Abstract
Background
Poor feeding, among other factors, predisposes neonates to hypoglycaemia. Early feeding is widely recommended to prevent hypoglycaemia in those at risk, but the effectiveness of this is uncertain. This review aimed to summarise and analyse the evidence on the effectiveness of early feeding for prevention of neonatal hypoglycaemia.
Methods
Four databases and three clinical trial registries were searched from inception to May 24, 2023. Published and unpublished randomised controlled trials (RCTs), quasi-RCTs, cluster randomised trials, non-randomised studies of interventions, and observational studies with comparison groups were considered for inclusion with no language or publication date restrictions. We included studies of neonates who were fed early (within 60 min of birth or study defined) versus delayed. Study quality was assessed using the Cochrane Risk of Bias 1 tool or Effective Public Health Practice Project Quality Assessment tool. Certainty of evidence was assessed using the Grading of Recommendations Assessment, Development and Evaluation approach. RevMan 5.4.1 or R was used to synthesise results in random-effects meta-analyses. This review was registered prospectively with PROSPERO (CRD42022378904).
Results
A total of 175,392 participants were included across 19 studies, of which two were RCTs, 14 cohort studies, two cross-sectional studies, and one a case-control study. Most studies (13/19) were conducted in low- or lower-middle-income countries. Early feeding may be associated with reduced neonatal hypoglycaemia (four cohort studies, 744 infants, odds ratio [OR] 0.19 (95% CI: 0.10–0.35), p < 0.00001, I2 = 44%) and slightly reduced duration of initial hospital stay (one cohort study, 1,673 infants, mean difference: –0.20 days [95% CI: −0.31 to −0.09], p = 0.0003), but the evidence is very uncertain. One RCT found early feeding had little or no effect on the risk of neonatal mortality, but three cohort studies found early feeding may be associated with reduced risk (136,468 infants, OR 0.51 [95% CI: 0.37–0.72]; low certainty evidence; p <0.0001; I2 = 54%).
Conclusion
We found that early feeding may reduce the incidence of neonatal hypoglycaemia, but the evidence is very uncertain. Given its many other benefits, early feeding should continue to be recommended. This review was primarily funded by the Aotearoa Foundation and the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health.
Keywords: Infant, Newborn, Glucose, Hypoglycemia, Breastfeeding, Systematic review
Introduction
Following birth, neonates must rapidly switch from the continuous supply of glucose from the placenta to metabolic independence. Delay in this transition can result in hypoglycaemia. Persistent or severe hypoglycaemia is associated with neurosensory impairment including executive dysfunction and developmental delay [1–4]. Hypoglycaemia affects up to 39% of all neonates [5] and 50% of at-risk neonates [6]. Poor feeding is a risk factor for neonatal hypoglycaemia [7, 8], and early feeding has been widely recommended to prevent hypoglycaemia [9]. For example, clinical practice guidelines from Queensland Health [10], the British Association of Perinatal Medicine [11], and World Health Organization [12] all recommend that breastfeeding be initiated within an hour of birth for the prevention of hypoglycaemia.
However, the evidence supporting an association between early feeding and blood glucose concentrations or hypoglycaemia is limited, and the results are mixed. One Israeli study found that, among term infants born to mothers with gestational diabetes, infants breastfed within 30 min of birth were at lower risk of hypoglycaemia (10%) than those whose first feed was delayed for 30 min or longer after birth (28%) [13]. However, another study of infants born to mothers without diabetes in Ireland concluded that there was no significant association between early feeding and blood glucose concentrations at 1 h of age [14]. The aim of this systematic review and meta-analysis was to assess the effect of early feeding on the incidence of neonatal hypoglycaemia.
Methods
This review was reported in line with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist [15] (online suppl. file 1; for all online suppl. material, see https://doi.org/10.1159/000535503) and prospectively registered with PROSPERO (registration number CRD42022378904).
Search Strategy and Selection Criteria
The following electronic databases were searched from inception to May 24, 2023: MEDLINE (Ovid), Embase (Ovid), CINAHL Plus, and the Cochrane Central Register of Controlled Trials (CENTRAL). Registered trials were also searched for in: Current Controlled Trials (www.controlled-trials.com), Clinical Trials (www.ClinicalTrials.gov), and WHO ICTRP Search Portal (https://apps.who.int/trialsearch/). Conference abstracts were included if they provided usable summary data. Reference lists of included studies were also screened. See online supplementary file 2 for the full search strategy.
Inclusion criteria were newborn infants (term or preterm), fed (breastfeeding, expressed breast milk feeding, or formula) either early (within 60 min of birth, or study defined) or delayed (more than 60 min after birth, or study defined). We included published and unpublished randomised controlled trials (RCTs), quasi-RCTs, cluster randomised trials, non-randomised studies of interventions, and observational studies with comparison groups. There were no language or publication date restrictions.
The primary outcome was neonatal hypoglycaemia (study defined). Secondary outcomes were hypoglycaemia (any blood glucose concentration <2.6 mmol/L) during initial hospital stay, receipt of treatment for hypoglycaemia (study defined), number of episodes of hypoglycaemia (study defined), severity of hypoglycaemia (study defined), breast milk feeding exclusively (baby only receives breast milk without any other drink or food) at discharge, developmental impairment at follow-up (study defined), adverse events (study defined), duration of hospital stay, cost of intervention (as measured by study), cost of neonatal care (as measured by study), and feed intolerance (study defined).
Data Collection and Analysis
Three review authors (L.F.R., E.W., and L.L.) independently screened titles and abstracts of identified records, then assessed potentially eligible full-text articles for inclusion using Covidence [16]. The same review authors then independently extracted data into a pre-specified data extraction form. The extracted data included the authors, study setting, study methodology, ethics approval, conflicts of interest, funding sources, information for the assessment of the risk of bias, participant characteristics, intervention and control details, and outcome data. Three authors (L.F.R., E.W., L.L.) independently assessed the risk of bias for RCTs with the Cochrane Risk of Bias 1 tool (RoB 1) [17] and for non-randomised studies of interventions and observational studies with the Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies [18]. Discrepancies at each step were resolved by discussion between the reviewers. Articles not written in English were translated by a colleague where possible and otherwise by DeepL [19] or Google Translate [20].
