Abstract
Background
Human milk feeding proportions among preterm infants remain unsatisfactory, and donor human milk is recommended as the best alternative when mother’s own milk is insufficient. Although established human milk banks provide essential donor human milk services, some settings may face challenges related to donor continuity and the timely alignment of donor milk supply with recipient demand. We therefore developed an in-hospital peer-supported human milk bank management model featuring a structured recipient-to-voluntary-donor pathway within the same neonatal unit.
Methods
This peer-supported model was implemented at Shenzhen Maternity and Child Healthcare Hospital in January 2024. In this retrospective analysis, extremely preterm infants admitted from January to December 2023 served as the control group and received the usual feeding practice, with mother’s own milk supplemented by preterm formula when insufficient. Infants admitted from January to December 2024 comprised the human milk bank group and received donor human milk from the in-hospital peer-supported bank when maternal milk was insufficient. Clinical outcomes including human milk feeding proportions, necrotizing enterocolitis, and bronchopulmonary dysplasia were compared between the two groups to examine the associations between implementation of the model and human milk feeding proportions and clinical outcomes.
Results
Human milk feeding proportions were significantly higher in the human milk bank group than in the control group during the first 7 days of life (100% vs. 49%, P < 0.001) and during the first 28 days of life (100% vs. 98%, P < 0.001). Total enteral intake volume was significantly higher in the human milk bank group than in the control group during the first 7 days of life (44.5 mL vs. 38.5 mL, P = 0.002), and during the first 28 days of life (1377.8 mL vs. 1079.0 mL, P < 0.001). The incidence of moderate to severe bronchopulmonary dysplasia was significantly lower in the human milk bank group (OR = 0.533, 95% CI 0.291–0.978, P = 0.042), whereas the adjusted association with stage II or above necrotizing enterocolitis was not statistically significant (OR = 0.226, 95% CI 0.042–1.224, P = 0.084).
Conclusions
The in-hospital peer-supported human milk bank model was significantly associated with higher human milk feeding proportions during the first 7 days of life and during the first 28 days of life, as well as lower odds of moderate-to-severe bronchopulmonary dysplasia in extremely preterm infants. This model may represent a simple, peer-supported, and potentially scalable strategy for optimizing human milk feeding support in neonatal intensive care units.
Keywords: Human milk feeding, Clinical outcomes, Donor human milk, Extremely preterm infants, Human milk bank
Introduction
Preterm birth remains a major global public health challenge [1]. Extremely preterm infants (EPI) with gestational age < 28 weeks account for approximately 0.6% of all live births but face higher risks of mortality and long-term complications in the neonatal intensive care unit (NICU) [2]. Although survival rates have improved with advances in perinatal medicine, enhancing quality of life for these infants continues to be a significant challenge. Human milk is the preferred enteral nutrition for preterm infants and has been shown to improve both short- and long-term outcomes [3, 4]. However, mothers of extremely preterm infants frequently experience delayed lactogenesis and insufficient milk supply, with reported rates of delayed lactogenesis ranging from 35.67% to 42.40% [5]. When mother’s own milk is unavailable, international organizations including the American Academy of Pediatrics (AAP), ESPGHAN, WHO, and UNICEF recommend donor human milk (DHM) as the best alternative [6]. DHM has been demonstrated to reduce major complications such as necrotizing enterocolitis (NEC) and bronchopulmonary dysplasia (BPD) compared with preterm formula [7].
Although established human milk banks provide essential donor human milk services, some settings may experience challenges related to donor continuity, operational logistics, and the timely alignment of donor milk supply with recipient demand. These include donor instability, high operational costs, complex logistics, and the fact that milk from mothers of full-term infants often has lower protein and energy content than preterm milk, which may not fully meet the nutritional demands of extremely preterm infants [8–11]. In China, human milk banks operate under a “voluntary donation, free use” model, yet public concerns regarding safety and low donation willingness further restrict sustainable supply.
Hospital-based human milk banks and lactation support services are well established in many countries and regions. Conventional human milk bank models generally follow a donor–milk bank–recipient pathway, in which donor recruitment and recipient allocation are managed as separate processes. Donors must complete the serological screening required by human milk banking guidelines, and donated milk must undergo appropriate quality testing before it can be accepted for use [12]. In contrast, our model involves mothers of preterm infants cared for within the same neonatal unit. After the feeding needs of their own infants have been met, these mothers may voluntarily donate surplus milk; similarly, mothers whose infants initially receive donor human milk may later become voluntary donors once their own milk supply is sufficient. This creates a virtuous cycle of receiving support, achieving sufficient milk supply for one’s own infant, and subsequently supporting others through voluntary donation. With donor consent, valid prenatal serological test results already available in the medical record may also be reviewed in accordance with institutional screening criteria, potentially reducing the need for repeat testing and the associated procedures and costs. This approach may help sustain donor participation and stabilize the supply of donor human milk. We therefore aimed to describe the development, implementation, and operational features of this in-hospital peer-supported human milk bank model and to evaluate its association with human milk feeding proportions and major clinical outcomes in extremely preterm infants using a retrospective before-and-after design.
Methods
This retrospective cohort study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.
