Diabetes in Pregnancy, Neonatal Morbidities, and Early Growth in Moderate or Late Preterm Infants

This is a cohort study from the Pediatrix Clinical Data Warehouse (CDW) of infants admitted to neonatal intensive care units (NICU) 2008 to 2019. The CDW is a prospective database that contains information from infants admitted to NICUs managed by the Pediatrix Medical Group across 35 states and Puerto Rico.¹²  Data for this study was extracted from the CDW for all infants born between 32 0/7 weeks and 36 6/7 weeks gestational age. Infants who died, had major congenital anomalies, congenital infection, were discharged after 40 weeks postmenstrual age, or had undetermined sex were excluded from the final analysis. Infants with undefined sex at birth were excluded because sex was included as a variable in all adjusted analyses of growth outcomes. The cut-off of 40 weeks postmenstrual age was chosen because most infants born moderate or late preterm are discharged by term age.^13,14  This study was approved by the Yale University institutional review board and determined to be exempt from review because the data set provided was de-identified.

The primary exposure of interest was diabetes mellitus in pregnancy, which was defined as any report of maternal diabetes, as indicated in the CDW. Other demographic, delivery, and maternal health covariates examined include maternal age, maternal race and ethnicity (Asian, non-Hispanic black, non-Hispanic white, Hispanic, and other), gravida, parity, delivery mode, hypertensive disorders in pregnancy (defined as gestational hypertension, chronic hypertension, pre-eclampsia and/or eclampsia), maternal smoking (defined as maternal report of smoking during pregnancy), multiple births, gestational age at birth, and infant sex.

Daily infant weight, length, and head circumference data were obtained from the CDW during the newborn hospitalization, and infant BMI, weight-for-length ratio, and weight gain velocity were calculated.^15,16  Nutritional support during the newborn hospitalization was defined as any use of human milk (mother’s own milk and/or donor human milk) during the hospitalization, total days of human milk received, any human milk at hospital discharge, any use of parenteral nutrition during the hospitalization, and total days of parenteral nutrition during the hospitalization. Other infant health variables examined include: hospital length of stay, postmenstrual age at discharge, diagnosis of hyperbilirubinemia, hypoglycemia, patent ductus arteriosus, neonatal withdrawal symptoms (as a marker of neonatal withdrawal syndrome), need for inhaled nitric oxide (as a marker for pulmonary hypertension), need for therapeutic hypothermia (as a marker for moderate to severe neonatal encephalopathy), necrotizing enterocolitis (defined as Bell’s stage 2 or 3), early onset sepsis (defined as a positive blood culture ≤72 hours after birth), late onset sepsis (defined as a positive blood culture >72 hours after birth), any report of antibiotics during the newborn hospitalization, and the need for respiratory support in the first 3 postnatal days (categorized as none, noninvasive support [oxygen, high flow nasal canula, and/or noninvasive positive pressure ventilation], or invasive ventilation [high frequency or conventional ventilation by endotracheal intubation]).

In a bivariate analysis, maternal and infant characteristics were compared between maternal diabetes (DM) exposure groups (DM group versus non-DM group) using the Wilcoxon Rank Sum test for continuous variables and the χ-square test for categorical variables. Next, average change in weight, length, head circumference, weight-for-length, and BMI from birth to hospital discharge were compared between diabetes exposure groups using multivariable linear regression models, with covariate adjustment as described below.

In the longitudinal analysis, trends in mean weight over time were estimated using a linear mixed effects modeling approach with a random intercept for each infant, with maternal DM status as the primary exposure variable, and with covariate adjustment as described below. The expected physiologic pattern of weight in a newborn is an initial diuresis associated with weight loss followed by an increase in weight back to birth weight by postnatal day 14.^17,18  Therefore, in the linear mixed effects, time specific trajectories of weight were modeled using piece-wise linear slopes with knots on days 3, 7, and 14. These models were additionally stratified by gestational age: (1) to evaluate the influence of gestational age at birth and diabetes exposure on early weight patterns, we estimated average percent weight loss from birth with surrounding 95% confidence intervals (95% CI) at postnatal days 3, 7 and 14, and plotted these by maternal DM status; (2) weight gain velocity (change in weight per day) was estimated for the following time periods: between birth and postnatal day 3, postnatal day 3 to 7, postnatal day 7 to 14, and postnatal day 14 to discharge, and the between group differences (DM group – non-DM group) in slopes using means, and 95% CIs were then estimated for weight gain velocity in each time specific period.

In the adjusted models, covariates included were chosen based on the prior literature and from the initial results of the bivariate analysis (using the 2-sided α of .15). The fully adjusted models included maternal age, maternal race and ethnicity, parity, maternal hypertensive disorders, maternal smoking, multiplicity, delivery mode, infant sex, gestational age, infant length of hospital stay, infant respiratory support, infant antibiotic exposure, hyperbilirubinemia, hypoglycemia, human milk at discharge, parenteral nutrition use, neonatal withdrawal symptoms, and the diagnosis of patent ductus arteriosus. For all analyses, all tests were conducted at the 2-sided α level of .05 and conclusions regarding clinically meaningful effects relied upon effects sizes and 95% CIs.^19–21  All analyses were conducted in SAS Version 9.4 (Cary, NC).

