Abstract
OBJECTIVE:
To examine the effects of early breastfeeding (eBF) or early formula feeding (eFF) on hypoglycemia and on BF initiation in infants born to women with pregestational diabetes mellitus (PGDM) who intended to BF.
METHODS:
Retrospective cohort investigation of 554 women with PGDM and their infants (IDMs) who delivered during 2008–2016. The first feeding (BF or FF) was considered early if given within 4 hours from birth.
RESULTS:
282 (51%) IDMs were admitted to the Well Baby Nursery. Of the 134 IDMs whose early feeding was BF, hypoglycemia affected 30% which was corrected with oral feedings in 78% of the cases. At discharge, 49% BF exclusively while 45% BF partially. Of the 148 IDMs whose early feeding was FF, hypoglycemia affected 40% which was corrected with oral feedings in 69% of the cases. At discharge, 14% BF exclusively while 48% BF partially. There were 272 (49%) IDMs admitted to the NICU. Their early feeding was BF (14%) and FF (86%). Hypoglycemia developed in 50% and 43% of these groups, respectively. Benefits of early feedings on hypoglycemia were masked by the routine use of IV dextrose infusions. At discharge, early BF led to exclusive BF in 45% and partial BF in 50% of the cases. Early FF led to exclusive BF in 17% and partial BF in 42% of the cases.
CONCLUSIONS:
Early and continued feeding (BF preferably or FF if BF is not feasible) should be the first line of treatment for hypoglycemia. Early BF is paramount for BF initiation. Early FF is an obstacle, albeit not absolute, to BF initiation, thus it should not deter continued efforts to start or resume BF.
Background
Infants (IDMs) born to women with pregestational diabetes mellitus (PGDM) are at risk for neonatal morbidities including hypoglycemia [1–5] and BF initiation failure [6–8]. Since cord blood glucose levels do not identify IDMs at risk for hypoglycemia, clinicians must rely on blood glucose screening [1–5, 9]. Early breastfeeding (eBF) or early formula feeding (eFF) may facilitate glycemic stability in IDMs born to women with PGDM and may prevent or correct neonatal hypoglycemia [3, 10–13].
While BF is preferable, maternal complications and/or neonatal morbidities that affect IDMs often delays or precludes BF [14–16] and may lead to formula supplementation [17, 18]. In addition, recognized barriers to BF among non-diabetic populations (i.e., lack of intention to BF, low socioeconomic status, obesity, smoking, low level of education, race, and maternal separation) also affect pregnancies complicated by PGDM [7, 18–21]. The potential effects of eBF on neonatal glucose homeostasis and the long term benefits of BF on mothers and infants health make BF the best choice especially for women with PGDM [10, 20–23].
Early BF is predictive of BF at discharge and BF duration afterward [17, 24]. Conversely, formula feeding shortly after birth or during the hospitalization is a strong predictor of BF initiation failure [17, 25]. The World Health Organization (WHO), the American Academy of Pediatrics (AAP) and the Academy of Breastfeeding Medicine (ABM), have strongly recommended BF for all healthy infants. However, they recognize that in special circumstances other nutritional options may be needed to temporarily replace or supplement BF [17, 27].
Multiple demographic and clinical factors have been identified as barriers to BF initiation and BF continuation in women with PGDM [7, 20]. However, the effect of eBF or eFF on neonatal hypoglycemia and BF initiation in this specific population deserves further attention [4, 17].
Objective
The primary objective was to examine the effects of eBF or eFF on hypoglycemia and on BF initiation at discharge from the hospital in infants born to women with PGDM who intended to BF. The secondary objective was to determine which clinical or demographic factors predict whether the first feeding will be BF or FF.
Subject and methods
This retrospective cohort investigation was approved by the Institutional Review Board of Wexner Medical Center at The Ohio State University. Hard copy (2008–11) and electronic maternal and neonatal records (2013–16) were reviewed. The transition from hard copy to electronic medical records occurred in 2012 thus, data from 2012 were not collected. Women were diagnosed with PGDM (Classes B-C-DRFH) according to clinical and laboratory criteria using the modified White’s classification [7, 8]. Obese was defined by a body mass index (BMI) of 29–34 kg/m2, and very obese by a BMI≥35 kg/m2 [16]. The study population consisted of 554 women with PGDM and their infants if delivered at≥34 weeks, a point in gestation at which most infants can feed orally. Pregnancies affected by major or fatal malformations were excluded. Upon arrival to labor and delivery, each woman declared her infant feeding preference, i.e., BF, FF or a combination.