Facilities and management
The peer-supported human milk bank for preterm infants at our institution was established in 2024 and is based in the neonatal department. This model is operated and managed by a multidisciplinary team of eight members, including neonatologists, nurses, researchers, and social workers, responsible for running the program and ensuring quality management. The facilities are functionally divided into a milk expression room, a processing room, a laboratory, and a milk storage room. Equipment includes hospital-grade breast pumps, milk storage containers, a fully automated human milk analyzer (SD-120T), a pasteurizer (water bath), medical refrigerators (standard and ultra-low temperature), as well as computers and an information management system.
A Hazard Analysis and Critical Control Points (HACCP) system was introduced during implementation, together with a dedicated electronic information system. This system allows for the electronic recording and tracking of the entire process of donor milk usage and related data, thereby ensuring traceability and safety in milk utilization.
Donor selection and milk sourcing
The donor milk provided through the in-hospital peer-supported human milk bank is obtained from mothers of preterm infants cared for in the same neonatal department. Eligible mothers voluntarily provide written informed consent and may donate surplus milk only after the nutritional needs of their own infants have been fully met. After the infant is discharged, any remaining surplus milk may be donated to the human milk bank. Donated milk is stored and transported in accordance with standardized procedures. Under the peer-supported model, mothers whose infants initially receive donor human milk may later voluntarily become donors once their own milk supply becomes sufficient and the feeding needs of their own infants have been met. This structured recipient-to-voluntary-donor pathway is intended to support donor continuity and stabilize the supply of donor human milk within the neonatal unit. Donors undergo standardized eligibility screening, and all donated milk is pasteurized, bacteriologically tested, stored, and traced in accordance with human milk bank procedures to ensure its quality and safety.
Human milk handling procedures
The operational model of the in-hospital peer-supported human milk bank consists of five key steps. All donor screening, human milk collection, processing, storage, and distribution procedures are conducted in accordance with standardized human milk bank procedures. The first step involves donor recruitment and screening, including education about human milk donation, obtaining informed consent from potential donors, initial health screening, and training in donation procedures. The second step involves the standardized collection and transportation of donated human milk, including equipment cleaning and disinfection, appropriate storage, accurate labeling and documentation, and transportation in accordance with established protocols. The third step involves inspection of donated milk upon receipt, control of pre-storage and storage conditions, pasteurization using the standard Holder method at 62.5 °C for 30 min in a dedicated water-bath pasteurizer, as recommended by the European Milk Bank Association (EMBA) [12], followed by bacteriological testing and ongoing quality tracking. The fourth step involves the standardized allocation and use of donated human milk according to predefined eligibility criteria and clinical indications, together with monitoring and documentation of feeding tolerance and clinical responses. The final step involves integrated information and quality management through an electronic information system, regular staff training, and routine quality-control procedures to ensure adherence to established standards. The specific operational steps are illustrated in Fig. 1.
Fig. 1.
Workflow of the in-hospital peer-supported human milk bank model
Donor eligibility and screening
The recruitment and screening of donors for the peer-supported human milk bank for preterm infants involve a multi-tiered standardized process. Educational outreach occurs through various channels, including hospital bulletin boards, breastfeeding manuals, teaching screens in the neonatal department, parent classes, and phone communications. Researchers explain the study’s purpose, process, and information security to potential donors, who then sign an informed consent form. To ensure donor safety, medical records of mothers of preterm infants are reviewed at the time of delivery, focusing on serological test results to confirm negative screenings for HIV, hepatitis B, hepatitis C, syphilis, and other infectious diseases. Additionally, potential donors are screened for risk factors such as smoking, alcohol use, excessive caffeine intake, history of blood transfusions or organ transplants, and exposure to high-risk infectious diseases. Qualified donors receive training in hand hygiene, as well as the proper methods for collection, storage, and transportation of milk. Donations must be postponed in cases of acute infections, after vaccinations, or if the donor is on medication.
Human milk collection, storage, and transport
To standardize the collection and transportation of human milk and ensure safety, neonatal nurses provide unified training and assessments for donors. Proper hand hygiene and breast cleanliness are required, with a preference for using medical-grade electric breast pumps. Donors are instructed to pump at least eight times daily, including at night. All equipment must be thoroughly cleaned and boiled for disinfection after use. Milk is stored in clean, dry, and disinfected food-grade containers with screw caps, not filled to the top. It should be sealed and refrigerated immediately after collection, stored separately from other food items, and clearly labeled with the donor’s name, hospital number, and pumping time. Family members are required to record the daily temperature of the refrigerator, and the milk is transported to the human milk bank using an insulated container with ice packs. Hospitalized mothers can directly deliver fresh milk to the bank.
Human milk processing and storage
The processing and quality control of human milk at the peer-supported human milk bank for preterm infants follow a fully standardized management process. All equipment is regularly validated, calibrated, and maintained, with records kept on file. When received, the labels, collection time, refrigeration conditions, container integrity, and appearance of the human milk are verified. Qualified human milk is immediately stored and labeled accordingly. Mother’s own milk is prioritized for feeding the donor’s own infant. After the infant is discharged, any remaining surplus milk intended for donation is collected by designated staff for standardized pasteurization and bacteriological testing. Storage conditions are determined according to time and temperature: freshly expressed human milk may be stored at room temperature (≤ 25 °C) for up to 4 h, refrigerated at 4 °C for up to 96 h, or frozen at − 18 °C for up to 6 months. Frozen milk should be slowly thawed in a refrigerator, stored for no more than 24 h after thawing, and heated using a water bath (37–40 °C) [13] while being gently agitated. Microwaving for thawing is prohibited to avoid water contact with the bottle caps [14].