We intended to determine the certainty of evidence for each key outcome (neonatal hypoglycaemia (study defined), receipt of treatment for hypoglycaemia during initial hospital stay (study defined), severity of hypoglycaemia (study defined), developmental impairment at follow-up (study defined), duration of initial hospital stay, breast milk feeding exclusively at discharge, and adverse events (study defined)) using the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach [21]. A “Summary of Findings” table was compiled using the GRADEpro Guideline Development Tool (GDT) [22].
Statistical Analysis
We carried out meta-analyses using RevMan 5.4.1 [23] or R [24] using random-effects models, calculating risk ratios (RRs) or odds ratios (ORs) with 95% confidence intervals (CIs) for dichotomous outcomes. For continuous outcomes, we calculated mean differences (MDs) with 95% CIs. To pool the adjusted ORs from cohort studies, we used R [24]. To convert RRs to ORs and vice versa, we used Zhang and Yu’s method [25]. For all models, a p value of <0.05 denoted statistical significance. We calculated I2 and χ2 for each analysis to determine the percentage of variability in effect estimates that may be due to heterogeneity. Where we observed substantial heterogeneity (I2 > 50% and p < 0.10 in the χ2 test), we intended to explore possible causes in sensitivity analyses.
We planned to assess publication bias by visual inspection of each funnel plot, plotting the study effect size against the sample size when there were 10 or more trials. We planned to undertake subgroup analyses to determine if the effect of early feeding differed by New Zealand ethnic group (Māori vs. non-Māori), duration of delay before feeding, mode of feeding (breastfeeding vs. expressed breast milk vs. formula), gestational age (preterm vs. term), babies at risk of hypoglycaemia versus not at risk, mode of delivery (vaginal vs. caesarean), birth setting (home birth vs. primary, secondary, or tertiary health care setting), and maternal diabetes diagnosis versus no diabetes. Minority and indigenous ethnic groups often have poorer pregnancy and childbirth outcomes. In New Zealand, Māori babies have higher perinatal mortality [26] and are less likely to be breastfed [27]. We therefore planned an additional subgroup analysis to determine if the effect of early feeding differed by New Zealand ethnic group (Māori vs. non-Māori). All analyses were pre-planned unless specified otherwise.
Results
A total of 9,768 records were identified by searching (Fig. 1). After automatic removal of duplicates, title and abstract screening was conducted for 6,349 records, followed by full-text screening of 110 records. Three records could not be retrieved. A total of 19 studies (20 records) met our inclusion criteria and were included in this review, with one additional study awaiting classification.
Fig. 1.
Flow diagram of included studies.
The 19 included studies were two RCTs, 14 cohort studies, two cross-sectional studies, and one case-control study (Table 1). A total of 175,392 participants were included, with four studies each having more than 10,000 participants. The studies were published between 1978 and 2023; one (5%) was conducted in a low-income country, 12 (63%) in a lower-middle-income, 2 (11%) in an upper-middle-income, and four (21%) in a high-income country [45]. Infants were born at home in 5 (26%) studies. Included infants were preterm in one study [40], late-preterm to term in two studies [14, 30], term in four studies [13, 31, 33, 39], all different risks in one study [36], and unspecified in 11 studies [28, 29, 32, 34, 35, 37, 38, 41–44].
Table 1.
Characteristics of included studies
| Author, year | Country | Participants description | Participants, n | Intervention (timing) | Comparison (timing) | Outcomes |
|---|---|---|---|---|---|---|
| Randomised controlled trials | ||||||
| Bullough et al. [28] (1989) | Malawi | Inclusion: babies delivered at home by trained birth assistants (TBAs) | 4,271 (intervention: 2,129; comparison: 2,142) | Early feeding (breastfed immediately after delivery) | Late feeding (timing not stated) | Adverse event – postpartum haemorrhage at birth |
| Exclusion: babies delivered by TBAs who failed the data collection test | Adverse event – neonatal mortality | |||||
| Salariya et al. [29] (1978) | United Kingdom | Inclusion: singleton, healthy infants, mothers chose to breastfeed, primiparous | 111 (intervention: 56; comparison: 55) | Early feeding (breastfeeding initiated ≤10 min after birth) | Late feeding (breastfeeding initiated 4–6 h after birth) | Breastfeeding (any) at 6 and 12 weeks |
| Exclusion criteria: not described | ||||||
| Cohort studies | ||||||
| Chertok et al. [13] (2009) | Israel | Inclusion: GDM mother, otherwise healthy, term (≥37 weeks GA), mother aged 18–45 years, uncomplicated vaginal delivery, infant minimum Apgar of 7/10 | 84 (intervention: 44; comparison: 40) | Early feeding (breastfed early in the delivery room) | Late feeding (not breastfed in the delivery room) | Neonatal hypoglycaemia (BGC <2.5 mmol/L, time unspecified) |
| Exclusion: not described | Neonatal BGC within 3 h of birth | |||||
| De et al. [30] (2011) | India | Inclusion: healthy, 1.5–3.9 kg birthweight, >34 weeks GA | 150 (intervention: 100; comparison: 50) | Early feeding (initiated ≤2 h after birth) | Late feeding (>2 h after birth) | Neonatal hypoglycaemia (BGC <2.2 mmol/L, measured at 1, 6, 12, 24, and 48 h after delivery which was independent of feeding time) |
| Exclusion: infant of mother with diabetes, birth asphyxia, congenital malformation, admitted to NICU, early onset septicaemia | ||||||
| Diwakar and Sasidhar [31] (2002) | India | Inclusion: healthy, term (37–42 weeks GA), AGA. | 200 (intervention: 142; comparison: 58) | Early feeding (breastfed ≤3 h after birth) | Late feeding (breastfed >3 h after birth) | Neonatal BGC at 3 h of age |
| Exclusion: respiratory distress, perinatal asphyxia, meconium aspiration syndrome, polycythaemia, receiving antibiotics, maternal history of hypertension, anaemia, or diabetes | ||||||
| Edmond et al. [32] (2007) | Ghana | Inclusion: singleton, initiated breastfeeding, survived to day 2, mothers visited in neonatal period | 10,947 (intervention: 4,763; comparison: 6,184) | Early feeding (breastfed ≤1 h after birth) | Late feeding (breastfed >1 h after birth) | Adverse event – neonatal mortality within 2–28 days of birth |
