Introduction
Human milk is the ideal nutrition and recommended first feed for all newborns, including sick and preterm neonates [1]. In low- and middle-income countries (LMICs), exclusive breastfeeding and early human milk feeding in the NICU remain suboptimal [2], even though the evidence for lower rates of necrotising enterocolitis (NEC) and late-onset sepsis and for better neurodevelopmental outcomes is strong [3]. When mother's own milk (MOM) is unavailable or insufficient, pasteurised donor human milk (PDHM) is the next best option for vulnerable neonates [4].
In India, the National Guidelines on Lactation Management Centres in Public Health Facilities (MoHFW 2017) recommend setting up Comprehensive Lactation Management Centres (CLMCs) in tertiary facilities, bringing breastfeeding promotion, milk expression support and donor human milk banking together under one service [5]. A CLMC differs from a stand-alone donor milk bank because it embeds counselling, bedside lactation care, milk expression support and donor milk services within the routine NICU workflow.
Despite this policy endorsement, implementation in high-volume public-sector tertiary centres is often uneven. Common obstacles are too few dedicated lactation staff, competing clinical demands in overcrowded NICUs, gaps in infrastructure, and the need for continued behavioural support for mothers coping with separation from their infants [6].
Published implementation experiences from North Indian public-sector tertiary NICUs remain scarce, and these settings differ substantially from urban private units in delivery volumes and resource constraints.
We report the operationalisation of a CLMC in our NICU between July 2024 and July 2025. Our objectives were to (1) describe how the CLMC was set up and adapted, using prospectively maintained operational records; (2) evaluate its effect on breastfeeding and counselling coverage through a retrospective before-after comparison; and (3) analyse trends in donor human milk collection and utilisation, the microbiological quality of donor milk, and donor mothers’ profiles.
Materials and methods
Study design and setting
This was a retrospective implementation study with a before-after comparison of feeding outcomes. It was conducted in the 40-bedded Level-III NICU of a public-sector tertiary teaching hospital that serves eastern Uttar Pradesh and the adjoining districts of Bihar. The institution records 4,200-4,500 deliveries annually, with a high proportion of preterm, low-birth-weight and referred-in sick neonates.
Ethical approval was obtained from the Institutional Ethics Committee, Institute of Medical Sciences, Banaras Hindu University (IMS-BHU) (Ref: IMS/IEC/FEBRUARY/2026/319, 12 March 2026).
The baseline period was April-June 2024. Structured counselling coverage was derived from mothers whose babies required NICU admissions (353 at baseline and 1,841 during implementation). Exclusive breastfeeding at NICU discharge predominantly included neonates who were sicker in the initial days and required growth monitoring and supportive supervision in the mother-newborn care unit before discharge (56 neonates at baseline and 534 during implementation).
The implementation period was July 2024-July 2025. We defined structured breastfeeding counselling as at least one counselling session with a hands-on demonstration of milk expression [7]. Before CLMC operationalisation, structured breastfeeding counselling was not routinely offered to all mothers, and the counselling provided followed a passive rather than proactive approach.
Exclusive breastmilk feeding at NICU discharge was defined as the neonate receiving only breastfeeding and/or MOM in the 24 hours before discharge [8,9].
Description of the Intervention
The CLMC comprised the following components.
Provision of training to staff working in the NICU, postnatal wards, labour rooms and operation theatres. This included small, focused discussions, knowledge-sharing sessions and practical skills training using breast models. Correct latching, infant positioning, manual milk expression, operation of hospital-grade breast pumps, and recognition and management of common breastfeeding problems, including breast engorgement, mastitis and inverted nipples, were also taught.
Structured delivery-room and postnatal breastfeeding counselling for mothers of neonates admitted to the NICU. Bedside and dedicated-room milk expression support, with emphasis on demonstrating hand expression alongside mechanical expression where available. A functional donor human milk pathway, including donor screening and serology (full criteria in Table 1), hygienic milk collection, Holder pasteurisation, microbiological safety checks, cold-chain storage and prescription-based issue of PDHM to eligible neonates when MOM was insufficient or unavailable.
