History of Donor Milk
Current State of Milk Banking
Nutrient and Bioactive Composition of Donor Milk
Strengths and Challenges of Conducting and Comparing Donor Milk Trials
Outcomes of Randomized Clinical Trials
Non-randomized Studies of Donor Milk Ongoing Trials
KeywordsDonor milk, Preterm infant, Very low birth weight infant
The authors wish to acknowledge the funding received from the Canadian Institutes of Health Research (MOP-102638 and FDN-143233) that supports their research in relation to the use of donor milk for VLBW infants.
Care of the preterm infant has undergone several major transformations over the past 120 years, and an integral part of this has been nutrition. This was outlined in an excellent review by Greer, in which he points out that from the turn of the 20th century, feeding human milk has been recognized as a priority in the care of the preterm infant or “weakling”. In a publication from 1922, mother’s milk or donor milk was the “gold standard” practice initiated at 140 mL/kg/d and increased to 200 mL/kg/d over the first 3 weeks of life, with gavage feeding provided to infants unable to nurse. If no human milk was available, a buttermilk and skim milk mixture or boiled milk with water and sugar added was recommended. The 1940s and 1950s brought two major changes to the nutritional care of preterm infants. First, withholding of feedings for the first 12 to 72 hours of life became common practice, as it was believed that delaying oral intake would reduce the incidence of aspiration pneumonia and prevent retention of extracellular fluid. Second, investigators began studying the effects of feeding cow’s milk formulations to preterm infants. Various diets, including evaporated milk, skimmed or partially skimmed milk, diluted milk, and with various additives, such as dextran-maltose, sugar, and olive oil, were studied. Feeding some of these preparations was demonstrated to result in accelerated growth compared with human milk feeding, leading to a change in clinical practice to a predominance of artificial formula feeding. Artificial formula feeding of preterm infants was widespread in the 1950s and 1960s.
Beginning in the 1970s, a body of research began to challenge the artificial formula feeding practices of the mid-20th century, bringing about a resurgence in the interest in human milk feeding. Indeed, it is now increasingly standard practice to feed very low birth weight (VLBW; <1500 g) infants donor milk when mother’s milk is unavailable. The World Health Organization (WHO) recently published a recommendation that low birth weight babies (<2500 g) who cannot be fed mother’s milk should be fed donor human milk and advocated for the establishment of safe and affordable milk-banking facilities. In 2012 the American Academy of Pediatrics (AAP) Breastfeeding Policy Statement recommended that preterm infants be given human milk and that donor milk be used if mother’s milk is unavailable. A clinical report on donor milk from the AAP in 2016 recommended that priority should be given to providing donor human milk to infants born weighing <1500 g. In a recently reaffirmed Position Statement of the Canadian Paediatric Society, it is similarly recommended that mother’s milk be the preferred nutrition for newborn infants and that pasteurized donor milk be the recommended alternative for hospitalized neonates when availability of mother’s milk is limited. Even in the settings of the authors’ home institutions with high rates of use of mother’s milk, for a variety of reasons, about two thirds of mothers are unable to provide a sufficient volume of their own milk; these findings are similar to those in other reports in the literature. Although a growing body of evidence indicates that use of mother’s milk increases when mothers are offered lactation support, it is clear that a supplement is frequently required.
This chapter will describe the history of the use of donor milk and review the randomized clinical trials (RCTs) that have been conducted over the past 40 years informing modern recommendations for the feeding of the preterm infant ( Table 5.1 ).
|Author, Year||Country||Participants||Intervention Groups||Outcomes Measured||Results|
|O’Connor, 2016||Canada||VLBW infants (<1500 g)||Double blind RCT |
MM, DM, BBF (n = 181)
MM, PF, BBF (n = 182)
Intervention: 90 days or hospital discharge
|Primary : Neurodevelopment (18 month BSID III) |
Secondary : BSID III language and motor scores, growth, mortality and morbidity index
|No difference in cognitive score (BSID III) |
DM 92.9; PF 94.5
Mean difference -2.0 (95% CI −5.8 to 1.8)
No difference in language, motor (BSID III)
No difference in growth
No difference in morbidity and mortality index
Lower incidence of NEC stage ≥II in DM group (1.7%) compared with PF group (6.6%), P = .02
|Corpelejin, 2016||The Netherlands||VLBW infants (<1500 g)||Double blind RCT |
MM, DM (n = 183)
MM, PF (n = 190)
Intervention: 10 days
|Primary: Serious infection or NEC within first 60 days||No difference in composite outcome (serious infection or NEC to 60 days): |
DM 42.1%; PF 44.7%
Hazard ratio 0.87 (0.63-1.19), P = 0.37
|Hair, 2014||USA||Preterm infants |
|Unmasked RCT |
MM, DM, HBF (n = 39)
MM, DM, HBF, with human-based cream when human milk <67 kcal/dL (n = 39)
Intervention: to 36 weeks corrected age
|Primary : Growth velocity||Faster weight (14.0 ± 2.5 vs. 12.4 ± 3.9 g/kg/d, P = .03) and length (1.03 ± 0.33 vs. 0.83 ± 0.41 cm/wk, P = .02) growth in cream-fortified group |
No difference in head growth
No difference in rate of sepsis
No occurrences of NEC or death
|Cristofalo, 2013||USA, Austria||Preterm infants (500-1250 g)||Double blind RCT |
PF (n = 24)
DM, HBF (n = 29)
Intervention: 91 days, hospital discharge or attainment of 50% oral feeds
|Primary : Duration of parenteral nutrition |
Secondary : Growth, respiratory support, NEC
|Fewer parenteral nutrition days in DM group (27 vs. 36 days, P = .04) |
Reduced incidence of NEC in DM group (3% vs. 21%, P = .08)
No difference in duration of mechanical ventilation or oxygen, ROP or death
|Sullivan, 2010||USA, Austria||Preterm infants (500-1250 g)||RCT |
MM, DM, HBF at 100 mL/kg/d (n = 67)
MM, DM, HBF at 40 mL/kg/d (n = 71)
MM, PF, BBF (n = 69)
Intervention: 91 days, hospital discharge or attainment of 50% oral feeds
|Primary : Duration of parenteral nutrition |
Secondary: Select morbidities and growth
|No differences in duration of parenteral nutrition |
No difference in sepsis, ROP, BPD or growth
Significant difference between groups in rate of NEC (1.7% HBF at 100 mL/kg/d, 3.2% HBF 40 mL/kg/d, 15.3% PF; P = .006)
|Schanler, 2005||USA||Preterm infants (<30 weeks)||Double blind RCT |
MM, BBF (n = 70)
MM, DM, BBF (n = 81)
MM, PF, BBF (n = 92)
Intervention: 90 days or hospital discharge
|Primary : Infection related events (sepsis, NEC, meningitis, UTI) |
Secondary : Milk intake, growth, duration of hospital stay
|No difference in infection related events between DM (77 ± 103 death or infection related event per 100 infants) and PF (85 ± 111) but fewer in MM only (47 ± 70); P = .012 |
21% of DM group switched to PF for poor weight gain
Greater enteral intake but lower weight gain in DM group but no difference in length or head circumference gains
Shorter duration of stay for MM group
|Morley, 2000||UK||Preterm infants (<1850 g) |
Patients from Lucas, 1990
