Donor milk intake and infant growth in a South African neonatal unit: a cohort study

Pasteurized donor milk (DM) availability in South African hospitals has increased substantially during the last twenty years with the establishment of the Human Milk Banking Association of South Africa (HMBASA) [1]. Low-cost technology supports pasteurization in resource limited settings [2]. Increased access to DM through these milk banks coincides with recommendations that DM be provided to preterm and low birth weight infants when a mother’s own milk (MOM) is not available or sufficient for feedings [3, 4]. Donor milk availability can allow feedings of critically ill infants born to women in resource limited settings and with high rates of human immunodeficiency virus (HIV) to be predominantly composed of human milk [5]. Its use should be expected to occur simultaneously with lactation support given a priority for providing MOM [6, 7].

Using DM in place of formula in conjunction with mother’s milk reduced the risk of prematurity associated morbidities including necrotizing enterocolitis (NEC) and acute and chronic respiratory disorders, although not in all situations [8‐12]. Heating milk alters its composition. Loss of bile salt-stimulated lipase activity, which occurs during any heating method, reduces infant fat absorption [13]. Trace elements, growth factors and cytokines are preserved during pasteurization [14, 15]. However, bioactive hormones including insulin and adiponectin are reduced [16], as may be oligosaccharides depending on the heating method [17]. The cumulative impact of these alterations on infant health, including growth patterns, remains unclear.

The relative proportions of DM versus MOM used for feedings influence clinical outcomes resulting from DM exposure [18]. Insufficient research has evaluated the impact of DM feedings on the outcomes of hospitalized infants in resource limited settings with high rates of maternal HIV. Identifying any impact on infant growth and implications of DM availability on volumes of MOM fed to high-risk infants would establish important foundations upon which nutrition provided to these infants may be improved. Thus, we aimed to measure growth in high-risk infants exposed to DM and the associated nutritional intake patterns when cared for in a sub-Saharan secondary care center managing women from a resource limited background with a high burden of HIV.

Donor milk consent existed for one hundred five infants during the study period (Table 1). Amongst mothers of these infants (n = 99; 27.9 ± 6.3 years old), 40 % were HIV positive (Table 1). Infant birth weight differed by maternal HIV status (HIV positive: 1790 ± 859 g; HIV negative: 1109 ± 390 g; p < 0.001). The most common infant indication for prescribing DM was prematurity (Table 2). Noteworthy infant outcomes included low rates of NEC and infrequent use of PN (Table 1).

While concern remains that DM feedings negatively impact growth for human milk-fed infants, this analysis suggests that increasing amounts of DM to > 50% of feedings in the first month does not slow infant growth. However, it is noteworthy that the GV of all infants in this South African cohort falls below fetal reference ranges [22]. For those infants with a signed consent in the MMH NU, DM availability facilitated high rates of feedings as human milk during the first 28 days. We detected no difference in GV related to maternal HIV status. We believe the difference in infant characteristics, including birth weight, based on maternal HIV status to be a consequence of infant inclusion criteria for DM at MMH. Infants of mothers with HIV may receive DM while an infant with a comparable degree of illness, born to a mother without HIV, might not qualify. Specifically, an ill infant born to a mother with HIV could receive DM while DM would not be used for the ill infant born to a mother without HIV if sufficient MOM were available. In a population of women with high rates of HIV infection, DM availability did not appear to decrease amounts of MOM fed to their infants. Availability of DM has previously improved rates of breastfeeding at discharge [9].

Implementation and acceptance of DM feedings in communities with high rates of maternal HIV and other clinical settings appears feasible [23]. Still, clinician sensitivity to women’s concerns towards use of DM feedings is important. Parents relay concerns about disease transmission through milk, including HIV, yet increasing parental familiarity with screening and pasteurization processes may reduce such concern [23]. Independent of safety concerns, using DM creates stress for many mothers resulting from the need to rely on milk expressed from other women [24]. Without surprise, affording parents sufficient time and opportunities to discuss DM with clinicians helps clarify misunderstandings and may ease concerns. As families continue to consider DM as a viable option for feedings, detailing growth and outcomes for infants using DM in sub-Saharan Africa is also necessary for improving infant care.