| Exclusion: mortality before day 2, not successfully breastfed, mother moved out of study area before second infant interview | ||||||
| Franco-del Rio and Paredes-Melesio [33] (2022) | Mexico | Inclusion: term, normal birth weight (>2,500 g), Apgar at 5 min >7 | 1,673 (intervention: 1,175; comparison: 498) | Early feeding (initiated ≤1 h after birth) | Late feeding (initiated >1 h after birth) | Duration of initial hospital stay |
| Exclusion: multiple pregnancies, newborns with any condition after birth or conditions that prevented lactation | ||||||
| Iarŭkov et al. [34] (1992) | Bulgaria | Inclusion: normal pregnancy and delivery | 79 (intervention: 60; comparison: 19) | Early feeding (initiated ≤2 h after birth) | Late feeding (initiated >2 and ≤6 h after birth) | Neonatal BGC on day one and on day two |
| Exclusion: not described | ||||||
| Klemm et al. [35] (2014) | Bangladesh | Inclusion: breastfed, singleton, survival to 48 h | 25,979 (intervention: 7,068; comparison: 18,911) | Early feeding (breastfed ≤1 h after birth) | Late feeding (breastfed >1 h after birth) | Adverse event – neonatal mortality by 1 month of age |
| Exclusion: not described | ||||||
| Kumar et al. [36] (2021) | India | Inclusion: infant of mother with diabetes, LGA, SGA, preterm (<37 weeks GA), LBW | 483 (intervention: 314; comparison: 169) | Early feeding (initiated ≤1 h after birth) | Late feeding (initiated >1 h after birth) | Neonatal hypoglycaemia (BGC <2.2 mmol/L, measured at 2, 6, 12, 24, 48, and 72 h after delivery) |
| Exclusion: major congenital malformations, hypoxic-ischaemic encephalopathy stage 3, meningitis, acute bilirubin encephalopathy, hospitalised. Parents who did not consent, not willing to attend follow-ups | ||||||
| Mullany et al. [37] (2008) | Nepal | Inclusion: live-born, survival to 48 h, mother reported breastfeeding, breastfeeding initiation time could be estimated | 22,838 (intervention: 771; comparison: 22,067) | Early feeding (breastfed ≤1 h after birth) | Late feeding (breastfed >1 h after birth) | Adverse event – neonatal mortality within 3–28 days of birth |
| Exclusion: early deaths (before 48 h) | ||||||
| NEOVITA Study Group [38] (2016) | Ghana, India, and Tanzania | Inclusion: mother reported they were likely to stay in study area for next 6 months, newborn <3 days old, could feed orally, primary caregiver gave consent | 99,632 (intervention: 56,981; comparison: 42,651) | Early feeding (breastfed ≤1 h after birth) | Late feeding (breastfed >1 h after birth) | Exclusively breastfed at 4 days of age |
| Exclusion: infant never initiated breastfeeding or initiated after 96 h | Adverse event – neonatal mortality within 5–28 days of birth | |||||
| Samayam et al. [39] (2015) | India | Inclusion: term, normal birth weight, healthy, Clinical Assessment of Nutrition (CAN) score ≥25, asymptomatic, singleton, delivered by vaginal route or LSCS | 100 (intervention: 76; comparison: 24) | Early feeding (breastfed ≤1 h after birth) | Late feeding (breastfed >1 h after birth) | Neonatal hypoglycaemia (BGC <2.2 mmol/L, measured at 1, 6, 24, and 48 h after delivery) |
| Exclusion: SGA, foetal malnutrition (CAN score <25), LGA, preterm, birth asphyxia, sepsis, requiring NICU admission, parenteral fluids/other modes of feed, infant of mother with PIH/diabetes | ||||||
| Smith et al. [40] (2017) | Tanzania | Inclusion: singleton, women presenting to labour wards or antenatal care before 32 weeks GA | 4,203 (intervention: 3,667; comparison: 536) | Early feeding (initiated ≤1 h after birth) | Late feeding (initiated >1 h after birth) | Adverse event – all-cause infant mortality at 3, 6, 12, and 24 months of age |
| Exclusion: congenital abnormality preventing feeding, mother did not intend to stay in study area | ||||||
| Sweet et al. [14] (1999) | Ireland | Inclusion: >37 weeks GA | 75 (intervention: 46; comparison: 29) | Early feeding (initiated ≤1 h after birth) | Late feeding (initiated >1 h after birth) | BGC at 1 h after birth |
| Exclusion: infant of mother with diabetes, low Apgar scores or requiring NICU admission | ||||||
| Zhou et al. [41] (2017) | USA | Inclusion: ≥1 BGC measurement during early newborn period, AGA, 5-min Apgar score ≥7, haematocrit <65%, survival to hospital discharge | 887 (intervention: 315; comparison: 572) | Early feeding (initiated before initial glucose screen [mean 0.9 h SD 0.6]) | Late feeding (initiated after initial glucose screen [mean 3.8 h SD 2.0]) | Neonatal BGC at 1–3 h after birth |
| Exclusion: received care in NICU, infant of mother with diabetes, received parenteral fluids, stayed in hospital >4 days, major congenital anomalies, inborn metabolic errors, chromosomal abnormalities | ||||||
| Case-control studies | ||||||
| Sasidharan et al. [42] (2004) | India | Inclusion: not described | 302 (intervention: 142; comparison: 160) | Early feeding (initiated ≤2 h after birth) | Late feeding (initiated >2 h after birth) | Neonatal hypoglycaemia (BGC <2.2 mmol/L measured within 48 h of delivery) |
| Exclusion: stillbirth or refusal by parents | ||||||
| Cross-sectional studies | ||||||
| Akter et al. [43] (2016) | Bangladesh | Inclusion: live-born, delivered at home, ever-married women aged 12–49 | 3,182 (intervention: 1,636; comparison: 1,546) | Early feeding (breastfed ≤1 h after birth) | Late feeding (breastfed >1 h after birth) | Adverse event – neonatal mortality within first 28 days of life |
| Exclusion: not described | ||||||
| Jatav et al. [44] (2023) | India | Inclusion: not described | 196 (intervention: 134; comparison: 62) | Early feeding (timing not stated) | Late feeding (timing not stated) | Neonatal hypoglycaemia (BGC <2.2 mmol/L or plasma glucose <2.5 mmol/L measured within 48 h of delivery) |
| Exclusion: neonatal asphyxia, mother with diabetes, sepsis, on intravenous fluids for any other cause, factors not allowing for breast feeding (e.g., cleft palate), post term, IUGR, prelacteal feeds given, no consent, not all seven BGC readings could be taken | ||||||
AGA, appropriate for gestational age; BGC, blood glucose concentration; GA, gestational age; GDM, gestational diabetes mellitus; LBW, low birth weight; LGA, large for gestational age; LSCS, lower uterine segment section; NICU, neonatal intensive care unit; PIH, pregnancy-induced hypertension; RR, risk ratio; SD, standard deviation; SGA, small for gestational age; TBA, trained birth assistant.