|Long-term follow-up n = 781||Primary: Late growth (7.5-8 y)||No differences in weight, height, head circumference, skinfold thickness or BMI between groups, at age 7.5-8 y|
|Lucas, 1994||UK||Preterm infants (<1850 g) |
Overlap of patients with Lucas, 1990
|RCT with blinded outcome assessor |
DM (sole or supplement) (n = 212)
PF (sole or supplement) (n = 210)
Intervention: to 2000 g or discharge, transfer to nonparticipating center or death
|Primary : Neurodevelopment (18 month BSID)||No differences in neurodevelopmental indices between groups at 18 months |
DM: MDI 98.6 ± 1.3
PTF: MDI 100.1 ± 1.5
Advantage for PF: 1.5 (95% CI −2.4 to 5.4)
DM: PDI 92.2 ± 1.2
PTF: PDI 90.9 ± 1.3
Advantage for PF: −1.3 (95% CI −4.8 to 2.3)
|Lucas, 1990||UK||Preterm infants (<1850 g) |
Overlap of patients with Lucas, 1989
Formula only: PF or TF (n = 236)
MM, PF, or TF (n = 437)
Human milk only:
MM or DM (n = 253)
|Primary : NEC||Significant difference in the incidence of NEC: PF or TF: 7.2% |
MM, PF, or TF: 2.5%
MM, DM: 1.2%
OR of formula only compared with human milk only 6.5 (95% CI 1.9-22; P < .001)
|Lucas, 1989||UK||Preterm infants (<1850 g) |
Overlap of patients with Lucas, 1984
|RCT with blinded outcome assessor |
DM (sole or supplement) (n = 195)
PF (sole or supplement) (n = 174)
Intervention: to 2000 g or discharge, transfer to nonparticipating center or death
|Primary : Neurodevelopment at 9 months corrected age (Knobloch et al. developmental screening inventory) |
Secondary : Neurologic examination at 9 and 18 months corrected age
|Significantly better overall developmental quotient in PF group : PF: 100.4 ± 10.7 |
DM: 97.9 ± 9.6
(95% CI 0.4-4.6; P < .025)
Difference greater for subgroup where supplement was >50% of intake:
PF: 101.4 ± 10.5
DM: 96.5 ± 9.9
(95% CI 1.3-8.5; P ≤ .01)
No difference between groups for diagnosis of neurologic impairment at 9 or 18 months
|Lucas, 1984||UK||Preterm infants (<1850 g)||RCT |
DM (n = 29)
PF (n = 33)
MM, DM (n = 67)
MM, PF (n = 65)
Intervention: to 2000 g or discharge, transfer to nonparticipating center or death
|Primary : Days to regain birth weight |
Weight, length, and head circumference gains while in hospital
|More days to regain birth weight for DM fed as sole diet or as supplement |
Slower weight, length gains for DM fed as sole diet or as supplement
Slower head circumference gain for DM as sole diet
|Tyson, 1983||USA||VLBW infants (<1500 g)||Unblinded RCT |
DM (n = 34)
PF (n = 42)
Intervention commenced at day 10 to exclude babies receiving MM
|Primary : Early growth (10-30 postnatal days) |
Secondary : milk intake, biochemical indices, neonatal responsiveness (Brazelton Neonatal Behavioural Assessment Scale at 37 weeks corrected age), maternal interaction
|Slower growth in DM group compared with PF group (weight gain 15 vs. 30 g/d; length gain 0.7 vs. 1.1 cm/wk; head circumference gain 0.8 vs. 1.2 cm/wk) |
Greater milk intake in DM group
No clinically significant differences between groups in biochemical indices (serum chemistry, amino acids)
Increased responsiveness in PF infants compared with DM
Brazelton orientation scale: PF: 3.4 ± 1.4 vs. DM: 2.6 ± 1.0; P < .10
Brazelton inanimate stimuli:
PF: 7.5 ± 3.0 vs. DM: 5.0 ± 2.1; P < .02
No reported differences in mother–infant interaction
|Gross, 1983||USA||Preterm infants (<1600 g and 27-33 weeks)||RCT |
TF (n = 20)
Preterm DM (n = 20)
Term DM (n = 20) Intervention until weight of 1800 g
|Primary : Early growth||Mean time to regain birth weight was significantly longer in the mature DM (18.8 ± 1.7) group compared with preterm DM (11.4 ± 0.8) or TF (10.3 ± 0.8), P < .001 |
All growth parameters significantly better in TF and preterm DM compared with term DM
Weight gain: TF 27.0 ± 0.8, preterm DM 23.7 ± 1.1, term DM 15.8 ± 0.8 g/d, P < 0.001
4 cases of NEC all withdrawn from the study, 3 in TF group and 1 in mature DM group
|Svenningsen, 1982||Sweden||Preterm infants (<2000 g)||RCT |
MM (n = 12)
DM (n = 6)
PF1 (2.3 g/100 kcal) (n = 14)
PF2 (3.0 g/100 kcal) (n = 16)
Intervention: weeks 2-15
|Primary: Growth until 2 years |
Secondary : Metabolic responses at 3, 5, and 7 weeks, Neurodevelopment examination at 2 years
|No differences between groups in early growth |
Higher weight gain between 15 and 20 weeks in PF2
No differences in growth at 2 years
Higher BUN and metabolic acidosis in PF1, PF2, compared with MM ± DM
No differences in neurodevelopment
|Schultz, 1980||Hungary||Preterm infants (<2000 g)||RCT |
DM (n = 10)
PF (n = 10)
Intervention: 4 weeks
|Primary : Metabolic indices at 4 weeks |
Secondary : Growth
|PF group had lower fasting blood glucose compared with DM |
PF group had significantly higher BUN, higher plasma free amino acids and metabolic acidosis compared with DM group
No difference in weight gain
|Davies, 1977||Wales||Preterm infants (<36 weeks)||RCT |
DM (n = 14)
PF (n = 14)
DM (n = 20)
PF (n = 20)
Intervention: 2 months
|Primary : Growth (birth to 1 month, 1 month to 2 months)||Slower linear growth and head growth in group born at 28-32 weeks fed DM compared with PF in the first month of life with similar growth in second month |
No difference in growth for either period for babies born at 33-36 weeks
|Raiha, 1976||Finland||LBW infants (<2100 g)||RCT |
DM (n = 22)
PF1 (60:40 whey/casein ratio, 1.5 g/dL protein) (n = 21)
PF2 (60:40, 3.0 g/dL) (n = 20)
PF3 (18:82, 1.5 g/dL) (n = 22)
PF4 (18:82, 3.0 g/dL) (n = 21)
Intervention: to 2400 g
|Primary : Early growth (up to 2400 g) |
Secondary : Metabolic responses
|No differences among groups in growth parameters with exception of longer time to regain birth weight and longer time to reach 2400 g in those born 34-36 weeks and fed DM |
BUN significantly higher in infants fed 3.0 g/dL protein formula compared with DM
Blood ammonia significantly higher in casein predominant and 3.0 g/dL whey predominant formula compared with DM group
Total protein higher in 3.0 g/dL groups compared with DM or 1.5g/dL
Late acidosis (5 weeks) observed with 3.0 g/dL casein predominant formula
History of Donor Milk
From the once-common practice of “wet nursing,” to the current rapid expansion of not-for-profit and commercial donor milk banks, donated human milk has long been a crucial aspect of feeding sick and vulnerable infants. From as early as 2000 bc until the early 20th century, mothers who were unable to breastfeed their infants employed the services of “wet nurses.” There are ancient depictions of wet nursing in many areas of the world, including Egypt, India, China, and Israel, among others. Eventually, advances in technology and hygiene permitted the collection and storage of donated human milk, and facilities to bank and distribute donor milk became feasible. This is detailed by Frances Jones in her paper entitled “History of north american donor milk banking: one hundred years of progress.” Briefly, the first documented human milk bank opened in 1909 in Vienna, Austria, followed by milk banks in the United States (1910) and Germany (1919). These early banks were established primarily in response to the diminishing number of available wet nurses. Donor mothers were initially compensated for their contribution of their milk, but concerns regarding the possibility of mothers denying their own babies milk or tampering with milk volumes led to the cessation of this practice which had continued until recently.
