(1251) Risk Stratification Model Identifies Unplanned Deliveries <34 Weeks in Placenta Accreta Spectrum

Optimal timing of delivery in pregnancies complicated by placenta accreta spectrum (PAS) remains challenging. Neonatal benefit of prolonging pregnancy must be weighed against risk of maternal morbidity with emergent delivery. Our objective was to create a risk stratification model for prediction of unplanned delivery < 34 weeks in pregnancies complicated by PAS.


Study Design:

Retrospective cohort study of pregnancies with suspected PAS by antenatal ultrasound from 2015–2025 and delivered within our hospital system. Pregnancies were evaluated in two groups: unplanned delivery < 34 weeks and planned delivery ≥ 34 weeks. Data was categorized and modified Poisson regression models were utilized for crude and adjusted incidence risk ratios. A subset of predictor variables was utilized to create a risk scoring system with binary variables adding to a total score ranging from 0 to 7. Pregnancies were stratified into low (score< 2), moderate (score=2), or high risk (score >2) for unplanned delivery < 34 weeks. Receiver operator characteristic (ROC) analysis was utilized to assess predictive ability of this risk stratification model.


Results:

Among 131 pregnancies with suspected PAS, adjusted models identified a significantly increased risk of unplanned delivery < 34 weeks in pregnancies BMI >30, Hispanic ethnicity, cervical length < 3cm in third trimester, ultrasound suspicion percreta, admission for preterm labor (Table 1). A predictive model was developed for risk stratification using variables listed in Table 2. This model resulted in a ROC area under the curve (AUC) of 0.79 (95% CI: 0.69 – 0.89), with 81% of cases correctly classified at the threshold of “high risk” and a specificity of 93%. Pregnancies classified as high risk were 2.5x more likely (RR: 2.47, 95% CI: 1.17-5.21) to have an unplanned preterm delivery compared to low-risk pregnancies.


Conclusion:

Our model had an AUC of 0.79 and correctly identified 81% of unplanned preterm < 34 week deliveries, with a low false negative rate. This predictive ability holds significant clinical value when faced with this life-threatening obstetric condition.