Donated human milk utilization
The distribution and use of donor human milk from the in-hospital peer-supported human milk bank are managed according to standardized protocols. Written informed consent is obtained from the infant’s parent or legal guardian before donor human milk is used. A physician submits a request based on clinical indications, after which the human milk bank allocates donor human milk according to the recipient’s clinical needs. Specialized nurses administer the milk while monitoring feeding tolerance and documenting the feeding volume and duration. This model incorporates a structured recipient-to-voluntary-donor pathway. Mothers whose infants initially receive donor human milk may later voluntarily donate surplus milk once their own milk supply becomes sufficient, the nutritional needs of their own infants have been met, and they meet the donor eligibility criteria. Ongoing lactation support and follow-up facilitate voluntary donation when eligible mothers have surplus milk available.
Data management and quality assurance
An integrated information management system has been established for the human milk bank. The information system digitally tracks donor and recipient data, as well as the entry and exit of donated human milk and quality events, with records maintained for at least five years. All personnel undergo standardized training and documentation. The system relies on multidisciplinary collaboration among neonatology, infectious diseases, laboratory testing, obstetrics and gynecology, nutrition, and nursing departments to regularly assess human milk usage demands and monitor adverse reactions. Ensuring the safe operation of the human milk bank involves inspections of operational protocols, equipment maintenance, and key processes such as pasteurization, heating, cooling, and thawing.
The donor model for the peer-supported human milk bank for preterm infants was established in January 2024 and has been continuously operational in the neonatal department of the hospital. Based on clinical feedback, the working group members have made ongoing improvements and optimizations to the model, which has now reached a stable operational state. Through an iterative improvement process based on feedback from healthcare providers, the donation system has achieved sustainable functionality in the neonatal intensive care environment. This study was approved by the Ethics Committee of Shenzhen Maternity and Child Healthcare Hospital affiliated with Southern Medical University (LLYJ2022-080-038) and was conducted in accordance with the Declaration of Helsinki. Informed consent was obtained from all individual participants included in the study.
Outcome measures
Participants
A retrospective before-after observational study was conducted on all eligible extremely preterm infants admitted to the Neonatal Department of Shenzhen Maternity and Child Healthcare Hospital between January 2023 and December 2024. The inclusion criteria were: (1) gestational age < 28 weeks; (2) admission to the neonatal department within 2 h after birth; and (3) written informed consent provided by a parent or legal guardian. Infants were excluded if they: (1) died before discharge or were transferred to another hospital; (2) had unavailable or incomplete medical records; (3) were discharged against medical advice before completion of outcome assessment; or (4) did not receive enteral feeding during the first 28 days after birth.
Model implementation
The control group consisted of EPI admitted from January to December 2023. During this period, the in-hospital peer-supported human milk bank had not yet been established. Therefore, no donor human milk was used. Infants received the usual feeding practice, with mother’s own milk as the primary source of enteral nutrition and preterm formula provided when mother’s own milk was insufficient. The human milk bank group included EPI admitted from January to December 2024, who received human milk from the peer-supported human milk bank as a supplement when their own human milk was insufficient.
Data collection
During the study, demographic and clinical characteristics of the extremely preterm infants were collected, including sex, gestational age, birth weight, length of hospital stay, Apgar scores at birth, and other relevant parameters. Researchers also retrospectively collected feeding-related indicators, including total enteral intake volumes, total human milk feeding proportions, mother’s own milk (MOM) feeding proportions, and donor human milk (DHM) feeding proportions during the first 7 days of life and during the first 28 days of life. The total human milk feeding proportion for each period was calculated as the combined volume of MOM and DHM divided by the total enteral intake volume during the corresponding period, multiplied by 100%. The MOM and DHM feeding proportions were calculated separately by dividing the respective milk volume by the total enteral intake volume during the corresponding period, multiplied by 100%. Growth and nutritional outcomes, including discharge weight and weight gain at 4 weeks, and major complications, including bronchopulmonary dysplasia (BPD), retinopathy of prematurity (ROP), necrotizing enterocolitis (NEC), intraventricular hemorrhage (IVH), early-onset sepsis (EOS), and late-onset sepsis (LOS), were also collected.
Complication diagnostic criteria
BPD [15] was classified based on the NICHD criteria, with oxygen dependency graded at 36 weeks corrected gestational age as mild (room air), moderate (oxygen < 30%), and severe (oxygen ≥ 30% or positive pressure ventilation). ROP [16] was graded according to the international classification standard ICROP-3, divided into stages 1–5 by severity and zones I-III by anatomical position, including Plus disease classification. NEC [17] was diagnosed using a modified Bell staging system: stage I (suspected), stage II (confirmed, divided into IIA/IIB), and stage III (progressive, divided into IIIA/IIIB) based on clinical, radiological, and laboratory evaluations. IVH [18] was classified using the Papile grading system as grade I (subependymal hemorrhage), grade II (intraventricular hemorrhage without dilation), grade III (intraventricular hemorrhage with dilation), and grade IV (periventricular hemorrhagic infarction), diagnosed via cranial ultrasound. EOS and LOS were defined as sepsis occurring within or after 72 h of life, with the gold standard being a positive blood culture, combined with clinical symptoms and inflammatory markers [19].
