Background Pregnant women are currently offered two ultrasound scans, one at booking (around 12 weeks’ gestation) and one at around 20 weeks’ gestation. No further scans are offered unless there are clinical indications. Ultrasound has an important role in the management of high-risk pregnancies. However, there is no clear evidence that it is effective in screening low risk and unselected women. The majority of complications, such as stillbirth and shoulder dystocia occur in low-risk pregnancies, first because most pregnancies are classified as low-risk and second, possibly due to inadequate screening. An effective ultrasound screening programme in late pregnancy combined with an intervention, like induction of labour, for the screen positives could potentially improve pregnancy outcomes. However, the diagnostic accuracy of many ultrasonic features is unknown in low-risk populations and there is a possibility of iatrogenic harm by intervening when it is not necessary.

Objectives

1. To assess the diagnostic effectiveness of late pregnancy ultrasound in nulliparous women based on the existing research literature.
2. To analyse the prospective cohort study, Pregnancy Outcome Prediction Study, for the above ultrasound findings and combine the results with the meta-analyses.
3. Finally, use the results to provide inputs for health economic analyses of the cost-effectiveness of universal ultrasound screening and assess the need, potential design, and acceptability of a future randomised controlled trial.

Methods The following key ultrasound measurements were identified which might be used in late pregnancy screening: (i) suspected small for gestational age (SGA), (ii) suspected large for gestational age (LGA), (iii) high resistance pattern of umbilical artery Doppler flow velocimetry, (iv) low cerebro-placental ratio (CPR), (v) severe oligohydramnios, (vi) borderline oligohydramnios. I found that there was an on-going Cochrane Diagnostic Test Accuracy review for SGA, hence I focused on the other five measures. The protocol was registered with the PROSPERO register of systematic reviews (CRD42017064093). Medline, EMBASE, Clinical Trials.gov and the Cochrane library were searched from inception. Studies that performed an ultrasound scan ≥24 weeks of gestational age in unselected, low or mixed risk populations were included, excluding studies which only included high risk pregnancies. The risk of bias in each included study was assessed using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS 2) tool. Meta-analysis was performed using the hierarchal summary receiver operating characteristic curve (HSROC) analysis and bivariate logit-normal models. I also performed new analyses on previously unpublished data of the Pregnancy Outcome Prediction (POP) study which was one of the few studies that blinded ultrasound scan results to the clinicians.

Results 41 studies of LGA met our inclusion criteria involving 112,034 patients in total. Ultrasonic suspicion of fetal macrosomia was strongly predictive of the risk of delivering a large baby with the positive LRs (LR+) ranging from 7 to 12. However, it was only weakly predictive of the risk of shoulder dystocia with LR+ around 2. 13 studies of umbilical artery (UA) Doppler that met our inclusion criteria including 67,764 patients in total. UA Doppler had weak/moderate predictive accuracy for detecting SGA and severely SGA (<3rd percentile) infants (LR+ between 2.5 and 3.0). However, it did not predict neonatal morbidity at term. The results were very similar in both the POP study and the meta-analysis (which included the POP study) with the only notable difference being that the association with severe SGA in the POP study was slightly stronger. 16 studies of CPR met the inclusion criteria involving 121,607 patients in total. CPR may be slightly more predictive than UA Doppler in identifying pregnancies at an increased risk of adverse outcome. In the case of SGA, the positive LRs were in the region of 3.5 to 4.0. Moreover, unlike UA Doppler, a low level of CPR was associated with an increased risk of neonatal morbidity. However, the association with morbidity was weaker with positive LRs of <2.0. Furthermore, in both analyses, there was very significant heterogeneity in relation to both SGA and neonatal morbidity. 14 studies of severe oligohydramnios that met our inclusion criteria involving 109,679 patients in total. Diagnosis of severe oligohydramnios was associated with a positive LR for SGA of between 2.5 and 3.0. It was also associated with positive LRs for admission to NICU and emergency caesarean section for fetal distress of between 1.5 and 2.5. However, the study quality was variable and only two studies containing <5% of the patients included in the meta-analysis blinded the results of the scan. 11 studies of borderline oligohydramnios (including the POP study) met our inclusion criteria involving 37,848 patients in total. Borderline oligohydramnios was weakly/moderately predictive of SGA (positive LRs 2.5 to 3.0). This was observed in the meta-analysis of multiple studies of variable quality. There was also a comparable association between borderline oligohydramnios and severe SGA in the only study where the scan result was blinded, the POP study. Finally, by analysing of the POP cohort We identified the 4.6% of women who had a breech presentation, and for more than half of these, it had not previously been clinically suspected. Most of these women were delivered by planned Caesarean section. No woman in the cohort had a vaginal breech delivery or experienced an intrapartum Caesarean for undiagnosed breech. An introduction of a policy of third trimester ultrasound for fetal presentation would prevent about 5000 emergency Caesarean sections and 8 perinatal deaths annually in the UK. The policy would be cost-effective at a cost of £19.80 per scan.

Conclusion There is a strong clinical and health economic case for implementing late pregnancy ultrasound screening to assess fetal presentation. Universal ultrasound screening for macrosomia would increase the detection of LGA infants at birth but is unlikely to increase the detection of shoulder dystocia or associated neonatal morbidity in a clinically significant way. Umbilical artery Doppler, CPR, severe oligohydramnios, and borderline oligohydramnios were all weakly predictive of the risk of delivering an SGA infant but either non-predictive or weakly predictive of the risk of neonatal morbidity. They should not be used alone to screen for neonatal morbidity, however a positive result would justify further fetal monitoring due to the association of all above markers with SGA.