Randomized Controlled Trial of Multisensory Early Oral Administration of Human Milk (M-MILK) for Very Preterm Infants

Randomized Controlled Trial of Multisensory Early Oral Administration of Human Milk (M-MILK) for Very Preterm Infants is sponsored by National Institute of Child Health and Human Development, NIH. A trial evaluating the effects of multisensory early oral administration of human milk on stress regulation, neurodevelopment, and oral feeding skills in very preterm infants.

Study Protocol for a Randomized Controlled Trial of Multisensory Early Oral Administration of Human Milk (M-MILK) for Very Preterm Infants: Enhancing Stress Regulation, Neurodevelopment, and Oral Feeding Skills - PMC As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. .

Author manuscript; available in PMC: 2026 Mar 11. Published in final edited form as: Biol Res Nurs. 2026 Jan 19;28(3):464–478.

doi: 10.

1177/10998004261418708 Study Protocol for a Randomized Controlled Trial of Multisensory Early Oral Administration of Human Milk (M-MILK) for Very Preterm Infants: Enhancing Stress Regulation, Neurodevelopment, and Oral Feeding Skills 1 Marcella Niehoff School of Nursing, Loyola University Chicago, Maywood, IL, USA Find articles by Thao Griffith Linda Janusek , PhD, RN, FAAN 1 Marcella Niehoff School of Nursing, Loyola University Chicago, Maywood, IL, USA Find articles by Linda Janusek Rosemary White-Traut , PhD, RN, FAAN 2 Nursing Research, Children’s Wisconsin, Milwaukee, WI, USA 3 Women, Children and Family Health Science, College of Nursing, University of Illinois at Chicago, Chicago, IL, USA Find articles by Rosemary White-Traut 4 Genomics and Microbiome Core Facility, Rush University Medical Center, Chicago, IL, USA Find articles by Stefan J Green Sachin Amin , MD, MBA, FAAP 5 Division of Neonatology, Loyola University Medical Center, Maywood, IL, USA Find articles by Sachin Amin 6 Stritch School of Medicine, Loyola University Chicago, Maywood, IL, USA Find articles by Cara Joyce 1 Marcella Niehoff School of Nursing, Loyola University Chicago, Maywood, IL, USA 2 Nursing Research, Children’s Wisconsin, Milwaukee, WI, USA 3 Women, Children and Family Health Science, College of Nursing, University of Illinois at Chicago, Chicago, IL, USA 4 Genomics and Microbiome Core Facility, Rush University Medical Center, Chicago, IL, USA 5 Division of Neonatology, Loyola University Medical Center, Maywood, IL, USA 6 Stritch School of Medicine, Loyola University Chicago, Maywood, IL, USA ✉ Corresponding author: Thao Griffith, PhD, RN; Marcella Niehoff School of Nursing, Loyola University Chicago, 2160 S First Avenue, Maywood, IL 60153, USA; tgriffith1@luc.

edu PMCID: PMC12974252 NIHMSID: NIHMS2140642 PMID: 41554032 The publisher's version of this article is available at Biol Res Nurs Early life stress results in disrupted stress regulation, and less optimal neurodevelopment and oral feeding skills in very preterm infants. Despite these associations, there are few evidence-based interventions to help these fragile infants during critical stages of development.

This article describes an ongoing research protocol of a randomized controlled trial (RCT) to evaluate the m ultisensory early oral administration of human milk (M-MILK) as an early NICU intervention to improve stress regulation, support optimal neurodevelopment, and promote competent oral feeding skills in very preterm infants.

M-MILK is an infant-led early NICU intervention beginning on day 3 of life to provide very preterm infants with small droplets of milk orally while engaging their innate senses. Infants (N = 124) born <32 weeks gestational age are randomized to either the M-MILK or control group (standard of care). M-MILK begins on day 3 of life.

We assess stress regulation, neurodevelopment, and oral feeding skills over time via salivary cortisol, buccal cell DNA methylation, NeoNatal Neurobehavioral Scale, Ages and Stages Questionnaire, Early Feeding Skills Assessment, and Neonatal Feeding Assessment Tool, respectively. Endpoints will be compared by treatment arm using linear regression and mixed-effects models for cross-sectional and longitudinal analyses, respectively.

Findings from this study will advance knowledge as to the efficacy of M-MILK as an epigenetically-informed intervention to enhance stress regulation, neurodevelopment, and oral feeding skills in very preterm infants during critical periods of neuroplasticity. The study has been registered at ClinicalTrials. gov ( NCT07216664 ).

