Nutritional care practices for very preterm (VP) infants have changed dramatically over the past several decades. Changes have mirrored scientific discovery with increased attention to the health benefits of nutrient density, the immunobiology of human milk, and the role of nils per os (NPO) in disease risk. Preclinical and clinical observational study and trial data have shaped neonatal nutrition practices, but questions remain, and, therefore, significant clinical variation persists in enteral feed initiation, composition, and method of delivery [1]. Healthcare providers must appraise the growing evidence as to how enteral feeding parameters relate to outcomes such as infant growth and risk for necrotizing enterocolitis (NEC) to develop an evidence-based best practice approach to optimize VP infant outcomes. Specifically for feed initiation, consideration of feed type [mother’s own milk (MOM), donor human milk (DHM), formula], time of initiation, and method by which feeding is initiated is complex as these decisions are distinct yet interdependent. For example, evidence-based “best practice” to feed as soon as possible and to feed MOM become competing interests since most mothers of VP infants experience delays in lactation [2,3,4,5] which limits MOM availability in the first postnatal days. When DHM is available, human milk feeds are available immediately post-birth, but there is no clear evidence as to which is more advantageous to the infant—starting feeds immediately with DHM to minimize the NPO duration or continuing NPO until the critical gut immunoregulatory components of maternal colostrum are available. This perspective is written to present a non-systematic literature review of the evidence that has been utilized to guide nutrition approaches for two centers who prioritize optimizing infant nutrition and MOM feeding.

The preterm infant gastrointestinal tract (GIT) is uniquely sensitive to imbalances in immunoregulation. As the largest immune organ, ontogeny of the GIT includes development of complex pro- and anti-inflammatory pathways to protect it from pathogens and toxins [6]. Bowel rest or NPO status was initially preferred for VP infants as a method to avoid taxing the immature metabolic processes of the GIT. However, studies of GIT maturation, as measured by intestinal permeability, show that the NPO GIT has marked immaturity compared to a fed GIT [7, 8]. MOM, especially colostrum, contains many immunomodulators and growth factors which contribute to intestinal growth and maturation and protect from NEC. MOM-fed preterm infants exhibit earlier GIT maturation compared to formula-fed infants in a dose-dependent manner [9]. This and other benefits of MOM feeding make it the ideal feeding type for VP infants. Although clinical trials demonstrate a significantly lower rate of NEC when DHM is given rather than formula [10], no studies have been performed to show whether early DHM is beneficial compared to NPO when MOM is not available in the early postnatal days.

When to initiate feeds

Over the past 30 years, postnatal age for initiation of enteral feedings in VP infants has decreased significantly, with units of measure changing from days and weeks to hours and days. A recently updated meta-analysis of randomized controlled trials compared delayed introduction of progressive feeds, defined as 4–7 days, with earlier introduction of progressive feeds, defined as less than 4 days. No difference in NEC or mortality was observed between comparison groups. Feeding intolerance risk was found to be reduced [relative risk 0.81 (95% confidence interval (CI) 0.68–0.97)] with delayed introduction of progressive feeds, but invasive infection risk was higher [relative risk 1.44 (95% CI 1.15–1.80)] [11]. With numerous neonatal intensive care units (NICUs) prioritizing efforts to reduce central line associated blood stream infections (CLABSI), many NICUs have adopted a strategy to initiate feeds by 4 postnatal days. A notable exception in the contemporary literature is one standardized slow enteral feeding (SSEF) protocol in which no enteral feeding is given for the first 10–14 days for infants with birthweight less than 750 g [12]. While this case/control study conducted over 12 years demonstrated the SSEF protocol was associated with a reduction in NEC and NEC/death outcomes, its results would require confirmation via randomized controlled trials conducted in the setting of current nutritional care practices which prioritized enteral nutrition and MOM feedings before implementation. In contrast to the SSEF protocol, adoption of early feeding initiation, based on meta-analyses of randomized controlled trials, has not led to increased NEC or mortality in most NICUs. In fact, one NICU’s quality improvement work led to initiation of feedings by a median of 14 rather than 33 postnatal hours, and this change was associated with less CLABSI, less feeding intolerance, higher weight gain velocity, and less death with earlier median initiation of feeds, although causation could not be shown in this cohort study [13].

