Extensive, consistent evidence of benefits was identified from recent systematic reviews (one including a meta-analysis), a subsequent cohort study, and a benefit–risk analysis for the pro-argument on the benefits of raw breastmilk. In contrast, limited evidence was identified for the contra-argument on the risks of enteric infections from potential pathogens in raw breastmilk.
3.1.1. Benefits: Pro-Argument
This analysis focused on two recent systematic reviews/meta-analyses [
6,
7] and a benefit–risk assessment of breastmilk for infant health in Norway [
28], summarizing an earlier report [
27]. Miller [
6] noted considerable overlap with the studies considered by Villamor-Martínez [
7]. The three studies applied and cited standard methods for assessing quality of the evidence and presenting pooled results. We refer readers with interest in the original research to these studies.
Miller et al. [
6] considered clinical trials and observational studies published between 1990 and 2017. These focused on the benefits of breastmilk in reducing morbidity and mortality in neonates, including the primary cause of mortality in NICU infants, a severe inflammatory disorder of the intestine of pre-term infants termed necrotizing enterocolitis (NEC). Other adverse health endpoints documented for neonates and children included late onset sepsis (LOS), retinopathy of prematurity (ROP), bronchopulmonary dysplasia (BPD), and neurodevelopment in infants born ≤28 weeks gestation, and/or studies on infants of mean birth weight of ≤1500 g. Additional information on these primary diseases of infants is provided below.
Miller identified 49 studies, of which 44 were included in their meta-analysis. Six studies were randomized trials (representing a total of 1472 infants), and 43 observational studies (representing a total of 14,950 infants). Risk of bias was judged low for all randomized trials, and low (26), moderate (14), and high (3) for observational studies. The authors noted that many studies had low sample size and were not designed or powered to detect small differences.
The authors concluded that breastmilk provided: (i) a clear protective effect against NEC (~4% reduction in incidence and 2% reduction in severity); and (ii) a possible reduction in late onset sepsis (LOS), severe retinopathy of prematurity (ROP), ROP, and severe NEC. Evidence was judged inconclusive regarding potential benefits to neurodevelopment. The authors noted that any volume of breastmilk is better than exclusive pre-term formula, and the higher the breastmilk dose, the greater the protective benefit to infants. The authors considered the evidence for benefits for pasteurized donor breastmilk inconclusive.
Villamor-Martínez and colleagues [
7] considered randomized controlled trials and observational studies (cohort and case–control) published before 2017 that compared pasteurized donor milk with pre-term formula and raw breastmilk (mother’s own milk) in reducing BPD in very preterm neonates (<32 weeks gestational age) or very low birth weight (<1500 g) infants. Other adverse health endpoints documented were days of mechanical ventilation and days on oxygen.
Villamor-Martínez included 18 studies (7 randomized and 11 observational) in their analyses. The authors noted a significant reduction in bronchopulmonary dysplasia (BPD) risk for raw breastmilk compared to pasteurized donor milk from two random trials. The authors also noted that raw breastmilk is more effective in reducing risk of multiple morbidities than pasteurized donor milk.
The findings of the Villamor-Martínez study from observational studies comparing raw breastmilk with pasteurized donor milk groups and pasteurized donor milk vs. preterm formula groups are discussed in the Attenuating section below.
Meltzer et al. [
28] presents evidence and analysis on benefits of breastmilk against possible risks from exposure to contaminants in breastmilk, focusing on conditions relevant in Norway. The benefit assessment was generally based on systematic reviews and meta-analyses published between 2003 and 2013. Although the authors conducted a full literature review for possible risks associated with breastmilk, the risk assessment portion of the study is less relevant to infectious and non-infectious disease effects consistent with the focus of the current study.
Three outcomes characterized by Meltzer and colleagues have direct relevance to the current study: (i) convincing evidence of beneficial effects of breastmilk on neurodevelopment; (ii) convincing evidence for a protective effect of breastmilk on infections (acute otitis, and gastrointestinal and respiratory infections), immune response-associated diseases (celiac disease, Crohn’s disease, inflammatory bowel disease, and ulcerative colitis), and type 1 diabetes; and (iii) convincing evidence of a protective effect of breastmilk on growth abnormalities (overweight and obesity in childhood). Overall, the study concludes that:
‘the benefits of breastmilk clearly outweigh the possible risks from contaminants’ for neurodevelopment, defense against infections, and growth abnormalities. Probable beneficial effects later in life were noted for type 1 and type 2 diabetes, as well as high blood pressure. No conclusions were drawn regarding other diseases of the immune system (allergy, asthma and wheezing, atopic dermatitis) due to ‘inconclusive results on the benefit side and few and disperse studies on the risk side’.