Preservation of milk in the absence of refrigeration presented a significant challenge, and many strategies were attempted in those first decades of milk banking, including chemical preservation with peroxides, boiling, autoclaving, and spray drying. All of these methods posed substantial problems, either with maintenance of bacteriologic safety or with extensive destruction of nutritional and bioactive components. In spite of these challenges, milk banking was widespread in the 1930s and 1940s, with more than 12 North American milk banks existing in 1939.
A decline in the use of donor milk occurred in the 1950s and 1960s as artificial formula feeding became more common. Once the shortcomings of artificial formula feeding were recognized, milk banking once again increased in popularity, peaking in the 1980s with 53 operational donor milk banks existing in North America. The human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS) crisis of the mid-1980s and awareness that the virus could be secreted into human milk subsequently resulted in the closure of the majority of these banks and a cessation in donor milk research. A resurgence in milk banking began in the early 2000s with the ability to thoroughly screen donor mothers and the safety of donor milk being further ensured through standardized donor milk processing and pasteurization techniques. In this current era of neonatal care and feeding the preterm infant, there has been an unprecedented growth in donor milk banking.
Current State of Milk Banking
There are two large international groups that are at the forefront of publishing guidelines for human milk banking: the European Milk Bank Association and the Human Milk Banking Association of North America (HMBANA). Member banks are nonprofit and follow their respective organization’s established guidelines and protocols to ensure the consistent quality and safety of donor milk. The Program for Appropriate Technology in Health, funded by the Gates Foundation, also provides guidance, with an international perspective, on the core requirements and quality principles for human milk banks. Last, for-profit suppliers of donor milk are now established in the United States and operate as food manufacturers under regulations that vary by state.
Both nonprofit and for-profit banks, according to the applicable guideline and/or regulatory requirements, conduct medical screening and blood testing of potential donors to reduce the risk of disease transmission. Pooling of milk from several mothers is generally recommended to ensure uniformity in nutritional content, and then the milk is pasteurized to remove known pathogens. Donor milk is cultured after pasteurization, and these cultures must show negative results before distribution. Typical donor milk processing is depicted in Fig. 5.1 .
Nutrient and Bioactive Composition of Donor Milk
Most donor milk used in the hospital setting globally is pasteurized using the Holder Method (30 minutes at 62.5° C; see Fig. 5.1 ). Heat treatment, per se, does not substantially compromise the macronutrient composition of donor milk, although transferring milk from container to container does result in loss of lipid content. Pasteurization partially inactivates some of the bioactive components found in human milk ( Table 5.2 ). Obliteration of all cellular activity (T cells, B cells, macrophages, neutrophils) occurs, along with a reduction in antibody (immunoglobulin A reduced by 0%-48%), lactoferrin (reduced by 57%-80%), lysozyme (reduced by 0%-60%), erythropoietin, growth factor, and cytokine concentrations (interferon-γ, tumor necrosis factor-α, interleukin [IL]-1β, IL-10). Despite these heat-induced alterations to the bioactive components of human milk, pasteurized milk maintains a degree of bacteriostatic and immune-stimulating properties.
|Component||Maintained (>90%)||Maintained (50%−90%)||Maintained (10%−50%)||Abolished (<10%)|
|Macronutrients||Carbohydrate (lactose, oligosaccharides)||Protein |
|Vitamins||Vitamin A||Folate |
Vitamin B 6
|Biologically active (immune)||IL-8, IL-12p70, IL-13 |
|IgA, sIgA |
IGF-1, IGF-2, IGF-BP2,3
IL-1β, IL-4, IL-5, IL-10
|CD14 (soluble) |
|Biologically active (metabolism)||Epidermal growth factor |
Heparin binding growth factor
Hepatocyte growth factor
|Bile salt–dependent lipase |
Human milk, whether mother’s own milk or donor milk, does not contain adequate protein to meet the elevated needs of the very preterm infant. To address this deficiency, fortifiers are added to donor milk, either by using ready-made multinutrient fortifiers, or by adding modular components of each macronutrient. Three fortification strategies are currently being used in neonatal intensive care units (NICUs) in North America.
Standard fortification is a fixed-dose approach that assumes the uniform macronutrient composition of all donor milk.
Adjustable fortification uses the infant’s blood urea nitrogen (BUN) as an indicator of protein requirement; fortifier strength is adjusted to maintain BUN within a target range.
Target fortification involves the regular assessment of macronutrient composition of milk samples and adjustment to target levels.
Most HMBANA milk banks do not provide a nutritional analysis for their donor milk and normal concentrations must be assumed from the literature. Typical estimates for protein are 0.9 g/dL compared with 1.2 g/dL for mature mother’s milk. Further, most NICUs utilize the standard fortification approach and target protein intakes based on these references ranges for human milk.
There are numerous fortification products on the market, currently available in liquid or powder form. The majority are concentrated bovine products, although a human milk-based fortifier now exists. Carbohydrates, fat, electrolytes, calcium, phosphate, vitamins, and minerals, in addition to protein, are usually present in fortifier formulations.
Strengths and Challenges of Conducting and Comparing Donor Milk Trials
The potential for bias in open-label and observational studies examining the impact of supplemental donor milk versus infant formula is well known. Views on which is the superior form of nutrition may influence, for example, how long clinicians wait until mother’s milk is available before feeding a supplement and/or advancement of enteral feeds. A blinded RCT with random allocation of infants between supplement groups should address these biases and ensure even distribution of known and unknown confounding variables (e.g., birth weight, early acuity, etc.) between the two groups. In this way, differences in outcomes can reasonably be ascribed to the intervention. Feeding RCTs in neonates are subject to a host of unique challenges. Foremost are ethical considerations for randomizing infants to various feeding protocols. If NICUs are already using donor milk as a supplement, the health care team may not feel there is sufficient clinical equipoise to ethically randomize infants. Given the recent data, this could be an issue for future research in this area. Second, clinical outcomes of interest, such as necrotizing enterocolitis (NEC), have a low prevalence and recruiting a sufficient sample size to yield meaningful interpretation may be difficult and time consuming, particularly for the VLBW baby (1.4% of births in the United States ).
The blinding of supplemental feeding assignments is a challenge, necessitating preparation in a designated milk preparation area by a technician who does not participate in the care of the neonate. Feeds must be placed in individual opaque syringes because those caring for the neonate can discern differences in color, smell, and appearance between various feed types. A strategy needs to be developed to ensure provision of vitamin and mineral drops that meets the needs of infants in both arms of the study as a differential approach may unblind the study. Not unexpectedly, these trials are very costly to run, especially so if the sample size requires a multicenter approach with further costs accruing if the outcome of interest requires long-term follow-up.
Quigley and McGuire published a Cochrane review in 2014 on the use of donor milk for feeding preterm infants. Examination of the studies included emphasizes the challenges of comparing the results of different RCTs now available. This meta-analysis includes nine trials, of which seven were conducted before 1985, reflecting a different era in the care and feeding of preterm infants. Only the two most recent trials used nutrient fortified donor milk. The authors highlighted weaknesses in the included studies, particularly the lack of blinding in many and the lack of long-term follow-up in most studies. Differences in subject inclusion criteria alone can heavily impact outcomes as the possible benefit, or risk, of using a supplement will likely be affected by the vulnerability of the sample studied. For example, some studies have included infants with a birth weight as high as 2100 g, where recent trials have more intensively studied higher-risk infants in the 500 to 1250 g range. The timeline for inclusion of infants in the study may substantially influence findings, with some studies enrolling infants immediately after birth and other protocols including infants only after a predetermined feeding tolerance has been achieved. This results in a “healthier” and more homogeneous study population through the exclusion of very ill infants.
