Factors associated with higher risk of small‐for‐gestational‐age infants in women treated for gestational diabetes

Previously, management of gestational diabetes (GDM) has focused largely on glycaemic control, with a view to reduce the occurrence of large‐for‐gestational‐age (LGA) infants. However, tight glycaemic control in GDM is associated with a higher incidence of small‐for‐gestational‐age (SGA) infants, which has been linked to higher rates of adverse outcomes.


INTRODUCTION
2][3][4][5][6] However, tight glycaemic control in GDM is associated with a higher incidence of small-for-gestational-age (SGA) infants. 7SGA has been linked to immediate and long-term adverse outcomes, including increased rates of neonatal death, seizures and sepsis, 8 and poor school performance, 9 ischaemic heart disease, 10 hypertension and diabetes 11 in adulthood.
Several studies have shown that ultrasound scan (USS)-guided management of GDM in conjunction with glycaemic control may lead to better outcomes, including lower rates of LGA infants, 12,13 reduced need for insulin therapy 13,14 and the potential for decreased frequency of blood glucose monitoring in low-risk groups. 15Relaxing glycaemic targets for women with GDM based upon low-risk growth parameters on USS for fetal macrosomia may also lead to lower rates of SGA infants. 13st guidelines for fetal USS monitoring in GDM only describe making changes to management based on a fetus measuring as LGA. 16,17A similar set of risk factors may be helpful in assessing women with GDM who are at an increased risk of an SGA infant, which would help to inform guidelines such that providers can modify GDM management accordingly.Therefore, the aim of this study was to characterise risk factors associated with having an SGA infant in women being treated for GDM.

Subjects
This retrospective observational cohort study was based on women diagnosed with GDM who delivered between August 2015 and August 2019 at a quaternary teaching hospital in Queensland, Australia.This study utilised the cohort from Davidson et al. 18 but was expanded by 90 subjects to gain more statistical power.Ethics approval was obtained from the

Inclusion criteria
Primiparous women diagnosed with GDM via the oral glucose tolerance test (OGTT) between 24 and 34 weeks' gestation of a singleton pregnancy who were otherwise medically well were included.GDM diagnosis on OGTT was based on the following criteria from the International Association of Diabetes and Pregnancy Study Groups Consensus Panel: one or more of the following blood glucose levels (BGL) identified on a 2-h 75-g OGTT: fasting >5.1 mmol/L, 1 h >10.0 mmol/L and 2 h >8.5 mmol/L. 17

Exclusion criteria
Women with hospitalisations (eg, cholecystitis in pregnancy, hyperemesis gravidarum) and medical comorbidities (eg, previous bariatric surgery, long-term steroid use) that may affect glucose metabolism were excluded from the study.Women were also excluded if their infant had a major congenital anomaly, if they were incorrectly coded as GDM or if they had not been seen consistently in the hospital's GDM clinic (Fig. 1).
Due to confines of resources, such as the hospital's primary use of paper charts, a narrowed random sample of the remaining women was selected.A clinical chart audit was then undertaken by two independent reviewers.

Study design
A woman was said to have received USS-guided management of her GDM based on the following two scenarios: (i) USS data were provided in the clinical notes, and fetal growth was normal, or (ii) there was documentation of changes in management if growth was abnormal.Women who had non-USS-guided management fit into one of the following three scenarios: (i) no USSs were conducted after 28 weeks' gestation; (ii) there was no mention of USS findings in the clinical notes; or (iii) there was no documentation of changes in management if growth was abnormal.
SGA was defined as a birthweight of <10th percentile based on the 2013 Fenton Preterm Growth Chart. 19Growth that was considered high risk for SGA on USS was characterised as an abdominal circumference (AC) <25th percentile or an estimated fetal weight (EFW) <10th percentile.Given the lack of clinical guidelines, these criteria were chosen based on discussion with expert clinicians and ultrasound parameters used in previous research. 20,21Additionally, these cut-offs are comparable to those used in the current guidelines for identifying those at risk of LGA (AC >75th percentile). 16her information that was collected included maternal and fetal demographics, anthropometric and medical data, pregnancy and delivery information and neonatal outcomes.Initial weight and body mass index (BMI) were calculated at the booking-in visit which primarily occurred before k10.Gestational weight gain (GWG) was calculated based on the booking in weight and then the final weight taken before delivery.GWG was assessed as low, appropriate or high for a woman's pre-pregnancy BMI based upon the Institute of Medicine Guidelines. 22The women received multidisciplinary care with regular obstetric, dietitian, and diabetic educator visits.Depending on the case, patients may also have been seen by a midwife or a specialist endocrinologist/obstetric medicine physician.Patient self-reported BGLs were monitored at least fortnightly by one of these professionals such that management could be adjusted.