Supporting
Two additional cohort studies [
8,
29] published after the two systematic reviews [
6,
7] are summarized here in support of the pro-argument on benefits of raw breastmilk.
Ford and colleagues [
8] conducted a prospective cohort study to compare gut microbiota and health outcomes of 74 preterm infants (<1500 g birth weight) fed mothers’ own milk (raw breastmilk) with 43 similar infants fed pasteurized donor milk in the NICU at Texas Children’s Hospital. Significant increases associated with raw breastmilk were measured: diversity of the infant gut microbiota (
p < 0.001; abundance of protective genera
Bifidobacterium (
p = 0.02) and
Bacteroides (
p = 0.04)), whereas pasteurized donor milk was associated with increased abundance of potentially pathogenic
Staphylococcus (
p = 0.02). Raw breastmilk was associated with a 60% reduction in feeding intolerance (
p = 0.03 by multivariate analysis) and greater growth (
p < 0.01, weight gain;
p = 0.03 length;
p = 0.02 head circumference;
p < 0.01, growth velocity). Pasteurized donor milk was associated with higher composite outcome of severe morbidity (necrotizing enterocolitis (or NEC), spontaneous intestinal perforation (or SIP), sepsis, severe BPD, or death;
p = 0.02 adjusted).
Sun and colleagues [
29] conducted a multi-center prospective cohort study to address the feasibility and safety of providing fresh mothers’ own milk within 4 h of expression (raw breastmilk) at least once daily for a study duration of 32 weeks. The study excluded a formula feeding intervention. Sun enrolled 207 very preterm infants at NICUs in China (221 mother–infant pairs), and all infants were initially provided intravenous feeding (total parenteral nutrition). Oral feeds were introduced as soon as possible after birth using a nasogastric tube. Fourteen of the mothers recruited for the intervention group were unable to produce sufficient breastmilk to continue in the study.
Once full enteral feeding was possible for both groups, fortifiers and other nutritional supplements including probiotics were added and, subsequently, vitamins and iron were added. The intervention group (n = 98 infants) was fed raw breastmilk at least once daily for 32 weeks, and 109 control infants were fed exclusively defrosted human milk from donors or mothers (all frozen). Donor milk was reported only pasteurized if cytomegalovirus (CMV) density in donor milk exceeded 104 copies per mL. Infants were moved from enteral feeds to breastfeeding when clinically advised. The authors noted that 10 preterm infants (three in the intervention treatment, seven in the control group) died before discharged from the NICU.
The outcomes for the study included feasibility (measured as percentage of mothers providing daily breastmilk sample and other metrics) and metrics for morbidity and mortality (infant growth; mortality; sepsis; NEC; ROP; BPD; intraventricular hemorrhage; and a post hoc metric, composite outcomes of NEC and mortality).
Statistically significant decreases in adverse health effects (
p < 0.05) were noted for the intervention group for the following metrics: ROP; BPD; sepsis; and the combination of NEC and mortality. Note that although the authors reported that mortality alone was similar between intervention and control groups, the trend of lower mortality for the intervention group was significant at
p < 0.10. Additionally, the duration of mechanical ventilation and total parenteral nutrition were lower for the intervention group. The authors concluded that feeding raw breastmilk once daily to preterm infants in the NICU was ‘safe, feasible, and may reduce morbidity’. The study raised the concern that practices in NICUs and human milk banks (freezing, pasteurization) deprives infants of benefits of the cellular content of natural breastmilk, predominantly bacterial and immune cells killed by thermal challenges [
29].
Attenuating
No studies were identified that attributed health benefits to specific raw milk microbes or microbial consortia, independent of other bioactive factors present in raw milks.
Two subsets of the data considered in the systematic review/meta-analysis of Villamor-Martínez and colleagues [
7] provided some evidence attenuating the pro-argument for benefits of raw breastmilk. Analysis of data from eight observational studies (including either human or bovine fortifier) yielded high overall significance for a protective effect against BPD due to pasteurized donor milk compared to preterm formula. Analysis of data from four observational studies did not find significant overall differences between raw breastmilk and pasteurized donor milk in prevention of BPD, although the largest of these studies did demonstrate protection for the raw breastmilk group at
p = 0.07. Thus, some evidence supports benefits for pasteurized donor milk over formula, apparently associated with heat-stable components of breastmilk. However, the studies did not compare raw and pasteurized breastmilk directly.
