Predictive performance of fetal growth restriction criteria for adverse perinatal outcomes in a hospital in Popayán, Colombia

ObjectivesTo determine the predictive performance of fetal growth restriction by Maternal Fetal Medicine Society (MFMS) definition of ultrasound, the Delphi consensus (DC) and the Barcelona Fetal Medicine (BFM) criteria for adverse perinatal outcomes, and to identify whether there is an association between the diagnosis of fetal growth restriction (FGR) and adverse perinatal outcomes. Material and methodsA retrospective cohort study was conducted including women with singleton pregnancies between 24 and 36 weeks of gestation seen at the maternal fetal medicine unit for ultrasound assessment of fetal growth and delivery care in a public referral hospital in Popayán, Colombia. Pregnancies with ultrasound findings of congenital abnormalities were excluded. Convenience sampling was used. Sociodemographic and clinical variables were measured on admission; additional variables were gestational age, FGR diagnosis and adverse composite perinatal outcome. The predictive ability of three fetal growth restriction diagnostic criteria for poor perinatal outcomes was analyzed and asociation between FGR and adverse perinatlal outcomes. ResultsOverall, 228 pregnant women with a mean age of 26.8 years were included; FGR prevalence according to the three criteria was 3.95 %, 16.6 % and 21.9 % for DC, BFM and MFMS, respectively. None of the criteria resulted in an acceptable area under the curve for the prediction of the composite adverse neonatal outcome; FGR diagnosis by DC and MFMS were associated with adverse perinatal outcomes with a RR of 2.6 (95 % CI: 1.5-4.3) and 1.57 (95 % CI: 1.01-2.44) respectively. No association was found for BFM RR: 1.32 (95 % CI: 0.8-2.1). ConclusionsGiven a positive result for FGR, the Delphi method is significantly associated with adverse perinatal outcomes. The proportion of false negative results for a poor perinatal outcome is high for the three methods. Prospective studies that reduce measurement and attrition bias are required.