Resources & Evidence Contact us

Benefits of an exclusive human milk diet: conclusions, controversies, and a critical review of the evidence

Neonatal Intensive Care

January 30, 2024

Sandra Sullivan, MD, IBCLC, FAAP and Brian Scottoline, MD

Caring for any patient involves caution, attention to detail, and an understanding of risks, benefits, and limitations of knowledge. It stands to reason that adherence to evidence-based practices, when the evidence is available, should form the backbone of care practices.

Poised near the top of the evidence pyramid is the randomized controlled trial (RCT), which typically aims to evaluate the impact of one variable while keeping all other variables constant, thus providing the opportunity to evaluate the benefits and risks of a specific intervention with minimal risk of bias. All research, whether trials, studies, or experiments, has limitations based on design, endpoints, and outcome measures, and thus we need to carefully consider how that study was designed, analyzed, interpreted, and reported. By no means does this signify that research is inherently flawed; it is just to say that consideration of research should be balanced and impartial.

When applying evidence to clinical practice, new or change, we must look at all the available evidence, keeping the relative pros and cons of each type in mind. While RCTs provide high-quality data, medical and biomedical research also recognizes the potential value of non-RCT data: prospective, retrospective, and real-world evidence that can reflect the interventions in neonatal intensive care unit (NICU) settings over long periods of time, with many patients—with, of course, their own limitations in what can be concluded. This data affords the opportunity to evaluate extensive patient populations to several thousand, compared to the typically much smaller numbers of patients in RCTs driven by realities of power calculations and resource limitations.

The role for an Exclusive Human Milk Diet (EHMD) is an example of an intervention for preterm infants in the NICU for which there is both a growing body of published clinical research and increasing controversy over how to apply that research. Critical Evaluation of the Data Is Essential Before applying conclusions from RCTs to clinical practice, we must consider the trial design, endpoints, or outcome measures, and the resulting statistical analyses performed. For instance, RCT trial design can be constrained by ethical considerations, as all groups must receive, at minimum, the best level of care available at the time of trial design lock.

In addition, confusion regarding the meanings of and differences between clinical and statistical significance can conceal potentially important findings, particularly in the case of studies that are underpowered. Understanding power and how sample sizes were calculated is crucial for interpretation of findings. RCTs generally select for a narrowly defined patient population, which may mean the findings have limited generalizability to specific NICUs or populations of patients.

While RCTs are considered very high level of evidence, it is important to recognize that many practices and interventions have been adopted into NICU practice despite a lack of RCT data showing benefit. One key example is use of inhaled nitric oxide for preterm infants <34 weeks’ gestation with hypoxemic respiratory failure and/or pulmonary hypertension.1

Design Implications
An EHMD is an example of an intervention for which RCT data has led to incomplete or simplified conclusions. In 2018, O’Connor et al published an RCT in which infants born weighing <1,250 g recruited from neonatal units in Ontario, Canada, were fed mother’s own milk (MOM) and donor milk (DM) as required. The infants were randomized to a human milk–based fortification (HMBF, 0.81 kcal/mL) or cow milk–based fortification (CMBF, 0.72 kcal/mL), commencing at >100 mL/kg/day and increased at 140 mL/kg/day to 0.88 and 0.78 kcal/mL, respectively.2

The study failed to demonstrate benefit of an EHMD compared with a CMBF strategy with regard to feeding tolerance, morbidity, or mortality. However, it was designed in such a way that made it unlikely to detect other important differences between the interventions, even if one was superior. One of the key benefits of an EHMD is that it allows for earlier fortification without increasing the risk of complications such as necrotizing enterocolitis (NEC), and earlier fortification is associated with improved outcomes.3,4,5,6 As an example, a recent head-to-head RCT involving term babies, who tolerate non-human protein better than premature infants, found improved growth and reduced NEC risk in infants with single ventricle physiology fed an EHMD post-surgery compared to a mixed human/cow milk-based diet.7