During the study period, 336 271 moderate or late preterm infants were admitted to neonatal intensive care units (Fig 1). We excluded 34 772 (10.3%) infants with major congenital anomalies, infants who died, had congenital infections, unknown sex at birth, and/or were discharged after 40 weeks postmenstrual age from the primary analysis. Among the 301 499 infants included in the primary analysis, 42 519 (14.1%) were exposed to diabetes in pregnancy (DM group). Including all infants born during the study period, any type of congenital anomaly, and the diagnosis of major cardiac anomalies were more frequent among infants in the DM group compared with infants in the non-DM group (Table 1).

Maternal and infant characteristics by exposure to diabetes in pregnancy are shown in Table 2. Mothers of infants in the DM group were older and more likely to have coexisting hypertensive disorders during pregnancy, cesarean section delivery, and report Hispanic ethnicity and/or Asian race (P < .001). Most mothers of infants in the DM group had the diagnosis of gestational diabetes (70.4%, N = 29 945), and 39.6% (N = 16 850) of mothers of infants in the DM group received insulin during the pregnancy. Infants in the DM group had shorter median length of stay in the hospital (11 days versus 12 days, P < .001). Infant health and nutritional outcomes during the newborn hospitalization are shown in Table 3. Birth and discharge anthropometry measures, including weight, length, head circumference, weight-for-length ratio, and BMI were higher among infants in the DM group compared with infants in the non-DM group (P < .001).

Infants in the DM group also had increased incidence of neonatal morbidities, including hyperbilirubinemia, hypoglycemia, need for nitric oxide therapy, and diagnosis of patent ductus arteriosus compared with infants in the non-DM group. Additionally, infants in the DM group were more likely to require noninvasive respiratory support in the first 3 postnatal days, including need for oxygen, high-flow nasal cannula, and/or noninvasive positive pressure ventilation (DM-group 49.6% versus non-DM group 47.8%). Differences in the use of parenteral nutrition and breastfeeding outcomes (mean days human milk and human milk at discharge) between groups were different, but small in magnitude (Table 3).

In the analysis of growth trajectory over the birth hospitalization, we found that patterns of weight change from birth weight differed across both gestational age and DM exposure groups (Fig 2). On postnatal day 14, DM-group infants born at 35- and 36-weeks gestational age remained on average 0.8% (95% CI: 0.48 to 1.12) and 2% (95% CI: 1.57 to 2.43) below birth weight, respectively. In contrast, DM-group infants born at 32-, 33-, and 34-weeks gestational age were on average 2.1% (95% CI: 1.69 to 2.51), 2.1% (95% CI: 1.79 to 2.41), and 1% (95% CI: 0.76 to 1.24) above birth weight on postnatal day 14, respectively.

In the adjusted regression models, weight trajectories over time also differed across gestational age and diabetes exposure groups (Table 4). In all gestational age groups, DM-group infants exhibited faster weight loss in the first postnatal week compared with non-DM group infants in the same gestational age group. The adjusted difference in weight gain velocity between DM exposure groups from birth 0 to postnatal day 3 was −7.2 (95% CI: −8.2 to −6.2), −6.5 (95% CI: −7.4 to −5.7), and −4.5 (95% CI: −5.1 to −3.9) g per day for infants born 36-, 35-, and 34-weeks gestational age, respectively. After postnatal day 7, DM-group infants born at 32-, and 33-weeks gestational age had on average accelerated weight gain, whereas infants in the DM group born at 35- and 36-weeks gestational age continued with on average weight loss compared with non-DM group infants born at the same gestational age. For example, the adjusted difference in weight gain velocity between DM groups from postnatal days 7 to 14 was 0.81 (0.29 to 1.33), 0.63 (0.15 to 1.10), −0.74 (−1.46 to −0.02), and −1.41 (−2.61 to −0.21) g per day for infants born 32-, 33-, 35- and 36-weeks gestational age, respectively. The overall difference in growth parameters from birth to discharge were higher among infants in the DM-group, however this difference was small in magnitude (Table 5).

In this large, multicenter cohort of moderate or late preterm infants admitted to newborn intensive care units, infants exposed to diabetes in pregnancy were more likely to be diagnosed with congenital anomalies and common neonatal morbidities, including need for respiratory support in the first postnatal days. Additionally, growth trajectories over the first postnatal weeks differed according to gestational age and diabetes exposure in pregnancy. The current study represents the largest multicenter cohort to date that examines short-term morbidities, including growth changes, among preterm infants exposed to diabetes in pregnancy and demonstrates gestational-age-specific differences in growth velocity according to diabetes exposure at birth. One prior study of late preterm infants (34 to 36 weeks gestational age) admitted to the both the NICU and well-baby nursery found that those exposed to diabetes in pregnancy had accelerated growth in the first postnatal week and some small changes in weight gain in later infancy.¹¹  In the current study, we postulate the differences in early weight loss patterns according to gestational age and diabetes exposure status may be related to differences in nutritional management among infants born at different gestational ages. Limitations in the data include the inability to assess in detail certain nutritional exposures, including formula supplementation, and enteral and intravenous fluid volumes. This study did not include infants who did not require neonatal intensive care admission and lack of this data may have biased the results.

