Small for gestational age at preterm birth identifies adverse neonatal outcomes more reliably than antenatal suspicion of fetal growth restriction

Abstract Background Fetal growth restriction (FGR) is an important reason for premature delivery and a leading cause of perinatal morbidity and mortality. We aimed to evaluate whether classification as small for gestational age (SGA; <10th centile) at birth or antenatal suspicion of FGR was more strongly associated with neonatal morbidity and mortality in preterm infants. Methods A retrospective audit of infants born between 24 + 0 and 32 + 6 weeks of gestation from 2012–2019 and admitted to the Neonatal Unit at Mercy Hospital for Women (MHW). Infants were categorized according to whether FGR was listed as an antenatal complication in the medical records and whether they were SGA (<10th centile on Fenton chart) or appropriate for gestational age (AGA) at birth, and comparisons for neonatal outcomes were made. Results 371/2126 preterm infants (17.5%) had antenatal suspicion of FGR, and 166 (7.8%) were SGA at birth. No differences in any neonatal outcomes were found between infants with or without suspected FGR, except decreased intraventricular hemorrhage (IVH) in the FGR group. SGA classification was associated with increased rates of all morbidities other than IVH, including bronchopulmonary dysplasia, retinopathy of prematurity, and necrotizing enterocolitis, compared with the AGA group. Death was significantly higher in the SGA group (7.2%) compared with the AGA group (3.5%). Conclusion SGA by Fenton chart more reliably identified neonates at risk of adverse neonatal outcomes than antenatal suspicion of FGR, suggesting it is a reasonable clinical proxy. This most likely reflects the much lower tenth centile weight cutoffs on the Fenton charts compared to in-utero charts used antenatally to diagnose FGR based on ultrasound estimated fetal weight. SGA classification by Fenton approximately equates to <3rd centile on in-utero charts at our institution, therefore identifying the most severe FGR cases.


Introduction
Fetal growth restriction (FGR) is a common pregnancy complication and is a major leading cause of stillbirth [1], neonatal morbidity [2], adverse neurodevelopmental outcomes, and poor long-term adult health [3].When infants are born prematurely with FGR, these risks are compounded [4].In Australia, 9.4% of singleton infants were born small for gestational age (SGA; <10 th centile) as determined by national percentiles [5] compared with low-and middle-income countries with rates as high as 30% of pregnancies [6].The epidemiology of infants born SGA changes based on the population references, which is why using a global versus local growth chart is debated.In Victoria, Australia, public reporting of hospital performance in detecting severe FGR (BW < 3 rd centile) has been associated with reducing stillbirth rates statewide.However, the corollary to these efforts to reduce stillbirth rates has been an increased rate of normally grown babies being delivered early due to suspicion of FGR, and a corresponding increase in neonatal intensive care unit (NICU) admissions [7].
The terms FGR and SGA are often used synonymously in the medical literature, but there exists an important difference between the two.FGR generally refers to a fetus that has failed to reach its biological growth potential -most often due to placental dysfunction.The definition of SGA is based on the crosssectional evaluation (either prenatal or postnatal) of size.This term has been used for neonates whose birth weight is less than the 10 th percentile for that particular gestational age or two standard deviations below the population norms on the growth charts [8].It includes a proportion of babies (18-22%) who are constitutionally small but healthy [9,10].In summary, the definition of SGA considers only the birth weight without considering the in utero growth and physical characteristics at birth [8].
FGR is challenging to define, and there is no gold standard for the diagnosis of FGR.The definition of FGR should be based on a combination of measures of fetal size percentile and Doppler abnormalities [11].A consensus-based report for FGR diagnosis (Delphi criteria), including biometric and functional parameters, was published in 2016 [12].
FGR is a well-described risk factor for stillbirth.An extensive and growing literature has focussed on the importance of early detection and surveillance of FGR, intending to reduce the associated risk of stillbirth [13].Morbidity following FGR has been described during the neonatal period, including increased cardiovascular, respiratory, neurological, and endocrine morbidity risks [14,15].FGR has also been associated with poor neurodevelopmental outcomes, which are influenced by the timing of the onset of FGR, the severity of FGR, and gestational age at birth [16].Further research has explored the impacts in the same domains into adulthood as part of the work investigating the developmental origins of health and disease [17,18].
Because, in our current practice, we frequently use FGR and SGA as a proxy, we aimed to evaluate whether classification as SGA at birth or antenatal suspicion of FGR was more strongly associated with neonatal morbidity and mortality in preterm infants.