Risk of Bias or Quality of Included Studies
Both RCTs were at high risk for performance bias (blinding of participants and personnel was either not reported or not possible), and one was at high risk for reporting bias (for not reporting baseline characteristics and whether they differed between groups) [28, 29]. Both were at low risk for other bias and at least one study had unclear risk of selection, detection, or attrition bias. The possibility of selection bias arose from lack of reporting on how the allocation sequence was generated and whether it was adequately concealed in one study. Detection bias was unclear in both RCTs since neither reported whether outcome assessors were blinded to the intervention allocation. Attrition bias was unclear in one study as the exclusion of participants after randomisation was ambiguous [29]. See Figure 2a and b for risk of bias ratings for included RCTs.
Fig. 2.
Risk of bias assessment or quality assessment. a Risk of bias for randomised trials using Cochrane Risk of Bias 1 (RoB 1) tool: review authors’ judgements about each source of bias presented as percentages across all included studies. b Risk of bias summary for randomised trials using Cochrane Risk of Bias 1 (RoB 1) tool: review authors’ judgements about each source of bias for each included study. c Quality assessment for observational studies using Effective Public Health Practice Project (EPHPP) Quality Assessment Tool for Quantitative Studies.
Among the observational studies, 5 (29%) were of strong quality overall (no weak ratings) [32, 38, 40, 41, 43], 6 (35%) were of moderate quality (1 weak rating) [13, 30, 31, 36, 37, 42], and 5 (35%) were of weak quality (two or more weak ratings) [14, 33–35, 39, 44]. The domain most often rated strong was “data collection method” with most studies using valid and reliable methods for outcome measurement. The domain most often rated weak was “confounders” with several studies either not reporting how confounders were addressed or having insufficient inclusion of appropriate confounders in statistical models. See Figure 2c for quality ratings for included observational studies.
Primary Outcome: Neonatal Hypoglycaemia
Of the six studies reporting neonatal hypoglycaemia [13, 30, 36, 39, 42, 44], 5 (83%) were conducted in India [30, 36, 39, 42, 44]. Early feeding was defined as within 1 h of birth in two studies [36, 39], within 2 h in two studies [30, 42], and was undefined in two studies [13, 44]. Neonatal hypoglycaemia was defined as <2.2 mmol/L in five studies [30, 36, 39, 42, 44] and <2.5 mmol/L in one study [13]. Evidence from four cohort studies [13, 30, 36, 39] showed that early feeding may be associated with a reduced incidence of neonatal hypoglycaemia, but the evidence is very uncertain (OR 0.19 [0.10, 0.35], p < 0.00001; I2 = 44%; 744 infants, Fig. 3a). Very uncertain evidence from one case-control study [42] showed the odds of neonatal hypoglycaemia were lower in the early feeding than in the delayed feeding group (OR 0.38 [0.22, 0.65], p = 0.0005; 302 infants, Fig. 3a). Findings from the one cross-sectional study [44] showed that early feeding may be associated with reduced neonatal hypoglycaemia, but the evidence is very uncertain (OR 0.48 [0.24, 0.96], p = 0.04; 196 infants, Fig. 3a). Four cohort studies [13, 14, 31, 41], 3 (75%) conducted in high-income countries [13, 14, 41], found little to no evidence that early feeding was associated with an increase in mean blood glucose concentration, measured 1–3 h after birth, compared to delayed feeding (MD 0.00 mmol/L [−0.27, 0.28], p = 0.98; I2 = 79%; 1,218 infants, Fig. 3b).
Fig. 3.
Association between early feeding and neonatal hypoglycaemia. a Neonatal hypoglycaemia (blood glucose concentration <2.5 mmol/L or <2.2 mmol/L). b Neonatal blood glucose concentration measured between 1 to 3 h after birth.
Secondary Outcomes
Breast Milk Feeding Exclusively at Discharge
One cohort study [38] conducted in three lower-middle-income countries reported that the early feeding group may have higher odds of breast milk feeding exclusively at 4 days of age than the delayed feeding group (OR 7.76 [7.54, 7.99], p < 0.00001; low certainty of evidence; 99,632 infants). Since most included births were home births, we considered this approximated this outcome at discharge. The evidence is very uncertain about the effect of early feeding on breast milk feeding (any) at 6 weeks (RR 1.15 [0.88, 1.50], p = 0.31; 109 infants) and at 12 weeks (RR 1.18 [0.84, 1.66], p = 0.35; 1 RCT; 108 infants) [29].
