Performance of a local reference curve for predicting small for gestational age fetuses in pregnant women with HIV/AIDS

AbstractPurposeTo compare the performance of a local estimated fetal weight curve with curves established for other populations to predict small for gestational age (SGA) fetuses.MethodsA retrospective and cross‐sectional study involving 231 fetuses in which the performance of a local curve (proposed model) was compared with the Hadlock and Intergrowth‐21st curves in the prediction of SGA fetuses, by applying them to a population of high‐risk pregnant woman with HIV/AIDS. For each model, a receiver operating characteristic curve was adjusted, considering the SGA classification by the neonatal Intergrowth method as the gold standard, and the area under the curve (AUC) was calculated.ResultsThe models presented linear correlations with each other. The agreement of the proposed model with Hadlock was very good (kappa = 0.83), whereas the proposed model and Intergrowth‐21st had moderate agreement (kappa = 0.44). The SGA fetus detection sensitivities of the proposed model and Hadlock were 61.9% and 57.1%, with specificity of 84.1% and 86.2% and accuracy of 80.1% and 81%, respectively, without statistical difference. The sensitivity of the Intergrowth‐21st model was 33.3%, while the accuracy was 85.7% and the specificity was 97.4%. The AUC estimated values for the Hadlock, proposed, and Intergrowth‐21st models were 0.834, 0.832, and 0.835, respectively.ConclusionThe proposed model and Hadlock were interchangeable in the prediction of SGA fetuses and superior to the Intergrowth‐21st model.