This was an analytical retrospective cohort study conducted in the Neonatology Department of Chettinad Hospital and Research Institute, Tamil Nadu, India. The study was approved by the Institutional Human Ethics committee (No: 296/IHEC/JAN 2021). Case records of LGA infants {birth weight more than 90^th percentile as per World Health Organisation (WHO) growth standards [14]} born between December 2018 to May 2020 were retrospectively reviewed by the investigators from December 2020 to January 2021.

Inclusion and Exclusion criteria: The LGA infants <35 weeks of gestation and case records with incomplete data were excluded. LGA infants were further classified as large infants born to gestational and LIDM and LnIDM. GDM was diagnosed if blood sugar was ≥140 mg/dL, 2 hours after a 75 g oral glucose challenge test, typically done between 24-28 week of gestation based on the WHO guidelines [15]. Infants born to mothers with pregestational or early GDM (Glucose intolerance diagnosed before pregnancy or in the first trimester or early second trimester) were also included in this group [16]. The LGA infants born to mothers without a diagnosis of GDM during pregnancy were chosen as controls, irrespective of the cause of LGA.

Clinicodemographic profile of the two groups at birth was recorded including socio-economic status according to Modified Kuppusamy scale [17], mode of conception, period of gestation, any antenatal comorbidities, mode of delivery, gender, birth weight, resuscitation details and any other significant antenatal history was also recorded.

As per the hospital policy, all sick infants and macrosomic infants >4000 g were admitted to the NICU. Other non sick LGA infants were roomed in with mother and initiated on breast feeding at birth as per WHO guidelines [18] and baby friendly hospital initiative policy [19]. All LGA infants underwent the following investigations: capillary blood glucose monitoring using glucometer every 6 hours for 3 days, serum ionised calcium in neonates with symptoms of jitteriness, seizures, lethargy, stridor or cardiac dysfunction, haemoglobin and haematocrit at 6 hours of life, serum bilirubin at 72 hours of life, or earlier, if the neonate was found icteric during routine clinical examination or if discharge of neonate is being planned. Initiation and termination of phototherapy was done as per the American Academy of Paediatrics guidelines [20]. Other investigations were done based on the clinical condition of the infant, as deemed necessary by the treating physician.

The National Neonatology Forum of India definitions used to draft the National Neonatal Perinatal Database 2002-03, were utilised in the study [21]. The data from case records were then abstracted and the short-term morbidities of all LGA infants during hospital stay till discharge were recorded. These included need for respiratory support; respiratory morbidities like transient tachypnoea of newborn, respiratory distress syndrome, meconium aspiration syndrome; requirement for NICU admission; perinatal depression, hypoxic ischemic encephalopathy, seizures; significant hyperbilirubinemia requiring phototherapy, peak bilirubin, duration of phototherapy and need for exchange transfusion; polycythemia (haematocrit ≥65%) and thrombocytopenia (platelet count <1,50,000/c.mm); echocardiography confirmed PPHN; cardiac morbidities and malformations; hypoglycaemia (capillary blood glucose ≤45 mg/dL after 4 hours of life) and hyperinsulinemia (any detectable insulin in the presence of critical glucose value ≤45 mg/dL); hypocalcaemia (serum total calcium <7 mg/dL); birth injuries like erb’s palsy, significant cephalhaematoma/subgaleal haemorrhage, fractures; acute kidney injury; and sepsis. Any other significant short-term morbidities were also recorded. All recorded data was stratified for GDM status in mother.

The primary outcome analysed was the need for respiratory support in LGA infants, stratified for GDM. Other outcomes studied were the incidence, demographic profile and other clinical morbidities. Previous data had shown that the rate of respiratory distress and requirement of respiratory support varied from 5-8% in LGA infants [9,22]. To determine an increase in respiratory distress and requirement for respiratory support from 5% in the LnIDM group to 15% in the LIDM group, with 80% power and an alpha error of 0.05, 120 subjects were required in each group (STATA IC, ver. 13).

Baseline clinicodemographical profile and outcome variables were presented as frequency and proportion for categorical variables. Distribution of numerical variables was assessed, using Shapiro wilk test, as skewed and hence, represented as median {Interquartile Range (IQR)}. The categorical outcomes between IGDM or non-IGDM groups were compared using Chi-square test/Fisher’s exact test, while numerical variables were compared using Mann-Whitney U test. Odds ratio and their 95% CI was obtained as appropriate. Multivariate logistic regression, controlling for potential confounders, was done to derive the adjusted odds ratio. The p-value <0.05 was taken as significant. All analysis was performed using SPSS software version 22.0. (IBM Corp. Released 2013. IBM SPSS Statistics for Windows, Version 22.0. Armonk, NY: IBM Corp, California, USA).

