Maternal age and the predictive performance of the uterine artery pulsatility index for composite adverse perinatal outcomes in late-onset fetal growth restriction: a retrospective cohort study

Abstract

Background This study aimed to evaluate whether the predictive performance of the uterine artery pulsatility index (UtA-PI) for composite adverse perinatal outcomes (CAPO) in late-onset fetal growth restriction (FGR) differs according to maternal age. Methods This retrospective cohort study included singleton pregnancies with late-onset FGR (>32 weeks of gestation) managed at a tertiary care center. Mean UtA-PI measured at diagnosis was normalized for gestational age and evaluated for its association with CAPO across predefined maternal age groups. Predictive performance was assessed using receiver operating characteristic (ROC) analysis and multivariable logistic regression, with formal testing for age-related effect modification. Results A total of 234 pregnancies with late-onset FGR were included, comprising 62 women <25 years, 88 aged 25-34 years, and 84 aged >= 35 years. CAPO occurred in 29.9% of all pregnancies. Mean UtA-PI values and the proportion of UtA-PI >= 90th percentile increased progressively with advancing maternal age (22.6%, 26.1%, and 36.9%, respectively; p = 0.04 ). The predictive performance of UtA-PI for CAPO improved across age groups, with area under the curve values for the >= 90th percentile threshold increasing from 0.60 in women < 25 years to 0.74 in those >= 35 years. After multivariable adjustment, elevated UtA-PI remained independently associated with CAPO, with the strongest association observed in women aged >= 35 years (adjusted OR 3.5, 95% CI 1.6-7.9). Conclusions These findings suggest that maternal age may provide additional clinical context when interpreting UtA-PI-based risk estimates in late-onset FGR. UtA-PI demonstrated modest prognostic performance in women aged >= 35 years, whereas discriminative ability was lower in younger maternal age groups. Prospective validation is required before these findings influence clinical practice.