Daily multidisciplinary NICU rounds incorporating feeding-plan review for every admitted neonate, with explicit assessment of MOM availability, volume and mode of delivery. A WhatsApp-based communication channel among clinicians, nursing staff and the CLMC manager enabled real-time problem-solving around donor availability, mother-side barriers, feeding challenges and milk-handling processes, especially during evening and night hours. Structured behaviour-change support addressed perceived low milk supply, sustaining milk expression during prolonged separation of mother and infant, and integrating kangaroo mother care (KMC) to support lactation.
The team included a paediatrician, a neonatologist, neonatology fellows, junior residents, nursing officers, a CLMC manager, a laboratory technician, hygiene support staff, and a microbiology team comprising a microbiologist and a laboratory technician. One dedicated lactation counsellor and one staff nurse were available during the daytime in each shift. At night, only one staff nurse was available, serving as both counsellor and nurse.
The implementation was supported by a system-strengthening facilitatory programme (LIFT-UP), which provided comprehensive training, supportive supervision and mentorship for CLMC-related activities in the unit.
CLMC operational pathway is presented in Figure 1.
Operational Review Mechanism
Implementation was monitored through weekly/monthly review meetings involving the NICU team and CLMC staff. These meetings examined service volumes, donor availability, mother-side bottlenecks and operational challenges. Adaptations such as revised counselling content, workflow adjustments for night-shift expression support and measures to address milk contamination rates were made iteratively in response to these reviews.
No formal quality-improvement methodology was applied.
Data Collection
For the baseline period, data were retrospectively abstracted from NICU admission registers and discharge records before CLMC operationalisation. For the post-implementation period, data were drawn from prospectively maintained CLMC service registers, postnatal round registers and feeding records. All data were de-identified before analysis.
EBF and counselling analyses included mothers and babies with documented details. Donor analysis excluded 29 donors with missing data on maternal age, parity, gestational age at delivery, maternal risk factors and need for NICU admission.
Primary outcomes were (1) the proportion of mothers receiving structured breastfeeding counselling and (2) the proportion of neonates exclusively breastfed at NICU discharge. Secondary indicators were the number of donors, the number of neonates receiving PDHM, the peak monthly donation volume, the microbiological status of PDHM and donor mothers’ profiles.
Before-After Comparison
EBF and structured breastfeeding counselling proportions from the baseline period were compared with those from the post-implementation period. Because this was a retrospective implementation design, the comparison is descriptive and intended to show the direction and magnitude of change associated with CLMC operationalisation, not to establish a causal effect.
Statistical Analysis
EBF and structured breastfeeding counselling proportions were compared descriptively using absolute differences between the baseline and post-implementation periods. Inferential testing for before-after feeding comparisons was not applied because differences in data quality between the baseline and post-implementation periods precluded valid causal inference; the RRs quantify magnitude only. Interrupted time-series regression was fitted to the 16 monthly data points (three baseline and 13 implementation), estimating the pre-intervention level and slope, the immediate level change at CLMC onset, and the post-intervention change in slope.
The within-implementation trend was also examined using a binomial segmented regression model. The monthly post-pasteurisation contamination rate was plotted on a statistical process control (p-) chart with 3-sigma control limits.
For donor-characteristic comparisons, categorical variables are presented as numbers (%) and continuous variables as mean ± SD for normally distributed data or median (IQR) for non-normally distributed data. Between-group comparisons used the chi-square or Fisher’s exact test for categorical variables and the independent-samples t-test for normally distributed continuous variables. All tests were two-tailed, with statistical significance set at P < 0.05.
Multivariable binary logistic regression was applied to assess factors associated with donation volumes above and below the sample median (above 185 mL versus 185 mL or below), adjusting for maternal age, risk factors, mode of delivery, NICU admission and gestational age. Monthly operational trends are described using tabular data (absolute counts and volumes) and graphs. Donor-characteristic comparisons and the multivariable logistic regression of donation volume were performed in JASP version 0.96.0 (JASP Team).
The remaining analyses were performed in Stata version 19.5 (StataCorp, College Station, TX). The Standards for Reporting Implementation Studies (StaRI) and the Template for Intervention Description and Replication (TIDieR) are reported in Appendices A-B.
