Beyond Necrotizing Enterocolitis Prevention: Improving Outcomes with an Exclusive Human Milk–Based Diet

Prematurity remains the leading cause of neonatal mortality in the United States. One of the most difficult aspects of the management of the premature infant is related to nutrition and gastrointestinal complications such as necrotizing enterocolitis (NEC) and intestinal perforation. The development and progression of NEC is known to cause significant morbidity and mortality. However, the exact pathophysiology of NEC has yet to be defined, thus making preventive strategies more difficult.^1–3 Epidemiology shows that the highest-risk infants are those weighing <1,250 g, born at a gestational age of <28 weeks, and with congenital heart defects.¹ It has been long established that breast milk is the best source of nutrition for all infants when it is safe and available. The continued gain in knowledge over the years has led the American Academy of Pediatrics (AAP) and Surgeon General to recommend in published statements the use of human milk (HM), including donor milk when needed, for premature infants.^4,5

Over the last 10 years, the implementation of donor HM for very low birth weight (VLBW) infants has become more widespread. Hair et al. demonstrated that the use of an exclusive human milk–based diet (HUM) using mother's own milk, donor HM, and a donor HM-derived fortifier was safe and did not hinder growth of infants.⁶

There may be multiple benefits in using HM for premature infants such as lower rates of NEC, late-onset sepsis, feeding tolerance, and mortality, along with improved neurodevelopmental outcomes.^7–11 Sullivan et al. demonstrated, in a prospective, randomized trial, that the use of breast milk and donor HM with a donor HM-derived fortifier was associated with a reduction in NEC.¹² Given the increasing research showing reduction in NEC rates with an exclusive HUM diet, this study sought to evaluate this trend in multiple centers. The objectives were to determine whether these trends hold true across four Level 3 Neonatal Intensive Care Units at different hospitals in the United States and to evaluate other important outcomes for extremely premature infants.

In a multicenter retrospective cohort study, infants who received a bovine-based diet (BOV) of mother's milk fortified with bovine fortifier and/or preterm formula were compared to infants who received an exclusive HUM diet feeding protocol. Infants with a birth weight (BW) <1,250 g were included. Infants were excluded if they had major congenital anomalies, died within the first 12 hours of admission, or if they were transferred in from an outside hospital after 1 week of age. Data were collected before and after the introduction of a HUM diet. For each institution below, the specific feeding protocol and time frame for data collection are detailed. In the BOV group, infants preferentially received mother's milk fortified with bovine fortifier over formula. Primary outcomes were NEC and mortality. Secondary outcomes included late-onset sepsis, retinopathy of prematurity (ROP), bronchopulmonary dysplasia (BPD), patent ductus arteriosus (PDA), intraventricular hemorrhage (IVH), ventilator days, and postmenstrual age (PMA) at discharge.

The four centers had 1,587 infants eligible for inclusion in the study. There were 768 infants in the BOV group and 819 infants in the HUM group. The number of infants from each study site were as follows: Baylor College of Medicine 978, Good Samaritan San Jose 91, Northwestern 317, and Winnie Palmer Hospital for Women and Babies 201. The characteristics of the two groups were similar (Table 1). However, there was statistically significant difference in BW and birth head circumference. The BOV group had a 5% lower percentage of mothers receiving antenatal steroids. There were no differences in APGAR scores between the two groups (Table 1).

The primary outcomes, NEC and mortality, were statistically significant (Table 2). The BOV group mortality rate was 17.2% (132/768), and the HUM group rate was 13.6% (111/819; p = 0.04). There was a reduction in NEC and/or death (p = 0.0002) that persisted when adjusted by study site. There was a significant reduction in NEC cases in the HUM group (Fig. 1). The number needed to treat to prevent one case of NEC is 10. With the degree of reduction in NEC cases, a subgroup analysis of NEC patients was completed (Table 3). There was a significant reduction in medical and surgical NEC (p < 0.00005 and p < 0.00002). The distribution of the number of NEC cases by gestational age is shown in Figure 2.

