Late-onset FGR is easy to miss clinically because of its insidious onset and mild symptoms. This study showed that a combined model, including maternal characteristics and second-trimester ultrasonographic parameters, was able to predict 52.6% of cases of late-onset FGR. On the other hand, we found no associations among assisted conception, previous stillbirth, an adverse obstetrical history, the hemoglobin level and late-onset FGR, which was inconsistent with the findings of other studies [10, 11, 21].
The detection rate for late-onset FGR was 52.6%, which is slightly lower than the DR of 59.6% reported by A. Sotiriadis et al[10] and the DR of 64% reported by J. Miranda et al[13] when using maternal characteristics and third-trimester EFW percentiles. F. Crovetto et al[14] used a combination of maternal risk factors, the first-trimester MAP, the mean UtA-PI and the soluble Fms-like tyrosine kinase-1 (sFlt-1)/placental growth factor (PLGF) ratio and detected 66% of cases of late-onset FGR, at a 10% FPR. These combined models were capable of detecting late-onset FGR more frequently or earlier. However, the application of UtA-PI, PLGF and sFlt-1 may be clinically difficult since these tests are not a component of routine prenatal care. The combined use of maternal factors and second-trimester markers in our study could achieve similar outcomes, and the data were from the default test for chromosomal abnormalities and fetal growth monitoring, making the model simpler, cheaper and easier to apply in local hospitals.
Additionally, it is noteworthy that the addition of the EFW Z-score increased the DR from 30.1–52.6%, highlighting the importance of the EFW in the prediction of late-onset FGR. Similarly, a recent study that included 30,849 singleton pregnancies found that a screening approach that combined the maternal characteristics and history with the EFW Z-score increased the DR from 30–80% for SGA neonates that were delivered within < 5 weeks of assessment, with a FPR of 10%[22]. The third-trimester customized estimated fetal weight percentiles could moderately predict FGR[23]. The EFW is objectively measured and not affected by ethnicity, so it is necessary to define its role in the clinical management of the prediction of late-onset FGR in pregnancy. However, some women are not routinely scanned in late pregnancy but are selected for third-trimester ultrasonography based on prepregnancy factors and serial measurements of the symphyseal-fundal height [24, 25]. This approach identifies only a few SGA babies, whereas a large proportion of pregnant women who are at increased risk for delivering babies with late-onset FGR cannot be detected. Thus, it is important to routinely monitor fetal growth by ultrasonic examination during pregnancy.
The strength of the study is the large number of pregnant women receiving routine care in a well-defined gestational-age range that is widely used for screening chromosomal defects and for assessments of fetal anatomy and growth. The screening markers were default tests from routine care, which makes the model very convenient and cheap. In China and other countries with inadequate medical resources, this model can significantly save time and reduce economic costs.
However, we acknowledge that there were some limitations in our study. First, given the nature of the retrospective design, data pertaining to many meaningful parameters, such as previous history of FGR, UtA-PI, PLGF and sFlt-1, which have high contributions to the prediction of FGR[13, 14, 26], were unavailable in the present study; otherwise, the detection rate would have been higher. Thus, we attempt to add the test of UtA-PI in pregnant women from the first trimester throughout the third trimester to further improve the detection rate of late FGR. In addition, nearly all participants reported no tobacco use or alcohol use, preventing us from controlling for differences in smoking or alcohol use.