Keywords: Preterm infants, stress, neurodevelopment, oral feeding, epigenetics, human milk Neurodevelopmental Impairments and Oral Feeding Difficulties Up to 60% of the 60,000 very preterm infants (<32 weeks gestational age) born annually in the United States exhibit neurodevelopmental impairments ( Jois, 2019 ; Lester et al. , 2015 ; Ream & Lehwald, 2018 ; Song, 2023 ) and 80% experience oral feeding difficulties ( Grabill et al.

, 2023 ; Kamity et al. , 2021 ; Pados et al. , 2021 ; Patton et al.

, 2022 ; Pineda et al. , 2020 ). Neurodevelopmental impairments exacerbate the fragility of suck-swallow-breathe coordination ( Rinat et al.

, 2022 ; Zhang et al. , 2017 ), hinder alertness and engagement ( Aho et al. , 2021 ; Als et al.

, 2004 ), and alter sensory processing in the oral regions ( Chung et al. , 2020 ; Crozier et al. , 2016 ; Ryckman et al.

, 2017 ), which can inhibit the process of oral feeding ( Griffith et al. , 2017 ; White-Traut et al. , 2022 ; White-Traut et al.

, 2017 ). Oral feeding is a critical milestone for growth, nutrition, and bonding with caregivers ( Griffith et al. , 2018 ; Griffith et al.

, 2020 ; Griffith et al. , 2019 ). Preterm infants lack crucial physiological and behavioral maturity ( Kamity et al.

, 2021 ; Lau, 2016 ; Viswanathan & Jadcherla, 2020 ). In addition, they are prone to gastrointestinal, pulmonary and cardiac issues ( Darnall et al. , 2017 ; Erickson et al.

, 2021 ; Hunt, 2017 ; Indrio et al. , 2022 ; Kamity et al. , 2021 ; Patton et al.

, 2022 ; Viswanathan & Jadcherla, 2020 ), further compromising their ability to attain safe and efficient oral feeding. Neurodevelopmental impairments and oral feeding difficulties are associated with long-lasting adverse effects on health, including malnutrition, developmental delays and mental health issues ( Bowe et al. , 2023 ; Chung et al.

, 2020 ; Halbmeijer et al. , 2023 ; Kamity et al. , 2021 ; McGowan et al.

, 2022 ; Pados et al. , 2021 ). Early Life Stress and Its Impact on Stress Regulation To ensure survival, all very preterm infants require admission to the neonatal intensive care unit (NICU) ( Griffith et al.

, 2020 ), exposing them to early life stress ( Bergman, 2019 ; de Magalhães-Barbosa et al. , 2022 ; National Association of Neonatal Nurses, 2022 ), and risk for epigenetic modifications ( van Dokkum et al. , 2023 ).

Epigenetics is the study of how behaviors and environmental factors modify phenotypes without changing the DNA sequence ( Provenzi et al. , 2018 ). Epigenetic modifications can increase or decrease gene expression and may occur by three mechanisms, including DNA methylation (DNAm), histone modifications, and micro-RNA.

DNAm, the focus of this study, consists of the addition of a methyl group to DNA. Epigenetic modifications are dynamic and capable of being attenuated or reversed, opening up the opportunity for epigenetically based interventions ( Fiorito et al. , 2021 ; Fitzgerald et al.

, 2021 ). Early life stress is associated with epigenetic modifications of two glucocorticoid regulating genes, NR3C1 and HSD11B2 ( Griffith et al. , 2025 ; Griffith et al.

, 2020 ; Lardenoije et al. , 2020 ; Lester & Marsit, 2018 ; Provenzi et al. , 2018 ; van Dokkum et al.

, 2023 ). These two genes influence the hypothalamic-pituitary-adrenal (HPA) axis via a complex negative feedback loop. The HPA axis is central to stress regulation and homeostasis of the glucocorticoid hormone, cortisol, the body’s key stress hormone ( Leistner & Menke, 2020 ; Spencer & Deak, 2017 ).

NR3C1 encodes glucocorticoid receptors, facilitating the stimulation and inhibition of cortisol secretion by the HPA axis. HSD11B2 encodes for the 11-beta-hydroxysteroid dehydrogenase type 2, an enzyme that converts cortisol to its inactive form, cortisone. Higher early life stress is also associated with higher salivary ( Pourkaviani et al.