Depending on the condition of mother and infant following delivery, interactions (holding, skin to skin contact, BF) were encouraged. Delivery room and postpartum interactions were observed and recorded by the nursing staff. Our institutional guidelines for care of the IDM recommend that any symptomatic infant, regardless of the mothers’ class of diabetes, be directly transferred from the delivery room to the NICU. Indications for NICU admission included respiratory distress [respiratory distress syndrome (RDS) and transient tachypnea of the newborn (TTNB)], prematurit y (apnea, severe hypotonia and poor sucking behavior), perinatal depression, and birth trauma. NICU admission for prevention of hypoglycemia was also an option for IDM whose mothers had poorly controlled diabetes and/or severe illness (i.e., classes DRFH). Asymptomatic IDM able to feed were transferred to the Well Baby Nursery (WBN) for routine care and glucose monitoring. Our family-centered care system has full-time lactation consultants and rooming-in available.
Screening for neonatal hypoglycemia (blood glucose <40 mg/dl) was done via serial point of care testing (Accu-Chek®) or by plasma glucose measurement in the laboratory (Beckman Coulter AU5800, Beckman Coulter Inc., Brea, CA, USA) starting within the first hour of life after the first feeding and every 2–4 hours as needed thereafter [1, 5]. The first feeding (BF or FF) was considered early if given within 4 hours from birth. Asymptomatic IDMs in the WBN with hypoglycemia were promptly BF or FF and those with recurrent hypoglycemia were treated with intravenous (IV) dextrose (4–6 mg/kg/min). On admission to the NICU, most IDMs were started on IV dextrose and those who could feed were BF or FF.
BF was defined as initiated if, at the time of hospital discharge, the infant was exclusively BF (all feedings during the 24 preceding hours were BF) or partially BF (BF and FF in combination). Exclusive BF combined with partial BF was considered any BF. Due to the study design, no follow-up information was available on infant feeding practices beyond hospital discharge.
Statistical analysis
Comparisons between patients according to the type of first feeding were made with two-sample t-tests for continuous variables and Chi square tests for categorical variables. Non-normally distributed continuous variables were compared using the Wilcoxon rank sum test. Significance was established at a p value < 0.05. Univariate and multivariate logistic regression were used to ascertain the strength of association of eBF or eFF and BF status at discharge, controlling for maternal (age, diabetes class, parity, BMI, race, smoking, mode of delivery, length of stay in the hospital, and infant feeding preference) and neonatal variables (late prematurity [10], birth weight, fetal growth, macrosomia, gender, place of admission, hypoglycemia, whether infant was discharged with mother, and length of stay in the hospital). A secondary analysis was designed to ascertain the reasons that determine the type of first feeding for IDMs born to women with PGDM who intended to BF.
Results
Comparison of the demographic and clinical characteristics of 554 women with PGDM who intended to BF according to their infants early feeding is presented in Table 1. The distribution of White’s classes was similar in both groups. Women in the eBF group were slightly older, had lower hemoglobin A1C values and delivered vaginally more often than those in the eFF group. Primary cesarean delivery and smoking were more common among mothers whose infants were in the eFF group. Rates of very obese women remain unchanged. Infants in the eBF group were slightly more mature, and were more often admitted to the WBN. There were no maternal or neonatal deaths.
Demographic and clinical characteristics of women with PGDM who intended to breastfeed
and their infants according to infants first feeding
Demographic and clinical characteristics of women with PGDM who intended to breastfeed and their infants according to infants first feeding
Two hundred and eighty-two infants were admitted to the WBN following vaginal (48%) or cesarean delivery (52%). For 134 (48%) of the 282 infants the first feeding was BF and for 148 (52%) was formula (Table 2). The time from birth to the first oral feeding showed that 70% of BF and 71% of FF were fed within 1 hour from birth, 14% of BF and 18% of FF between 1 and 2 hours, and 16% of BF and 11% of FF between 3 and 4 hours.