Statistical analysis
Statistical analyses were performed using SPSS 27.0 software. The Shapiro–Wilk test was used to evaluate the normality of continuous variables and the two-sided independent-samples t test and the Mann–Whitney U test to analyze continuous variables according to the normality. Chi-square or Fisher’s exact tests were used to compare categorical data, as appropriate. To control for potential confounders, multivariable logistic regression analysis was performed. Covariates were selected a priori based on clinical relevance and established literature identifying them as key confounders for BPD and NEC in extremely preterm infants [20, 21], as well as clinical relevance to our human milk bank intervention. A post hoc power analysis was performed for the human milk feeding proportion during the first 7 days of life using G*Power 3.1. Based on the observed between-group difference, an estimated effect size of Cohen’s d ≈ 0.51, sample sizes of 84 and 128, and a two-sided α of 0.05, the achieved power (1 − β) was 0.951. The significance level for all tests was set at α = 0.05.
Results
Population clinical characteristics
A total of 251 extremely preterm infants were identified during the study period. After 39 infants were excluded, 212 infants were included in the final analysis (Fig. 2). In the human milk bank group, the gestational age was 26.1 (25.1–27.1) weeks, and in the control group it was 25.8 (24.3–27.1) weeks. The birth weight was 815 ± 189 g in the human milk bank group and 827 ± 208 g in the control group. There were no significant differences between the two groups in terms of sex, gestational age, birth weight, mode of delivery, 1 min, 5 min Apgar scores, intrauterine distress, length of hospital stay, or amniotic fluid status (Table 1).
Fig. 2.
Flowchart of participant selection
Table 1.
Clinical characteristics of EPI in both groups
| Characteristics | Control Group | Human Milk Bank Group | P |
|---|---|---|---|
| (n = 84) | (n = 128) | ||
| Male (%) | 52 (61.9) | 74 (57.8) | 0.553 |
| Gestational Age (wk) | 25.8 (24.3–27.1) | 26.1 (25.1–27.1) | 0.343 |
| Birth Weight (g) | 827 ± 208 | 815 ± 189 | 0.662 |
| Cesarean Section (%) | 31 (36.9) | 64 (50.0) | 0.061 |
| 1 min Apgar Score | 6 (5–8) | 7 (5–8) | 0.110 |
| 5 min Apgar Score | 9 (7–9) | 9 (8–9) | 0.349 |
| Intrauterine Distress (%) | 3 (3.6) | 12 (9.4) | 0.107 |
| Length of Hospital Stay(days) | 86 (60.0-101.8) | 90.5 (75.3–103.8) | 0.135 |
| Clear Amniotic Fluid, n (%) | 72 (85.7) | 101 (78.9) | 0.211 |
Note: Data are presented as mean ± SD, median (IQR), or n (%), as appropriate
Human milk feeding proportions were significantly higher in the human milk bank group than in the control group during the first 7 days of life (100% vs. 49%, P < 0.001) and during the first 28 days of life (100% vs. 98%, P < 0.001). There were no significant between-group differences in MOM feeding proportions during either the first 7 days of life (P = 0.429) or the first 28 days of life (P = 0.464). In the human milk bank group, the median DHM feeding proportions were 30.7% during the first 7 days and 1.4% during the first 28 days, whereas DHM was unavailable in the control group. Total enteral intake volume was significantly higher in the human milk bank group than in the control group during the first 7 days of life (44.5 mL vs. 38.5 mL, P = 0.002) and during the first 28 days of life (1377.8 mL vs. 1079.0 mL, P < 0.001). The incidence of stage II or above NEC was lower in the human milk bank group than in the control group (1.6% vs. 7.1%, P = 0.037), and the incidence of moderate-to-severe BPD was also lower in the human milk bank group (25.8% vs. 40.5%, P = 0.024). No significant differences were observed between the two groups in early-onset sepsis, late-onset sepsis, stage III or above IVH, ROP, discharge weight, or weight gain at 4 weeks (Table 2).
Table 2.