Another consideration in initiation of feeds is whether progressive or trophic feeds are initiated. Historically, trophic feeds or minimal enteral nutrition (MEN) were considered a way to avoid the atrophy associated with an NPO gut while giving time for postnatal metabolic maturation prior to advancement of feed volume. MEN has been defined as <20 or 12–24 mL/kg/day and is commonly continued for several days. A recent study conducted in the United States randomized 60 extremely preterm infants (<29 weeks gestation) to MEN for 4 days versus progressive enteral feedings without a MEN phase [14]. Trial results demonstrated shorter duration to achieve full enteral feedings, TPN and central line and a greater number of days receiving full enteral feeds in the first month of life with progressive feds versus MEN; however, both randomized groups initiated enteral feeds at 3 (1–3) days. Meta-analysis of nine trials comparing MEN demonstrated no benefit or harm [15]. With the potential benefit of progressive feeds, MEN has been phased out or the duration of MEN has decreased in many NICUs.

Although VP infants have initiation of feeds prior to 4 days in most NICUs, investigation of the best time, such as postnatal day 1, 2, 3, or 4, has had minimal research. Preclinical studies in mammals, specifically preterm piglets indicate that reduction in mucosal mass occurs within 48 h of NPO with total parenteral nutrition (TPN) while villus atrophy is complete within 72 h [16]. In humans, only two retrospective cohort studies are available. In one cohort study [17] comparing initiation of feeds before or after 72 h post-birth in infants born less than 33 weeks’ gestational age, infants fed earlier were larger and more mature, had lower severity of illness acuity scores, were less likely to be growth restricted, and had less exposure to antibiotics. The late group was more likely to receive MOM as their first feeding at 82% compared to 33%. Despite the higher intake of MOM, infants in the late group were more likely to have retinopathy of prematurity and chronic lung disease after adjusting for confounders. The proinflammatory marker, fecal IL-8, was higher, and anti-inflammatory marker, fecal IL-10:IL-8 ratio, was lower in the late compared to the early group after adjustment for potential confounders [17]. In a second retrospective cohort study [18] of infants with birthweights 1000–1500 g, early initiation was defined as <48 h and late as >72 h. The early group exhibited shorter time to return to birthweight, shorter length of stay, and fewer central line days but did not exhibit a difference in sepsis rate [18].

Even though initiation of feeds on the first postnatal day has been minimally investigated, three randomized trials of exclusive enteral feedings have been conducted in low resource areas, enrolling larger very low birthweight infants, weighing 1–1.5 or 1.2–1.5 kg [19,20,21]. In these studies, feeds were initiated in the first 24 h post-birth in both groups; however, the intervention group avoided all intravenous fluids or parenteral nutrition by the initiation of 60–80 mL/kg/day or exclusive enteral nutrition (MOM or formula) on the first postnatal day. In comparison, the control groups received parenteral nutrition along with initiation of 20–30 mL/kg/day of enteral feeds on the first day. In all studies, the exclusive enteral nutrition group demonstrated early attainment of full feeds and decreased length of stay without an increase in GIT pathology, although these studies were underpowered for NEC as an outcome. These studies point to the potential for VP infants to tolerate feeds early after birth despite the lack of evidence comparing the time of initiation of feeds within the first 4 days.

What to feed

Since VP infants do not have sufficient oral skills to feed by mouth at birth, they are not able to breastfeed initially. Therefore, parents of VP infants rely on breast pumping and hand expression to obtain MOM, and the volume of MOM is typically low until secretory activation occurs around postpartum day 4 [22]. Due to difficulties in obtaining MOM in the first postnatal days for VP infants, prioritization of the initiation of progressive feeds in the first postnatal day decreases the likelihood of the first feeds being MOM, specifically early MOM or colostrum. This is a concern as preclinical studies indicate that colostrum is critical to early GIT maturation and to counteracting unchecked inflammation. MOM provides immunologic, anti-inflammatory and growth factors that are absent in the VP infant GIT and are essential to the developing immature GIT. A recent study randomly allocated preterm piglets to a diet of bovine colostrum, DHM or formula initiated on the first postnatal day that was gradually advanced over the first 11 days [23]. Preterm piglets that received an initial diet of colostrum demonstrated superior outcomes, with better GIT growth and enzymatic function, greater absorptive capacity, higher bacterial diversity, and lesser GIT inflammation than formula- or DHM-fed piglets. The GIT responses in DHM-fed piglets were generally intermediate between those in colostrum- and formula-fed piglets.