Supplemental information on plausible mechanisms for benefits of breastmilk microbiota summarized below were noted by Ojo-Okunola and colleagues [
34], with the suggestion that these mechanisms operate optimally in synergy rather than as discrete independent mechanisms, in humans and other mammals. Further information from supplemental studies listed in
Figure 1 are briefly tabulated in
Supplemental Table S1. These plausible mechanisms include the following five examples:
Vertical transmission of microbes from mother to infant, including in utero microbes, anaerobes typical of the GI tract, strains of oral probiotic supplements during pregnancy and lactation, and microbiota of breast tissue, vaginal tissues, and skin;
Anti-infective functions, including colonization resistance by commensals for resistance to acute infections and induction of oral tolerance;
Immunomodulatory activities of T-regulatory cells, peripheral blood mononuclear cell subsets, cytotoxic T-cells (CD8+), natural killer (NK) cells (non-specific), and cytokines for milk microbiota. Decrease exaggerated inflammatory responses to colonizing bacteria (commensals and opportunistic pathogens under certain conditions);
Anti-allergic properties attributed to LABs and commensals of milk microbiota that decrease the occurrence and severity of allergic responses in animal models and some human studies; and
Metabolic activities of LABs and commensals essential for digestion of oligosaccharides into short chain fatty acids (SCFAs) that become an energy source for host cells in the colon, thus increasing nutrient availability and absorption for the host. Further, Sozańska [
35] cited studies demonstrating that SCFAs in the GI tract enhance the epithelial barrier function of the gut, influence bone marrow dendritic cell maturation, and inhibited Th2-dependent response, interconnecting metabolic functions to functions 2, 3, and 4 above.
In addition to the five plausible mechanisms noted above, anti-tumor properties have been observed [
34], but this class of activity is not discussed further herein.
3.1.2. Risks: Contra-Argument
No systematic reviews, cohort studies, or quantitative microbial risk assessments (QMRAs) were identified that estimated the likelihood of infectious diseases transmitted to infants from raw breastmilk.
Supporting
A recent policy statement by the American Academy of Pediatrics (AAP) [
17] included some evidence supporting the contra-argument for raw breastmilk risks. This report [
17] cited studies documenting microbial contamination [
36,
37] in breastmilk donated via milk banks. Keim and colleagues [
36] also compared prevalence and levels of microbial contamination between breastmilk purchased from from the Internet (101 samples) and breastmilk from milk banks (20 samples).
However, these studies also include some evidence attenuating the contra-argument under certain conditions. In spite of potential microbial contamination of raw breastmilk, Keim and colleagues [
36] stated that milk banks typically screen donors or donor milk for some viral diseases. In their view, the benefits of feeding unpasteurized mother’s own breastmilk to hospitalized infants ‘likely outweigh the risk of bacterial disease’. The AAP [
17] also acknowledged that breastfeeding (mother’s own milk) during hospitalization is always preferred and should be encouraged, citing five studies demonstrating loss of cells, macronutrients, anti-inflammatory factors, and potential probiotic organisms with pasteurization.
The AAP recommends only pasteurized donor milk collected from screened donors and distributed through established human milk banks. The AAP recommends against use of unpasteurized donor milk from direct, internet-based, or informal milk sharing.
The AAP cited the Human Milk Banking Association of North America (HMBANA) with overseeing nonprofit human milk banks in the US and Canada. Most HMBANA milk banks use Holder pasteurization (heating at 62.5 °C for 30 min) to process raw donor milk.
The key points developed by the AAP [
17] include preference for feeding mothers’ own milk (raw breastmilk) when available and pasteurized donor milk for preterm or ill infants whose mothers are unable to provide sufficient raw breastmilk. Other key points include screening of donors, pasteurization and post-pasteurization testing, discouragement of use of internet donor milk or raw breastmilk sharing, and access to pasteurized donor milk to appropriate high-risk infants based on medical necessity, regardless of an individual’s financial status or ability to pay.
A subsequent cohort study conducted at NICUs in Germany [
30] documented significantly decreased rates of CMV in very preterm infants whose mothers were seropositive for CMV. Mother’s own breastmilk treated with ‘short-term pasteurization’ (heating at 62 °C for 5 s; 87 infants) was associated with CMV infection in 2 of 87 (2.3%) infants in the treatment group (recruited from 2010–2012) comparted to 17 of 83 (20.5%) infants in the historical control group fed untreated (raw) mother’s breastmilk (retrospective cohort from 1995–1998).