Adverse Events (Study Defined)
Seven studies (one RCT [28], five cohort studies [32, 35, 37, 38, 40], and one cross-sectional study [43]), all conducted in low- or lower-middle-income countries, studied the association between early feeding and neonatal or infant mortality, with most (n = 6, 86%) reporting neonatal mortality within the first month [28, 32, 35, 37, 38, 43]. Three of the cohort studies excluded infants who died prior to 2 days old to avoid reverse causality [32, 37, 38]. One study [35] did not report mortality and could not be included in the meta-analyses. The RCT [28] suggested that early feeding may result in little to no difference in neonatal mortality compared to delayed feeding, but the evidence is very uncertain (RR 1.01 [0.14, 7.14], p = 1; 4,271 infants; Fig. 4a).
Fig. 4.
Association between early feeding and infant mortality. a Results from randomised controlled trials. b Results from observational studies (unadjusted). c Results from observational studies (adjusted).
Evidence from three cohort studies [32, 37, 38] suggests that early feeding may be associated with reduced neonatal mortality (OR 0.51 [0.37, 0.72], p = <0.0001; I2 = 54%; low certainty of evidence; 136,468 infants, Fig. 4b). The meta-analysis using adjusted ORs (aOR) from the same studies demonstrates a similar outcome (aOR 0.55 [0.46, 0.65], p < 0.0001); I2 = 0%; 98,480 infants, Fig. 4c). The cohort study not included in the meta-analysis [35] found the risk of neonatal mortality may be reduced among infants fed within an hour compared to those first fed 2–23 h after birth (RR: 0.86 [0.69, 1.08]) but not compared to those first fed after 24 h (RR: 1.10 [0.84, 1.68]). Evidence from one cross-sectional study [43] suggests early feeding may be associated with reduced neonatal mortality (OR 0.54 [0.32, 0.92], p = 0.02; 3,182 infants, Fig. 4b), but certainty of evidence is low.
Evidence from one cohort study suggests there may be little to no difference in infant mortality at 3, 6, 12, and 24 months between early feeding and delayed feeding groups [40]. Evidence from one RCT [28] is very uncertain about the effect of early feeding on postpartum haemorrhage (RR 0.94 [0.77, 1.16], p = 0.58; 4,271 infants).
Duration of Initial Hospital Stay
One cohort study [33], including only term infants, suggested early feeding may be associated with reduced duration of initial hospital stay, but the evidence is very uncertain (MD −0.20 days [−0.31, −0.09], p = 0.0003; 1,673 infants). No data were available for the following outcomes: hypoglycaemia (any blood glucose concentration <2.6 mmol/L during initial hospital stay), receipt of treatment for hypoglycaemia, number of episodes of hypoglycaemia, severity of hypoglycaemia, developmental impairment at follow-up, cost of intervention, cost of neonatal care, and feed intolerance.
Subgroup Analyses
Subgroup analyses from one small cohort study [14] showed no interaction between the effect of early feeding on neonatal blood glucose concentration and feeding method (MD for breastfed infants: −0.24 mmol/L, 95% CI: −0.75, 0.27 and MD for bottle-fed infants: −0.06 mmol/L, 95% CI: −0.54, 0.42, p = 0.42 for interaction). It is unclear whether the infants were bottle-fed mother’s breast milk, donor breast milk, or formula milk. We were unable to undertake other planned subgroup or sensitivity analyses due to insufficient data.
Certainty of Evidence (GRADE Assessment)
The certainty of evidence was assessed as low for breast milk feeding exclusively at discharge, neonatal mortality, and postpartum haemorrhage, and as very low for neonatal hypoglycaemia and duration of initial hospital stay (see Table 2). There were no data for receipt of treatment for hypoglycaemia, severity of hypoglycaemia, and developmental impairment at follow-up.
Table 2.
Certainty of evidence (GRADE assessment)
| Outcomes | Participants (studies) at follow-up, n | Certainty of the evidence (GRADE) | Relative effect (95% CI) | Anticipated absolute effects | |
|---|---|---|---|---|---|
| risk with delayed feeding | risk difference with early feeding | ||||
| Neonatal hypoglycaemia (study defined) | 744 (4 cohort studies) | ⊕○○○ Very lowa, b | OR 0.19 (0.10–0.35) | 385 per 1,000 | 278 fewer per 1,000 (326 fewer to 205 fewer) |
| 70 cases 302 controls (1 case-control study) | ⊕○○○ Very lowc | OR 0.38 (0.22–0.65) | / | ||
| 196 (1 cross-sectional study) | ⊕○○○ Very lowc, d | OR 0.48 (0.24–0.96) | 323 per 1,000 | 137 fewer per 1,000 (220 fewer to 9 fewer) | |
| Duration of initial hospital stay | 1,673 (1 cohort study) | ⊕○○○ Very lowd | – | The mean length of initial hospital stay was 2.3 days | MD 0.2 days fewer (0.31 fewer to 0.09 fewer) |
| Breast milk feeding exclusively at discharge | 101,305 (1 cohort study) | ⊕⊕○○ Low | OR 7.76 (7.54–7.99) | 390 per 1,000 | 442 more per 1,000 (438 more to 446 more) |
| Adverse events – neonatal mortality | 4,271 (1 RCT) | ⊕⊕○○ Lowd, e | RR 1.01 (0.14–7.14) | 1 per 1,000 | 0 fewer per 1,000 (1 fewer to 6 more) |
| 136,468 (3 cohort studies) | ⊕⊕○○ Low | OR 0.51 (0.37–0.72) | 11 per 1,000 | 5 fewer per 1,000 (7 fewer to 3 fewer) | |
| 3,182 (1 cross-sectional study) | ⊕⊕○○ Low | OR 0.54 (0.32–0.92) | 25 per 1,000 | 11 fewer per 1,000 (17 fewer to 2 fewer) | |
| Adverse events – postpartum haemorrhage | 4,271 (1 RCT) | ⊕⊕○○ Lowd, e | RR 0.94 (0.77–1.16) | 83 per 1,000 | 5 fewer per 1,000 (19 fewer to 13 more) |
CI, confidence interval; MD, mean difference; OR, odds ratio; RCT, randomised controlled trial; RR, risk ratio.