Data analysis
Statistical analysis was performed using the IBM SPSS Statistics Subscription, Build 1.0.0.1508 (IBM, Armonk, NY).Demographical data were compared between women who had an SGA and AGA infant.For continuous variables that had normally distributed data, independent t-tests were used to compare means.For continuous variables that had skewed data, independent samples Mann-Whitney U-tests were performed.For categorical variables, χ 2 tests were used to compare groups.
Descriptive statistics were utilised to identify which USS measurement (AC, EFW or both) categorised the fetus as having growth that was high risk for SGA and how management was then changed.They were also performed to calculate the absolute risks of having an SGA infant when each of the following risk factors was present (ie, low pre-pregnancy BMI (<18.5),low fasting BGL (<5.1) or growth that was high risk for SGA at baseline USS).
Review of the literature and discussion with expert clinicians aided in the determination of several potential predictors of women with GDM delivering an SGA infant.These variables included pre-pregnancy BMI, diagnostic OGTT levels, GWG, fetal growth at baseline USS and type of GDM treatment. 7,23,24All factors with P-values <0.1 in the bivariate analysis were included in a logistic regression model to control for confounding variables.

Maternal demographical and medical findings
A total of 308 women aged 18-49 years (mean: 30.50, standard deviation: 4.82) with GDM were included in this analysis.More women delivered SGA infants than LGA infants (n = 25, 8.1%, vs n = 19, 6.2%, respectively), and 264 women delivered AGA infants (85.7%).The results relating to the LGA infants are presented in Table S1 in the Supporting Information.Table 1 presents the demographical and medical data relating to the SGA and AGA groups.There were significantly more Caucasian women in the AGA group compared to the SGA group (53.8%, n = 142, vs 32.0%, n = 8, respectively, P = 0.04), and there was a greater number of women of 'other' ethnicities in the SGA group (14.8%, n = 39, vs 36.0%,n = 9, P = 0.01).The mean maternal height was also significantly greater in the AGA group compared to the SGA group (162.5 vs 159.2, respectively, P = 0.02).There was a higher proportion of women in the SGA group who had a hypertensive disorder when compared to the AGA group (24.0%, n = 6, vs 9.5%, n = 25, respectively, P = 0.04); however, when focusing on only preeclampsia, this difference was not statistically significant (12.0%, n = 3, vs 3.8%, n = 10, respectively, P = 0.09).The women in the SGA group also received their diagnostic OGTT earlier than the women

Metabolic risk factors for SGA infant findings
A comparison of the metabolic factors that may put women with GDM at risk of having an SGA infant is also presented in Table 1.
Women with an SGA infant had a significantly lower median prepregnancy BMI (21.5 vs 24.4,P = 0.01) and median OGTT fasting level (4.30 vs 4.80 mmol/L, P = 0.001).Further, a higher percentage of women in the SGA group had a fetus with growth that was high risk for SGA at their baseline USS (54.2%, n = 13, vs 13.3%, n = 35, P < 0.001).Of note, there were no cases where the fetal growth was measured as low in the baseline scan that then resulted in an LGA infant at delivery.Finally, a significantly larger proportion of women in the SGA group were managed on diet when compared to the AGA group (96.0%, n = 24, vs 75.8%, n = 200, P = 0.02).
The results of the logistic regression are summarised in Table 2.
A lower pre-pregnancy BMI, lower OGTT fasting level and growth that was high risk for SGA at baseline USS were significantly related to increased odds of having an SGA infant even when adjusting for confounding variables of maternal height, hypertensive disorders, ethnicity and GA at OGTT.The absolute risk of having an SGA infant in women with a low pre-pregnancy BMI was 27.3% compared to 7.9% in those who had a normal to high BMI (absolute risk increase (ARI) = 19.4%).The absolute risk of having an SGA infant in those who had a lower fasting OGTT level was 12.5% compared to 2.7% in those who had a higher fasting OGTT level (ARI: 9.8%).Finally, the absolute risk of an SGA infant for those who had fetal growth that was high risk for SGA at their baseline USS was 27.1% compared to 4.6% in those who had normal or high-risk growth for LGA (ARI: 22.5%).

USS-guided management findings
Table 3 presents the percentage of fetuses that would be identified as having growth that was high risk for SGA based on either low AC, low EFW or both, as well as a comparison of the proportion of these women who received USS-guided management.
This showed that more women with normal fetal growth at baseline USS received USS-guided management when compared to women with growth that was high risk for SGA (84.6%, n = 126, vs 37.5%, n = 18, respectively, P < 0.001).Between the women who had normal growth and those who had growth that was high risk for SGA, there was not a significant difference in the prevalence of pre-eclampsia (P = 0.153).
Of the 48 women with fetal growth that was high risk for SGA at their baseline scan, the most common changes made to treatment were a relaxation of the standard blood glucose targets (38.9%, n = 7; eg.standard fasting target of ≤5.0 mmol/L increased to ≤5.5-6.0 mmol/L; standard 2-h postprandial target of ≤6.7 mmol/L increased to ≤7-10 mmol/L; for those managed on a diet, this often meant relaxing their caloric restriction) or a change in pharmacotherapy (22.2%, n = 4) (ie, decreased dose (5.6%, n = 1) or delayed commencement (16.7%, n = 3) of insulin).

