Summary of main results

The trial data included in this review provide low‐certainty evidence that delaying the introduction of progressive enteral feeds beyond about four days after birth may not reduce the risk of NEC in very preterm or VLBW infants. The boundaries of the 95% CI for the estimate of effect are consistent with either two fewer cases or three more cases of NEC in every 100 infants who have delayed introduction of progressive enteral feeds. Meta‐analysis of data from these trials did not show evidence of an effect on all‐cause mortality, with the 95% CI boundaries being consistent with either three fewer or three more deaths in every 100 infants who had delayed introduction of progressive enteral feeds. Prespecified analyses did not show subgroup effects on risk of NEC or death among infants with growth restriction or evidence of absent or reversed end‐diastolic flow velocities in the fetal aorta or umbilical artery. Meta‐analysis showed that delayed introduction of progressive enteral feeds may result in a slight reduction in feed intolerance. Data from seven trials showed a higher risk of late‐onset infection among infants who had delayed introduction of progressive enteral feeds. The point estimate suggests that an extra episode of late‐onset infection occurs for every 10 infants who have delayed rather than early introduction of progressive enteral feeds. None of the included trials has reported effects on long term growth or neurodevelopmental outcomes.

Overall completeness and applicability of evidence

These data are relevant to current practice since most of the included trials were conducted since the year 2000. Six of these trials (767 infants) specifically recruited infants thought to be at high risk of developing NEC due to intrauterine growth restriction and abnormal fetal circulatory distribution or flow. This widens the applicability of the findings since this is the population for which most clinical uncertainty and variation in practice with regard to early feeding strategies exists (Klingenberg 2012). Previously, this population of infants has been specifically excluded from participating in many trials of early enteral feeding practices (Tyson 2007).

Artificial formula feeding increases the risk NEC (Quigley 2019). The risk‐benefit balance of enteral feeding strategies may differ between human milk‐fed and formula‐fed very preterm or VLBW infants (Young 2020). Subgroup analyses at trial‐level by type of milk did not show evidence of differences in effect. Most trials, however, included infants who received either human milk or cow milk formula or both, and subgroup data were not reported. If such subgroup data were available, an individual participant‐level meta‐analysis could be conducted to explore this issue further. It is also unclear whether the findings can be applied to infants who receive continuous infusion of intragastric feeds, as most of the infants in the included trials received enteral feeds as interval gastric boluses. Randomised controlled trials have reported conflicting findings about the effect on continuous enteral infusion on feed tolerance in VLBW (and especially ELBW) infants (Premji 2021).

The included trials were mainly undertaken in neonatal care centres in middle‐ or high‐income countries. It is less clear how applicable this evidence is to neonatal care practices in low‐income countries. Conservative strategies, such as delayed introduction of enteral feeds, may confer substantial nutritional disadvantage in settings with less technologically developed healthcare provision where adjunctive parenteral nutrition is not readily and safely available (Akindolire 2020). In some low‐ or middle‐income countries where severe infection is a much more important cause of mortality and morbidity, the nutritional and immunological advantages of early feeding, particularly with breast milk, may outweigh any risks associated with enteral feeding for very preterm or VLBW infants (de Silva 2004). A recent Cochrane Review has assessed the data from six trials undertaken in India since the late 2000s that compared exclusive enteral feeding (no parenteral fluid) from birth with gradual introduction of enteral feeds over several days in VLBW infants with birth weight greater than 1000 grams (Walsh 2020). While these trials were not eligible for inclusion in this review, none found evidence of an effect on NEC or other adverse outcomes.

Quality of the evidence

The GRADE‐assessed certainty (quality) of evidence for primary outcomes was downgraded because of lack of masking in the included trials and imprecision of estimates of effect (summary of findings Table 1). Although these trials were otherwise of good methodological quality, in common with other trials of feeding interventions in this population, it was not possible to mask parents, caregivers and clinical assessors to the nature of the intervention. Lack of masking may have resulted in surveillance and ascertainment biases. It is more likely, however, to have caused an overestimation of feed intolerance and NEC among infants whose feed volumes were advanced faster. Assessment of abdominal radiographs for signs of NEC was masked in some trials to try to ensure that the diagnosis of severe NEC (confirmed by radiological detection of gas in the bowel wall or portal tract) was not prone to bias. As microbial generation of gas in the bowel wall is substrate dependent, however, infants who received more enteral milk (substrate) may have been more likely to demonstrate this radiological sign than infants with equally severe bowel disease who had less intraluminal substrate. This 'substrate effect' is also more likely to cause over‐ascertainment of NEC among infants who had faster rates of feed volume advancement (Tyson 2007).