Results
The initial operating model, implementation experience, challenges and adaptations were derived from prospectively maintained CLMC service registers, weekly/monthly review meetings, postnatal round registers, the WhatsApp group archive and the event log initiated in December 2024. No formal qualitative methods (e.g., structured interviews, focus groups or thematic analysis) were used; the narrative reflects only the documented operational records.
Establishment and initial operating model
The CLMC began functioning in July 2024 in the dedicated CLMC area within the existing NICU. As per the national guidelines, our centre had designated areas for counselling, handwashing, milk expression, pasteurisation, storage, thawing and dispensing, equipment cleaning and sterilisation, and record maintenance. A laminar flow system maintained sterility during sample pooling and microbiological culture sampling.
According to the CLMC service register, early work focused on setting aside physical space for milk expression, establishing donor screening protocols in accordance with the 2017 National Guidelines (Table 1) [10,11], assigning dedicated counselling staff, and incorporating CLMC-specific items into the daily NICU and postnatal ward rounds. According to the milk-bank log, when a single donor’s milk volume was below 100 mL, milk from more than one donor was pooled to reach the minimum sample volume. The Holder method of pasteurisation was used.
The cold chain was strictly maintained at −20 °C for pre- and post-pasteurisation samples and during transport. Microbial testing was performed only after pasteurisation, and a sample from each batch of PDHM was sent to the microbiology laboratory to confirm sterility before dispensing. Samples reported as unsterile were discarded, as documented in the milk-bank log.
Review records show that the initial model emphasised process strengthening through structured counselling, routine bedside expression support, and explicit feeding-plan documentation for every admitted neonate.
Implementation experience
Review records from the first two to three months noted that starting lactation counselling and expression support within the first hour and the first day of admission was associated with higher documented milk output in the CLMC service register. Structured 2-3 hourly expression cycles, reinforced by bedside demonstration by staff nurses, were recorded as the standard operating procedure from the first month onward. Hand expression was the main method used for mothers of very preterm or sick neonates in the early postpartum days.
The CLMC functioned as a referral point where clinicians, staff nurses and the CLMC manager worked together to address individual mother-side barriers to expression and breastfeeding, rather than as a stand-alone milk-collection facility.
Challenges and adaptations
Separation of mother and infant because of NICU admission was the most frequently documented barrier to early and sustained lactation, despite gains in early initiation. Perceived low milk supply was the maternal concern most often recorded in counsellor notes. Postnatal round registers showed that donor recruitment required sustained counsellor-led outreach in the postnatal wards and did not become self-sustaining during the study period.
Heavy microbial contamination and errors in handling PDHM were additional challenges documented.
Adaptations during the implementation period included WhatsApp-based coordination for after-hours queries, revised counselling content that directly addressed perceived insufficient milk, and stronger night-shift expression support. The event log records that responsibility for microbiology-related processes was handed over to the CLMC technician in October 2024. This followed microbial growth detected in post-pasteurisation samples in September (20/47, 42.6%, of batches were unsterile) and October 2024 (13/46, 28.3%).
As the pasteuriser temperature had been consistently well maintained, the heavy contamination was attributed instead to inadequate cleaning of the valve diaphragm in the breast-pump suction kit, and the hygiene staff were subsequently retrained. From December 2024, a structured event log was established to document, audit and correct data-entry discrepancies in CLMC records and errors in handling PDHM, particularly at night. Problems with CLMC equipment and its functional status were also recorded in this log.
Quantitative outcomes
The proportion of mothers receiving structured breastfeeding counselling rose from 87/353 (25%) at baseline to 1,613/1,841 (88%) after implementation. Beyond coverage, counselling was delivered at a high dose, with 2,155 CLMC counselling contacts recorded over the period and a further 6,864 in-ward counselling sessions documented from October 2024. Exclusive breastfeeding at NICU discharge increased from 25/56 (45%) to 436/534 (82%) (Table 2).
Interrupted time-series analysis suggested an immediate increase in exclusive breastfeeding of approximately 36 percentage points at CLMC implementation (95% CI 24.6-47.2; P < 0.001), with binomial segmented regression yielding an OR of 5.38 (95% CI 1.61-17.9; quasi-binomial dispersion 0.60). However, the availability of only three pre-implementation months limited reliable estimation of the baseline trend, making the estimated change dependent on baseline extrapolation (Figure 2).