During the duration of retrospective review of case records, between December 2018 to May 2020, 2653 infants were delivered in institution, out of which 268 were LGA infants ≥35 weeks (9.72%). Ten case records had incomplete data and were excluded. Hence, 258 infants were included in the final analysis.

The LIDM were 126 (48.8%) and LnIDM were 132 (51.2%). Among mothers of LIDM, 69 (54.76%) mothers had gestational diabetes controlled on diet, 15 (11.9%) were on oral hypoglycaemic agents, 30 (23.8%) were on insulin and overt pregestational diabetes were seen in 12 (9.53%). The clinico-demographic profile of both the groups of LGA infants at birth is shown in [Table/Fig-1].

The LIDM’s weighed higher (p=0.001) and were more macrosomic (p=0.008) compared to LnIDM’s [Table/Fig-1]. The clinical outcomes of the two study groups and their comparison are shown in [Table/Fig-2].

On univariate analysis, the primary outcome of need for respiratory support was not significantly different between the two groups (aOR 1.62; 95% CI 0.92 -2.83; p=0.08), but NICU admission rates {100 (79.4) vs 89 (67.4)}, neonatal hyperbilirubinemia {79 (62.7) vs 66 (50)}, thrombocytopenia {9 (7.14) vs 2 (1.5)} and PPHN {18 (14.3) vs 4 (3.03)} n(%) were significantly higher in the LIDM’s compared to the LnIDM’s. On multivariate logistic regression, controlling for covariates like gestational age, birthweight, gestational hypertension and polycythaemia; NICU admission rates (aOR 2.15; 95% CI 1.17-4; p=0.01), neonatal hyperbilirubinemia (aOR 1.70; 95% CI 1.01-2.84; p=0.04) and PPHN (aOR 4.43; 95% CI: 1.41 -13.82; p=0.004) were still significantly higher in the LIDM’s. None of the other morbidities were different between the study groups.

The LGA infants constitute a significant proportion of live births but this population has not been well studied, at least in LMIC’s like India, where GDM is widely prevalent and is a major cause of LGA births [2]. In present study, the LGA incidence was 9.72% and GDM contributed to 48.8% of LGA births. LIDM’s required more neonatal admission (aOR 2.15; 95% CI 1.17-4; p=0.01*), and had significantly higher neonatal hyperbilirubinemia (aOR 1.70; 95% CI 1.01-2.84; p=0.04*) and PPHN (aOR4.43; 95% CI: 1.41-13.82, p=0.004*), compared to LnIDM’s, after adjusting for confounding factors.

The LGA incidence of 9.72% reported in this study was close to another single centre study from South India where Jeyaseelan L et al., had reported an LGA incidence of 9.4% in their analysis of 35,718 deliveries over a 15-year period [2]. But Malik M et al., in a community based study of a rural population from North India, had reported a LGA prevalence of 1.3% [23]. The difference in rates could be explained by the differing patient characteristics, demographic factors and heterogeneity in screening criteria used. A prevalence ranging from 2.3% to 19.1% has been reported from other LMIC’s [13,24-27].

In our study, GDM contributed to 48.8% of all LGA births, which was higher than previous studies (10.3%-18.7%) [9,22,28,29]. The GDM is a known risk factor for LGA births and GDM has a high prevalence among pregnant women in India [13,30-33]. Hence, it is not surprising that a higher proportion of LGA births are related to GDM. Nevertheless, in current study, the pre-pregnancy weight and weight gain during pregnancy were not documented in all case records and hence, the role of maternal obesity and excessive weight gain during pregnancy as an additional contributor to LGA births in GDM mothers cannot be ruled out.

In present study, LIDM’s had a significantly higher median birth weight (3.92 kg vs. 3.89 kg, p=0.001**) and were more likely to be macrosomic (35.7% vs. 20.5%, p=0.008*), compared to LnIDM’s. Other similar studies had also reported a higher birth weight in the GDM group compared to the non-GDM [9,28,34]. But, unlike these studies, which reported a higher rate of caesarean section (78-87.7%) in the LIDM’s, compared to LnIDM’s (38-68.5%), in this study, the caesarean section rates were similar in both the groups (73.8% vs. 70.5%, p=0.28). This finding was not surprising, as the need for caesarean section is determined by various obstetric and patient driven factors, and not necessarily by standard indications [35].

