Predicting the risk of respiratory distress in newborns with congenital pulmonary malformations
Résumé
Objectives Most children with prenatally diagnosed congenital pulmonary malformations (CPMs) are asymptomatic at birth. We aimed to develop a parsimonious prognostic model for predicting the risk of neonatal respiratory distress (NRD) in preterm and term infants with CPM, based on the prenatal attributes of the malformation. Methods MALFPULM is a prospective population-based nationally representative cohort including 436 pregnant women. The main predictive variable was the CPM volume ratio (CVR) measured at diagnosis (CVR first) and the highest CVR measured (CVR max). Separate models were estimated for preterm and term infants and were validated by bootstrapping. Results In total, 67 of the 383 neonates studied (17%) had NRD. For infants born at term (>37 weeks, n=351), the most parsimonious model included CVR max as the only predictive variable (receiver operating characteristic (ROC) curve area: 0.70±0.04, negative predictive value: 0.91). The probability of NRD increased linearly with increasing CVR max and remained below 10% for CVR max <0.4. In preterm infants (n=32), both CVR max and gestational age were important predictors of the risk of NRD (ROC: 0.85±0.07). Models based on CVR first had a similar predictive ability. Conclusions Predictive models based exclusively on CVR measurements had a high negative predictive value in infants born at term. Our study results could contribute to the individualised general risk assessment to guide decisions about the need for newborns with prenatally diagnosed CPM to be delivered at specialised centres.