An Enhanced Method for Identifying Hospital-Based Obstetric Unit Status, Version 2 with Addendum
Health services research has used a range of methods for measuring whether hospitals have an obstetric unit. Variation across methods may lead to inaccurate or inconsistent findings and to different conclusions about access to obstetric services. Having a clear and consistent measurement of obstetric care access is important for policymaking on rural health as well as maternal and infant health. In 2022, we established and published a transparent and replicable process for consistently defining whether a hospital has an obstetric unit. The purpose of this methodology brief is to add updated information on our enhanced method for identifying hospital-based obstetric unit status and for identifying closures of obstetric units. Version two of the methodology was used to identify obstetric care access in the 2010-2023 data resource.
Key Findings:
- We developed an enhanced, two-stage method for identifying hospital-based obstetric unit status. This method involved 1) single-year assessments using four American Hospital Association (AHA) variables and one Provider of Services (POS) variable, followed by 2) multi-year assessments to check for and correct status inconsistencies over time and account for hospital mergers.
- There were 529 (11.2%) hospitals that had inconsistencies in obstetric unit status across the study period (2010-2018), resulting in obstetric unit status changes in 200 of those hospitals (37.8%).
- We compared our method to a primary survey sample of rural hospitals asking about their obstetric unit status. Had we only used POS data, we would have captured 61.0% of obstetric unit closures, on average two years late, and incorrectly identified one additional closure. Had we only used AHA data, we would have captured 87.8% of closures, on average one year late, and incorrectly identified eight additional closures. Our enhanced method was the most accurate, capturing 92.7% of closures 0.3 years late with six incorrectly identified additional closures.
- Had we ignored hospital mergers, we would have identified 32 fewer hospitals with current obstetric services, but 44 additional hospitals would have looked like they had experienced an obstetric unit closure when they had not.