Other population-based cohorts that have examined the association of diabetes in pregnancy with neonatal morbidities have not specifically differentiated gestational age in the analyses.^22–26  Billionet et al demonstrated an association of gestational diabetes in pregnancy with cardiac malformations, macrosomia, and the diagnosis of respiratory distress in population-based cohort of 700 000 births in France.²²  The prematurity rate in this cohort was approximately 7%, and outcomes according to gestational age at birth were not reported. Other studies of extremely preterm infants have reported no differences in mortality or severe morbidities among diabetes exposure groups.^23,27,28  An international cohort of infants born <1500 g reported higher birth weight in the group exposed to diabetes in pregnancy, however there were no differences in severe morbidity or infant mortality in that cohort.²³  Similarly, Bental et al demonstrated no association between diabetes in pregnancy and the diagnosis of respiratory distress or bronchopulmonary dysplasia among infants in the Israel National Very Low Birth weight Infant Database.²⁷ 

Overall, other studies of moderate or late preterm infants have not consistently demonstrated an increased risk of respiratory distress and/or the need for mechanical ventilation in diabetes exposed groups. In one study, offspring of mothers with gestational diabetes did not have increased risk of respiratory morbidity in the first 72 hours after birth.²⁹  In another study, late preterm and term infants of mothers with diabetes were less likely to have respiratory morbidities after birth if there was exposure to labor.³⁰  Additional studies are needed to further examine respiratory outcomes in preterm infants of mothers with diabetes. As work continues to optimize both the diagnosis and treatment of diabetes in pregnancy, focusing on how this may influence short term outcomes in offspring is an important public health measure.^31,32 

Regarding growth outcomes, prior population-based studies that include preterm infants have demonstrated increased risk of higher birth weight, higher adiposity, and large-for-gestational age status with exposure to insulin dependent diabetes in pregnancy.^10,22–24  In studies of term infants, infants exposed to any hyperglycemia in pregnancy appear to exhibit a “catch down” period of growth, or decreasing weight z-scores, through early infancy.^33,34  Although it is known that offspring of mothers with diabetes exhibit increased risk of obesity and cardiometabolic disease in childhood and adolescence, it is unclear how early growth patterns and nutritional management may relate to these cardiometabolic risks, especially among infants born preterm.^6,35,36  This may include the use of gavage feeding as oral feeding maturity develops, use of supplemental intravenous fluids and/or parenteral nutrition, and fortification of breast milk and formula with additional calories. Although these factors were not able to be explored in detail in this study, variation in the nutritional care of moderate or late preterm infants exists.^14,37,38  Future studies should consider detailed examination of the relationship of these early nutritional factors with growth outcomes in moderate or late preterm infants of mothers with diabetes in pregnancy.

A major strength of this study is the large, multicenter data that allowed for the largest analysis to date of moderate or late preterm infants to examine growth outcomes and morbidities related to prenatal exposure to diabetes in pregnancy. The availability of daily growth data during the newborns’ hospitalization allowed for robust modeling of early growth trajectories despite the relatively short length of stay in this group of preterm infants. Additionally, maternal prepregnancy weight, weight gain during pregnancy, and specific diabetes diagnosis classifications and treatment regimens were not able to be controlled for in the regression models. Measures of diabetes control may be an important factor in the observed differences in neonatal outcomes in this study. Lastly, weight z-scores, weight percentile at birth, and classification of small- and/or large-for-gestational age were not assessed in this study. Although most clinical guidelines regarding preterm infant growth focus on targets for weight gain velocity, future studies in prospective cohorts are needed to further examine how weight z-scores and more specific measures of adiposity and body composition in this subset of preterm infants may be associated with short- and long-term growth outcomes.

Short- and long-term outcomes among offspring of mothers with diabetes in pregnancy have been studied in infants born at term but not widely examined among infants born preterm. In this large, multicenter study of moderate or late preterm infants admitted to newborn intensive care units, exposure to diabetes in pregnancy was associated with short-term morbidities during the newborn hospitalization, including hyperbilirubinemia, hypoglycemia, need for early respiratory support, patent ductus arteriosus, and diagnosis of congenital anomalies. Additionally, exposure to diabetes in pregnancy was associated with differences in early growth trajectories in this group of preterm infants. We postulate that variation in the intensive care management of moderate or late preterm infants may influence these outcomes. Future studies in moderate or late preterm infants exposed to metabolic conditions during pregnancy are needed to identify additional perinatal and newborn environmental and nutritional factors that may be associated with early growth trajectory and to examine longitudinal outcomes in this subgroup of preterm infants.:

CDW

Clinical Data Warehouse

DM

diabetes mellitus

GA

gestational age