Poster Session 1
Juliet Musabeyezu, MD
Resident in Obstetrics and Gynecology
Massachusetts General Brigham
Boston, MA, United States
Kaitlyn E. James, MPH, PhD (she/her/hers)
Massachusetts General Hospital
Boston, Massachusetts, United States
Christina M. Duzyj, MD, MPH (she/her/hers)
Director, Maternal Fetal Medicine Fellowship Program
Massachusetts General Hospital
Boston, MA, United States
William H Barth, Jr., MD
Massachusetts General Hospital
Boston, Massachusetts, United States
Thomas H. McCoy, MD
Massachusetts General Brigham
boston, Massachusetts, United States
Roy H. Perlis, MD
Massachusetts General Hospital
Boston, Massachusetts, United States
Anjali J. Kaimal, MD (she/her/hers)
Professor and Vice Chair of Clinical Operations, Department of OBGYN
University of South Florida
Tampa, Florida, United States
Mark A. Clapp, MD, MPH (he/him/his)
Physician Investigator
Massachusetts General Hospital
Boston, MA, United States
PPH is one of the most common causes of morbidity during labor and delivery. Most current risk stratification tools rely on manual completion of checklists at multiple time points, increasing non-patient-facing tasks for the care team and relying on information recorded in multiple places in the EHR. Our objective was to automate a common PPH risk stratification tool using data routinely captured in the EHR and assess its performance to predict PPH.
Study Design:
We automated the assignment of the AWHONN PPH Risk Tool designation using structured, real-time data in the EHR at the time of admission, pre-delivery, and postpartum. Individuals were classified as low, medium, or high risk using the AWHOON PPH Tool risk factor framework (v1.2). We assessed the performance of the automated tool to predict PPH (EBL≥1,000 mL) by measuring sensitivity, specificity, PPV, and NPV for the high-risk group at each timepoint using data from all individuals who delivered at a single academic institution between July 2023 and April 2024.
Results:
1208 participants were included, of which 141 (11.8%) had a PPH. The distribution of patients between the risk categories at the three timepoints and the rate of PPH are shown in the Figure. 32.8%, 42.5%, and 45.0% were high-risk on admission, pre-birth, and post-birth, respectively. The rate of PPH increased with corresponding risk categories at each time point, demonstrating the tool’s value in stratifying risk. On admission, the rate of PPH in the low risk compared to the high-risk group was 6.5% vs. 17.4%. Among those classified as high risk, the sensitivity increased from 48.9% to 61.0% from admission to post-birth, and the PPV decreased from 17.4 to 15.8%.
Conclusion:
A real-time, automated version of a common PPH risk assessment tool can stratify the PPH risk using clinical EHR data. Future work is focused on the ability of this tool to reduce administrative burdens of the team and improve team awareness and communication through the triggering of real-time, automated alerts for PPH risk.