Infants admitted to the WBN according to their first feeding
Infants admitted to the WBN according to their first feeding
Hypoglycemia preceded BF in 14 (10%) of 134 cases and followed BF in 26 (20%) of the remaining infants. Of the 120 infants whose blood glucose was tested after feeding, 26 (22%) developed hypoglycemia. Nine of the 40 (23%) hypoglycemic infants required oral feedings and IV dextrose. Of the thirty-one (77%) treated successfully with oral feeding alone, 4 IDMs repeated BF and 27 received BF and formula supplementation.
Hypoglycemia preceded FF in 46 (31%) of 148 cases and followed feeding in 13 (9%) of remaining infants. Of the 102 infants whose blood glucose testing was done after feeding, 13 (13%) developed hypoglycemia. Eighteen of the 59 (31%) hypoglycemic infants required oral feedings and IV dextrose. Of the 41 (69%) corrected with oral feeding alone, 14 IDMs repeated FF, 15 continued with BF and supplemental formula and 12 resumed BF exclusively.
Regardless of the type of the initial feeding, 99 of the 282 (35%) infants admitted to the WBN developed hypoglycemia. Eighty-six of the 99 (87%) hypoglycemic episodes occurred within the first 2 hours of life, 11 (11%) between 3 and 6 hours and 2 (2%) occurred between 6 and 24 hours. Forty-one of the 99 (41%) hypoglycemic events were single low glucose episodes. Hypoglycemia was diagnosed in 18 (18%) of the instances by a glucose value of <25 mg/dl and in the remaining 81 (82%) by values between 25 and 39 mg/dl. Correction of hypoglycemia with oral feedings and IV dextrose was needed for 55% of IDMs in the lower glucose group and for 21% of IDMs in the higher glucose group.
Comparison of maternal and neonatal characteristics between 99 IDMs who developed hypoglycemia and 183 who did not demonstrate that the only difference was a higher number of large for gestational age (LGA) infants (47% vs 34%, p 0.02) among those who became hypoglycemic.
Two hundred and seventy-two infants were admitted directly to the NICU following cesarean (67%) or vaginal delivery (33%). First feeding was BF for 14% and FF for 86% of the infants (Table 3). Time from birth to the first oral feeding showed that 18% of BF and 47% of FF were fed within 1 hour from birth, 8% of BF and 14% of FF between 1 and 2 hours, 11% of BF and 9% of FF between 3 and 4 hours, and 63% of BF and 30% of FF after 4 hours.
Infants admitted to the NICU according to their first feeding
Infants admitted to the NICU according to their first feeding
Comparison between eBF and eFF groups showed that the admission diagnoses were similar. Respiratory distress syndrome (6 vs 3%) and transient tachypnea of the newborn (46 vs 35%) were the most common diagnoses, followed by prematurity (14 vs 20%), prevention of hypoglycemia (17 vs 31%) and miscellaneous (17 vs 11%).
Of 272 IDM admitted to the NICU, 120 (44%) developed hypoglycemia. Fifty-six (47%) of the hypoglycemia cases were single low glucose episodes. One hundred and eight of the 120 (90%) hypoglycemic episodes occurred within the first 2 hours of life, 10 (8%) between 3 and 6 hours and 2 (2%) occurred between 6 and 24 hours. Initial low glucose values were <25 mg/dl for 39 (33%) infants and were 25–39 mg/dl for the remaining 81 (67%) infants. Eighty (67%) of the 120 infants were treated with IV dextrose.
Comparison of maternal and neonatal characteristics between 120 IDMs who developed hypoglycemia and 152 who did not, showed that the only differences were a higher number of LGA (60% vs 40%, p 0.0015) and macrosomic infants (28% vs 14%, p 0.0097) among those who became hypoglycemic.
There were 134 IDMs admitted to the WBN whose early feeding was BF (Table 2). Of them, 38 (28%) infants continued BF throughout the hospitalization and 96 (72%) required formula supplementation during the hospitalization. At discharge 66 (49%) BF exclusively, 60 (45%) BF partially and 8 (6%) were FF.
There were 148 IDMs admitted to WBN whose first feeding was FF. Thirty-six (24%) received FF through discharge and 112 (76%) supplemented BF. At discharge 21 (14%) BF exclusively, 71 (48%) BF partially and 56 (38%) were FF.