Methods
We retrospectively reviewed data of infants born between 24 þ 0 and 32 þ 6 weeks of gestation admitted to the Neonatal Unit at Mercy Hospital for Women (MHW) in Victoria, Australia, between 1 January 2012 and 31 December 2019.We included infants born at less than 33 weeks' gestation because this is the gestation at which the 10 th centile on the Fenton Chart equates to 1500 g.Using a threshold of 33 weeks gestation would allow for the inclusion of some babies with suspected FGR who were subsequently born above the 10 th centile, i.e. above 1500 g, to reduce bias.
Each infant was categorized according to whether suspected FGR (sFGR) was listed or not as an antenatal complication in the medical records.The in-utero growth chart in our institution is based on the fetal weight equation proposed by Hadlock et al. [19] and adapts the customization proposed by Gardosi et al. [20] (supplemental image).While we defined suspected FGR according to whether this was listed as an obstetric complication in our medical records, the obstetric unit at our hospital often suspects fetal growth restriction if an ultrasound reports an estimated fetal weight <10 th centile according to this chart.
We defined SGA as birth weight less than the 10 th percentile on the Fenton Chart and appropriate for gestational age (AGA) as birth weight between the 10 th and 90 th percentiles on Fenton.We applied the research bulk calculator from The University of Calgary, which uses completed weeks for weight, head circumference, and length [21].Fenton Charts are used worldwide.They are designed for growth monitoring of preterm infants (< 37 weeks at birth).In 2013, a systematic review and metanalyses revised the 2003 Fenton Chart to harmonise the preterm growth chart with the new World Health Organization (WHO) Growth Standard [22].
Data were collected retrospectively from the MHW neonatal database.We then compared associated neonatal morbidities: bronchopulmonary dysplasia (BPD), which was defined as oxygen requirement at 36 weeks of corrected gestational age; any grade of intraventricular hemorrhage (IVH); periventricular leukomalacia (PVL); any degree of retinopathy of prematurity (ROP); proven necrotising enterocolitis (NEC), defined as a diagnosis at the surgery or post mortem, or radiological diagnosis (e.g.pneumatosis intestinalis, portal vein gas, persistent dilated loop on serial X-rays) with a clinical history or clinical diagnosis (a clinical history plus abdominal wall cellulitis and palpable abdominal mass); and proven bacterial, viral, or fungal infection.Finally, the outcome of death refers to neonatal deaths, i.e. a live-born infant dying before discharge from the hospital.Stillbirths were not included in our analysis.
The maternal database characteristics included maternal age, ethnicity (based on Australian & New Zealand Neonatal Network data), spontaneous preterm labor, evidence of preeclampsia, fetal distress, antenatal diagnosis of congenital malformations, and multiple births.Infant database characteristics included gestational age (GA), birth weight, male sex, antenatal steroids, cesarean section, Apgar at 5 min, and whether endotracheal intubation occurred after birth.Prenatal steroid exposure was reported if at least one dose of steroids was administrated during pregnancy.
Descriptive data included mean ± standard deviation (SD) and median (interquartile range (IQR)) for parametric and non-parametric data, respectively.Categorical data were compared between groups of sFGR and non-sFGR, and SGA and AGA using the Chisquare test and Fisher's exact test as appropriate.Continuous variables were compared between groups using t-tests or Wilcoxon Rank-Sum tests as appropriate.A p-value of <0.05 was considered statistically significant.Stepwise multivariable logistic regression analysis was performed to examine the influence of maternal and infant confounders about neonatal outcomes.Maternal and neonatal variables associated with differences between sFGR/non-sFGR and SGA/ AGA, respectively, on univariate analyses (Tables 1 and  2) were then used to derive maternal and neonatal models for each neonatal outcome identified as being significantly different for sFGR/non-sFGR and SGA/AGA (Table 3).Final models were generated for each neonatal outcome combining significant associated maternal and infant variables.

