Prospective risk of stillbirth according to fetal size at term

Introduction

Fetal size is a major determinant of the risk of stillbirth [1, 2], and one of the key considerations for clinicians regarding advice to mothers on the best time to deliver. Royal College of Obstetricians and Gynaecologists guidelines [3] recommend delivery by 37 weeks if a fetus is small for gestational age (SGA, <10th centile), based largely on findings of the DIGITAT [4] trial. However early term delivery has been associated with risk of delayed effects including an increase in infants with special educational needs [5].

Therefore current versions of the NHS England saving babies lives care bundle [6] and the care pathway of the Growth Assessment Protocol (GAP) [7, 8] recommend a distinction between fetuses with estimated fetal weight below the third centile, which are more likely to have fetal growth restriction and should be delivered by 37 weeks, and those between the third and tenth centile, where delivery can be delayed until 39 weeks if Doppler investigations are normal [9]. At the other end of the size spectrum, fetuses that are large for gestational age (LGA, >90th centile) have an increased risk of shoulder dystocia and other birth injuries [10, 11] but guidelines for management are likely to follow completion of the ‘Big Baby Trial’ [12] expected in 2023.

To help clinicians with informed advice to expectant mothers on timing of birth, we wanted to investigate the predictive value of different fetal size bands on prospective risk of stillbirth at each week at term.

Materials and methods

Results

The study cohort consisted of 2,094,702 term deliveries (≥37 weeks) and included 4,670 stillbirths (rate 2.2/1,000). Table 1 displays the characteristics of this population. The incidence of underweight (<18.5 BMI) pregnancies was 2.7% and 23.4% were classified as obese (BMI>30). The largest ethnic group was British European (68.4%) followed by South Asian (8.9%) and East European (6.5%). The SGA rate was 10.0% and LGA rate was 8.2%.

Table 2 displays the prospective stillbirth risk and risk ratios from 37 to 41 completed weeks by centile band. The baseline stillbirth risk (10–90 centile at 37 weeks; per 1,000) was 0.29 and rose to 1.39 at 41+ weeks. In comparison, stillbirth risk in the 3<10 centile band rose from 0.72 to 2.43, and in fetuses <3 centile from 1.62 to 6.16. The table also shows the risk ratios in reference to a 10–90 centile fetus at 37 weeks: for example, a fetus <3rd centile at 37 weeks has a 5.6 fold higher risk of stillbirth, which progressively increases to 21.2 fold at 41+ weeks. Stillbirth risk and risk ratios for LGA >97 centile babies showed a less accelerated increase with gestational age, while in the 97>90 centile band they rose at a similar rate to the 10–90 centile control. These findings are illustrated in Figure 1A and 1B.

Discussion

This is to our knowledge the first study of prospective risk of stillbirth by fetal size as assessed by customised centiles, the standard used in most pregnancies in the UK national health service. We found that SGA and LGA fetuses have an increasing risk of fetal death after 37 weeks which increases further with advancing gestation and is highest for babies below the third centile.

Our findings are consistent with previous studies of large databases. Muglu and colleagues’ systematic review showed a steady increase in prospective stillbirth risk with advancing gestational age [17]. Pilliod et al. [18] stratified their cohort by uncustomised fetal size and also found the highest risk in fetuses that were <3rd centile.

A strength of our study is its large size with over two million term pregnancies, which allowed us to divide the cohort into different gestational age and centile bands. Pregnancies were routinely dated by scan, and size was assessed by customised centiles, which improves the distinction between pathologic growth restriction and the constitutionally small fetuses.

A potential limitation of our study was the use of birthweight rather than fetal weight to determine the size centile of the fetus. However ultrasound scans are less precise than birthweight due to systematic and random error [19]. Furthermore, assessment of fetal weight is not done routinely at term, and selecting only the cases where it had been performed is likely to reduce generalisability of our findings. Detection of SGA or LGA at scan may have influenced the timing of delivery, and it could be speculated that at the LGA end of the distribution, the observed drop in stillbirth rate at 39 weeks in the >97 centile group (Figure 1B) may have been due to elective deliveries to avoid birth complications such as shoulder dystocia. We did not have information on co-morbidities and ultrasound and Doppler investigations in this large cohort, and most growth scans in our population are carried out before 37 weeks. However these are additional factors which need to be taken into consideration when deciding on the optimal time for delivery.

Figure 1:

Prospective risk of stillbirth.

Appropriate for gestational age (AGA, 10–90 centile) is compared with (A) small for gestational age (SGA; <3 and 3<10) and (B) large for gestational age (LGA; >97 and 97>90) centile bands.

Our study provides real world, routinely collected data of customised birthweight centiles as proxy for fetal growth in term pregnancies. The findings support the concept that, while there is a gradual increase in stillbirth risk for all pregnancies at term, small and large babies have a sharper, weekly increase in risk which is steepest when fetal size is <3rd centile.

These findings may assist clinicians in their advice to mothers on timing of delivery. The advice will need to balance the antenatal risk of a potentially compromised baby continuing in an unfavourable intrauterine environment, against increasing evidence that ‘early term’ delivery presents neurodevelopmental risk for the infant [5]. Currently, such developmental studies relate infant risk to gestational age at delivery alone, but it is hoped they will also be able to stratify risk according to subcategories of fetal size representing slow and accelerated growth.