These differences in outcomes may result from the numerous differences in bioactive factors between colostrum and DHM, due to the stage of lactation during which DHM is typically donated (mature milk) and the pasteurization process which reduces many bioactive components and abolishes live cells [24,25,26,27,28,29,30]. Notably, provision of colostrum may be particularly important for the VP infant who misses the third trimester transplacental transfer of nutrients and immunoglobulins as well as the trophic and protective effects of amniotic fluid [31, 32] due to preterm birth. The composition of colostrum and the trajectory of immunologic and nutritive changes that occur longitudinally during lactation suggest that colostrum has a primary immunological and trophic role rather than a nutritive role [3, 33,34,35]. Colostrum contains greater concentrations of immunoglobulin, growth factors, lactoferrin, leukocytes and stem cells than mature MOM [36, 37]. The milk microbiome in colostrum serves as a programming function for the infant gut microbiota [38], which is eradicated in DHM due to pasteurization. In addition, early colostrum feedings correspond to the increased permeability of the immature GIT after birth that is conducive to transfer of cellular components and macromolecules [39]. Preclinical piglet studies demonstrate colostrum feedings result in maximal macromolecular absorption in the first 24 h after birth compared to mature milk feeding and that lack of colostrum feeding may result in an earlier cessation of macromolecule transport or “intestinal closure” [40]. These preclinical and cohort studies must be considered when determining the risks and benefits to the VP infant GIT and overall health between awaiting colostrum or avoiding NPO by initiating early feeds with DHM (Fig. 1).

Fig. 1: Effects of Colostrum vs NPO on the very preterm infant’s gastrointestinal tract.
figure 1

The figure illustrates the effects of maternal colostrum feeds (left panel) on the neonatal GIT, including supporting commensal microbiota through human milk oligosaccharides (HMO), immunomodulation, growth and repair of gut mucosa through growth factors (e.g. EGF) and stem cells, enhancing tight junctions through upregulation of occludin, protection from pathogenic bacteria through immunoglobulins (Ig), lactoferrin (LF) and leukocytes, and upregulating enzymes to enhance nutrient absorption. The figure illustrates the effects of maintaining NPO (right panel) on the neonatal GIT, including bacterial overgrowth with potentially pathogenic bacteria, mucosal atrophy and flattened villi, increased intestinal permeability, and reduced nutrient absorption. Adapted from “Vitamin D Deficiency Impacts Immunity in the Gut”, by (2022). Retrieved from

Although DHM does not offer all the bioactives of MOM and especially colostrum, when MOM is not available, DHM is the preferred feeding for VP infants. Only one contemporary small trial has evaluated VP infant health outcomes based on randomized diet of DHM versus formula without any MOM [41]. A 2018 Cochrane review including that trial and trials where DHM and formula are provided as supplements to MOM demonstrated that while infants have better growth on formula compared to DHM, infants fed formula had significantly higher risk of NEC [10].

How to feed

Significant variation exists in how nasogastric enteral feedings are provided to VP infants, specifically whether in continuous or bolus form and the frequency of feedings. Potential advantages and limitations of each approach exist with a small number of trials and experiments providing data to guide practice.

Potential benefits of continuous feedings include reduced feeding intolerance associated with the administration of smaller volumes in the setting of reduced gastrointestinal motility and improving nutrient absorption [42]. However, nutrient loss of calcium, phosphorus and fat were noted when continuous feedings were provided due to adherence to the tubing [43]. This nutrient loss can be reduced by using air to clear the milk from the tubing. Other potential concerns with continuous feedings include alteration of the normal cyclical pattern of release of GIT hormones and exacerbation of gastroesophageal reflux and aspiration. A recent review of nine RCTs concluded that infants receiving continuous feeding may reach full enteral feeding slightly later than infants receiving intermittent bolus feeding (mean difference 0.84 days, 95% CI –0.13 to 1.81), although the evidence is of low certainty [42]. No other differences in clinical outcomes were noted between continuous and bolus feedings. Based on this low quality of evidence and lack of harm demonstrated with bolus feeds, bolus feedings are generally recommended for VP infants [44]. However, practice may vary based on specific circumstances, such as hypoglycemia or severe feeding intolerance which may be ameliorated with continuous feedings.