Feeding Outcomes and Major Complications in Extremely Preterm Infants
| Variables | Control Group | Human Milk Bank Group | P |
|---|---|---|---|
| (N = 84) | (N = 128) | ||
| Human milk feeding proportion during the first 7 days (%) | 49 (0, 73) | 100 (100, 100) | < 0.001 |
| MOM feeding proportion during the first 7 days of life (%) | 48.7 (0.0, 73.1) | 62.4 (0.0, 79.4) | 0.429 |
| DHM feeding proportion during the first 7 days of life (%) | 0.0 (0.0, 0.0) | 30.7 (9.5, 83.2) | — |
| Human milk feeding proportion during the first 28 days (%) | 98 (95, 99) | 100 (99, 100) | < 0.001 |
| MOM feeding proportion during the first 28 days of life (%) | 98.1 (95.0, 99.2) | 98.0 (84.9, 99.3) | 0.464 |
| DHM feeding proportion during the first 28 days of life (%) | 0.0 (0.0, 0.0) | 1.4 (0.5, 5.1) | — |
| Total enteral intake during the first 7 days of life (mL) | 38.5 (16.4, 50.0) | 44.5 (27.0, 59.8) | 0.002 |
| Total enteral intake during the first 28 days of life (mL) | 1079.0 (584.9, 1413.8) | 1377.8 (985.3, 1773.8) | < 0.001 |
| EOS, n (%) | 14 (16.7) | 12 (9.4) | 0.113 |
| LOS, n (%) | 3 (3.6) | 2 (1.6) | 0.346 |
| Stage II or above NEC, n (%) | 6 (7.1) | 2 (1.6) | 0.037 |
| Stage III or above IVH, n (%) | 8 (9.5) | 5 (3.9) | 0.095 |
| Moderate-to-severe BPD, n (%) | 34 (40.5) | 33 (25.8) | 0.024 |
| ROP, n (%) | 11 (13.1) | 9 (7.0) | 0.140 |
| Discharge Weight (g) | 2795 (2166.25–3255.00) | 2885 (2476.25-3427.50) | 0.077 |
| Weight gain at 4 weeks (g) | 352.0 (272.0, 419.8) | 372.5 (310.0, 440.0) | 0.105 |
Note: Data were displayed as median (Interquartile Range, IQR) or number (%)
Abbreviations: MOM, mother’s own milk; DHM, donor human milk. EOS, early-onset sepsis; LOS, late-onset sepsis; NEC, necrotizing enterocolitis; IVH, intraventricular hemorrhage; BPD, bronchopulmonary dysplasia; ROP, retinopathy of prematurity. Donor human milk was unavailable in the control group; therefore, the DHM feeding proportion was structurally zero in this group, and no between-group statistical comparison was performed. Weight gain at 4 weeks was calculated as body weight at day 28 minus birth weight
Multivariable logistic regression analysis showed that, after adjusting for gestational age, sex, 1 min Apgar score, early-onset sepsis, late-onset sepsis, and weight gain at 4 weeks, the human milk bank group was associated with lower odds of moderate to severe BPD (OR = 0.533, 95% CI 0.291–0.978, P = 0.042). After adjustment for potential confounders, early-onset sepsis was associated with higher odds of stage II or above NEC (OR = 6.53, 95% CI 1.42–30.05, P = 0.016), whereas the association between the human milk bank group and stage II or above NEC was not statistically significant (OR = 0.226, 95% CI 0.042–1.224, P = 0.084). Given the small number of NEC events, these estimates should be interpreted cautiously because of potential model instability (Table 3).
Table 3.
Multivariable logistic regression analyses of BPD and NEC
| Outcome Variable | Predictor Variable | B | Wald | OR | 95% CI | P |
|---|---|---|---|---|---|---|
| Moderate-to-severe BPD | Human Milk Bank Group | -0.629 | 4.13 | 0.533 | 0.291–0.978 | 0.042 |
| Male | 0.384 | 1.49 | 1.47 | 0.792–2.72 | 0.222 | |
| 1 min Apgar Score | -0.077 | 1.08 | 0.926 | 0.801–1.071 | 0.299 | |
| Gestational age | -0.151 | 1.67 | 0.860 | 0.684–1.081 | 0.197 | |
| Early-onset sepsis | 0.075 | 0.026 | 1.08 | 0.435–2.673 | 0.871 | |
| Late-onset sepsis | -0.561 | 0.232 | 0.571 | 0.058–5.592 | 0.630 | |
| Weight gain at 4 weeks | 0.001 | 0.552 | 1.001 | 0.999–1.002 | 0.457 | |
| Stage II or above NEC | Human Milk Bank Group | -1.49 | 2.98 | 0.226 | 0.042–1.224 | 0.084 |
| Gestational Age | 0.305 | 0.970 | 1.36 | 0.739–2.49 | 0.325 | |
| Male | -0.238 | 0.072 | 0.788 | 0.139–4.47 | 0.788 | |
| 1 min Apgar Score | -0.140 | 0.711 | 0.869 | 0.628–1.204 | 0.399 | |
| Early-onset Sepsis | 1.88 | 5.81 | 6.53 | 1.42–30.05 | 0.016 | |
| Late-onset Sepsis | -17.99 | 0.000 | 0.000 | — | 0.999 |
Note: BPD, bronchopulmonary dysplasia; NEC, necrotizing enterocolitis; OR, odds ratio; CI, confidence interval. The BPD model was adjusted for gestational age, sex, 1 min Apgar score, early-onset sepsis, late-onset sepsis, and weight gain at 4 weeks. The NEC model was adjusted for gestational age, sex, 1 min Apgar score, early-onset sepsis, and late-onset sepsis. OR < 1 indicates lower odds of the outcome, whereas OR > 1 indicates higher odds. Statistical significance was set at P < 0.05
Discussion
This retrospective before-and-after study showed that implementation of the in-hospital peer-supported human milk bank model was associated with higher human milk feeding proportions during the first 7 days of life and during the first 28 days of life, as well as lower adjusted odds of moderate-to-severe bronchopulmonary dysplasia in extremely preterm infants. Specifically, compared with the control group, the human milk bank group had a higher median human milk feeding proportion during the first 7 days of life (100% vs. 49%, P < 0.001) and was associated with 46.7% lower adjusted odds of moderate-to-severe BPD (OR = 0.533, 95% CI 0.291–0.978, P = 0.042). Although the unadjusted incidence of stage II or above NEC was lower in the human milk bank group, the association was not statistically significant after adjustment for potential confounders (OR = 0.226, 95% CI 0.042–1.224, P = 0.084), likely because of the small number of NEC events (n = 8).