Ethical Implications
Early fortification could not ethically be attempted in the CMBF arm of the O’Connor study, as the early fortification strategies commonly used with EHMD have not been well-tested with CMBF in very preterm (VPT; <28 weeks gestational age (GA) at birth) infants and thus might have put the infants at risk for undefined complications. To keep the two arms identical for as many variables as possible, early fortification was not part of the trial design for either arm. Therefore, it was highly improbable to see any benefits of early fortification with EHMD that had thus far been demonstrated. Despite this design limitation, patients in the EHMD arm had an adjusted absolute 14.5% reduced morbidity and mortality index, which approached statistical significance, at p < 0.07, and may represent a meaningful clinical effect.2

Statistical Implications
Recently, results became available for the prospective NFORTE trial comparing HMBF (Prolacta, Duarte, CA, USA) with a standard CMBF commonly used in Sweden involving 228 infants born between 22 weeks 0 days and 27 weeks 6 days GA who were fed MOM or DM with individualized targeted fortification. All centers in this multicenter study fortified milk based on measured protein content of breast milk (target of 4.0-4.5 gm/kg/ day with a gradual decrease in protein intake with approaching term equivalent age). The primary outcome identified was the composite of NEC, sepsis, and mortality.8

No statistically significant differences between the two groups were reported for primary or secondary outcomes (in unadjusted analyses). However, it appears that the study was underpowered for some measures, making it difficult to interpret the findings. An a priori interim analysis was performed to evaluate the adequacy of the sample size (a predetermined upper limit of 322 was set during sample size calculations based on incidence of 47.7% for the primary outcome). However, the actual incidence in the control group was much lower (34.5%); a smaller difference between control and target intervention incidence means it would take more subjects to show a true difference. To reach statistical significance of any difference, a total sample size of about 1,600 infants would have been required, yet the study sample size remained at 222. Finding no difference does not necessarily mean that no difference exists; it means that one cannot say based on these results.8

Closer review of the data suggests that a larger sample size likely would have identified potential benefits of the HMBF. Mortality was decreased by 47% in the HMBF arm (7 in HMBF group vs. 13 in CMBF), which is an interesting trend and potentially clinically important even though not statistically significant in this underpowered study. After adjusting for GA, the HMBF group had a statistically significant decreased risk of developing bronchopulmonary dysplasia (BPD) at 36 weeks GA (P = 0.049).8

Protocol Implications
Like the O’Connor study, fortification in the Jensen trial did not start until infants reached at least 100 mL/kg/day of feeding volume and differed between the two treatment groups. Fortification started at a significantly higher feeding volume in the CMBF group (P = 0.001), than in the HMBF group. In the CMBF arm, 25% of the study subjects were not fortified until reaching at least 114 mL/kg/day. This could be interpreted as a sign that providers were reluctant to start CMBF sooner, given that the two treatment groups were not blinded. Initiating fortification at this volume of feeding may now be considered late fortification and may not be generalizable. Based on 15 years of real-world clinical data, an EHMD feeding pathway for Prolacta HMBF was introduced in 2022 and indicates that to achieve the best short- and longterm outcomes with an EHMD, fortification should begin as soon as possible, ideally in the first few days of life.3,4,9,10,11,12,13,14,15

As is almost inevitable in clinical trials, the Jensen trial suffered some protocol deviations that could have significant impact on conclusions. For instance, survival for at least 3 days of life was an inclusion criterion in the study, yet one infant in the HMBF arm died at 2 days of life. In trials with outcome measures that may be underpowered, one such instance can greatly affect statistical significance. The study was analyzed as intent-to-treat, as is standard for reporting such outcomes. It warrants discussion that three patients in the HMBF group and seven in the CMBF group never received fortification and would be useful to see additional analyses comparing outcomes exclusive of these subjects.