aDowngraded one level for risk of bias: two studies were rated weak for confounders, one was rated weak for data collection, and two were rated weak for withdrawals and dropouts. One of the four studies was rated weak quality overall, while the other three were rated moderate quality overall.
bDowngraded one level for indirectness due to heterogeneity in feeding timings across studies.
cDowngraded one level for imprecision due to insufficient sample size.
dDowngraded one level for risk of bias: the only study in this category is rated high risk of bias or weak quality in two domains.
eDowngraded one level for imprecision due to wide confidence interval that suggests both harm and benefit.
Discussion
The evidence summarised in this systematic review suggests that early feeding, within one to 2 h of birth, may be associated with reduced risk of neonatal hypoglycaemia. However, this evidence is very uncertain due to heterogeneity across studies, insufficient sample sizes, and poor study quality. This finding differs from that of a 2020 review of preterm and low birth weight (LBW) infants fed with milk either “soon after birth” or delayed which found no evidence of a difference in hypoglycaemia between groups, in part due to very low event numbers [46]. However, infants at higher risk of hypoglycaemia, such as infants born preterm or LBW, are often admitted to the neonatal intensive care unit (NICU) and treated with intravenous dextrose and/or infant formula [6], and we were unable to assess, due to lack of data, whether infants who are at risk of hypoglycaemia benefit most from early feeding.
Nevertheless, we found that early feeding may provide other benefits and no evidence of harm. We found that early feeding may be associated with a reduced risk of neonatal mortality, consistent with findings of a previous review [47]. In our review, mortality was reported only in studies from low-resource settings, where neonatal mortality can range from 1.5 to 22.5% across hospitals [48] and up to 43% in all birth settings [49]. It may be that the extent to which early feeding is associated with neonatal mortality is relatively lower in high-income settings, but we have no data to verify this. Breast milk contains antibodies, lactoferrin, and oligosaccharides that have immunologic benefits [40] which may in turn contribute to reduced mortality risk.
We found that early breastfeeding, within an hour of birth, was associated with exclusive breastfeeding at 4 days of age. Although there are many benefits of breastfeeding, we found no evidence that the effect of early feeding on neonatal blood glucose concentrations differed between infants who were breastfed and infants who were bottle-fed. We also found that early feeding may slightly reduce duration of initial hospital stay, in line with findings from an earlier review, although this difference (0.2 days) is unlikely to be clinically important [46]. In addition, vaginal births are associated with a shorter initial hospital stay [50], and in the one study that reported this outcome, infants born vaginally were significantly more likely than those born by Caesarean section to feed within an hour of birth (87% vs. 61%, respectively) [33]. Therefore, it is possible that the relationship between early feeding and duration of initial hospital stay was confounded by mode of delivery.
A strength of this review is that, to our knowledge, no previously published review has synthesised the evidence on the effect of early feeding on neonatal hypoglycaemia and other clinically important health outcomes without restrictions on infant gestational age, birthweight, or language. Our review used a broad and highly sensitive search strategy to ensure studies from a range of settings were discovered.
However, despite inclusion of 19 studies and more than 175,000 participants, few studies reported on our primary outcome, neonatal hypoglycaemia, or on many of the other pre-specified outcomes. We hoped to compare data for Māori and non-Māori populations in New Zealand to understand whether early feeding unequally affects health outcomes, but we found no relevant data. Additionally, we were unable to carry out sensitivity analyses for GRADE outcomes with significant heterogeneity due to insufficient studies. The evidence base is also not geographically representative, with relatively few studies in high-income countries. Our wide inclusion criteria may have increased heterogeneity, including due to such factors as different definitions of hypoglycaemia, different timings of glucose measurement, and different feeding strategies. Because early feeding is widely considered beneficial for neonates, ethical issues restrict the conduct of RCTs, such as the randomisation of infants to a delayed feeding intervention. Consequently, almost all of the included studies were observational studies, reducing the certainty of the evidence base.
Conclusion
There is very low certainty evidence that early feeding may be associated with reduced neonatal hypoglycaemia and neonatal mortality. We consider that it is reasonable to support current recommendations for early feeding for the prevention of neonatal hypoglycaemia. However, were further studies to be undertaken, they should report outcomes in subgroups including babies of different ethnicity (Māori vs. non-Māori in New Zealand); gestational age; risk factors for hypoglycaemia; mode of birth; type of feeding; and birth setting.
Acknowledgment
We thank librarian Rayna Dewar at University of Auckland Library for supporting development of the search.
Statement of Ethics
An ethics statement is not applicable since this review is based exclusively on published literature.
Conflict of Interest Statement
The authors declare that they have no competing interests.
Funding Sources
This work was funded in part by grants from the New Zealand Ministry of Business and Employment (E.W.), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health (NIH R01HD091075, L.F.R.), the Health Research Council of New Zealand (19/690, J.E.H., C.A.C.), and the Aotearoa Foundation (9909494, L.L.). The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NICHD or the NIH.
Author Contributions
J.E.H., C.A.C., and L.L. planned the systematic review. L.F.R. designed the search strategy. L.F.R., E.W., and L.L. conducted title and abstract screening, full-text assessment, data extraction, quality and bias assessment, and assessment of certainty of the evidence. Z.W. conducted statistical analyses. L.F.R., L.L., and J.E.H. drafted the manuscript. All authors read and approved the final manuscript.