Of the 241 donor mothers, descriptive analysis was performed for 212 (88.0%), comprising 117 term (55.2%) and 95 preterm (44.8%) mothers; the remaining 29 (12.0%) were excluded because of incomplete study records (Figure 3).
Of the 241 donor mothers, descriptive analysis was performed for 212 (88.0%; 117 term and 95 preterm); the remaining 29 (12.0%) were excluded because of incomplete study records (Figure 3). Mean maternal age was 27.06 ± 3.39 years in the term group and 27.26 ± 4.42 years in the preterm group. Most donors had a lower-segment caesarean delivery, and 76/95 (80%) of mothers who delivered prematurely had health problems.
Maternal age, parity, occupation and residence were comparable between groups. NICU admission and maternal risk factors were significantly more common among preterm donor mothers. Skin-to-skin contact (term: 80/117, 68.4%, vs. preterm: 25/95, 26.3%; P < 0.001) and early initiation of breastfeeding (term: 66/117, 56.4%, vs. preterm: 18/95, 18.9%; P < 0.001) were significantly less common among preterm donor mothers.
Mothers of preterm infants contributed a larger share of the total milk volume (92,952/133,542 mL, 69.6%) than mothers of term infants (40,590/133,542 mL, 30.4%), although per-donor volume did not differ significantly between groups (overall median 185 mL, IQR 83.75-391.25; P = 0.184) (Table 3).
Although mothers of NICU-admitted neonates donated more milk in the univariable analysis (median 218 vs. 128 mL; Mann-Whitney P = 0.004), this association did not persist after adjustment (aOR 1.61; P = 0.186). This indicates that the crude association was not independent and that the sample may have been underpowered for multivariable modelling of donor milk volume (Table 4).
Monthly donation volume peaked at 30,009 mL. This sudden rise was driven largely by a few mothers who donated substantially more than the average donor, although it also reflected the maturation of donor recruitment, pasteurisation throughput and clinical uptake of PDHM in July 2025 (Figure 4).
During the implementation period, 241 donors donated milk. In all, 541 neonates received PDHM as a bridge when MOM was insufficient or temporarily unavailable. Most beneficiaries were preterm infants admitted to our NICU.
Bacterial growth was detected in 58/591 (9.8%) processed samples on microbiological culture. Heavy microbial growth in PDHM samples in September (20/47, 42.6%) and October (13/46, 28.3%) affected team motivation and disrupted workflow; however, corrective actions restored normal operations (Table 5; Figure 5).
Discussion
This retrospective implementation study describes the operationalisation of a CLMC in a high-volume public-sector tertiary NICU in North India and documents improvements in EBF and structured breastfeeding counselling coverage. The proportion of neonates exclusively breastfed at discharge nearly doubled, and a functional donor human milk pathway was established and used.
The gains are best understood as the combined effect of several operational components rather than any single input. Early expression, particularly hand expression, enabled mothers of sick and preterm neonates to build and maintain their milk supply during the first few days, when separation from the infant is unavoidable.
Structured support within a dedicated service also helps the unit adhere consistently to the revised Baby-friendly Hospital Initiative (BFHI) [9]. BFHI practices, structured post-partum lactation counselling and NICU feeding protocols are central to the CLMC's function and align with breastfeeding-centred care [12]. In a systematic review and meta-analysis by Sinha et al., interventions to improve feeding outcomes increased EBF rates by 44%.
Among the interventions, counselling had the greatest effect on EBF, whether provided in hospital (RR 1.66, 95% CI 1.43-1.92) or at home (RR 1.58, 95% CI 1.39-1.80) [13]. Another systematic review of interventions in LMICs found that programmes spanning the antenatal to postnatal period and delivered through multiple contacts also had larger effects on EBF at 6 months (RR = 2.19) [14]. The reductions in formula use and uptake of PDHM that we observed are consistent with reports from public-sector facilities in India [15].
We did not report neonatal outcomes such as necrotising enterocolitis (NEC) or sepsis.