Results
Over eight years, 2132 preterm infants between 24 þ 0 and 32 þ 6 weeks were admitted to the Neonatal Unit.Six infants were excluded due to insufficient data regarding suspicion or not of FGR during pregnancy.371 (17.5%) had a documented antenatal suspicion of FGR, and 166 (7.8%) were classified as SGA on the Fenton Chart.Around 1/3 (33.7%) of the sFGR group were also born SGA.In the group of infants who were born SGA, 3/4 (75.3%) of them had a diagnosis of sFGR during pregnancy.
Maternal characteristics are summarized in Table 1.There were more mothers of Asian background in the sFGR group as well as in the SGA group.There were significantly more women with preeclampsia and fetal distress but fewer with spontaneous preterm labor in the sFGR and SGA groups.Multiple births were also more common in the sFGR group but less common in the SGA group.
Infant characteristics are described in Table 2; all groups were born at similar gestation (29 weeks).As expected, birth weight was significantly higher in the Non-SFGR group and AGA group.The sFGR and the SGA infants were most likely born by cesarean section and had been exposed to antenatal steroids.There were slightly more males in the Non-sFGR group and the SGA group.
With respect to neonatal outcomes there was a decrease in IVH in the sFGR group compared to those without sFGR (Table 2).There were no other differences between sFGR and Non-sFGR groups for any other neonatal outcome.In contrast, SGA classification was associated with increased rates of all neonatal morbidities (BPD, ROP, NEC) other than IVH, the rates of which increased in the AGA group compared with SGA infants.However, rates of PVL were similar between each group.Multivariable logistic regression analysis showed that even when accounting for multiple maternal and infant confounders the relationships between SGA and the outcomes of BPD, ROP, NEC, and death remained significant (Table 4).However, the reduced rates of IVH in sFGR and SGA groups were no longer apparent once accounting for confounders.
Eighty-six premature infants under 33 weeks gestation died before being discharged from the hospital.Twelve preterm infants died in the sFGR group, of whom eight were SGA.The percentage of deaths in the SGA group (7.2%) was double that of the AGA group (3.5%; p-value ¼ 0.005).However, no significant difference was seen in death between the sFGR and non-sFGR groups.

