(513) Comparison of sepsis case definitions among obstetric delivery encounters

Objective:
Accurate retrospective labeling of obstetric sepsis is critical for performance improvement, reporting and epidemiology. Multiple billing and electronic health record case definitions of sepsis exist for non-pregnant adults. We sought to compare incidence rates of sepsis across four case definitions.

Study Design:
A retrospective cohort study of all delivery encounters across a large, academic health system of 8 hospitals from 1/1/2023 to 12/31/2024 that met at least one of four sepsis case definitions: Centers for Disease Control (CDC) clinical definition, and Centers for Medicare & Medicaid (CMS), New York State Department of Health (NYSDOH), or Vizient ICD-10 diagnosis code definitions. Identified charts were reviewed by obstetricians to determine if sepsis was present. The incidence rate of obstetric sepsis for each case definition was calculated and visualized and the frequency of intersection combinations of case definitions was visualized using an upset graph and Venn diagram.

Results:
Of 49,083 delivery hospitalizations, 171 met criteria for at least one case definition. Encounter characteristics are summarized in the Table. Sepsis incidence rates varied by case definition: CDC 2.2%, CMS 1.9%, NYSDOH 0.5%, Vizient 2.4%, and clinical review 2.1%. Of the 171 cases, 60.8% were deemed to meet sepsis criteria on clinical chart review. The intersections of case definition combinations are shown in the Figure. There was large variability, with only 18 cases (10.5%) meeting the four case definitions and clinical review.

Conclusion:
Discrepancies in incidence rates among different sepsis definitions applied to obstetric sepsis cases highlights variability in retrospective sepsis identification and suggests potential under- or overestimation of obstetric sepsis incidence. The use of any of the four case definitions may lead to variable identification of obstetric sepsis cases and estimates of incidence. Tailored criteria for obstetric sepsis clinical and coding definitions are warranted.