Funding Statement
This work was funded in part by grants from the New Zealand Ministry of Business and Employment (E.W.), the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health (NIH R01HD091075, L.F.R.), the Health Research Council of New Zealand (19/690, J.E.H., C.A.C.), and the Aotearoa Foundation (9909494, L.L.). The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. The content is solely the responsibility of the authors and does not necessarily represent the official views of the NICHD or the NIH.
Data Availability Statement
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.
Supplementary Material
References
- 1. Lucas A, Morley R, Cole TJ. Adverse neurodevelopmental outcome of moderate neonatal hypoglycaemia. BMJ. 1988;297(6659):1304–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Edwards T, Alsweiler J, Gamble G, Griffith R, Lin L, McKinlay C, et al. Neurocognitive outcomes at age 2 years after neonatal hypoglycemia in a cohort of participants from the hPOD randomized trial. JAMA Netw Open. 2022;5(10):e2235989. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. McKinlay C, Alsweiler J, Anstice NS, Burakevych N, Chakraborty A, Chase JG, et al. Association of neonatal glycemia with neurodevelopmental outcomes at 4.5 years. JAMA Pediatr. 2017;171(10):972–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Duvanel CB, Fawer CL, Cotting J, Hohlfeld P, Matthieu JM. Long-term effects of neonatal hypoglycemia on brain growth and psychomotor development in small-for-gestational-age preterm infants. J Pediatr. 1999;134(4):492–8. [DOI] [PubMed] [Google Scholar]
- 5. Harris D, Weston P, Gamble G, Harding J. Glucose profiles in healthy term infants in the first 5 days: the GLucOse in Well babies (GLOW) study. J Pediatr. 2020;223:34–41.e4. [DOI] [PubMed] [Google Scholar]
- 6. Harris D, Weston P, Harding J. Incidence of neonatal hypoglycemia in babies identified as at risk. J Pediatr. 2012;161(5):787–91. [DOI] [PubMed] [Google Scholar]
- 7. Alsweiler J, Harris D, Harding J, McKinlay C. Strategies to improve neurodevelopmental outcomes in babies at risk of neonatal hypoglycaemia. Lancet Child Adolesc Health. 2021;5(7):513–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Edwards T, Harding J. Clinical aspects of neonatal hypoglycemia: a mini review. Front Pediatr. 2020;8(8):562251. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. UNICEF . Guidance on the development of policies and guidelines for the prevention and management of Hypoglycaemia of the Newborn. Available from: https://www.unicef.org.uk/babyfriendly/wp-content/uploads/sites/2/2010/10/hypo_policy.pdf [accessed September 7, 2023]. [Google Scholar]
- 10. Queensland Health . Queensland clinical guidelines. Hypoglycaemia-Newborn. Guideline No. MN19.8-V12-R24 Available from: https://www.health.qld.gov.au/__data/assets/pdf_file/0043/881899/g-hypogly.pdf [accessed September 7, 2023].
- 11. British Association of Perinatal Medicine . Identification and management of neonatal hypoglycaemia in the full term infant: a framework for practice. Available from: https://www.bapm.org/resources/40-identification-and-management-of-neonatal-hypoglycaemia-in-the-full-term-infant-2017 [accessed September 7, 2023]. [DOI] [PubMed]
- 12. World Health Organization . Hypoglycaemia of the newborn: review of the literature. Available from: https://apps.who.int/iris/handle/10665/63362 [accessed September 7, 2023]. [Google Scholar]
- 13. Chertok IRA, Raz I, Shoham I, Haddad H, Wiznitzer A. Effects of early breastfeeding on neonatal glucose levels of term infants born to women with gestational diabetes. J Hum Nutr Diet. 2009;22(2):166–9. [DOI] [PubMed] [Google Scholar]
- 14. Sweet DG, Hadden D, Halliday HL. The effect of early feeding on the neonatal blood glucose level at 1-hour of age. Early Hum Dev. 1999;55(1):63–6. [DOI] [PubMed] [Google Scholar]
- 15. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372:n71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Veritas Health Innovation . Covidence systematic review software. Melbourne, Australia. Available from: https://www.covidence.org/. [Google Scholar]
- 17. Higgins JP, Altman DG, Gøtzsche PC, Jüni P, Moher D, Oxman AD, et al. The Cochrane Collaboration’s tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Effective Public Health Practice Project . Quality assessment tool for quantitative studies. Hamilton (ON); 2007. Available from: https://www.ephpp.ca/quality-assessment-tool-for-quantitative-studies/. [Google Scholar]
- 19. DeepL . Available from: https://www.deepl.com/translator.
- 20. Google Translate . Available from: https://translate.google.com/.
