Introduction of Enteral Feeds in Preterm Infants

To the Editor. —In the context of major uncertainty about the optimum approach to enteral feeding in the preterm infant, Berseth et al1 addressed one important issue: that of the rate of feed advancement. We are concerned that the results of this study, which has major limitations, will be overinterpreted as to its implications for clinical practice.The significant and competing risks of necrotizing enterocolitis (NEC) with milk feeds and line-related sepsis from the use of parenteral nutrition prior to the attainment of full enteral feeds are poorly quantified by existing data. A recent snapshot survey in Scotland over 1 week (14 units, 81 infants with a median birth weight of 1080 g [range: 575–2230 g] and gestational age of 29 weeks [range: 24–32 weeks]) showed variation in practice ranging from complete parenteral feeding for the first week to establishment of full feeds within a few days, with a significant difference between centers in time to full feeds even after allowing for case mix.2In the Berseth et al article, although the age of starting enteral feeds is not specified, the data in Table 2 suggest that day 10 or 11 was the earliest time of milk feeding, with some infants starting feeds as late as day 31 (which is very different from practice in any of the centers in the above-mentioned survey). We cannot understand the cause of such delays in a center that seems to support minimal enteral feeding. The conclusions are thus inapplicable in the many centers in which practice falls outside the boundaries of the two arms of this study.It is also unclear why this study explicitly excluded small-for-gestational-age infants (a group with a high incidence of NEC) rather than using stratification or minimization to prevent any confounding affect of growth restriction.The methodology for study-group assignment is unorthodox and inadequately explained (drawing lots suggests a method open to bias).The use of expressed breast milk (a factor thought to have a major influence on the incidence of NEC) is not described adequately: Was donor breast milk used in addition to the mother’s own milk, and was there stratification for infants receiving solely breast milk, formula only, or a combination?Infants assigned to minimal volumes had a higher incidence of patent ductus arteriosus and a slightly higher incidence of antenatal steroid usage (their Table 1). The impact of these factors, along with the use of breast milk, on the incidence of NEC could have been examined further by multiple-regression analysis.The study was stopped early on the dubious basis of a significant statistic in a one-tailed Fisher’s exact test, with its greater inherent likelihood of showing significance, although there is no clear plausibility that the effect of the interventions was going to be in a particular direction. It is thus an overinterpretation to report that “advancing feeding volumes increase the risk of NEC,” particularly when the study stopped well short of the recruitment required by the power calculation.3 It is of interest that the more important combined outcome of NEC and death showed no significant difference between groups.Units with a faster feeding schedule than the “advancing” regimen in this study (including our units) can quote a lower incidence of NEC; could the high incidence of NEC in this study be related to the gastrointestinal effects of prolonged gut disuse or to the infrequent use of breast milk? It should be noted that, in the article by Lucas and Cole4 on milk type and NEC, the speed of milk advance was associated with incidence of NEC only in those infants who were formula fed.It is arguable that the mode of presentation of NEC as opposed to its incidence, and in particular the development of pneumatosis, may depend on the amount of milk substrate for gas-forming organisms within the gut and hence the rate of advancement of feeds.5 Thus, some more pragmatic outcome than Bells stage II NEC needs to be looked at.Nosocomial sepsis is an important outcome in such a feeding study (slow feeding leads to longer duration of intravenous line and parenteral nutrition use and may lead to poor maturation of the gut barrier). Studies of long-line–related sepsis suggest that the risk of line colonization, and hence infection, is directly related to the duration of use of such a line,6 and slow milk advancement would thus present a cumulative risk. Unfortunately, this article does not define “late sepsis,” and the lack of difference between groups could potentially be the result of loose interpretation of this (failure to exclude colonization/contamination).A large, multicenter, randomized, controlled trial is needed, comparing different rates of feed advancement in very low birth weight infants. Such a study should 1) be based on information about the range of currently accepted practice, 2) be sufficiently powered to look at a pragmatic but important outcome, namely death or disability at follow-up but also allow subanalysis for NEC and nosocomial infection (clearly defined), and 3) stratify infants according to type of milk such that the optimum speed of feed advancement can be defined for infants fed breast or formula milk.