Feeding protocols differed substantially among the trials included in the Cochrane meta-analysis. Early study protocols usually included a group that received strictly cow’s milk formula, with no mother’s milk provided. Once it became clear that mother’s milk exerted beneficial effects, it became unethical to randomize infants to receive no mother’s milk. In light of this, many recent studies employ a pragmatic approach that includes mother’s milk, whenever available, in all feeding groups. Others selected infants of mothers who did not intend to breastfeed, which may reflect a different demographic of infants. Donor milk fortification strategies varied significantly among trials.
Finally, no data on using donor milk in term infants are available; all studies, to date, have investigated health outcomes of feeding donor milk to preterm infants.
Outcomes of Randomized Controlled Clinical Trials of Donor Milk
Since the late 1970s, several trials have been conducted investigating the clinical impact of various strategies for feeding preterm infants human donor milk, initially with a focus on growth and metabolic outcomes; later trials investigated infection-related outcomes, NEC, and neurodevelopment. Trials in this review were identified by searching the Medline database by using the search terms “donor milk” or “pasteurized donor milk.” RCTs that compared feeding of donor milk (with or without fortification) to preterm infants with other feeding strategies (formula, mother’s milk, or a combination of these) are reviewed in this section and in Table 5.1 . Trials that did not randomize subjects are discussed in section 5, Nonrandomized Studies of Donor Milk.
Although a subject of debate, the goal of providing nutrition for preterm infants is to facilitate growth that mimics intrauterine fetal growth rates and body composition changes. There are multiple health concerns related to intrauterine and postnatal growth restriction, including both short- and long-term neurodevelopmental delays. Early studies that prompted the widespread feeding of cow’s milk-based formulas to preterm infants did so by demonstrating improved growth parameters for various formulas over human milk. One of the first trials that began to challenge the growth advantages of formula over donor milk was conducted by Raiha et al. between 1972 and 1975. This randomized controlled study investigated not only the quantity but the quality of protein provided. This study of 106 low birth weight babies (<2100 g) randomized infants to one of five feeding regimens: (1) 1.5 g/dL protein with a whey/casein ratio of 60:40; (2) 3 g/dL protein with whey/casein ratio of 60:40; (3) 1.5 g/dL protein with whey/casein ratio of 18:82; and (4) 3 g/dL protein with whey/casein ratio of 18:82, with all these four formulas containing the same quantities of minerals and vitamins; and (5) pooled donor human milk, supplemented with vitamins A, D, and C, and iron. No infants were fed mother’s milk. Subjects were also divided into three groups based on gestation: (1) 28 to 30 weeks; (2) 31 to 33 weeks; and (3) 34 to 36 weeks. This strategy yielded a final number of six to eight subjects per group. Feeding was initiated within the first 24 hours (typically 6-9 hours), and increased to 150 mL/kg/d within the first week of life. Protein intakes were 2.25 g/d for the 1.5 g/dL protein formula groups, 4.5 g/d for the 3 g/dL protein formula groups, and 1.63 g/d for the human milk group. No differences were observed among groups for initial weight loss, but in the group of infants born at 34 to 36-weeks’ gestation, all formula-fed infants regained birth weight more rapidly compared with donor milk-fed infants. Once birth weight was achieved, there were no differences among groups in attaining a body weight of 2400 g. No differences were observed in any groups in linear growth (measured by crown–rump length), head circumference, skinfold thickness, or hematocrit. BUN was significantly higher in infants fed the 3 g/dL protein formula, compared with those fed donor milk. Blood ammonia was higher in all formula groups compared with the donor milk group, except for the 1.5 g/dL whey-predominant formula. Feeding the casein-predominant 3 g/dL protein formula resulted in a late (5 weeks) acidosis. Overall no growth advantage was observed over the range and type of bovine protein fed to preterm infants, compared with donor milk. The authors identified the importance of the quality of the protein in infant formulas for preventing the development of metabolic acidosis arising from feeding large amounts of casein. An alternative explanation is that the sample size (n = 6-8 infants/group) was insufficient to detect a difference in growth. In planning a five-group study, 26 infants per group would be required to detect a 1 standard deviation (SD) difference in growth with 80% power an alpha-level of 0.05.
This study was quickly followed by a growth study of Welsh preterm infants (28-32 weeks, n = 28; 33-36 weeks, n = 40). Subjects were randomized to be fed either mature pooled, pasteurized donor milk (protein 1.1 g/dL), as assessed in a prior study, or a cow’s milk-based formula (protein 2.7 g/dL). Feeding was initiated at 50 mL/kg/d and increased by 15 mL/kg/d daily until a daily oral intake of 200 mL/kg/d was achieved. All infants received vitamin and iron supplementation. In infants born at 28 to 32 weeks’ gestation, there were no significant differences in the rate of weight gain over the first 2 months of life; however, linear growth and head circumference were higher in the formula-fed group from birth to 1 month, but not from 1 to 2 months. No differences in any growth parameter were observed between the two feeding groups in the infants born at 33 to 36 weeks’ gestation. The authors concluded that the slower growth of the head and length of the donor milk-fed infants in the lower gestational age group may have been related to term milk providing inadequate nutrition for early preterm infants. The sample size may again have been insufficient to detect growth differences.
Further evidence of preterm infant growth in response to human milk compared with formula came from an American study by Gross in 1983. In this study, preterm infants (<1600 g; 27-33 weeks) were randomized to be fed early milk from mothers of preterm infants (n = 20), term donor milk (n = 20), or a whey-based formula (n = 20). All donor milk was pasteurized according to the Holder technique. For the group randomized to preterm donor milk, the milk was collected from mothers who gave birth before 35 weeks of gestation, and the postpartum week of gestation for the study infant corresponded to the postpartum week during which the milk was collected. The composition of the milks differed, and preterm milk changed over the 12 weeks of the study. Protein content of term milk was 1 g/dL; formula was 1.93 g/dL; and the protein content of preterm milk decreased from 2.26 g/dL in week 1 to 1.14 g/dL in week 12. Mineral content was higher in the formula compared with donor milk. Feedings began at a mean of day 3 of life and were initiated at 24 mL/kg/d. By day 8, all subjects were receiving 180 mL/kg/d. The number of days to regain birth weight was different between groups and was highest in infants receiving term milk (18.8 ± 1.7 days), compared with preterm milk (11.4 ± 0.8 days) and formula (10.3 ± 0.8 days) ( P < .001). Rate of weight gain, linear growth, and head circumference were also higher in preterm milk-fed and formula-fed infants, compared with infants fed term milk. The author concluded that term milk alone is not a satisfactory source of nutrition for preterm infants.
In the same year, Tyson et al. published the results of a second American study of VLBW (<1500 g) infants who were randomized to receive either pooled, raw unfortified donor milk (n = 34) or an enriched formula designed for preterm infants (n = 42), on postnatal days 10 to 30. Infants whose mother had an adequate supply of milk after day 10 were excluded from the study. Donor milk and formula were reported to have very different macronutrient profiles, with much higher concentrations of protein, fat, and carbohydrate in the formula (2.22, 4.02, 8.84 g/dL, respectively) compared with donor milk (1.09, 2.21, 7.72 g/dL, respectively). Mineral content was also higher in the formula. Infants were fed by bottle as much as their appetite would allow or by intermittent gavage if their suck was inadequate. Daily milk intake was higher in the donor milk group (197 mL/kg/d) compared with the formula group (165 mL/kg/d). The infants fed donor milk grew more slowly between days 10 and 30, in weight, length, and head circumference. The authors concluded that donor milk must be analyzed to be certain that it contains adequate concentrations of macronutrients for the preterm infant.
