Identification of the optimal growth chart and threshold for the prediction of antepartum stillbirth

To evaluate the effect of the choice growth chart and threshold used to define small for gestational age (SGA) on the predictive value of SGA for placenta-related or unexplained antepartum stillbirth.A retrospective cohort study of all women with a singleton pregnancy who gave birth 24 week gestation in a single center (2000-2016). The exposure of interest was SGA, defined as birth weight 10th or 25th centile according to three fetal growth charts (Hadlock et al., Radiology 181:129-133, 1991; intergrowth-21st (IG21), WHO 2017, and a Canadian birthweight-based reference-Kramer et al., Pediatrics 108:E35, 2001). The outcome of interest was antepartum stillbirth due to placental dysfunction or unknown etiology. Cases of stillbirth attributed to other specific etiologies were excluded.A total of 49,458 women were included in the cohort. There were 103 (0.21%) cases of stillbirth due to placental dysfunction or unknown etiology. For cases in the early stillbirth cluster (≤ 30 weeks), the detection rate was high and was similar for the three ultrasound-based fetal growth charts of Hadlock, IG21, and WHO (range 83.3-87.0%). In contrast, the detection rate of SGA for cases in the late stillbirth cluster ( 30 weeks) was low, being highest for WHO and Hadlock (36.7% and 34.7%, respectively), and lowest for IG21 (18.4%). Using a threshold of the 25th centile increased the detection rate for stillbirth by approximately 15-20% compared with that achieved by the 10th centile cutoff.At 30 week gestation, the Hadlock or WHO fetal growth charts provided the best balance between detection rate and false positive rate for stillbirth.