There were 38 IDMs admitted to the NICU whose initial feeding was BF (Table 3). Eight (21%) infants continued BF throughout the hospitalization and discharge and 38 (79%) required formula supplementation. At discharge 17 (45%) BF exclusively, 19 (50%) BF partially and 2 (5%) were FF.
There were 234 IDMs admitted to NICU whose first feeding was FF. Forty-seven (20%) received FF through discharge and 187 (80%) supplemented BF. At discharge 39 (17%) BF exclusively, 98 (42%) BF partially and 97 (41%) were FF.
Regardless of place of admission, of the 554 infants 172 (31%) had eBF, while the remaining 382 (69%) had eFF. First feeding by BF predicts that the odds of BF exclusively at discharge are higher than that of partial BF or FF (OR 5.0, CI 3.3–7.5). Furthermore, initial FF is a strong predictor of exclusive FF at discharge (OR 11.1, CI 5.5–22.4).
Bivariate analysis showed that higher hemoglobin A1C, primiparity, cesarean delivery, LGA, late preterm delivery and admission to NICU characterized infants whose first feeding would be formula. Neither severity nor duration of diabetes, obesity, race and time from birth to first feeding were predictive of the type of first feeding. Multivariate logistic regression showed that smoking during pregnancy, delivery by primary cesarean, LGA, admission to NICU and hypoglycemia before feeding remain significant predictors of formula as the first infant feeding (Fig. 1).

Strongest predictors of formula as the first infant feeding among PGDM women who intended to breastfeed.
The incidence of diabetes complicating pregnancy continues to rise across the world [28]. Thus, it may be anticipated that maternal and neonatal morbidities that often prompt NICU admissions will remain a challenge for IDMs born to women with PGDM [12, 29]. Timing of the first feeding (BF or FF) is important because many critical maternal and neonatal physiological interactions occur during the first hours after birth (i.e., skin to skin contact, eBF) [30–33]. In normal as well as in in complex pregnancies, certain hospital practices may delay infant feeding (i.e., cesarean section, eye prophylaxis, vitamin K administration and relevant to this study, blood glucose monitoring). While some of these practices may be postponed, others may be unavoidable in women with PGDM and their infants [3, 35].
A recent world-wide study on early initiation of BF reported that half of healthy term infants did not BF within the first 60 minutes as recommended by the WHO [36]. The data reported here demonstrated that all infants who initiated BF or FF in the WBN did so within 4 hours from birth. Among IDMs admitted to the NICU, only 37% of eBF and 70% of eFF occurred within the first 4 hours of life. It is safe to assume that in the NICU the separation of mothers and infants was almost universal and that the majority missed the benefits that early uninterrupted contact provides [30–33]. The difference in timing of first feeding between WBN and the NICU is not surprising given the high cesarean rate affecting women with PGDM and the morbidities challenging their infants [31–35, 37].
The definition of neonatal hypoglycemia based on glucose measurements remains a problem for infants born to healthy women as well as those born to women with PGDM [3–5, 13]. Neonates at risk for developing hypoglycemia should be routinely monitored for blood glucose levels [1–5, 38]. Prevention and treatment of neonatal hypoglycemia without disruption of mother-infant interactions is essential to maternal and infant health, especially BF initiation. We agree with Harding et al that dextrose gel in conjunction with BF provides an attractive, non-invasive alternative to infant formula as first line of treatment for hypoglycemia [4]. However, until this new therapeutic modality is widely implemented, repeated BF and FF or combination will remain the most commonly used treatment for hypoglycemia [3, 17].
Like other clinicians faced with IDMs at risk for hypoglycemia, we adopted threshold values for the first 24 hours of life similar to those recommended by the AAP [1, 5]. Neonatal hypoglycemia in IDMs born to women with PGDM is reported at rates varying from 20 to 51% [2, 15]. It has been proposed that among healthy term IDMs, early and frequent BF is often adequate prophylaxis against hypoglycemia [9, 17]. In our study the rate of hypoglycemia following initial BF was 22% and following initial FF 13%. While there is no certainty of the preventive value of early feeding on neonatal hypoglycemia in IDMs, the therapeutic benefits of repeated feedings (BF or FF) have been validated by our data [4, 17]. We agree that the ideal treatment for hypoglycemia should support the establishment of BF while avoiding the unnecessary administration of formula or IV dextrose [4, 38]. We observed that hypoglycemia often preceded the initial feeding and prompted a response (per guidelines) of healthcare providers to assist with BF, or in cases when the mother is not available or ready to BF, to provide formula supplementation [10, 27].