The frequency of bolus feedings, specifically every 2 or 3 h, has been investigated to determine which is optimal for VP infants. Theoretical benefits associated with 2-h feedings include smaller feeding volumes which may be better tolerated, potentially resulting in fewer episodes of withholding feedings due to feeding intolerance and reaching full enteral nutrition sooner. In contrast, 3-h feedings may be beneficial to the infant as they result in greater time for gastric emptying, fewer interruptions and longer sleep cycles, along with reduced nursing workload. Six studies have compared 2- versus 3-h feeding intervals, mostly in larger VP infants (birthweights 1000–1750 g) [45,46,47,48,49,50], with only one study including infants with birthweight ≤1000 g [50]. Due to the inability to blind feeding intervals, all studies had potential bias. A recent systematic review and meta-analysis included these six studies, resulting in 872 subjects [51]. Overall, no significant difference in time to achieve full enteral feeds was found as well as no significant differences in other clinical outcomes including weight gain, duration of intravenous fluids, hypoglycemia, apnea, feed intolerance, or duration of hospitalization. Subgroup analysis of 84 infants with birthweight 501–1000 g demonstrated a shorter time to reaching full enteral feeds by 2.90 (95% CI 1.16–4.64) days with 2-h feedings, although the small size of this subgroup analysis limits the generalizability of the result [51]. Without larger studies, especially in the smallest and most immature infants, there is limited data to guide bolus feeding interval selection.

Special considerations

Intrauterine growth restriction

Studies in infants who experienced intrauterine growth restriction have demonstrated no benefit to delaying feeding initiation. A randomized, controlled trial of 404 infants who were born less than 35 weeks’ gestational age with birthweight below the 10th centile and with abnormal antenatal umbilical artery Doppler waveforms compared introduction of feeds by Day 2 versus Day 6. The cohort fed earlier exhibited early attainment of full, sustained enteral nutrition with no difference in NEC [52].

Severity of infant illness at birth

NICU feeding protocols commonly have specific exclusion criteria for feeding, including criteria for withholding the initiation of feeds. Unfortunately, minimal evidence is available to provide guidance in development of these criteria and, therefore, this portion of NICU feeding protocols relies heavily on healthcare provider experience and consensus. Common considerations include cardiovascular compromise, especially the requirement for vasoactive medications, perinatal hypoxic ischemic insult, and persistent hypoxia. Other less often considered factors that may influence GIT integrity and feeding tolerance include (1) antenatal conditions, such as preterm premature rupture of membranes, chorioamnionitis, funisitis, maternal inflammation, fetal inflammation, and fetal growth restriction, and (2) antenatal management, such as antenatal antibiotic and steroid therapy. These factors may impact GIT microbiome, motility, and may lead to abnormal GIT development and immunity [53,54,55,56,57,58]. While it would not be pragmatic to create specialized feeding protocols for each of these conditions, consideration of antenatal events may be advisable when determining tolerance of feeding protocols.


The WHEN and WHAT to feed VP infants are closely interrelated as the availability of MOM versus DHM may impact timing of feed initiation. While avoiding NPO status for 24–48 h by initiating feedings with DHM may protect against gut atrophy, this approach may prevent infants from reaping the benefits of colostrum as their first feeds. Colostrum as the first feed in preclinical studies is associated with maximal macromolecular absorption in the first 24 h after birth compared to mature milk feeding, potentially limiting absorption of bioactive and cellular components found in colostrum. Further research and clinical work are needed to afford colostrum for VP infants in the first postnatal hours (Table 1). Until that goal is routinely reached, NICU care teams must weigh the benefits of colostrum as the first feed versus the benefits of avoiding NPO status with DHM to develop evidence-informed approaches to nutritional practice. Examples of two feeding protocols from two NICUs which place high value on optimizing infant nutrition and MOM feeding demonstrate significantly varying approaches to handling this common situation (Table 2). These differ in volume, frequency and duration of trophic and initial feeding protocols. For example, the Yale protocol utilizes a 4-h feeding interval for the smallest infants due to slow gut motility in the most immature infants. These protocols highlight the significant variation in this universal and important practice of initiating enteral feedings in high-risk infants in the NICU.

Table 1 Research gaps.
Table 2 Comparison of two clinical protocols for very preterm infant feed initiation.