In another early 1980s study of growth and development, Swedish preterm infants were randomized to receive either a high-protein formula (3 g/100 kcal; n = 16), a lower-protein formula (2.3 g/100 kcal; n = 14) or human milk (donor, n = 6; mother’s milk, n = 12). Svenningsen reported no statistically significant difference in growth parameters between the groups. Differences may not have been detected as a result of the small sample size.
Lucas et al. conducted in the United Kingdom a series of elegant parallel, multicentered RCTs of varying feeding regimens for low birth weight (LBW) infants (<1850 g) in the 1980s. The first trial was published in 1984, with early growth data from 194 preterm infants born at a mean of 31 weeks’ gestation and a mean birth weight of 1364 g. Groups included exclusive feeding of a newly designed preterm formula (n = 33), exclusive donor milk (n = 29), mother’s milk + preterm formula (n = 65), and mother’s milk + donor milk (n = 67). The preterm formula contained 80 kcal/dL and 2 g/dL protein, and donor milk contained 46 kcal/dL and 1 g/dL protein. The median days to regain birth weight was greater in the exclusive donor milk group (16 days) compared with the preterm formula group (10 days). Linear growth, head circumference, and rate of weight gain was higher in the preterm formula group compared with the donor milk group. In the groups where mother’s milk was supplemented, supplementation with preterm formula resulted in faster weight gain and linear growth rate compared with those in infants supplemented with donor milk.
By this point, in the mid-1980s, it had become abundantly clear that the growth of preterm infants was impeded by a diet of unfortified human milk—whether mother’s milk or, most especially, pooled, pasteurized, term donor milk. The focus of much research at this point was on the commercial development of cow’s milk formulas that provided improved growth. This occurred concomitantly with the closure of the majority of milk banks throughout industrialized countries in the 1980s when it became apparent that HIV could be transmitted via human milk. Research in human donor milk ceased for nearly 2 decades, and preterm infants who required supplementation to mother’s milk were routinely fed preterm formulas.
As the study of the bioactive components of human milk evolved in the 1990s, and as evidence emerged regarding the benefits of human milk, including improved neurodevelopment and a reduction in the incidence of infection- and inflammation- related outcomes in preterm infants, such as NEC and sepsis, a focus on human milk emerged in trials of infant feeding. Investigators sought to determine how infant growth could be optimized with the use of human milk, rather than by developing more nutrient-dense bovine-based formulas. A more pragmatic approach to feeding evolved, with mother’s milk provided whenever possible but fortified with a multinutrient bovine based fortifier. The first of these more recent trials using donor milk and multinutrient fortifiers, published by Schanler in 2005, represented the beginning of a shift in thinking in the clinical care of preterm and low-birth weight babies. This study enrolled mothers who intended to breastfeed their preterm infants and randomly assigned the infants to receive either donor milk or preterm formula as a supplement, if the mother’s milk supply was insufficient. The study compared three groups: (1) infants receiving exclusively their mother’s milk (n = 70); (2) those receiving mother’s milk + donor milk (n = 81); and (3) those receiving mother’s milk + preterm formula (n = 92). Both mother’s milk and donor milk were fortified with a bovine-based milk fortifier. The study duration was 90 days or until discharge from hospital, whichever was earlier. Early growth results in this study demonstrated that preterm infants fed fortified donor milk gained weight at a slower rate than those fed preterm formula; there was, however, no difference between groups in length or head circumference gains. Secondary to poor weight gain, 21% of the infants in the donor milk group crossed over to the preterm formula group. There were no crossovers from the preterm formula group. Weight gain over the study duration was highest in preterm formula-fed infants and lowest in donor milk-fed infants.
Sullivan et al. hypothesized that health benefits (reduced duration of parenteral nutrition, sepsis, and NEC) might be observed if infants were fed an exclusively human milk-based diet, with fortification of mother’s milk with a human milk-based fortifier, instead of the bovine milk-based fortifiers that had been investigated to date. To evaluate a human milk-based fortifier, infants (500-1250 g) whose mothers intended to breastfeed were recruited from 12 centers and randomized to receive human milk-based fortifier when their enteral intake was 40 mL/kg/d (n = 71), 100 mL/kg/d (n = 67), or bovine milk-based fortifier when enteral intake was 100 mL/kg/d (n = 69). The first two groups received donor milk if a supplement to mother’s milk was required, whereas the final group received a preterm formula if a supplement was required. Feeding was initiated 1 to 4 days after birth, and once feeding at 10 to 20 mL/kg/d was tolerated for up to 5 days, milk volumes were increased by 10 to 20 mL/kg/d. No difference between groups in weight gain, length, or head circumference growth was observed. Because the majority of milk provided in this trial was mother’s milk, Cristofalo et al. conducted a subsequent multicenter trial of similar design but with infants whose mother did not intend to provide their own milk. The infants were randomized to be fed either a bovine milk-based preterm formula or pasteurized donor milk fortified with a human milk-based fortifier. Weight gain was not different over the study period between the two groups, but infants fed preterm formula had increased head and length growth compared with human milk-fed infants.
Recently, a human milk-derived cream has come on the market. Hair et al. investigated the impact on the growth of feeding this supplement to preterm infants (750-1250 g; n = 78). Infants received mother’s milk, supplemented with donor milk if required, and a human milk-based fortifier. Infants in the cream group also received human milk cream if the caloric content of the milk received was <67 kcal/dL based on a daily batched assessment. Weight and length gain was higher in the group receiving the cream supplement, and there were no differences in head circumference gains. The study did not include a preterm formula control group.
Finally, a large pragmatic, multicenter, double-blind, randomized trial of VLBW infants (<1500 g) was recently conducted in Canada. This trial was entitled DoMINO (Donor Milk for Improved Neurodevelopmental Outcomes). Infants with a birth weight of <1500 g were enrolled in the first 96 hours of life. They were randomized to receive a supplement to mother’s milk, as required, as either pasteurized donor milk with a bovine milk-based fortifier and modular protein supplement of 0.4 g/dL protein/dL (n = 181) or preterm formula (n = 182) for 90 days or until hospital discharge. No differences were observed between the groups in any anthropometric measures at the end of the feeding intervention even after controlling for the volume of mother’s milk fed.
The 2014 Cochrane review of quasi-controlled trials and RCTs by Quigley and McGuire included studies of now-obsolete methods of feeding preterm infants, such as unfortified donor milk and term formula. This systematic review and meta-analysis concluded that feeding donor milk resulted in slower gain in weight, head circumference, and length compared with formula feeding. The slower growth associated with the use of donor milk compared with formula in some early studies appears to have been attenuated or even ameliorated in recent clinical trials that employed various strategies for fortifying donor milk.
The large, multicenter randomized parallel trials led by Lucas and initiated in the early 1980s ultimately saw the enrollment of 926 preterm infants. In a series of reports, this group has subsequently examined the growth and health outcomes of these infants at 7.5 to 8 years of age and beyond. This larger cohort of now school-age children had been fed one of 4 feeding regimens during initial hospitalization when they were infants: (1) donor milk; (2) preterm formula; (3) mother’s milk + donor milk; and (4) mother’s milk + preterm formula. Lucas et al. presented data from several time points: 9 months, 18 months, and 7.5 to 8 years. Although significant differences in growth were observed in the neonatal period (as described previously), no significant differences in any measured parameters (weight, length, head circumference, skinfold thicknesses, body mass index) were identified at any of the follow-up points.