Study Population Implications
The Jensen trial authors conclude that routine fortification for infants born at less than 28 weeks GA using HMBF is not supported by their findings. It should be recognized that this conclusion is for the primary outcome measure (the composite of NEC, sepsis, and mortality), that there were power limitations for the outcome measures of interest, and that other interesting outcomes may not have been evaluated. For example, the trial does not address whether HMBF may be beneficial among select high-risk infants, such as those born small for gestational age, weighing <1,250 g, with significant fetal acidosis, hypoxia or ischemia, requiring cardiopulmonary resuscitation, or when used as rescue for infants who develop complications, or fail to thrive on standard manufacturer recommended feeds. These are the typical situations in which an HMBF strategy is currently being used in NICUs.

Non-RCT Evidence on EHMD
As even RCTs have limitations that impact how to best interpret and implement the findings, it is important to look to other evidence. There are numerous non-RCT studies of the effects of an EHMD of varying strength of evidence. Delaney-Manthe et al implemented an EHMD for all infants born weighing 1,250 g or less in 2016 with the goal of improving feeding tolerance and reducing complications (specifically NEC and PN days). Infants who received an EHMD maintained desirable weight trends, had statistically significantly fewer late-onset sepsis evaluations (P = 0.0027) and less BPD (P = 0.018) than historical controls. There was a statistically nonsignificant trend toward less surgical NEC (57% of total NEC cases vs. 14.3%), though possibly clinically and fiscally significant.12

Non-Evidence-Based Implications
One potential downside of HMBF and an EHMD is concern that HMBF displaces MOM, resulting in the infant missing out on important bioactive factors only available in MOM. While use of HMBF with high caloric density can greatly displace MOM,16 high caloric density HMBF is used sporadically and for short time periods in specific clinical instances. Typically, an infant fed with 100% MOM who receives Prolact+6 HMBF will have 30% of MOM displaced by the fortifier (30 mL fortifier to 70 mL MOM). In comparison, an infant fed with 100% MOM who receives a commercially available liquid CMBF will have approximately 17% of MOM displaced by the fortifier (5 mL of fortifier to 25 mL MOM). These calculations assume on-label use of Prolacta fortifiers and 2 commercially available liquid CMBF (Abbott, Reckitt/Mead Johnson). It should be noted that, when increasing calories, off-label mixing strategies for both HMBF and CMBF have been used and result in increased MOM displacement. Additionally, most preterm infants in the NICU require DM at some point due to insufficient MOM,17 and DM, which has reduced bioactive properties compared to MOM,18 would be displaced preferentially before displacing MOM. In many cases, there is minimal, if any, displacement of MOM. When it does occur, MOM is not discarded or wasted, but saved for use later, thus extending the availability of MOM to the infant, which is also associated with better outcomes.

While only fresh MOM would seem ideal for all infants, the reality is that nearly all preterm infants require fortifiers to reach nutritional targets. Both HMBF and CMBF result in some displacement of MOM or DM. While the volume of displacement may be less for CMBF, the substance added lacks protective bioactive properties. HMBF ensures that 100% of the protein fed to the infant is human-derived and also more than replenishes displaced bioactive components. Philip et al analyzed the biochemical and immunochemical properties of fresh and frozen MOM as well as pasteurized banked DM that were supplemented with either HMBF or CMBF. They demonstrated that DM has less lactoferrin and α-lactalbumin than MOM, but HMBF (not CMBF) not only reinstates both but also has higher antioxidant activity. They concluded that “freshly expressed MOM fortified with HMDF and given early, enterally, and exclusively appears to be an optimal nutritional choice for extremely premature infants.”18

Another concern with using an EHMD has been growth. This, too, is closely related to following recommended feeding guidelines. Those institutions that follow optimal feeding protocols report good growth.3,4,19 Those that delay fortification with HMBFs have more difficulty. As with feeding tolerance and infant morbidity, the feeding protocol and appropriate use of the HMBF is critically important to see the best outcomes. Finally, an EHMD is not a trivial cost for NICUs, which has led to concerns about return on investment. An analysis of 2019- 2022 data from more than 3,000 premature infants treated at more than 60 US hospitals found EHMD implementation improved health outcomes and reduced costs, generating a 2.6-fold dollar-for-dollar return on investment.20 In addition, a 2023 peer-reviewed report found EHMD implementation resulted in annual cost savings of $500,000 to $3.4 million per hospital from a reduction in comorbidities and shorter lengths of stay among very low birth weight infants.21

