A model to predict latency in preterm prelabor rupture of membranes

Unpredictability regarding latency following preterm premature rupture of membranes (PPROM) complicates counseling and contributes significant patient stress and uncertainty.

To develop a predictive model for latency duration in patients presenting with PPROM.

We conducted a retrospective cohort study of singleton gestations presenting with PPROM between 24w0d and 32w6d from 2017-2022 at two academic hospitals. Patients with active labor, clinical chorioamnionitis, contraindications to expectant management, or known fetal anomalies were excluded. Primary outcomes were latency ≥2 days and ≥7 days. Data on maternal history, prenatal course, and presentation characteristics-including symptoms, laboratory values, vital signs, and fetal data-were collected via chart review. Multivariable logistic regression was used to generate predictive models.

Among 218 patients presenting with PPROM, 192 (88%) met inclusion criteria; exclusions were primarily due to fetal anomalies or contraindications to expectant management. Mean gestational age (GA) at presentation was 30w2d, and mean GA at delivery was 31w5d. Overall, 71.4% achieved a latency of ≥2 days, while 40.6% achieved a latency of ≥7 days. For latency ≥2 days, GA at presentation (p = 0.02), presence of contractions (p = 0.02), and cervical dilation (p = 0.03) were significantly associated. A model incorporating these variables demonstrated moderate discrimination (AUC 0.72). Parity, obstetric history, maternal vital signs, basic laboratory markers, and fetal heart rate patterns were not significant. For latency ≥7 days, only GA (p = 0.01) and cervical dilation (p = 0.01) were significant, yielding a lower-performing model (AUC 0.68).

A predictive model for latency ≥2 days can be generated for patients presenting with PPROM and may support individualized counseling and expectation-setting. An equally accurate predictive model is not able to be generated for latency ≥7 days.