DISCUSSION
The aim of this study was to characterise the risk factors associated with SGA infants in women being treated for GDM.The findings indicate the following risk factors: a lower pre-pregnancy BMI, a lower fasting BGL and growth that was high risk for SGA at baseline USS.
Koren et al. 25 similarly found that insufficient GWG in women with GDM was associated with higher rates of SGA infants.
Comparably, Xiao and Zhang demonstrated that low to normal pre-pregnancy BMI and early insufficient GWG were associated with higher rates of SGA infants in women with both preeclampsia and GDM. 26Novel findings in our study included the determination that lower fasting glucose on OGTT and baseline fetal growth on USS may help to predict SGA infants.Our findings together with those of previous research suggest that some women with GDM may exhibit a low-risk clinical profile for having an LGA infant.This clinical profile may include many of the following factors: lower pre-pregnancy BMI, GWG, fasting OGTT level and fetal growth at baseline USS.
Interestingly, despite fetal growth that was high risk for SGA being identified at their baseline USS, most women did not  have any changes in management by the treating This is especially relevant considering the poorer outcomes of SGA infants [8][9][10][11] and the findings of Davidson et al., 18 which demonstrated that neonates with USS-guided management had fewer admissions to the Special Care Nursery/Intensive Care Nursery, fewer prolonged hospital stays and a lower risk of hypoglycaemia after birth.Although many of these women were not treated with pharmacotherapy, they were managed with the same tight glycaemic targets as those with AGA and LGA infants.
It highlights that clinicians may not carefully consider the risks of SGA infants in women with GDM, where the focus has been on LGA infants.
However, when discussing relaxing glucose targets in GDM, it is important to consider that hyperglycaemia may result in placental inflammation, which can also affect fetal growth. 27Thus, a fine balance may exist between relaxing glucose targets sufficient to allow for improved fetal growth while not inducing placental insufficiency.Currently, there is little evidence to support the safety of relaxing glycaemic targets, and thus, it would be a vital area for future investigation.This also highlights the importance of assessing an entire clinical picture, not a sole risk factor, when making management decisions related to glycaemic control.
The findings of this study need to be considered in the context of several limitations and strengths.The data for many secondary outcomes utilised coded data which can be incorrect or missing.Of note, none of the primary objectives used categorical coded data but instead numerical data which are more reliable.
Moreover, there is an inherent error in USS measurements, 28 which may have led to fetuses being misclassified as having abnormal growth.However, in a quaternary hospital setting such as in this study, the error for the raw EFW measurement would not be more than 10%. 28Another factor to consider is that only primiparous women without significant medical comorbidities were included, making the results less generalisable.
Furthermore, pre-eclampsia may be a confounding variable in the study.Although none of the outcomes in the analysis were significantly different when it came to pre-eclampsia rates, there were very few cases of pre-eclampsia overall.As pre-eclampsia is known to affect fetal growth, 29,30 a larger study is required to confirm or refute our findings.Further, if USS results were not documented in the patient notes, participants were allocated into the non-USS-guided management group.Thus, it is possible that some USS-guided decision-making was included in the non-USSguided group, leading to selection bias.
Nonetheless, this rich data set from a quaternary hospital in an urban centre featured women of different ages, races, ethnicities, socio-economic statuses, educational backgrounds and medical histories, making these findings more generalisable to other populations.We also utilised a larger sample size than other papers with similar aims. 18,26Additionally, it is one of the first studies to focus primarily on the risk factors related to SGA infants in women with GDM.
Women who have the risk factors identified in this study may be exhibiting a low-risk profile for adverse diabetes-related outcomes, specifically macrosomia.Thus, it would be beneficial for future research to determine whether this SGA-risk profile would be suitable for less-aggressive GDM management and what that management might entail.

TABLE 1 Maternal
baseline characteristics, gestational diabetes diagnosis and management data

TABLE 2 Odds
of an SGA infant at delivery based on fasting OGTT level and growth that was high risk for SGA at baseline USS † †Adjusted for maternal height, hypertensive disorders, ethnicity and gestational age at OGTT. ‡Abdominal circumference <25th percentile or estimated fetal weight <10th percentile.BMI, body mass index; CI, confidence interval; OGTT, oral glucose tolerance test; SGA, small-for-gestational age; USS, ultrasound scan.

TABLE 3 Identification
and treatment of GDM women based on fetal growth at baseline USS