Customised growth charts could improve how we identify infants who are small and large for gestational age

Abstract

Foetal growth restriction screening is a pillar of antenatal care, because it is an essential component of strategies to reduce stillbirth rates.¹ However, identifying growth-restricted infants is complicated by the difficulties of defining this condition. Foetal growth restriction refers to foetuses with insufficient growth compared with their intrinsic potential. Since this cannot be measured in clinical practice, healthcare professionals mainly rely on using a proxy, namely small for gestational age (SGA). During pregnancy, SGA is defined as an estimated foetal weight or an abdominal circumference under the 10th percentile of a growth chart. At birth, it is a birthweight under the 10th percentile. Similarly, large for gestational age (LGA), which is based on the 90th percentile threshold, is used as a proxy to identify foetuses and newborn infants with excess growth. These infants also face increased risk of neonatal mortality and morbidity.

SGA and LGA are used for daily clinical care and research, but they cannot distinguish between constitutional and pathological growth. In the 1990s, Gardosi et al. proposed customised growth charts that took account of the genetic and biological factors that physiologically impact individual growth potential.² These charts were adjusted for foetal sex and the mother's parity, height, pre-pregnancy weight and ethnicity. It has been shown that customisation reclassified about 2% of non-SGA births, defined using non-customised charts, as SGA.³ These births face an increased risk of stillbirth and neonatal death. At the same time, 20%–25% of SGA births, defined using non-customised charts, were reclassified as not SGA by customised charts and did not show a significantly increased mortality rate.³ In contrast, there were no significant differences in the risk of perinatal death when customised and non-customised charts were used at the population level, mainly because the births that were most at risk were defined as SGA in both charts. Some researchers have suggested that the limited added value of customisation may not be worth the complex implementation that is required. The concept holds promise but needs further improvement to increase its performances when screening SGA and LGA foetuses.

A paper published in Acta Pediatrica, by Zeegers et al., describes how the authors constructed Dutch charts customised on sex and maternal height. They evaluated the effect of customisation on the classification of SGA and LGA newborn infants by comparing their new customised charts to local non-customised birthweight charts.⁴ The authors used a sample of 21 094 term singleton pregnancies between 2012 and 2020. All women had spontaneous onset of labour and were cared for in midwifery-led units because they were considered to pose a low risk of complications. A subsample of 11 660 women who did not smoke and had a normal body mass index was used to create the customised charts.

The results showed that the new customised charts had good internal validity and were an appropriate fit for the population of interest. They also confirmed that the non-customised charts did not fit women who were shorter and taller than average. The study found that 19.7% of SGA and 1.4% of LGA newborn infants had mothers in the lowest height category, whereas these percentages were 3.4% and 21.8%, respectively, for mothers in the highest height category. When customised charts were used, these provided proportions of SGA and LGA in accordance with the thresholds used.

The choice of the best chart to identify foetuses and newborn infants with altered growth has been very actively debated worldwide for the past 10 years. It has been shown that the charts that are used can have a significant impact on the prevalence of SGA and LGA and that important differences in foetal growth can be observed between countries.⁵ The study by Zeegers et al. provides a new Dutch-customised chart for use in clinical practice that fits the local population.

The method that Zeegers et al. used to develop their chart was based on Gardosi et al.'s approach. However, the authors selected different characteristics for customisation, and parity, maternal pre-pregnancy weight and ethnicity were not included. There have been extensive debates on whether these factors really have a physiological impact on foetal growth or if pathological mechanisms could also be involved. For example, adjusting for parity has been associated with a decrease in the proportion of primiparous women who deliver SGA infants, but it does not improve the screening of those at high risk of complications.⁶ Similar concerns have been raised about maternal pre-pregnancy weight and the appropriateness of adjusting growth charts on pathological conditions such as obesity. Given the strong association between maternal pre-pregnancy weight and estimated foetal weight, this may lead to under-identifying LGA foetuses. In contrast, other studies have shown that adjusting or stratifying growth charts by foetal sex reduced the bias of under-identifying male foetuses and over-identifying female foetuses.⁷ Zeegers et al. showed similar results for maternal height when they compared customised and non-customised charts, with substantial reclassifications of infants with short and tall mothers. These characteristics supposedly have a strictly physiological impact on foetal weight and have the advantage of being widely available in daily practice.

Validation of these new Dutch charts is now essential. First, concerns have been raised about the models used for customised charts and the unverified hypotheses that underpin the construction method.⁸ Gardosi's method consists of quantifying the impact of the selected characteristics on birthweight at 40 weeks of gestation and then modelling individualised norms at each gestational age. This assumes that there is a proportional effect of foetal and maternal characteristics throughout pregnancy. However, their effect on weight may be stronger at birth than in the third or the second trimesters of pregnancy, which could limit the performance of these charts.⁸ That is why it is important to validate customised charts using estimated foetal weight data as well as birthweight data, as this directly evaluates their utility for identifying foetuses at risk during pregnancy.

Second, the growth charts should be evaluated using perinatal health outcomes. One study adapted Gardosi et al.'s method to create charts adjusted at the national level to allow international comparisons.⁹ Using national charts, adapted to the populations' characteristics, rather than a unique international chart, was better at identifying newborn infants at risk of stillbirth and neonatal death. Evaluating the charts' ability to identify births at increased risk of complications is of major importance to guide the choice of which chart to use in clinical practice and to underpin recommendations to adopt them.

Customised growth charts are recommended by the UK Royal College of Obstetricians and Gynaecologists, the New Zealand Maternal Fetal Medicine Network and the Institute of Obstetricians and Gynaecologists at the Royal College of Physicians of Ireland.¹⁰ These charts represent a real opportunity for more individualised care and better differentiation between constitutional and pathological growth. However, improvements are still necessary to deal with the methodological limitations raised about the hypotheses underlying the modelling of the customised charts, as well as the practical limits of this approach which relies on routinely available data on foetal and maternal characteristics. The study by Zeegers et al. contributes to providing new knowledge for the path forward.

The authors declare no conflicts of interest.