Balancing Evidence With Our Clinical Insights
As clinicians, it is our duty to fully consider all sources of data—RCTs, observational studies, and real-world data—for their utility and limitations. We must also keep in mind the continual evolution of these data. Because of the time and resources necessary to carry out larger scale RCTs, they may reflect clinical knowledge of the past, and can be confined in what they report because of the need for narrowly defined endpoints and protocols. This isn’t to say we should ignore RCT data, only to understand both their distinct advantages and limitations and thus what can be concluded. We should then integrate RCT conclusions with other data, possibly of lesser study design control but still with potential utility, to develop best practices for our patients.

While research regarding the utility of EHMD in preterm neonates has produced useful results for defining the potential risks and benefits of human milk fortification, there are limitations in the data and what can be interpreted from that data. A large enough RCT to tease out true head-to-head superior fortifier is unlikely to happen in the United States, due to lack of funding and ethical constraints as EHMD has become standard care in so many NICUs that randomizing patients to CMBF would be unacceptable to providers and families. In every study published evaluating EHMD, at least one beneficial effect of HMBF and EHMD has been shown. With all the evidence available, one can conclude that infants fed EHMD have at least as good outcomes as those fed CMBF, possibly better, certainly not worse. Some will interpret this as CMBF is doing a good enough job. We must ask ourselves, with all the information available now, is “good enough” good enough for the lives depending on us?