, 2020 ) and skin ( D’Agata et al. , 2019 ) cortisol levels. Epigenetic modifications of NR3C1 exon 1F and HSD11B2 promoters may be a potential mechanism for dysregulation of cortisol secretion ( Conradt et al.

, 2015 ; Oberlander et al. , 2008 ; Stroud et al. , 2016 ).

Early Life Stress and Its Impact on Neurodevelopment and Oral Feeding Skills Infants exposed to excessive acute and chronic stress during their NICU stay exhibit high-risk neurodevelopment ( Cong et al. , 2017 ; Zhao et al. , 2022 ).

High early life stress exposure is also associated with altered brain development ( Boggini et al. , 2021 ; Howell et al. , 2019 ; McCormack et al.

, 2022 ; Morin et al. , 2020 ), poor motor performance ( Cook et al. , 2023 ), and risk for autism ( Cook et al.

, 2023 ). Others report that infants with a high-risk neurodevelopment profile had high buccal cell DNAm of NR3C1 exon 1F and low DNAm of HSD11B2 promoters compared to low-risk infants ( Lester et al. , 2015 ).

Oral feeding skills require intact neurodevelopment and may be especially vulnerable to early life stress ( Grabill et al. , 2023 ; Griffith et al. , 2017 ; Griffith et al.

, 2019 ; White-Traut et al. , 2022 ; White-Traut et al. , 2017 ).

Yet, the potential impact of early life stress on oral feeding skills development in preterm infants is understudied.

Recent evidence, however, demonstrates that aberrant stress regulation and DNAm, i.e., high buccal cell DNAm at multiple specific cytosine-guanine (CpG) dinucleotide sites of NR3C1 exon 1F and HSD11B2 promoters, is associated with less competent oral feeding skills across 5 domains, including respiratory regulation, oromotor function, swallowing, engagement, and physiological stability ( Griffith et al. , 2025 ).

Need for a Novel NICU Intervention to Improve Stress Regulation, Neurodevelopment, and Oral Feeding Skills Despite findings linking early life stress with aberrant stress regulation, neurodevelopment, and oral feeding skills in very preterm infants, evidenced-based interventions to protect infants from these adverse effects are limited.

The m ultisensory early oral administration of human (M-MILK) is a novel, developentally informed NICU intervention. Our ongoing study evaluates M-MILK as an early intervention to improve stress regulation, optimize neurodevelopment, and promote competent oral feeding skills in very preterm infants.

Early administration of human milk, including oropharyngeal therapy with colostrum or mother’s own milk and milk drop interventions, shows potential efficacy. Evidence suggests that early human milk administration provides immune protection, supports oral feeding transition, and reduces ventilator-associated pneumonia, oxygen needs, NICU stay, and cost of care ( Abd-Elgawad et al. , 2020 ; Aggarwal et al.

, 2021 ; Chen et al. , 2021 ; Embleton & Chmelova, 2024 ; Kelich et al. , 2024 ; Muelbert et al.

, 2019 ; O’Rouke et al. , 2022 ; Rodriguez et al. , 2023 ; Silva et al.

, 2021 ; Snyder et al. , 2017 ; Wetzel et al. , 2020 ; Xiu et al.

, 2025 ). However, these earlier interventions administered milk directly into infants’ oral or oropharyngeal cavity using a prescribed volume, with little consideration for infants’ behavioral cues and responses. Adapting from these earlier interventions, M-MILK was designed to emphasize multisensory experiences and infant-led interactions .

M-MILK is an early NICU intervention and can be introduced as early as 23 weeks postmenstrual age (PMA), starting on day 3 of life. It involves the oral administration of up to 1 mL of milk in small droplets, allowing infants to lead the process based on their behavioral cues. The small droplets are placed on the infant’s upper lip, near the nose, gradually dripping into their oral cavity.

This method allows them to engage their tactile, olfactory, and gustatory senses while licking the milk and safely practicing tonguing, sucking, and swallowing skills without the risk of respiratory compromise. Infants also benefit from visual and auditory stimuli, such as seeing their caregiver and hearing their soothing voice.

The M-MILK intervention is not defined solely by the oral route of milk administration, however, it is grounded in the neuroprotective developmental care framework,( Altimier & Phillips, 2016 ; Johns Hopkins All Children’s Hospital, 2023 ) integrating multisensory experiences with infant-led interactions. While the adoption of M-MILK is based on robust evidence, a comprehensive and rigorous investigation of M-MILK is warranted.

