensive care unit (NICU) stay. We assessed preterm birth and low birth weight (LBW) with diagnosis codes and investigated birth weight data availability. We defined surgeries with procedure codes, and medications/vaccinations frequently administered to neonates with national drug codes and generic/brand names.

Results: We identified 762,220 neonates, of which 166,408 (22%) had evidence of NICU stays. All neonates and the subset with NICU stays were similar with regards to demographics (50% were male and majority were White). Higher proportions of neonates with NICU stays versus all neonates were preterm (38% vs. 11%) and had LBW (22% vs. 6%). Most neonates (99%) had at least one weight entry in recent vitals data (2021 onward). Compared to all neonates, those with NICU stays had more comorbidities, including hyperbilirubinemia (49% vs. 24%) and acute respiratory distress syndrome (17% vs. 4%), more frequent cardiac surgery (11% vs. 4%), and longer lengths of stay (average 11.4 days vs. 4.8 days). Endotracheal intubation, and exchange transfusion were not observed, although we observed surfactant use in 8% of those with NICU stays (2% overall). Most neonates had vitamin K administrations (96%), Hepatitis B vaccinations (87%), and use of antibiotics/antifungals (including topical and ophthalmic formulations) commonly used in the neonatal population (95%).

Conclusions: Data typically required for neonatal pharmacoepidemiology studies such as preterm status, birth weight, pharmacotherapy, and vaccination were frequently recorded within inpatient EHR data. Neonates admitted to the NICU had more comorbidities and treatments than the general neonatal population. However, procedures such as endotracheal intubation may be under captured. Future studies could consider augmenting these data with clinical notes and investigating additional components of the EHR specific to newborns.