In the previously described Swedish trial by Svenningsen et al., infants were randomized to receive one of two formulas or human milk, and no differences were found in growth in the neonatal period; those authors also followed up their patients at ages 8 and 24 months. There were similarly no differences in weight, length, or head circumference in any of the three feeding groups at the later time points, although the sample size was likely insufficient to detect differences that may have been present.
To date, no other studies of donor milk feeding in preterm infants have measured long-term effects on growth.
Early studies of donor milk measured a variety of metabolic indices, including BUN, serum ammonia, serum albumin, blood glucose, and metabolic acidosis. In an early study by Raiha et al., infants with birth weights <2100 g (but few infants <1250 g) were randomized to either unfortified donor milk (0.96 g/dL protein) or one of four different formula preparations as previously described (1.5 versus 3.0 g/dL protein, whey/casein ratio of 60:40 or 18:82). The feeding intervention lasted until the infants reached 2400 g. As expected, BUN values reflected the protein content of enteral feeds with infants in the unfortified donor milk group with the lowest values (mean values ≈3.6 nmol/L). Although the Raiha et al. study found little evidence that the protein content of enteral feeds alone impacted growth, Arslanoglu et al. reported more recently that individually adjusting the protein content of enteral feeds to achieve a BUN concentration in the range of 3.6 to 7 mmol/L (versus <3.6 nmol/L) is best associated with optimal growth. Raiha et al. further found that BUN levels resulting from feeding the 1.5 g/dL protein whey-predominant (60:40) formula most closely mimicked the BUN of donor milk fed infants. Casein-predominant formulas (18:82; 1.5 and 3 g/dL) and the 3 g/dL whey-predominant formula (60:40; 3 g/dL) resulted in higher blood ammonia levels compared with donor milk feeding. Blood ammonia concentrations, urine osmolarity, and total serum protein were highest in the 3 g/dL protein formula groups and lowest in the donor milk group. Infants fed a high-casein 3 g/dL formula developed late acidosis for as long as 5 weeks, and feeding whey-predominant formulas resulted in nearly normal acid–base balance. In all measured metabolic parameters, feeding a whey-predominant, 1.5 g/dL protein formula most closely resembled donor milk feeding. In North America, standard preterm formulas for in-hospital use are whey predominant and contain 2.0 g/dL (67 kcal/dL) and 2.4 g/dL (80 kcal/dL) protein.
In 1980 Schultz et al. randomized 20 preterm infants to be fed either unfortified pooled donor milk (1.2 g/dL protein) or formula (2.6 g/dL protein). Like the Raiha et al. study, the BUN values among preterm infants fed formula were greater than those of infants fed unfortified donor milk. Infants in the formula group were also more likely to develop late acidosis in postnatal weeks 2 and 3. Plasma free amino acid concentrations were generally higher in formula-fed infants compared with donor milk-fed infants, particularly phenylalanine and lysine. This remained true throughout the 4-week study period.
To try to support growth at intrauterine rates but address the metabolic disturbances resulting from high-protein formula feeding, Tyson et al. fed a 2.2 g/dL protein (60:40 whey/casein ratio)–mineral–calorie–enriched formula and compared metabolic responses to unfortified donor milk feeding (1.1 g/dL protein) in 76 preterm infants (<1500 g). No differences between the groups in serum albumin, protein, BUN, blood pH, or osmolarity were observed on postnatal days 20 and 30, although BUN did decrease from day 20 to day 30 in both groups. Few differences in plasma amino acids were noted (lower threonine on days 20 and 30, lower proline on day 20, and lower arginine and methionine on day 30 in the donor milk group). The authors attributed similarity in the metabolic response in the two groups to the use of whey-predominant formula and a more modest protein/calorie ratio compared with previous studies. In that same year, Gross reported similar findings after a feeding trial of unfortified term donor milk, unfortified early preterm donor milk, and a whey-predominant 1.9 g/dL protein infant formula (n = 20 per group). No differences in BUN, blood pH, bicarbonate, serum calcium, serum total protein, or albumin were observed among the groups.
Necrotizing Enterocolitis and Late-Onset Sepsis
In the 1970s and 1980s, the predominant research focus for preterm infant nutrition was the development of an infant formula that would achieve better intrauterine growth rates while minimizing metabolic disturbances. Feeding preterm formula became common practice in many North American NICUs. It was in these early studies that other ramifications of feeding preterm formula began to emerge, particularly in the rates of NEC and late-onset sepsis (LOS).
NEC affects preterm infants disproportionately, with the incidence highest in the earliest gestations (varying from 1.3%-12.9% in infants with <33 weeks’ gestation within the Canadian Neonatal Network) and a peak occurrence at weeks 31 to 33 corrected gestational age. NEC is marked by intestinal inflammation with or without infection and may lead to intestinal necrosis and perforation. It carries a mortality rate of 25% to 50% and is the leading cause for short bowel and multiorgan transplantation in childhood. Staging of NEC is defined by the Bell criteria, which have been modified as the understanding of NEC has evolved over the years. Emerging research suggests that very preterm infants who develop NEC have an altered microbiome (dysbiosis) compared with very preterm infants who remain healthy. The emergence of non–culture-based techniques for elucidating the intestinal microbiota have led to a clearer picture of the pathogenic events that precede NEC. The mode of feeding preterm infants impacts the development of the microbiome along with other factors, such as mode of delivery and antibiotic exposure.
By the 1980s, observational data became available, and these suggested that human milk may be protective against NEC. No trial of feeding donor milk to preterm infants had yet investigated outcomes beyond growth. Tyson et al. did mention in their discussion that two infants in the formula-fed group developed signs of NEC; they did not elaborate on the significance of this finding but suggested that future studies that “enroll sick infants shortly after birth” would be able to elaborate on the relationship between formula feeding and NEC.
Lucas et al.’s parallel multicenter clinical trials of the 1980s were the first to investigate the incidence of a number of sequelae to various feeding strategies for preterm infants beyond growth and biochemical indices. Ultimately the trials included 926 infants, and clinical features of NEC developed in 51 subjects and was confirmed in 31 subjects. The differences observed among the groups were striking, with confirmed cases of NEC in 7.2% of formula-only group, 2.5% of formula + mother’s milk group, and only 1.2% of the human milk only group (donor milk and mother’s milk). Donor milk as a sole diet appeared to be as protective as mother’s milk, but the sample size was limited for these analyses. Other risk factors identified for NEC included gestational age, respiratory disease, umbilical artery catheterization, and polycythemia, but significant differences in the incidence of NEC were observed between diet groups even after controlling for these variables.
Since 2005, there have been five published RCTs that examined the impact of either donor milk compared with preterm formula or an exclusive human diet on the incidence of NEC. In the study by Schanler et al. (n = 243), the rate of NEC alone was not different between groups (6% in mother’s milk + donor milk, 11% in mother’s milk + preterm formula, 6% in mother’s milk alone; P = .27 comparing donor milk group and preterm formula group). When comparing the sum of death or any infection-related events, fewer events cases per 100 infants) occurred in the preterm infants fed exclusively mother’s milk, but there was no difference between the donor milk supplemented group and the preterm formula supplemented group (77 ± 103 events in mother’s milk plus donor milk, 85 ± 111 in mother’s milk plus preterm formula, 47 ± 70 in mother’s milk alone; P = .012 comparing mother’s milk alone group and combined other groups). In this trial, mother’s milk and donor milk were enriched with nutrients with intact bovine-based fortifiers. The authors concluded that in a setting of high use of mother’s milk, there was little advantage of donor milk over preterm formula when a supplement was required.