This article describes the protocol for our ongoing randomized controlled trial (RCT), which addresses the following specific aims: 1) evaluate the extent to which M-MILK improves stress regulation, 2) evaluate the extent to which M-MILK supports optimal neurodevelopment, and 3) evaluate the extent to which M-MILK promotes competent oral feeding skills. Figure 1 .

This RCT uses a two-group, parallel-arm, and longitudinal design of infants from birth to two months corrected age (CA). Day of oral feeding initiation is defined as the first of at least 2 consecutive days when infants orally consume ≥ 10% of prescribed feedings ( Griffith et al. , 2024 ).

Day of full oral feeding is the first of at least 2 consecutive days when the infant orally consumes 100% of prescribed feedings ( Griffith et al. , 2024 ). Demand feeding (PO ad lib) is defined as when oral feeding is based on the infant’s cues; infants are fed orally as much (or as little) as they want, and as often as they demand, provided they meet minimal daily fluid and energy requirements ( Brigham and Women’s Hospital, 2020 ).

The clinical team decides oral feeding initiation and progression for each infant per the unit’s feeding protocol. illustrates the study flow diagram of the study, which employs a 2-group, equivalent, parallel design, and longitudinal RCT. This study is being conducted at a level III NICU.

A sample of 124 very preterm NICU infants are being enrolled. Inclusion criteria are: born less than 32 weeks GA and receiving mother’s own milk and/or donor milk. Exclusion criteria are: receiving only formula, gastrointestinal defects, chromosomal abnormalities, necrotizing enterocolitis requiring surgery, or congenital cardiac defects requiring surgery.

Sample Size and Statistical Power. Our estimate of 124 infants was informed by the rate of eligibility determination versus those recruited and enrolled (80%) in our pilot study. We anticipate loss of follow-up due to withdrawal or death to be no more than 20%.

The expected endpoint sample size is 100 ( n = 50 per group). This sample size was determined based on the primary endpoint of Early Feeding Skills Assessment (EFS). Our preliminary data indicated a standard deviation of 5 points, and in a RCT of oromotor intervention ( n = 40), Comuk Balci et al.

found a mean difference of 13 points (49. 5 ± 7. 2 in intervention compared to 36.

3 ± 8. 7 in control, p < 0. 001) ( Comuk Balci et al.

, 2023 ). Thus, we conservatively estimate a pooled standard deviation of 9 points, for which a mean difference of 5. 1 points at discharge has 80% power using a two-sample t-test.

This represents the lower bound for power, as adjusting for prognostic covariates may increase precision of effect estimates. Secondary endpoints also have 80% power for small to moderate mean differences at this sample size, assuming pool SD observed in the pilot and literature ( Lester et al. , 2015 ; Pados et al.

, 2018 ; Shuffrey et al. , 2022 ). Participant Screening, Recruitment, and Enrollment.

After infants are admitted to the NICU, we review their electronic medical records to screen for eligibility (pre-randomization screening). Parents of eligible infants are approached for recruitment. If parents agree to have their infant participate, they sign a written informed consent.

Parents are compensated up to $200 in gift cards. After enrollment, on-going screening continues until discharge. If the infant develops any of the exclusion criteria, the infant will discontinue the assigned intervention, though data will be collected as planned for intention-to-treat analyses if not otherwise contraindicated.

If the infant is withdrawn by parents or an infant’s death occurs, parents are fully compensated for their infant’s participation up to that point. Randomization and Blinding Procedures. After informed consent is obtained, the infant is randomly assigned to the M-MILK (group 1) or the control (group 2) in a 1:1 ratio based on a randomization scheme generated using SAS statistical software (SAS Institute, Cary, NC).

In families with multiple infants (e.g., twins, triplets), infants will be randomized to the same treatment arm, however, only data from a single, randomly selected infant of the set will be included in data analysis. The random numbers are printed on cards and placed in sequenced envelopes securely stored in the research office.

After parents complete the informed consent, they open the next available envelope to reveal their group assignment. The study team staff, staff nurses, and parents cannot be blinded to the group assignment. Study team staff know the group assignment to provide M-MILK.

Staff nurses know the assignment because they provide care for infants. Parents know the assignment as both groups are described in the informed consent. We will monitor bedside nursing activities and documentation for potential contamination or cross-over (e.g., M-MILK attempted in the control group).