Evidence suggests that greater human milk exposure and higher human milk feeding proportions may be associated with lower risks of BPD and NEC [7, 22, 23]. Mechanistically, York et al. reviewed evidence suggesting that bioactive growth factors in human milk may contribute to NEC prevention by promoting intestinal growth and epithelial barrier integrity, supporting microvascular development, and modulating immune maturation [24]. Although previous studies have primarily focused on these clinical associations and potential biological mechanisms, few have examined practical approaches to improving early human milk feeding proportions in extremely preterm infants. Following implementation of the in-hospital peer-supported human milk bank model, DHM became available as a supplement when MOM was insufficient. The median total human milk feeding proportion during the first 7 days of life reached 100% in the human milk bank group, compared with 49% in the control group. When MOM and DHM were examined separately, MOM feeding proportions did not differ significantly between groups at either time point. In the human milk bank group, DHM accounted for 30.7% of enteral intake during the first 7 days and 1.4% during the first 28 days, suggesting that the higher total human milk feeding proportion was mainly associated with early DHM supplementation when MOM was insufficient, without an apparent reduction in MOM provision. Total enteral intake volume was also significantly higher in the human milk bank group during the first 7 days of life (44.5 mL vs. 38.5 mL, P = 0.002) and during the first 28 days of life (1377.8 mL vs. 1079.0 mL, P < 0.001).
Human milk bank services may operate as hospital-based, community-based, regional, satellite, or blood bank–integrated systems, and formal services generally follow standardized procedures for donor screening, milk collection, pasteurization, storage, distribution, and quality assurance [25]. These regulated services should be distinguished from informal peer-to-peer milk-sharing networks, including those facilitated through social media platforms. Informal peer-to-peer milk sharing may not consistently include standardized donor screening, pasteurization, or controlled milk handling and storage, potentially increasing the risks of infectious exposure and bacterial contamination [26]. Establishing and operating formal human milk banks may require substantial infrastructure and operational resources [27], whereas satellite-center models may reduce equipment and maintenance costs through resource sharing with a central human milk bank [28]. Blood bank–integrated models represent another coordinated hospital-based approach but require integration across clinical, laboratory, and quality-management systems [29].
Compared with conventional donor–milk bank–recipient pathways, the peer-supported model incorporates a structured transition from recipient mother to voluntary donor within the same neonatal unit and may therefore offer several operational advantages. First, safety remains the primary consideration. Since all donors are mothers admitted to the hospital, their baseline health information has already been obtained during prenatal examinations, which may reduce the need for repeat serological testing when valid prenatal test results are available and meet institutional screening criteria. Second, because donor milk was collected and used within the same neonatal unit, this model may shorten the interval between donation and recipient use, facilitate rapid inventory turnover, and reduce the duration of frozen storage. Third, this model incorporates a structured recipient-to-voluntary-donor pathway: mothers whose infants initially receive donor human milk may later voluntarily donate surplus milk once their own milk supply is sufficient and the feeding needs of their own infants have been met. This mutual support mechanism may increase willingness to donate, improve the overall availability of donor human milk, and accelerate the turnover rate of milk bank usage. Fourth, because this model operates within the neonatal unit, it allows timely DHM supplementation when MOM is insufficient, thereby facilitating early access to human milk.
Finally, the in-hospital model may simplify some operational processes and support supply stability by minimizing external transportation and reducing the need for repeated donor recruitment. All donor milk was provided voluntarily by mothers whose infants were cared for in the same neonatal unit. In addition, existing prenatal serological results could be reviewed in accordance with institutional screening procedures when they remained valid, potentially avoiding unnecessary duplicate testing. Daili et al. [8] reported that the operating cost of a human milk bank in China is approximately US$168 per liter. However, because no formal cost-effectiveness analysis was conducted in our study, the economic impact of this model remains uncertain and should be evaluated in future research.
However, this peer-supported model also has potential drawbacks. Theoretically, donor mothers may experience a certain psychological burden. Under this model, however, donation is permitted only after mothers have fully met their own infant’s nutritional needs, participation is entirely voluntary, and ongoing psychosocial support is provided. During routine clinical interactions, some donor mothers informally expressed positive feelings about helping others; however, their psychological experiences were not systematically assessed in this study. Future studies should therefore formally evaluate the psychological responses and experiences of donor mothers. Furthermore, the current model was implemented in a single center, and its feasibility and scalability in multicenter or regional networks require further evaluation.
Limitations
Several limitations of this study should be acknowledged. First, although a historical control design was used, the time span was only two years, and during this period there were no major changes in medical and nursing practices that could have biased the comparison. Second, the sample size was relatively small, particularly for NEC, with only eight events in total. The multivariable analysis of NEC was therefore limited by the small number of events and potential model instability. Although the adjusted association between the human milk bank group and NEC did not reach statistical significance (P = 0.084), the adjusted estimates should be interpreted as exploratory. Third, we did not perform a formal cost-effectiveness or economic analysis, because the operational costs of our center’s donor human milk bank model could not be accurately estimated at this stage, precluding an objective economic comparison with other established human milk bank models. Future studies with larger cohorts, longer follow-up, and comprehensive economic evaluations are warranted to validate and extend our findings.