It was proposed that that the use of bovine milk-based fortifiers in donor milk may have impacted the findings in the Schanler et al. study and that an exclusively human milk-based diet that includes not only mother’s milk and donor milk as a supplement but also a fortifier derived from human milk may be of benefit. This hypothesis was tested in an RCT reported by Sullivan et al. in 2010. As described previously, infants with a birth weight of 500 to 1250 g and whose mothers intended to breastfeed were randomized to one of three groups: (1) HM100 (infants fed mother’s milk supplemented with donor milk and fortified with a human milk-based fortifier commencing at enteral tolerance of 100 mL/kg/d; n = 67); (2) HM40 (infants fed mother’s milk supplemented with donor milk and fortified with a human milk-based fortifier commencing at enteral tolerance of 40 mL/kg/d; n = 71); and (3) BOV (infants fed mother’s milk fortified with a bovine based fortifier at enteral tolerance of 100 mL/kg/d and supplemented with preterm formula as required; n = 69). When the rate of NEC was analyzed for only infants completing the study without any protocol violation, there were significant reductions in the human milk fortifier groups with rates of NEC of HM100 1.7%, HM40 3.2%, and BOV 15.3% ( P = .006). What was unclear in the outcome of this trial is whether it was the lack of preterm formula, the lack of bovine fortifier, or both that led to the reduction in NEC in the exclusive human milk diet groups.
The impact of an exclusive human milk diet was further studied in a trial of infants whose mother did not intend to breastfeed. As described earlier, infants were randomized to receive either preterm formula (n = 24), or donor milk (n = 29) with a human-milk–derived fortifier. The incidence of NEC in infants fed preterm formula was significantly higher (21%) than in the group receiving donor milk (3%) ( P = .04). In this study, the rate of NEC in the preterm formula group was much higher than the typical rates reported elsewhere and calls into question the difficulty of confirming cases of NEC in a blinded fashion. The previously described Hair et al. trial on human milk cream as a supplement had both study arms receiving an exclusively human milk-based diet (n = 78; birth weight 750-1250 g). There were no cases of NEC in either group.
Two new trials comparing donor milk as a supplement with mother’s milk were published in 2016, and these were the first trials to describe masking of the study feeds in amber-colored syringes. In the trial by Corpeleijn et al. from the Netherlands, 373 VLBW (<1500 g) infants were randomized to two supplement groups (183 to donor milk and 190 to preterm formula as a supplement) if mother’s milk was not sufficiently available for the first 10 days after birth. There was a high mean intake of mother’s milk during the intervention period (89.1% in the donor milk group and 84.5% in the preterm formula group). There was no difference in the combined incidence of NEC stage II or greater and/or serious infection to 60 days between groups (44.7% in the preterm formula group versus 42.1% in the donor milk group; mean difference 2.6%; 95% confidence interval [CI], −12.7% to 7.4%; adjusted hazard ratio [HR] 0.87; 95% CI 0.63-1.19; P = .37). The authors acknowledge that a longer duration of intervention should be studied.
The second donor milk trial published in 2016 was the Greater Toronto Area Donor Milk for Improved Neurodevelopmental Outcome (GTA DoMINO), a multicenter RCT by O’Connor et al. Confirmed cases of NEC were adjudicated by a panel blinded to feeding assignments using the modified Bell criteria; stage 1 was defined as presence of consistent symptomatology, as defined by Bell, along with treatment for a minimum of 7 days (suspension of enteral feeds and antibiotics). Stage II or greater was defined by the presence of pneumatosis; portal venous air or bowel perforation on any of the radiographs or ultrasound scans or at the time of surgery; or bowel ischemia on histology. Preterm infants who received donor milk with bovine fortifier exhibited a lower rate of any stage NEC (7 of 181; 3.9%) compared with those receiving preterm formula (20 of 182; 11%) resulting in a risk difference of −7.1 (95% CI −12.5 to −1.8; P = .01). The rate of stage II or higher NEC was similarly lower in the donor milk group (3 of 181; 1.7%) compared with those receiving preterm formula (12 of 182; 6.6%), a risk difference of −4.9 (95% CI −9.0 to −0.9; P = .02).
Overwhelming evidence favors provision of human milk over provision of preterm formula for prevention of NEC. Numerous mechanisms have been proposed to explain why bovine milk causes higher rates of NEC, including increased intestinal permeability, dysbiosis, and direct cytotoxicity to the intestinal epithelial cells.
The trial by Svenningsen et al. in 1982 with six donor milk-fed infants was the first to report long-term neurodevelopment in donor milk-fed infants. There was no difference noted at 2 years, although the specifics of the testing were not provided in the report and the number of infants studied was too small to detect meaningful clinical differences. Tyson et al. in 1983 studied 76 VLBW infants fed either unfortified donor milk or preterm formula and measured neonatal behavior at 37 weeks corrected age according to the Brazelton Neonatal Behavioral Assessment Scale, a measure of 27 behavior-related indices, and 20 elicited responses. The average score for the Brazelton orientation scales, which measure alertness and responsiveness to auditory and visual stimuli, was not different between the two groups (preterm formula 3.4 ± 1.4 versus donor milk 2.6 ± 1.0; P < .10). The groups did differ in their responses to inanimate objects, with a higher score observed in the formula group compared with the donor milk group (preterm formula 7.5 ± 3.0 versus donor milk 5.0 ± 2.1; P < .02).
One of the primary outcomes of the large multicenter feeding intervention trials of preterm infants conducted by Lucas et al. in the 1980s was neurodevelopment at 9 and 18 months. At 9 and 18 months corrected age, 502 infants underwent follow-up examinations, including a developmental screening inventory by Knobloch et al. This assessment tool tests five fields of behavior, for which the investigators adjusted the quotient for preterm birth: adaptive, gross motor, fine motor, language, and personal–social. Infants were also neurologically assessed and categorized as normal, equivocal, or impaired, according to the methods of Ameil-Tison and Grenier.
At 9 months, there were no differences observed in neurologic status among the four feeding groups (donor milk alone [6% impaired] or as a supplement to mother’s milk [5% impaired], preterm formula alone [14% impaired], or as a supplement to mother’s milk [9% impaired]). There were no differences in overall development between the groups fed donor milk as the sole diet versus the group fed preterm formula as the sole diet. When groups were combined, however, to include infants fed donor milk as the sole diet and those fed donor milk as a supplement (n = 195), there were significant differences between the group fed preterm formula as a sole diet and that fed preterm formula as a supplement (n = 174) (donor milk developmental quotient mean 97.9 [SD 9.6] versus preterm formula 100.4 [SD 10.7], difference 95% CI 0.4-4.6; P < .025). This difference became more significant when subgroups were analyzed on the basis of infants receiving more than half of intake as a supplement or being small for gestational age (<10th percentile for birth weight). Among babies who received >50% of their intake as a supplement, there were 68 in the donor milk group and 56 in the preterm formula group (donor milk developmental quotient mean 96.5 [SD 9.9] versus preterm formula 101.4 [SD 10.5]; difference 95% CI 1.3-8.5; P < .01). Among babies who were born small for gestational age, there were 62 in the donor milk group and 68 in the preterm formula group (donor milk developmental quotient mean 94.3 [SD 8.7] versus preterm formula 99.6 [SD 10.7]; difference 95% CI 2.0-8.6; P < .01). Although the predictive validity of neurodevelopment results at 9 months corrected age is uncertain, these statistically significant differences do raise concerns about the impact of feeding unfortified donor milk on the neurodevelopment of preterm infants.