Attenuating
Three studies include evidence attenuating the contra-argument for raw breastmilk risks. One is a brief review by Gribble and Hausman [
31] that pointed out the paucity of scientific evidence supporting the theoretical transmission of infectious disease to infants via raw breastmilk. Of viruses examined to date, cytomegalovirus (CMV), human immunodeficiency virus (HIV), and human T-cell leukemia virus (HTLV) appear to be transmitted via breastmilk, but the latter two appear to require repeated exposures over a long period of time to actually cause infection. Further, the authors state [
31] that while a majority of mothers are infected with CMV, the presence of CMV in breastmilk is only a problem for premature infants. Cohort studies were cited that estimated the rate of disease transmission at 0.6 to 4% of infants breastfed for 6 months by HIV-positive mothers. Bacterial pathogens (
Streptococcus,
Salmonella, or
Listeria) apparently rarely cause infant disease via breastmilk, and other potential bacterial pathogens that could be present in milk (
Klebsiella,
Pseudomonas,
Staphylococcus, or
Bacillus) have not been demonstrated to cause disease in infants via breastmilk.
The second attenuating study [
32] enrolled a cohort of 161 mothers and their 209 infants to test the hypothesis that the presence of microbes in expressed breastmilk was correlated with subsequent infections in preterm infants (<30 weeks post gestational age).
Milk samples (n = 813) were collected weekly and evaluated for presence and levels of Gram-positive and Gram-negative bacteria and presumptively identified using standard culture methods in the Manual of Clinical Microbiology published by the American Society for Microbiology. All isolates of potential pathogens from infant samples of blood, cerebrospinal fluid, or urine were identified. Relative risks were estimated from initial milk culture positives that appeared as a pathogen in the paired preterm infant(s).
Bacterial isolates in milk were classified first by Gram-stain reactions. Approximately 49% of milk samples had low densities of Gram-positives (<104 cfu/mL) and 5% high density Gram-positives (≥104 cfu/mL). Low density (<103 cfu/mL) Gram-negatives were not detected, and 0.4% of milk samples had high density Gram-negatives (≥103 cfu/mL). In addition, results for mixed Gram-positive and -negative milk samples were reported (17.5% low density Gram-positives and -negatives; 14.3% low Gram-positives and high Gram-negatives; 7% high density Gram-positives and -negatives; and 2.5% high density Gram-positives and low-density Gram-negatives).
Of 1963 isolates from milk, the predominant species (present at ≥5% total isolates) are summarized in
Table 1, collectively accounting for 87.3% of isolates. (Note that the pathogenic potential of the isolated strains was apparently not tested using in vitro or in vivo models or for sequence homology with known virulence genes.)
The authors reported that bacterial isolates identified in the initial milk samples were sporadic and not predictive of subsequent milk sample positives. Further, there were no significant relationships between total or specific bacteria and densities and NEC, surgery for NEC, duration of antibiotic use, time to full feeding, or duration of hospitalization. The odds of infection in infants before or after exposure to breastmilk containing potential pathogens were not significant for Gram-positive (Staphylococcus, Streptococcus) and most Gram-negative bacteria (E. coli, Enterobacter, and Klebsiella). The distributions of bacteria in milk did not match bacteria isolated from pathology samples (infant blood, cerebrospinal fluid, and urine).
Schanler and colleagues [
32] questioned the utility of screening breastmilk by traditional culture methods for predicting infectious potential. The authors and other cited studies observed that exposure to potential pathogen in breastmilk appears to cause adverse effects (infections) in few infants. Further, the authors suggest that co-occurrence of potential pathogens in breastmilk and infants may reflect common nosocomial exposures to mother and infant rather than transmission of potential pathogens from breastmilk.
The third study attenuating the contra-argument for raw breastmilk risks [
15] described a long tradition for milk banks in Norway to provide raw donor breastmilk, with strict controls and rigorous screening of donors and trace-back if needed for investigation of adverse events as practiced by blood banks.
Norwegian milk banking has continued since 1941, with rigorous donor screening comparable to screening of blood donors. Routine use of raw donor milk was reportedly based on screening of donors, a low incidence of CMV and hepatitis B and C via blood testing of donors, and the high standard of living in Norway. In addition, donors are screened for human T-cell leukemia virus (HTLV 1 and 2). Approved donors negative for these agents had not become positive upon regular re-testing, as of 2009. Norway continues to provide raw donor breastmilk as the best choice for premature or ill infants whose own mother’s milk production is insufficient.
The first author of this study provided updates on guidelines and practices used in Norway for screening donor breastmilk [
33]. Each container of donor milk (250–500 mL, received frozen) is screened for microbes (total counts and pathogens). In 2019, 698 of 4317 L of donor milk (16%) was discarded due to the presence of potential pathogens, rather than pasteurizing donor milk for feeding to preterm or full-term infants [
33].