Donor milk feedings should not be singled out as the sole contributor to the measured GV of these infants. Prompt initiation, advancement and fortification of feedings contributed to a weekly caloric intake that ultimately exceeded recommendations [21], yet not until after the first week. Also, as VLBW infants made up a considerable proportion of this cohort, average caloric intake suggests that infants with larger birth weights achieve higher caloric intake than smaller infants. Thus, early caloric intake appears to be a modifiable factor in efforts aimed at improving growth. Early PN is a commonly used method for minimizing early caloric deficits [25]. However, resources and personnel required to maintain safe intravenous infusions currently precludes the use of PN in many infants. Protein concentrations in DM, even when combined with human milk fortifiers, may not meet estimated requirements [21, 26]. Also, in consideration of the standards for processing human milk used in the MMH NU, much of the milk used (any DM, all stored MOM regardless of a mother’s HIV status and all expressed milk from HIV positive mothers) undergoes some method of heating prior to infant feeding. In addition, the preterm infant’s endogenous lipase capabilities are limited [27]. Cumulatively, these combined factors would be expected to exacerbate fat malabsorption as opposed to other clinical scenarios where MOM is not heated. Access to a safe adjuvant to facilitate fat absorption would be expected to mitigate the poor growth observed in comparable populations, however, a well-tolerated option is not currently available [28].

Delayed availability of MOM was the most common maternal indication for DM feedings. This is consistent with other reports [1]. Delayed onset of lactation after preterm delivery is not uncommon [29]. In this context, the establishment of DM availability for preterm infants during insufficient supply of MOM is consistent with global public health recommendations [3, 4]. Indicators such as rates of breastfeeding at time of discharge from neonatal units suggest DM availability may improve use of MOM [9, 30], but in some scenarios it appears to hinder intake of MOM [30]. It is possible that discussions of DM with parents emphasize the importance of human milk for feedings, and when combined with lactation support, increased intake of MOM should be expected in some settings [31]. Thus, ongoing lactation support is critical for women who deliver critically ill, preterm infants and for women who may be temporarily separated due to infant transfer to another facility in the immediate postnatal period [32].

In this cohort of infants with consent to use DM, a small but meaningful proportion (5%) of infants ultimately never used DM and received their mother’s milk exclusively. This scenario of anticipated need for supplemental milk without eventual need has been documented in the setting of large, multicenter randomized controlled trials. In a Dutch trial comparing DM versus formula to fill the gap when needed for feedings, 3.2% of randomized infants received all feedings as MOM [33]. In a separate Canadian trial comparing DM and formula, approximately 28% of infants ended up feeding only their mother’s milk [34]. These subsets of infants and their mothers appear worthy of attention in efforts to define clinical scenarios in which a clinician may predict the need for supplementation and yet ultimately there was no need.

Approximately half of infants received formula at some point during hospitalization. Feeding composition during the first month suggests that formula use occurred later in hospitalization when needed. High rates of human milk feedings and later exposure to formula may in part explain the low rates of NEC in this population [8]. In comparison, overall rates of NEC for VLBW infants cared for in the MMH NU are 1.6%.

The long-term implications of the measured growth outcomes remain to be determined, although infant growth rates act as surrogate markers of whether brain development is optimal. Greater weight gain and head growth during initial hospitalization were previously associated with higher developmental testing scores in preterm infants [35]. Although length data were not available for these infants, linear growth during hospitalization as well as post discharge is a valid predictor of neurodevelopmental impairment in preterm infants [36].

Inherent limitations must be acknowledged related to the retrospective study design, including missing data. Also, despite considerable detail of nutrition and fluid intake during the first 28 days, daily intake throughout hospitalization may have revealed additional nutritional factors that contributed to the observed growth patterns. Additionally, growth was only measured during the hospital stay. Post discharge outcomes are necessary to draw robust conclusions about any lasting implications of early nutrition. This investigation was unable to identify infants who may have qualified for DM feedings but parental preference was to not provide consent. Similarly, we could not ascertain reasons why parents may have preferred not to consent for DM in those scenarios. However, clinicians in the MMH NU observe this to be an uncommon occurrence. Regardless, a prospective investigation examining the percentage of guardians choosing not to consent for DM, their reasoning why, and how these infants grow relative to those who do have signed consents would be important for a comprehensive understanding of the implications of DM feedings.

In conclusion, the use of DM for feeding infants managed in a secondary South African Neonatal Unit was associated with suboptimal infant growth. As intended, DM availability allows delayed exposure to formula and can contribute to maintaining high proportions of human milk feedings for high-risk infants. The consideration of all circumstances surrounding the nutritional management in any single center appears vital to understanding how to optimize infant outcomes associated with DM feedings. Lactation support, relative amounts of a mother’s milk availability and how MOM is handled prior to feedings likely create unique and individual circumstances which require attention during clinical and research efforts aimed at improving feeding quality and infant outcomes.