The other reason for downgrading the certainty of evidence was the existence of substantial imprecision in estimates of effect, with meta‐analyses generating 95% CI that included benefit as well as no benefit or harm. Although the total number of participants in the 14 included trials was more than 1500, not all trials contributed data to all outcome estimates, and estimates of effect were consequently imprecise, especially for less common outcomes including NEC and mortality.

The definition of delayed introduction of progressive feeds varies between subpopulations of very preterm or VLBW infants who have different empiric risks for developing feed intolerance and NEC. The effects of enteral feeding are likely to be very different, for example, for an inotrope‐supported infant of birth weight less than 750 grams compared with a clinically stable infant of birth weight greater than 1000 grams. For this Cochrane Review, we defined delayed introduction as later than four days after birth since some observational studies have found the risk of NEC to be lower when feeds are introduced five to seven days after birth (Patole 2005). For ELBW or extremely preterm infants, it may be more appropriate to define delayed introduction as more than seven days after birth (or even later) since small‐intestinal motility is poorly organised before about 28 weeks' gestation resulting in a high risk of feed intolerance. In addition, enteral feeds are often delayed in this population because of respiratory or metabolic instability or because of other putative risk factors for NEC, such as the existence of a patent ductus arteriosus, the use of non‐steroidal anti‐inflammatory drugs or the presence of an umbilical arterial catheter (McGuire 2004).

Potential biases in the review process

The main concern with the review process is the possibility that findings are subject to publication and other reporting biases (Hopewell 2009). Data from trials which show statistically significant or potentially important effects tend to be more readily available for inclusion in meta‐analyses (Gale 2020). Publication bias, as well as other sources of small‐study bias, can inflate effect size estimates in meta‐analyses of interventions to improve outcomes in very preterm or VLBW infants (Young 2021). The Cochrane Review of probiotics to prevent NEC in very preterm or VLBW infants, for example, shows a large reduction in the risk of NEC, but the funnel plot and regression analysis indicate that publication bias is likely to have inflated the pooled effect size estimate (Sharif 2020). We attempted to minimise this threat by screening the reference lists of included trials and related reviews and searching the proceedings of major international perinatal conferences to identify trial reports that are not published in full form in academic journals. Inspection of funnel plots of meta‐analyses that included at least 10 data points did not show sufficient asymmetry to raise concerns about possible publication or small study bias.

In six trials (accounting for about half of all the participants), all participating infants needed to have evidence of intrauterine growth restriction or circulatory redistribution demonstrated by absent or reversed end‐diastolic flow velocities in the fetal aorta or umbilical artery (Abdelmaaboud 2012; Arnon 2013; Davey 1994; Karagianni 2010; Leaf 2012; Srinivasan 2017). Most infants who participated in these trials were very preterm or VLBW infants, but subgroup data were not available. We included data from these trials since intrauterine growth restriction or circulatory redistribution demonstrated by absent or reversed end‐diastolic flow velocities in the fetal aorta or umbilical artery have been associated with a high risk of developing NEC and associated complications, and the findings are applicable to enteral feeding policies and practices (Embleton 2017).

Agreements and disagreements with other studies or reviews

This review focused specifically on the comparison of delayed versus early introduction of progressive enteral feeds. Other Cochrane Reviews have assessed how (i) enteral fasting versus trophic feeding (minimal enteral nutrition), (ii) slow versus faster rates of feed volume advancement, and (iii) early full enteral feeding versus gradual introduction of feeds affects important outcomes in very preterm or VLBW infants (Morgan 2013; Oddie 2017; Walsh 2020). These reviews, consistent with the findings of this review, have found evidence that conservative feeding strategies probably do not reduce the risk of NEC, mortality, or associated morbidity.