Current AAP guidelines for healthy infants recommended eBF, exclusive BF for at least 6 months and continued BF for at least one year [1, 5]. Information about eBF, BF initiation and BF duration for healthy populations is abundant, [17, 39] however, data regarding eBF among women with PGDM remains scarce [6–8, 40]. The ABM strongly endorse exclusive BF and discourages unnecessary FF supplementation among healthy term and late preterm infants but provides specific examples of mother-infant dyads who due to illness would benefit from supplementation [27].
Early BF that continues uninterrupted during hospitalization predicts BF beyond the time of discharge from the hospital [17, 24]. Unfortunately, recent global data of healthy term infants showed that no more than half of them had established BF by the time of discharge [36]. More disturbing is the recognition that in most cases there was no medical justification for BF discontinuation [25, 36]. In our study, the opportunity for early maternal-infant interactions (including feeding) is provided to IDMs admitted to the WBN but it is seldom feasible for those admitted to the NICU. For those who had eBF, despite BF interruptions and resumptions in some cases, the rate of exclusive BF or any BF during hospitalization and at discharge was high. For infants who had eFF, BF initiation and BF continuation, as expected, was lower than for those with eBF.
The negative impact of eFF on BF initiation at discharge is clear, however, the fact that 66% of the infants from the WBN and 59% of those discharged from the NICU who supplemented had any BF at discharge is encouraging. Of course, to restrict supplementation to a minimum and to encourage BF resumption as soon as possible requires a full commitment from lactation consultants and healthcare providers [22, 27]. Physicians who prescribe formula supplementation should reassure parents that limited and temporary disruptions should not prevent the successful resumption of BF [22, 34].
Intention to BF remains the strongest predictor of BF initiation and BF continuation [6–8, 29]. In our experience, the rate of women with high risk obstetrical conditions who intended to BF is similar to that of the general population [7, 29]. However, there exists a discordance between women with complex pregnancies who intend to BF and exclusive or partial BF at discharge [7, 41].
For all women who intend to BF, the first infant feeding should be BF. Identification of factors that may prevent this and lead to FF as the first feeding highlight areas for potential intervention. Prenatal smoking cessation programs may be successful in reducing the number of women who smoke during pregnancy [7, 42]. It is unlikely that the cesarean delivery rate among women with PGDM will decline significantly, however, efforts to minimize mother-infant separation could be beneficial [31, 43]. NICU admissions for many IDMs may be indicated, however, unnecessary admissions of asymptomatic IDMs should be avoided [9, 15]. The need for FF as the first choice to correct hypoglycemia may be decreased if other therapeutic approaches such as dextrose gel are proven successful [3, 11–13].
Limitations of this investigation are those inherent to all retrospective designs and the lack of follow-up information regarding infant feeding after discharge. Also, the definition of BF initiation at discharge may only be applicable to women with complex pregnancies for whom early mother-infant contact may be delayed. The strength of this investigation rests on the size of the high-risk obstetrical and neonatal population and the fact that the data were obtained directly from hospital records not via post-delivery maternal questionnaires. Also, recognition of the importance of eBF or eFF and the identification of historical predictors of the type of early feeding in this high-risk population highlights opportunities for targeted interventions.
Conclusion
IDMs born to women with PGDM who intend to BF are at risk for maternal-infant separation, hypoglycemia and BF initiation failure. Early feeding decreases the rate of neonatal hypoglycemia and when continued as BF, FF or a combination, can correct hypoglycemia in most cases preventing the need for infusions. Smoking during pregnancy, cesarean delivery, admission to the NICU and neonatal hypoglycemia before feeding, decrease the chances that the first feeding will be BF. Initial BF predicts high rates of exclusive and partial BF at discharge. Conversely, first FF is a major obstacle, albeit not absolute to BF initiation at discharge. The judicious use of formula supplementation when medically indicated need not preclude successful BF initiation at hospital discharge.
