Icon Legend

This session is not in your schedule.

This session is in your schedule. Click again to remove it.

Presentation Icons

Additional registration fee required

Faculty have requested this content not be shared outside of the session

CME Credit Offered

Abstract Award

Recording available 2/16-5/2

AIUM Credit

Foundation Awardee

Poster Icons

Abstract Award

Foundation Awardee

Poster Session 3

Category: Obstetric Quality and Safety

Poster Session 3

(949) Developing a risk prediction model to understand cesarean delivery rates within a large health system

Thursday, February 12, 2026

10:30 AM - 12:00 PM

Submitting Author and Presenting Author(s)

Sarah Harris, MD, MS

Assistant Professor
Wake Forest University School of Medicine
Winston-Salem, North Carolina, United States

Coauthor(s)

Sarah White, MD (she/her/hers)

Assistant Professor
Wake Forest University School of Medicine
Winston Salem, North Carolina, United States
AA

Asma Ahmed, MD, PhD

Wake Forest University School of Medicine
Winston Salem, North Carolina, United States
RH

Robert Phillips Heine, MD

Wake Forest University School of Medicine
Winston Salem, North Carolina, United States

Objective:
Safely reducing the rate of cesarean birth is an important quality improvement initiative. Current definitions of low-risk cesarean deliveries do not account for maternal or fetal factors known to increase the risk for cesarean delivery. We aimed to develop and validate a risk prediction model for cesarean delivery among pregnant patients with no prior births.

Study Design:
We performed a retrospective cohort study within a large health system using Health Records data. All nulliparous patients with a live, term, singleton fetus in the vertex presentation delivered between September 2019 to February 2024 were included. We identified maternal and fetal factors as potential predictors based on literature review and clinical/biological plausibility. Cesarean delivery was defined using diagnostic and procedure codes. We developed a cross-validated mixed-effects logistic regression model (GLMM) to predict cesarean delivery using patient and hospital predictors, applying 5-fold cross-validation and Platt scaling to align predicted and observed rates. Calibration plots and observed versus expected comparisons by hospital were generated.

Results:
Our cohort included 23,724 deliveries occurring at 11 hospitals that had at least 500 deliveries during the study period. The model was well calibrated allowing for accurate site level predictions (Figure 1A). Overall, observed cesarean delivery rates were higher at large, academic hospitals serving higher risk populations and lower for smaller community hospitals (Figure 1B). While observed and predicted rates generally aligned for larger centers, rates differed for smaller hospitals where observed rates tended to be lower than predicted.

Conclusion:
We developed of a cesarean risk prediction model to improve understanding of cesarean delivery rates. The prediction model can be used to adjust expected cesarean delivery rates at the hospital level, allowing institutions to evaluate their cesarean delivery rates more accurately by accounting for patient level data. Future studies are planned to understand the variance observed within our sample.