Discussion
This study demonstrated that categorizing infants as SGA-identified neonates at risk of adverse neonatal outcomes was more reliable than antenatal suspicion of FGR, even when we accounted for maternal and infant confounders.This most likely reflects the much lower tenth-centile weight cutoffs on the Fenton charts compared to in-utero charts.SGA classification by Fenton approximately equates to <3 rd centile on in-utero charts at our institution, therefore identifying the most severe FGR cases.
The main strength of our study is the relatively large sample size of complete datasets extracted from a well-established database.There are several limitations to this study.This retrospective study has confounding variables we cannot control.For example, we lack knowledge of obstetricians' and midwives' criteria to report suspected FGR during pregnancy.We also have not compared the outcomes based on the underlying cause of FGR.Finally, we have not excluded congenital abnormalities in our cohort study.
Despite these limitations, this study offers some novel findings compared to others reported previously.A large cohort study compared three measures of FGR with a control group of well-grown infants: 1) antenatal diagnosis of FGR; 2) birth weight less than 10 th percentile (SGA); and 3) infants with either or both of these diagnoses.In their analyses, within each stratified week of GA from 25 to 32 weeks, neonates with all three categories of growth restriction were more likely to die, have NEC, require respiratory support at 28 days of age, and develop ROP.Severe IVH in neonates with growth restriction on any measure was similar to that of neonates without any signs of growth restriction [23].In contrast to the previous study, we reported a significantly decreased rate of IVH in the sFGR and SGA groups.However, existing literature has inconsistent evidence on whether placental insufficiency and FGR are directly linked to IVH [24].Some studies have shown FGR to be associated with an increased prevalence of IVH and white matter injury in preterm neonates [25][26][27].Conversely, other studies have reported a reduced rate of IVH in FGR infants or have shown no change in the incidence of neonatal cranial ultrasound abnormalities in FGR infants compared with appropriately grown preterm infants [23,28].A recent meta-analysis suggested that the risk of IVH may indeed be increased in FGR/SGA moderate to late preterm (32-36 þ 6 weeks gestation) and term infants compared to infants unaffected by FGR; the certainty of the evidence was very low due to non-randomised studies, methodological limitations, and betweenstudies Heterogeneity [29].
In our study, we have shown that preterm infants who were classified as SGA had an increased rate of BPD (44%) compared with the AGA group (26.9%).When we perform logistic regression analysis, the odds of BPD are 2:1 in SGA vs AGA (p-value <0.001), even when accounting for multiple births.Sassi et al. demonstrated similar results, showing that preterm infants with FGR-SGA (BW <10th centile and diagnosed with FGR in utero based on antenatal ultrasound scans) are 45% more likely to have BPD or die from respiratory complications after birth than well-grown neonates [30].
SGA is associated with an increased likelihood of any stage of ROP, severe ROP, and treated ROP in preterm infants [31].In the ELGAN study, which included infants under 28 weeks of gestation, both low gestation and severe growth restriction (infants whose birthweight Z-score was �2) were associated with an increased risk of pre-threshold ROP.The study's authors inferred that processes associated with true FGR influence the risk of severe ROP, providing additional support for the claim that ROP has antenatal origins [32].Our study showed similar results with a significantly increased rate of ROP in the SGA group.
True FGR infants demonstrate elevated intolerance rates to feeds, feeding difficulties, and NEC [3].Abnormal placental function, as suggested by abnormal Doppler indices, is known to cause a redistribution of circulation in the fetus to preserve blood supply to the brain at the expense of visceral organs, which has been implicated in the higher incidence of NEC.However, NEC has a multifactorial aetiology predominantly affecting preterm neonates [33].In our study, the odds ratio for NEC, accounting for multiple births, for sFGR versus non-sFGR infants was non-significant.On the other hand, for the SGA versus the AGA group, the odds ratio was significantly increased to 2.75 (p-value <0.001).
SGA preterm infants are at increased risk of postnatal infection compared to their age-matched AGA controls, particularly nosocomial infection, irrespective of the responsible pathogen.One possible mechanism is the delayed development of the immune system, which has been described in association with FGR.Indeed, SGA infants have a disproportionately small thymus and low leukocyte, lymphocyte, and macrophage counts compared with their AGA counterparts [34].In our study, a trend toward an increased rate of proven infection was not supported statistically.
SGA infants have an increased risk of mortality, NEC, late-onset sepsis, severe ROP and BPD, but these risks differ across gestation ranges [35].In a population-based study of nearly 1.7 million births, neonatal mortality was shown to be 100 times higher among preterm infants with severe SGA (BW <5 th centile) than those born at term with a BW greater than the 5 th centile, representing an excess of 59 neonatal deaths per 1000 live births.This risk pattern was especially prominent before 29 weeks gestation and was evident across various demographics [36].Our study showed higher neonatal mortality, more than double in the SGA group (7.2%) compared with (3.5%) in the AGA group.
It is unclear which antenatal growth charts perform best in identifying fetuses at increased risk of stillbirth and other adverse perinatal outcomes.Researchers previously showed that intrauterine charts classify a significantly greater proportion of the preterm population as SGA than birth weight charts [37].There is a considerable discrepancy between the intrauterine growth and birth weight charts.Birth weight distributions have lower means, medians, and 10th centiles than estimated fetal weight distributions, especially among preterm newborns -reflecting that among those with pathology leading to preterm births, there is a higher rate of concurrent pathologies leading to FGR.Some newborns whose birth weight is considered appropriate by neonatal standards are growthrestricted by fetal standards.This phenomenon appears to reflect the preferential preterm birth of growth-restricted infants [38].Future research is required in this area to balance the benefit of increasing SGA detection to decrease stillbirth with the risks associated with iatrogenic preterm delivery.

Conclusion
SGA by Fenton chart more reliably identified neonates at risk of adverse neonatal outcomes than antenatal suspicion of FGR, suggesting it is a reasonable clinical proxy for neonatal risk.This most likely reflects the much lower tenth-centile weight cutoffs on the Fenton charts than in utero charts.SGA classification by Fenton approximately equates to <3 rd centile on in-utero charts at our institution, therefore identifying the most severe FGR cases.

Table 4 .
Multivariable logistic regression analysis of neonatal outcomes.