The median volume donated per donor was 185 mL (IQR 83.75-391.25). Other Indian studies reported mean volumes per donor of 268 ± 386 mL and 77.62 ± 51.26 mL, respectively [16,17]. Because of extreme values in our dataset, we reported the volume per donor as a median; therefore, a meaningful comparison with these studies is not possible.
The lower volumes compared with those reported in developed nations reflect the structural realities of LMICs and highlight the need for sustained donor recruitment and effective lactation support.
Globally, the mean volume donated per donor is much higher (3L/donor-40L/donor) [18,19].
Programmes are better structured in developed nations, where donors are often recruited through awareness campaigns, are motivated to donate and frequently continue donating over the long term [20-22]. The post-pasteurisation contamination rate in our study was high compared with rates reported in other studies [16,17]. Simple handwashing techniques and proper cleaning and sterilisation of equipment used in milk collection, processing and storage help reduce contamination by hospital microbiota [23].
Educational interventions using video messages and awareness activities also reduced contamination, from 62/244 (25.4%) to 20/244 (8.2%), in the study by Karimi et al. [24].
Mean maternal age was 27.06 ± 3.39 years in the term group and 27.26 ± 4.42 years in the preterm group. Other Indian studies have also reported that most donors were in the 21-30-year age group [16,17]. Data from developed countries, by contrast, show older donors, with most mothers aged 30-39 years [25].
Early marriage in LMICs means that most pregnancies, and therefore most donors, occur among younger women. Studies from developing countries have reported a higher proportion of primiparous donors, but we did not observe this in our study [6,26]. In developed countries, mothers are often screened for health conditions, are educated and tend to lead healthy lifestyles.
In LMICs, the higher burden of malnutrition and pregnancy-related health problems increases the risk of premature and operative deliveries, while structured support during the hospital stay is associated with higher donation rates [16,17]. We found that mothers of preterm neonates donated a larger share of the total milk volume. In a study of donor profiles in an LMIC, most donor mothers had delivered preterm infants (1,018/1,553 (65.5%)); NICU stays were longer, and mothers of NICU-admitted infants donated significantly more milk than mothers in the postnatal ward (P = 0.0001) [16].
System-level insights
Although the CLMC was structured, its functioning depended on continuous reinforcement. Donor recruitment, adherence to milk expression and documentation all required ongoing counsellor engagement, and the gains would probably regress without sustained staffing and supervision. Mother-infant separation was the barrier most consistently documented in operational records, although we did not formally assess its quantitative effect on lactation outcomes.
This suggests that scaling up a CLMC in similar settings should include explicit provision for counselling time, hands-on expression support, coordination platforms and continuous mentorship, rather than focusing solely on equipment or physical infrastructure. The model could be replicated within the government healthcare system to address the initial operationalisation challenges of establishing a CLMC.
Strengths and limitations
Strengths of this study include its real-world tertiary public-sector setting, the relatively long implementation period of over a year, the use of both process (counselling coverage) and outcome (feeding practice at discharge) indicators, and the documentation of operational adaptations that are rarely reported in implementation literature from this context.
The main limitations are the retrospective design and the lack of a parallel control group. Baseline data were drawn from routine pre-CLMC clinical records, which captured counselling and feeding practices less completely than the prospective registers maintained during implementation. This data asymmetry introduces a risk of measurement bias and may overestimate the effect size.
The before-after comparisons are descriptive, and inferential testing was not applied. We could not extract neonatal clinical outcomes from baseline records with sufficient completeness for comparison. The proposed mechanisms of change were inferred from routine documentation, rather than from formal qualitative methods (e.g., interviews or focus groups), and should be interpreted as hypotheses.
Microbial identification was not performed because it was unavailable at our centre. The segmented interrupted time-series analysis was based on only three monthly baseline data points, limiting the precision of the estimated pre-intervention slope.
Conclusions
Operationalising a CLMC in a public-sector tertiary NICU was associated with improvements in structured breastfeeding counselling coverage and exclusive breastfeeding at discharge. Our experience suggests that counselling, early bedside expression support, KMC integration, daily coordination and responsive problem-solving may have contributed to the observed improvements. However, the study design precludes valid causal inference.
These findings may inform consideration of CLMC operationalisation in similar tertiary settings in LMICs, with attention to staffing, coordination and continuous mentorship.