Heterogeneous Associations of Latency Duration with Neonatal Outcomes After Preterm Prelabor Rupture of Membranes: A Nationwide Cohort Study

Importance: Expectant management for preterm prelabor rupture of membranes (PPROM) is guided primarily by gestational age (GA), yet whether the association between latency duration and neonatal outcomes varies among pregnancies remains uncertain. Objective: To characterize heterogeneous associations between latency duration and neonatal outcomes after PPROM and identify maternal and fetal characteristics associated with the largest estimated benefit. Design: Nationwide registry-based retrospective cohort study using restricted cubic spline regression and causal forest analysis. Data were analyzed from July 2025 to July 2026. Setting: The Korean Neonatal Network (KNN), a nationwide registry of infants born before 32 weeks' gestation or with a birth weight less than 1,500g, from 2013 to 2023. Participants: A total of 7,057 neonates born after PPROM were included. Exposures: Latency duration, defined as the number of days from PPROM diagnosis to delivery (0-118 days). Main Outcomes and Measures: Neonatal Health Index (NHI), an ordinal composite from 0 (death) to 8 (survival free of 7 major neonatal morbidities: intraventricular hemorrhage grades 3-4, periventricular leukomalacia, retinopathy of prematurity stage [&ge;]3, bronchopulmonary dysplasia, pulmonary hypertension, necrotizing enterocolitis stage [&ge;]2, and sepsis). Results: Among 7,057 neonates (median [IQR] maternal age, 34.0 [31.0-36.0] years; GA at PPROM, 27.0 [24.0-29.0] weeks; latency duration, 2.0 [0.0-8.0] days), longer latency was associated with better neonatal health, reflected by higher NHI (n = 6,872), in a non-linear, GA-dependent pattern (overall P < .001; nonlinearity P < .001). Causal forest analysis estimated an average partial effect of a 0.078-point increase in NHI per additional day of latency (P < .001), with significant heterogeneity across individuals (P < .001). GA at PPROM, oligohydramnios, and maternal age were the strongest modifiers; neonates with GA at PPROM of 26 weeks or less, no oligohydramnios, and maternal age younger than 30 years had an estimated treatment effect 28.0% greater than the rest of the cohort (0.121 vs 0.094 NHI points per latency-day). Conclusions and Relevance: Associations between neonatal outcomes and prolonged latency after PPROM were positive overall but heterogeneous in magnitude, varying by GA at PPROM, oligohydramnios status, and maternal age. These findings support tailoring expectant management after PPROM according to individual patient characteristics rather than a uniform, GA-based strategy.