In this study, LIDM’s had a higher need for NICU admission which was also shown in previous studies [9,22,29]. As most clinical morbidities were higher in LIDM’s compared to the LnIDM’s, it is only expected that the NICU admission rates in this group is higher. Also, LIDM infants had a significantly higher risk of neonatal hyperbilirubinemia in this study, consistent with previous reports [22,34,36]. Both LGA births and GDM are associated with greater risk of neonatal hyperbilirubinemia [37] and as GDM contributed significantly to LGA births in this study, the risk of neonatal hyperbilirubinemia was higher in the LIDM group.

The PPHN is an important cause of hypoxemic respiratory failure and mortality in term infants. In a California birth cohort, Steurer MA et al., found a significant higher prevalence of PPHN in LGA infants (aOR 1.8, 95% CI 1.6-2.0) and infants with maternal diabetes (14.2 vs 8.4%) [38]. Also, a recent meta-analysis showed a significant association between maternal DM and pulmonary hypertension (risk ratio-1.37; 95% C.I. 1.23-1.51) [39]. In our study, the LIDM’s had 4.43 times higher odds of developing PPHN, compared to LnIDM’s, after adjusting for confounding co-variates (p=0.004). Other studies, similar to ours, have not reported on PPHN [9,22,28,29,34]. As both LGA and GDM are independent risk factor for PPHN [38,39], GDM induced LGA is a double whammy and this could explain the higher odds of PPHN obtained in the study. In this study, a higher but non significant proportion of infants required respiratory support (aOR1.62; 95% CI 0.92-2.83; p=0.08), whereas respiratory morbidities like transient tachypnoea of newborn and meconium aspiration syndrome were not different between groups. Higher need for respiratory support was also seen in other similar studies [9,29].

In present study, other neonatal morbidities which were nearing statistical significance were polycythaemia (aOR3.73; 95% CI 0.98-14.22; p 0.053) and thrombocytopenia (aOR4.2; 95% CI 0.86-20.3; p 0.08). The risk of polycythaemia is high in GDM due to relative foetal hypoxia induced erythropoiesis, but whether this risk is increased in GDM induced LGA infants compared to non-GDM LGA infants is unknown [40]. Cordero L et al., and Onal EE et al., observed a higher risk of polycythaemia in LGA neonates of GDM mothers [9,28], but the same was not observed in other similar studies [22,34]. Polycythaemia increases the risk of neonatal thrombocytopenia and hyperbilirubinemia [40,41] and both have been reported in a higher proportion of LIDM’s in our study. Further large studies with a bigger sample size, are required to determine the association between polycythaemia and LGA births, and this could help in re-evaluating the policy of routine haematocrit estimation in all LGA irrespective of cause.

Previous studies have reported that neonatal outcomes like hypoglycaemia and birth injuries were more common in LIDM’s compared to LnIDM’s [9,34,42], but this was not seen in this study. A higher proportion of LIDM’s developed hypoglycaemia (aOR2.26; 95% CI 0.82-6.2; p=0.11), but statistical significance could not be obtained due to the small sample size. The lower rate of birth injuries in current study, could be explained by the lower proportion of vaginal births and macrosomia [43].

The retrospective study design and small sample size were potential limitations of the study. Present study was a single centre experience, which could limit its generalisability and external validity. Also, we could not access more antenatal data on known risk factors of LGA births like pre-pregnancy weight and weight gain during pregnancy. Thus, the exact burden of non-GDM induced LGA could not be ascertained. Also, certain significant outcomes reported in our study, like neonatal hyperbilirubinemia, had wide CI, and the long-term morbidities of LGA infants were not analysed. Lastly, as with any observational study, confounding bias is a problem, as all confounding covariates could not have been accounted for.

This study which analysed and compared LIDM {126 (48.8%)} and LnIDM {132 (51.2%)} clearly showed that GDM contributes significantly to LGA births, and is associated with further increase in short term morbidities, compared to non-GDM causes. These infants are more likely to require NICU admission and paediatricians caring for these infants, should recognise the need for respiratory support and echocardiography, in addition to screening for hyperbilirubinemia, hypoglycaemia and polycythaemia.

Further, large population-based studies are required to refine our knowledge about LGA infants and the influence of GDM. Also, a longer follow-up period is needed to capture any additional morbidities like abnormal anthropometry and cardiometabolic parameters.