The secondary outcomes considered include growth parameters as well as variables associated with neonatal morbidity (Table 4). There was no statistical increase in weight gain, head circumference growth, or being SGA at the time of discharge (Table 2). There was no significant decrease in severe IVH, length of stay, or PMA at discharge. However, there was a significant reduction in the incidence of late-onset infection, ROP, presence of PDA, and BPD (Table 4). Using the Wilcoxon rank sum test, there was a reduction in ventilator days from a mean of 32 days with a median of 17 days to a mean of 29 days with a median of 9 days (p = 0.003).

Both logistic (categorical outcomes) and multiple linear regression (quantitative outcomes) models were used to adjust for the use of antenatal steroids. The model consisted of the particular primary or secondary outcome as the dependent variable, and study group (HUM or BOV) along with antenatal steroid use as the independent variables. Significance was lost only for mortality (p = 0.27). All other outcomes remained statistically significant. These models were also used to adjust for study site, and all variables remained significant.

It was found that the use of an exclusive HUM diet in extremely premature infants (<1,250 g BW) decreased the incidence of both medical and surgical NEC, further supporting the results of Sullivan et al. in regards to reduction of NEC.¹² This study also shows an association with a reduction in mortality and late-onset infections. An association was demonstrated with decreased ROP, BPD, PDA, and ventilator days that has not been previously reported. In addition to the reduction in NEC, the additional reduction in morbidity may decrease the cost of a NICU hospitalization. Recent analysis found a significant cost reduction when using a HUM feeding protocol compared with using a bovine fortifier.¹⁶

To the authors' knowledge, this is the first study to compare rates of NEC after a feeding protocol implementation at multiple institutions with multiple years of follow-up while using an exclusive HUM diet. Additional strengths of this study include a large study population to evaluate differences in outcomes between the two groups. The multiple centers involved increases the generalizability of the study.

In addition, while many clinical studies have extensive exclusion criteria, including inability to obtain parental consent, this study only excluded infants with major congenital anomalies, infants who died within 48 hours of admission, or infants transferred to another hospital after 1 week of age. Therefore, all infants <1,250 g BW were included in this naturalistic study, including infants <23 weeks of age, <500 g BW, and growth-restricted infants. Due to the inclusion of high-risk infants, who are not often included in feeding or NEC studies, it is speculated that this may have contributed to the higher rates of NEC, sepsis, and mortality. However, a significant reduction was shown in all outcomes after the introduction of an exclusive HUM diet. Although three of the four centers fed infants <1,000 g BW a HUM feeding protocol, 62% of the infants in this study were from Baylor College of Medicine (Texas Children's Hospital), which fed infants <1,250 g BW a HUM feeding protocol. The effect of a HUM diet provides benefits across all weight categories consistent with the previous randomized clinical trials evaluating a HUM diet.^9,12

Some limitations to this study are acknowledged. A small difference was identified in BW and birth head circumference between study groups that is not believed to be clinically significant. The nature of a retrospective study lends to several biases such as misclassification biases, particularly in relation to the diagnosis of NEC, although the classification was determined by two researchers separately. It is possible that over the period this study occurred, there were changes in practice aside from type of milk and fortifier used that could confound these notable changes. It is acknowledged that mother's milk and pasteurized donor HM likely increased during the study period and potentially contributed to the improved outcomes. However, the study showed a 60% decrease in the incidence of NEC compared with only a 20% decrease in NEC in participating centers in the VON over the same time period. It is also acknowledged that at the time of the feeding protocol changes, there was increased awareness of central line infection initiatives to decrease central line associated bloodstream infections, which may have contributed to a reduction in late-onset infection. However, an all-HUM diet has been shown to be associated with decreased parenteral nutrition days, which serves as a surrogate for central line days.^6,9 In addition, with a decrease in the incidence of NEC, it was possible to reduce the central line days.

The use of an exclusive HUM diet is associated with significant benefits for extremely premature infants <1,250 g BW. The benefits include decreased NEC rates, mortality, late-onset sepsis, PDA, BPD, ventilator days, and ROP. Importantly, while evaluating the benefits of using an exclusive HUM-based protocol, it appears that there were no feeding-related adverse outcomes. This study demonstrates that an exclusive HUM diet provides important benefits beyond NEC.

We would like to thank David Rechtman, MD (Prolacta Bioscience), for reviewing the manuscript.