Conclusion
Implementation of the in-hospital peer-supported human milk bank management model was associated with higher human milk feeding proportions during the first 7 days of life and during the first 28 days of life, as well as lower adjusted odds of moderate-to-severe BPD in extremely preterm infants. As a simple and peer-supported approach, this model may have practical value for optimizing human milk feeding support in neonatal intensive care units.
Acknowledgements
We thank the multidisciplinary team at Shenzhen Maternity and Child Healthcare Hospital—including neonatal physicians, nurses, human milk bank staff, and administrative personnel—for their essential contributions to establishing and operating the in-hospital peer-supported human milk bank.
Abbreviations
- AAP
American Academy of Pediatrics
- BPD
Bronchopulmonary Dysplasia
- DHM
Donor Human Milk
- EPI
Extremely Preterm Infants
- ESPGHAN
European Society for Paediatric Gastroenterology, Hepatology and Nutrition
- HACCP
Hazard Analysis and Critical Control Points
- IVH
Intraventricular Hemorrhage
- EOS
Early-Onset Sepsis
- LOS
Late-Onset Sepsis
- NEC
Necrotizing Enterocolitis
- NICHD
National Institute of Child Health and Human Development
- NICU
Neonatal Intensive Care Unit
- ROP
Retinopathy of Prematurity
- STROBE
Strengthening the Reporting of Observational Studies in Epidemiology
- UNICEF
United Nations Children’s Fund
- WHO
World Health Organization
Author contributions
XS, AC, XX, and XC jointly designed the study. XS wrote the initial draft of the manuscript. XS, YD, and QY conducted clinical data collection. XS performed statistical analysis with support from XC. XX reviewed and edited the manuscript. All authors provided critical interpretation of the results and wrote the report. All authors reviewed and approved the final version.
Funding
This study was supported by the Maternal and Child Health Research Project (Maternal and Infant Nutrition and Health Special Program) of the National Center for Women and Children’s Health, National Health Commission (2025FYH21), Shenzhen Science and Technology Program (JCYJ20230807120213026, JCYJ20240813115035045, JCYJ20250604191408012), and Sanming Project of Medicine in Shenzhen (No. SZSM202211001).
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the Declaration of Helsinki for Ethical Principles of Research. Approval was obtained from the Ethics Committee of Shenzhen Maternity and Child Healthcare Hospital affiliated with Southern Medical University (LLYJ2022-080-038). Additionally, informed consent was obtained from all individual participants included in the study.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Xinxin Song and Aifen Cao contributed equally to this work and share first authorship.
Contributor Information
Xueyu Chen, Email: snowvsrain@smu.edu.cn.
Xiaoyun Xiong, Email: xxyun@smu.edu.cn.
References
- 1.Chawla D, Agarwal R. Preterm births and deaths: from counting to classification. Lancet Glob Health. 2022;10(11):e1537–8. [DOI] [PubMed] [Google Scholar]
- 2.Higgins BV, Baer RJ, Steurer MA, Karvonen KL, Oltman SP, Jelliffe-Pawlowski LL, et al. Resuscitation, survival and morbidity of extremely preterm infants in California 2011–2019. J Perinatol. 2024;44(2):209–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Israel-Ballard K, Cohen J, Mansen K, Parker M, Engmann C, Kelley M. Call to action for equitable access to human milk for vulnerable infants. Lancet Glob Health. 2019;7(11):e1484–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Group NS. Timing of initiation, patterns of breastfeeding, and infant survival: prospective analysis of pooled data from three randomised trials. Lancet Glob Health. 2016;4(4):e266–75. [DOI] [PubMed] [Google Scholar]
- 5.Li S, Wupuer T, Hou R. Factors Influencing Delayed Onset of Lactogenesis: A Scoping Review. Int J Gen Med. 2024;17:2311–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.García-Lara NR, Escuder-Vieco D, Cabrera-Lafuente M, Keller K, De Diego-Poncela C, Jiménez-González C et al. Clinical Impact of supplementation with pasteurized donor human milk by high-temperature short-time method versus holder method in extremely low birth weight infants: a multicentre randomized controlled trial. Nutrients. 2024;16(7). [DOI] [PMC free article] [PubMed]
- 7.Merino-Hernández A, Palacios-Bermejo A, Ramos-Navarro C, Caballero-Martín S, González-Pacheco N, Rodríguez-Corrales E et al. Effect of donated premature milk in the prevention of bronchopulmonary dysplasia. Nutrients. 2024;16(6):859. [DOI] [PMC free article] [PubMed]
- 8.Daili C, Kunkun Z, Guangjun Y. Cost Analysis of Operating a Human Milk Bank in China. J Hum Lact. 2020;36(2):264–72. [DOI] [PubMed] [Google Scholar]
- 9.Fengler J, Heckmann M, Lange A, Kramer A, Flessa S. Cost analysis showed that feeding preterm infants with donor human milk was significantly more expensive than mother’s milk or formula. Acta Paediatr. 2020;109(5):959–66. [DOI] [PubMed] [Google Scholar]
- 10.Karami M, Zaremohzzabieh Z, Zarean M, Ahmadi AR. Exploring barriers to human milk banking acceptability among nursing mothers in Iran using social cognitive perspectives. Int Breastfeed J. 2025;20(1):71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Nielsen SD, Beverly RL, Underwood MA, Dallas DC. Differences and similarities in the peptide profile of preterm and term mother’s milk, and preterm and term infant gastric samples. Nutrients. 2020;12(9):2825. [DOI] [PMC free article] [PubMed]
- 12.Weaver G, Bertino E, Gebauer C, Grovslien A, Mileusnic-Milenovic R, Arslanoglu S, et al. Recommendations for the Establishment and Operation of Human Milk Banks in Europe: A Consensus Statement From the European Milk Bank Association (EMBA). Front Pediatr. 2019;7:53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Çağlar S, Korkmaz N, Azak M, Dervişoğlu H, Kızılelma Yiğit A. The Effect of Breast Milk Temperature on Comfort and Physiological Parameters of Preterm Infants: A Randomized Crossover Study. Breastfeed Med. 2025;20(10):742–9. [DOI] [PubMed] [Google Scholar]
- 14.Breastfeeding. the use of human milk. Pediatrics. 2012;129(3):e827–41. [DOI] [PubMed] [Google Scholar]
- 15.Sahni M, Mowes AK. Bronchopulmonary dysplasia. statpearls. Treasure Island (FL): StatPearls Publishing Copyright © 2025. StatPearls Publishing LLC. 2025.