  1. Chandrasekharan P, Kozielski R, Kumar VH, et al. Early use of inhaled nitric oxide in preterm infants: is there a rationale for selective approach?. Am J Perinatol. 2017;34(5):428-440. doi:10.1055/s-0036-1592346
  2. O’Connor DL, Kiss A, Tomlinson C, et al. Nutrient enrichment of human milk with human and bovine milk-based fortifiers for infants born weighing <1250 g: a randomized clinical trial [published correction appears in Am J Clin Nutr. 2019 Aug 1;110(2):529] [published correction appears in Am J Clin Nutr. 2020 May 1;111(5):1112]. Am J Clin Nutr. 2018;108(1):108-116. doi:10.1093/ajcn/nqy067
  3. Hair AB, Hawthorne KM, Chetta KE, Abrams SA. Human milk feeding supports adequate growth in infants ≤ 1250 grams birth weight. BMC Res Notes. 2013;6:459. Published 2013 Nov 13. doi:10.1186/1756-0500-6-459
  4. Huston R, Lee M, Rider E, et al. Early fortification of enteral feedings for infants <1250 grams birth weight receiving a human milk diet including human milk based fortifier. J Neonatal Perinatal Med. 2020;13(2):215-221. doi:10.3233/ NPM-190300
  5. Wyble L, Ferry J, Vaughan E, Lee M. Earlier, more rapid feeding of exclusive human milk diet leads to lower incidence of bronchopulmonary dysplasia (BPD). Presented at: 3rd Congress of joint Neonatal Societies; Maastricht, Netherlands. September 17-21, 2019.
  6. Salas AA, Gunawan E, Nguyen K, et al. Early human milk fortification in infants born extremely preterm: a randomized trial. Pediatrics. 2023;152(3):e2023061603. doi:10.1542/ peds.2023-061603
  7. Blanco CL, Hair A, Justice LB, Roddy D, Bonagurio K, Williams PK, Machado D, Marino BS, Chi A, Takao C, Gordon EE, Ashrafi A, Cacho N, Pruetz JD, Costello JM, Cooper DS, & Cardiac Neonate Nutrition Study Group. A randomized trial of an exclusive human milk diet in neonates with single ventricle physiology. J Pediatr. 2022;256: 105–112. doi. org/10.1016/j.jpeds.2022.11.043
  8. Jensen GB, et al. Effect of human milk-based fortification in extremely preterm infants fed exclusively with breast milk: a randomised controlled trial. eClinicalMedicine, Volume 0, Issue 0, 102375
  9. Sullivan S, Schanler RJ, Kim JH, et al. An exclusively human milk-based diet is associated with a lower rate of necrotizing enterocolitis than a diet of human milk and bovine milkbased products. J Pediatr. 2010;156(4):562-7.e1. doi:10.1016/j. jpeds.2009.10.040
  10. Hair AB, Peluso AM, Hawthorne KM, et al. Beyond necrotizing enterocolitis prevention: improving outcomes with an exclusive human milk-based diet [published correction appears in Breastfeed Med. 2017 Dec;12 (10):663]. Breastfeed Med. 2016;11(2):70-74. doi:10.1089/bfm.2015.0134
  11. Visuthranukul C, Abrams SA, Hawthorne KM, Hagan JL, Hair AB. Premature small for gestational age infants fed an exclusive human milk-based diet achieve catch-up growth without metabolic consequences at 2 years of age. Arch Dis Child Fetal Neonatal Ed. 2019 May;104(3):F242-F247. doi: 10.1136/archdischild-2017-314547. Epub 2018 Nov 13. PMID: 30425116; PMCID: PMC6764250.
  12. Delaney Manthe E, Perks PH, Swanson JR. Teambased implementation of an exclusive human milk diet. Adv Neonatal Care. 2019;19(6):460-467. doi:10.1097/ ANC.0000000000000676
  13. Hair AB, Lee ML. The effectiveness of an exclusive human milk diet in premature infants <750g birthweight. Presented at jENS September 2021.
  14. Bergner EM, Shypailo R, Visuthranukul C, et al. Growth, body composition, and neurodevelopmental outcomes at 2 years among preterm infants fed an exclusive human milk diet in the neonatal intensive care unit: a pilot study. Breastfeed Med. 2020. 15(5):304-311. doi:10.1089/bfm.2019.0210
  15. Rahman A, Kase J, Murray Y, et al. Neurodevelopmental outcome of extremely low birth weight infants fed an exclusive human milk diet is not affected by growth velocity. Breastfeed Med. 2020;15(6):362-369. doi:10.1089/ bfm.2019.0214
  16. Perrin MT. Donor human milk and fortifier use in United States Level 2, 3, and 4 neonatal care hospitals. J Pediatr Gastroenterol Nutr. 2018 Apr;66(4):664-669. doi: 10.1097/ MPG.0000000000001790. PMID: 29045350
  17. Carroll K, Herrmann KR. The cost of using donor human milk in the NICU to achieve exclusively human milk feeding through 32 weeks postmenstrual age. Breastfeed Med. 2013 Jun;8(3):286-90. doi: 10.1089/bfm.2012.0068. Epub 2013 Jan 16. PMID: 23323965; PMCID: PMC3663453.
  18. Philip RK, Romeih E, Bailie E, Daly M, McGourty KD, Grabrucker AM, Dunne CP, Walker G. Exclusive human milk diet for extremely premature infants: a novel fortification strategy that enhances the bioactive properties of fresh, frozen, and pasteurized milk specimens. Breastfeed Med. 2023 Apr;18(4):279-290. doi: 10.1089/bfm.2022.0254. PMID: 37071630; PMCID: PMC10124176.
  19. Huston RK, Markell AM, McCulley EA, Gardiner SK, Sweeney SL. Improving growth for infants ≤1250 grams receiving an exclusive human milk diet. Nutr Clin Pract. 2018;33(5):671- 678. doi:10.1002/ncp.10054
  20. Data on file; hospital-provided outcomes analysis from 2019 to 2022.
  21. Swanson JR, Becker A, Fox J, et al. Implementing an exclusive human milk diet for preterm infants: real-world experience in diverse NICUs. BMC Pediatr. 2023;23(1). doi. org/10.1186/s12887-023-04047-5