- 21. Schünemann H, Brożek J, Guyatt G, Oxman A. GRADE handbook for grading quality of evidence and strength of recommendations. 2013. Available from: https://gdt.gradepro.org/app/handbook/handbook.html [accessed May 12, 2023]. [Google Scholar]
- 22. GRADEpro GDT: GRADEpro guideline development tool. McMaster University and Evidence Prime; 2022. Available from: https://gdt.gradepro.org/app/#projects. [Google Scholar]
- 23. Review manager 5 (RevMan 5). Copenhagen: The Cochrane Collaboration; 2020. Version 5.4.1 ed. [Google Scholar]
- 24. R: A language and environment for statistical computing. Vienna, Austria: R Core Team; R Foundation for Statistical Computing; 2021. Available from: https://www.R-project.org/. [Google Scholar]
- 25. Zhang J, Yu KF. What’s the Relative Risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA. 1998;280(19):1690–1. [DOI] [PubMed] [Google Scholar]
- 26. The New Zealand Ministry of Health . Fetal and infant deaths web tool. Wellington: Ministry of Health; 2021. Available from: https://www.tewhatuora.govt.nz/our-health-system/data-and-statistics/fetal-and-infant-deaths-web-tool/#:∼:text=By%20ethnic%20group&text=In%20the%20previous%20five%2Dyear,3.6%20and%203.3%2C%20respectively. [Google Scholar]
- 27. Castro T, Grant C, Wall C, Welch M, Marks E, Fleming C, et al. Breastfeeding indicators among a nationally representative multi-ethnic sample of New Zealand children. N Z Med J. 2017;130(1466):34–44. [PubMed] [Google Scholar]
- 28. Bullough CHW, Msuku RS, Karonde L. Early suckling and postpartum haemorrhage: controlled trial in deliveries by traditional birth attendants. Lancet. 1989;2(8662):522–5. [DOI] [PubMed] [Google Scholar]
- 29. Salariya EM, Easton PM, Cater JI. Duration of breast-feeding after early initiation and frequent feeding. Lancet. 1978;2(8100):1141–3. [DOI] [PubMed] [Google Scholar]
- 30. De A, Biswas R, Samanta M, Kundu C. Study of blood glucose level in normal and low birth weight newborns and impact of early breast feeding in a tertiary care centre. Ann Niger Med. 2011;5(2):53. [Google Scholar]
- 31. Diwakar KK, Sasidhar MV. Plasma glucose levels in term infants who are appropriate size for gestation and exclusively breast fed. Arch Dis Child Fetal Neonatal Ed. 2002;87(1):F46–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Edmond KM, Kirkwood BR, Amenga-Etego S, Owusu-Agyei S, Hurt LS. Effect of early infant feeding practices on infection-specific neonatal mortality: an investigation of the causal links with observational data from rural Ghana. Am J Clin Nutr. 2007;86(4):1126–31. [DOI] [PubMed] [Google Scholar]
- 33. Franco-del Rio GR, Paredes-Melesio N. Impact of early initiation of breastfeeding on exclusive breastfeeding. Ginecología y obstetricia de México. 2022;90(7):551–8. [Google Scholar]
- 34. Iarŭkov A, Nin’o A, Iarŭkova N, Doĭcheva E, Kolev D. Ranno zakŭrmiane pri novorodenite [The early breast feeding of newborn infants]. Akush Ginekol. 1992;31(1):13–5. [PubMed] [Google Scholar]
- 35. Klemm R, Wu L, Christian P, Shamim AA, Shaikh S, Ali H, et al. Neonatal survival appears unaffected by timing in the start of breast-feeding in rural Bangladesh. FASEB J. 2014;28(S1):1015–5. [Google Scholar]
- 36. Kumar P, Narayan Singh S, Tripathi S, Kumar M. Incidence, predictors, and outcomes of hypoglycemia among at-risk neonates. Perinatology. 2021;22(3):171–7. [Google Scholar]
- 37. Mullany LC, Katz J, Li YM, Khatry SK, LeClerq SC, Darmstadt GL, et al. Breast-feeding patterns, time to initiation, and mortality risk among newborns in southern Nepal. J Nutr. 2008;138(3):599–603. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. NEOVITA Study Group . Timing of initiation, patterns of breastfeeding, and infant survival: prospective analysis of pooled data from three randomised trials. Lancet Glob Health. 2016;4(4):e266–75. [DOI] [PubMed] [Google Scholar]
- 39. Samayam P, Ranganathan PK, Kotari UD, Balasundaram R. Study of asymptomatic hypoglycemia in full term exclusively breastfed neonates in first 48 hours of life. J Clin Diagn Res. 2015;9(9):SC07–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40. Smith ER, Locks LM, Manji KP, McDonald CM, Kupka R, Kisenge R, et al. Delayed breastfeeding initiation is associated with infant morbidity. J Pediatr. 2017;191:57–62.e2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Zhou Y, Bai S, Bornhorst JA, Elhassan NO, Kaiser JR. The effect of early feeding on initial glucose concentrations in term newborns. J Pediatr. 2017;181:112–5. [DOI] [PubMed] [Google Scholar]
- 42. Sasidharan CK, Gokul E, Sabitha S. Incidence and risk factors for neonatal hypoglycaemia in Kerala, India. Ceylon Med J. 2004;49(4):110–3. [DOI] [PubMed] [Google Scholar]
- 43. Akter T, Dawson A, Sibbritt D. What impact do essential newborn care practices have on neonatal mortality in low and lower-middle income countries? Evidence from Bangladesh. J Perinatol. 2016;36(3):225–30. [DOI] [PubMed] [Google Scholar]
- 44. Jatav A, Verma YS, Gaur A. Neonatal hypoglycemia: a problem not well addressed in postnatal wards. Int J Pharm Clin Res. 2023;15(4):756–62. [Google Scholar]
- 45. Hamadeh N, van Rompaey C, Metreau E, Eapen SG. New World bank country classifications by income level: 2022-2023: World bank blogs. Available from: https://blogs.worldbank.org/opendata/new-world-bank-country-classifications-income-level-2022-2023 [accessed July 30, 2023]. [Google Scholar]
- 46. Walsh V, Brown JVE, Copperthwaite BR, Oddie SJ, McGuire W; Cochrane Neonatal Group . Early full enteral feeding for preterm or low birth weight infants. Cochrane Database Syst Rev. 2020;12(12):CD013542. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Smith ER, Hurt LS, Chowdhury R, Sinha B, Fawzi WW, Edmond KM, et al. Delayed breastfeeding initiation and infant survival: a systematic review and meta-analysis. PLoS One. 2017;12(7):e0180722. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Narayanan I, Nsungwa-Sabiti J, Lusyati S, Rohsiswatmo R, Thomas N, Kamalarathnam CN, et al. Facility readiness in low and middle-income countries to address care of high risk/small and sick newborns. Matern Health Neonatol Perinatol. 2019;5:10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. World Health Organization . Fact sheets: newborn mortality. 2022. Available from: https://www.who.int/news-room/fact-sheets/detail/levels-and-trends-in-child-mortality-report-2021 [accessed August 2, 2023]. [Google Scholar]
- 50. Soto Vega E, Marta U-O, Fernando A-V, Yazmin LBI, Hazel HRC. The epidemic of the cesarean section in private hospital in Puebla, México. Obstet Gynecol Int J. 2015;2(6):184–7. [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.