- 16.Tan TF, Tay SA, Agarwal-Sinha S, Tan GSW, Wu WC, Tsai ASH. Persistent avascular retina in retinopathy of prematurity. Graefes Arch Clin Exp Ophthalmol. 2025;263(8):2177–90. [DOI] [PubMed] [Google Scholar]
- 17.Hu X, Liang H, Li F, Zhang R, Zhu Y, Zhu X, et al. Necrotizing enterocolitis: current understanding of the prevention and management. Pediatr Surg Int. 2024;40(1):32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Rees P, Gale C, Battersby C, Williams C, Carter B, Sutcliffe A. Intraventricular Hemorrhage and Survival, Multimorbidity, and Neurodevelopment. JAMA Netw Open. 2025;8(1):e2452883. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.De Rose DU, Ronchetti MP, Martini L, Rechichi J, Iannetta M, Dotta A et al. Diagnosis and management of neonatal bacterial sepsis: current challenges and future perspectives. Trop Med Infect Dis. 2024;9(9):199. [DOI] [PMC free article] [PubMed]
- 20.Peng HB, Zhan YL, Chen Y, Jin ZC, Liu F, Wang B, et al. Prediction Models for Bronchopulmonary Dysplasia in Preterm Infants: A Systematic Review. Front Pediatr. 2022;10:856159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Toth T, Borda A, Borka-Balas R, Cucerea M, Szasz EA, Gozar H et al. Risk factors and predictive parameters of necrotizing enterocolitis in preterm infants-a single-center retrospective study. Diseases. 2025;13(11):368. [DOI] [PMC free article] [PubMed]
- 22.Yang J, Li J, Wang H, Chen X, Yang C, ChaoMu L et al. Human milk and bronchopulmonary dysplasia have a dose-dependent effect: a cohort study. BMJ Paediatr Open. 2025;9(1):e002727. [DOI] [PMC free article] [PubMed]
- 23.Cohen M, Steffen E, Axelrod R, Patel SN, Toczylowski K, Perdon C, et al. Availability of Donor Human Milk Decreases the Incidence of Necrotizing Enterocolitis in VLBW Infants. Adv Neonatal Care. 2021;21(5):341–8. [DOI] [PubMed] [Google Scholar]
- 24.York DJ, Smazal AL, Robinson DT, De Plaen IG. Human milk growth factors and their role in nec prevention: a narrative review. Nutrients. 2021;13(11). [DOI] [PMC free article] [PubMed]
- 25.Israel-Ballard K, LaRose E, Mansen K. The global status of human milk banking. Matern Child Nutr. 2024;20(Suppl 4):e13592. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Donor human milk for the high-risk infant: preparation, safety, and usage options in the United States. Pediatrics. 2017;139(1):e20163440. [DOI] [PubMed]
- 27.Hoang MV, Nguyen TT, Tran AT, Luu TQ, Vu MQ, Tran HT, et al. Cost analysis of establishing and operating the first human milk bank at Da Nang Hospital for Women and Children in Vietnam: an activity-based costing ingredients study. Int Breastfeed J. 2024;19(1):47. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Affumicato L, Sánchez Tamayo T, Espinosa Fernandez MG, Peña Caballero M, Ruiz Morcillo CE, Acebes Tosti R, et al. [Satellite center of human milk: Analysis of cost reduction]. J Healthc Qual Res. 2018;33(1):3–9. [DOI] [PubMed] [Google Scholar]
- 29.Barin J, Touati J, Martin A, Fletgen Richard C, Jox RJ, Fontana S et al. Bridging the milk gap: integrating a human milk bank-blood bank model to reinforce lactation support and neonatal care. Nutrients. 2025;17(11):1765. [DOI] [PMC free article] [PubMed]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request.