In their parallel trials that compared preterm formula to unfortified donor milk at 18 months, Lucas et al. followed up 422 infants by using the Bayley Scales of Infant Development (BSID). The BSID is commonly used and is a validated tool to assess neurodevelopment in infancy and early childhood (from 1 to 42 months); version 1 yielded a composite score on the mental development index (MDI) and on the psychomotor development index (PDI). No differences were observed in the MDI or the PDI at 18 months for infants fed preterm formula compared with those fed donor milk as the sole diet or as a supplement (95% follow-up rate in the donor milk group [n = 212] and 97% in the preterm formula group [n = 210]). For the donor milk group, the mean MDI ± standard error was 98.6 ± 1.3 and for the preterm formula group 100.1 ± 1.5, with a non-statistically significant mean difference of 1.5 (95% CI 2.4-5.4). For the donor milk group, the mean PDI ± standard error was 92.2 ± 1.2 and for the preterm formula group 90.9 ± 1.3, with a non-statistically mean difference of 1.25 (95% CI −4.8 to 2.3). A subgroup analysis of infants small for gestational age similarly showed no difference in the MDI or the PDI at 18 months corrected age. The authors had expected to see improved developmental outcomes for infants fed preterm formula compared with those fed donor milk on the basis of a higher nutrient content of the preterm formula. They speculated that the lack of difference between groups may be related to other biologic advantages of human milk and recommended that future research focus on the fortification of donor milk to further improve outcomes.
The long-term neurodevelopmental outcome after the use of fortified donor milk was the goal of the GTA DoMINO trial. O’Connor et al. assessed neurodevelopment in preterm infants who had been randomized to receive either fortified donor milk or preterm formula as a supplement to mother’s milk, whenever required. The authors used the third edition of the Bayley Scales of Infant and Toddler Development (BSID III) to assess infants at 18 months of age. This tool is designed to assess the cognitive, language, and motor development of infants up to 42 months of age. All of the assessors were blinded as to the intervention group and recertified to ensure >80% agreement on BSID III measures. Of the 363 infants randomized to receive either donor milk (n = 181) or preterm formula (n = 182), 37 died during initial hospitalization, and 299 were assessed at 18 months corrected age (92% follow-up rate of survivors). The mean (SD) birth weight was 995 (273) g in the donor milk group and 996 (272) g in the preterm formula group. Neurodevelopmental outcomes were analyzed by using logistic regression and then adjusted according to two different models. In the first model, the analyses were adjusted according to randomization strata, including the recruitment center and the birth weight group. In the second model, the analyses were adjusted according to the maternal education level and the percentage of total enteral feeds consumed by the infant as mother’s milk during the intervention period. No significant difference was observed in mean (95% CI) cognitive composite scores with a score of 92.9 (range 89.8-95.9) in the donor milk group and 94.5 (range 91.4-97.5) in the preterm formula group. In the statistical model that adjusted for recruitment center and birth weight, the mean effect size was −1.6 (95% CI −5.5 to 2.2; P = .41). In the fully adjusted statistical model that included the maternal education level and the percentage of total enteral feeds as mother’s milk, an effect size of −2.0 was observed (95% CI −5.8 to 1.8; P = .31). Children with a cognitive composite score of <85 were classified as having a neuroimpairment and those with <70 as having a disability. There were significantly more children with a neuroimpairment in the donor milk-fed group (27.2% versus 16.2%) with an adjusted risk difference of 10.6 (95% CI 1.5-19.6; P = .02). There was no difference between the two groups in those children classified as having a disability. No differences in the language composite (adjusted scores 87.3 in the donor milk group versus 90.3 in the formula group; fully adjusted mean difference -3.1 [95% CI −7.5 to 1.3]) or the motor composite (adjusted scores 91.8 in the donor milk group versus 94.0 in the formula group; fully adjusted mean difference −3.7 [95% CI −7.4 to 0.09]) were identified.
In summary, neither Lucas et al. nor O’Connor et al., the only two groups to describe neurodevelopmental outcomes in infants at 18 months corrected age, demonstrated a long-term benefit when donor milk is fed to VLBW infants, particularly in a setting that supports high use of mother’s milk. This should therefore not be seen as a treatment goal in prescribing donor milk.
Retinopathy of Prematurity
Retinopathy of prematurity (ROP) remains a common morbidity in VLBW infants, with rates varying from 25% to 91% internationally, and about 25% of these infants require treatment with either laser or intravitreal antivascular endothelial growth factor. A recent meta-analysis of observational studies reported a possible role for human milk in the prevention of ROP. Results from RCTs comparing donor milk to formula do not support a benefit with donor milk in this outcome. O’Connor et al., Sullivan et al., and Cristofalo et al. all did not see a difference in the occurrence of ROP between infants receiving donor milk as supplement versus those receiving preterm formula as supplement. Schanler et al. observed that stage 3 ROP occurred less frequently in infants fed mother’s milk compared with those fed donor milk or preterm formula. There were no differences between the groups with regard to requirement for surgery for the treatment of ROP.
To date, none of the trials reporting mortality as an outcome has shown a difference between donor milk and formula; interpretation of these comparisons must, however, be done with caution because of the low event rate and sample size of available trials. Gross was the first to report on deaths in their trial in 1983, and they noted three deaths resulting from NEC, of which two occurred in infants who had been in the formula group and 1 in an infant in the mature donor milk group. In the Lucas et al. study, mortality was not a preplanned outcome measure, but these authors were the first to analyze the differences between groups. Of 502 infants randomized, those fed donor milk alone and those fed preterm formula alone had mortality rates of 8.4% and 11.8%, respectively. Infants fed donor milk and those fed preterm formula in addition to mother’s milk had mortality rates of 7.1% and 8.7%, respectively. Neither difference was statistically significant.
Schanler et al.’s 2005 trial revealed no differences in mortality among feeding intervention groups receiving mother’s milk (3%), donor milk (4%), or preterm formula (3%). Mortality rates from Sullivan et al.’s trial comparing donor milk fortified at two different periods with preterm formula were not statistically analyzed for differences among groups, but the mortality rate was reported as 3 of 138 in the exclusive human milk-fed group and 5 of 69 in the bovine milk-fed group. In the subsequent trial by Cristofalo et al., which only included infants whose mothers did not intend to breastfeed, differences in mortality were not significantly different (2 of 24 in the preterm formula-fed group and 0 of 29 in the human milk-fed group).
In the Dutch trial by Corpelejin et al., comparing infants fed donor milk or preterm formula as a supplement to mother’s milk (babies only fed mother’s milk excluded) for the first 10 days, there was no difference between groups in mortality rate. Mortality in the donor milk group was 13.7%, with a median day of death of 11.0 (interquartile range [IQR] 5.5-26.0) and for the preterm formula group 12.1%, with a median day of death of 9.0 (IQR 5.0-18.0). The adjusted HR was 1.15 (95% CI 0.64-2.05; P = .63).
Similarly, in the Canadian trial by O’Connor et al., which compared infants fed donor milk with those fed preterm formula as a supplement to mother’s milk for 90 days or to hospital discharge found no significant difference among groups in mortality. The mortality rate in the donor milk group was 17 of 181 (9.4%) and in the preterm formula group was 20 of 182 (11%), a risk difference of −1.0% (95% CI −9.7 to 7.6; P = .82). Further, there was no difference between groups in the mortality/morbidity index. The index was a dichotomous variable and was positive either for mortality or any one of a predetermined list of morbidities that included confirmed LOS (positive blood or cerebrospinal fluid culture), NEC (Bell stage ≥ II), chronic lung disease (oxygen support at 36 weeks) or retinopathy of prematurity (International stage 4/5, laser of intraocular antivascular injection). This index was positive for 43.1% of infants in the donor milk group and 40.1% in the preterm formula group, with a risk difference of 5% (95% CI −-2.7 to 12.7; P = .2).