Based on our pilot study, we anticipate high adherence to group assignments. If contamination is identified, individualized education will be provided to parents and/or bedside nurses to reinforce the intervention protocol. The study team staff who conduct the neurodevelopment and oral feeding assessments are blinded to the group assignment.

Intervention: M-MILK Group. Infants assigned to the M-MILK group receive the M-MILK intervention along with standard care. Parents in the M-MILK group are provided with educational materials on preterm infant care, i.e., Your Guide to Neonatal Intensive Care Unit and Your Guide to Breastfeeding (Baby360, Arlington, TX).

Parents in the M-MILK group receive an individual in-person 30-minute M-MILK training session within 1 week of enrollment, given by a clinical research nurse. M-MILK begins on day 3 of life regardless of respiratory support. M-MILK is provided by clinical research nurses or parents.

M-MILK is implemented at least 5 days a week, up to 4 times a day, during the day shift, after hands-on care and at the beginning of a gavage feeding session. Infants receive either mother’s own milk or donor’s milk based on availability. Infants receive M-MILK in small droplets via a 1-ml syringe.

Infants may receive up to 1 mL of milk each time based on their behavioral cues and responses. The 1 mL volume intake is included as part of their oral caloric intake. M-MILK is not offered when there is an oral feeding attempt.

M-MILK ceases when infants achieve PO ad lib status. Implementation of the M-MILK intervention. After the assigned staff nurse completes all hands-on care, the infant is swaddled.

The infant’ hands are placed by his/her face to encourage self-soothing, self-organizing, mouthing, and sucking. The infant may be (1) tucked and nested on their side with boundaries to support comfort, security, and flexion, or (2) cradled or held skin-to-skin by parent. The scent heart is placed near his/her face.

The clinical research nurse draws up 1 mL of milk into the 1-mL syringe from a prepared gavage feeding syringe. The gavage feeding is started. The clinical research nurse then proceeds to implement M-MILK while closely observing the infant’s responses.

To begin M-MILK, the clinical research nurse pauses and sets an intention for this caring encounter. The clinical research nurse greets the infant by gently talking and placing hands on him/her. The clinical research nurse talks to the infant with a soft calm voice while providing gentle touch throughout the intervention.

The clinical research nurse slowly places 1–2 drops of milk onto the infant’s lips, near the nose, allowing the infant to feel and smell the milk. The clinical research nurse gently strokes the infant’s lips with a pacifier/swab/infant’s fingers. The infant may respond to M-MILK, and the clinical research nurse’s actions in one of 3 ways: (1) Response: infant engages with mouthing or rooting.

Action: as the infant accepts the pacifier/swab/infant’s fingers and begins to suck, slowly give more milk, a drop at a time, onto the pacifier/swab/infant’s fingers, and support the infant to participate by sucking and guiding the process. (2) Response: the infant engages with subtle mouthing movements, tonguing, or licking.

Action: continue to offer milk, a drop at a time, onto the infant’s lips, and support the infant to participate by licking the milk and guiding the process. (3) Response: the infant does not engage at all. Action: no additional milk is offered, and the infant is supported to rest.

M-MILK is concluded if the infant is no longer engaged or has finished the 1 mL. Intervention: Control Group. Infants assigned to the control group receive the current standard care and no M-MILK intervention.

Parents in the control group are provided with educational materials on preterm infant care, i.e., Your Guide to Neonatal Intensive Care Unit and Your Guide to Breastfeeding (Baby360, Arlington, TX). Stress Regulation: Salivary cortisol. Saliva samples are collected by study team staff in the morning between 0600 to 1200 to control for the diurnal cortisol rhythm ( Iwata et al.

, 2013 ). Collection is completed before feeding to avoid milk contamination. Samples are collected using the SalivaBio Infant’s Swab (Salimetrics, Carlsbad, CA).

The swabs are placed in the Swab Storage Tube (Salimetrics, Carlsbad, CA) and stored at −70°C until analysis. Salivary cortisol are measured in duplicate using the Salivary Cortisol ELISA Kit, a high sensitivity enzyme immunoassay (Salimetrics, Carlsbad, PA) ( Griffith et al. , 2024 ).

Test volume is 25 μl of saliva per determination. Assay sensitivity is . 007 μg/dL.

The average intra-assay coefficient of variation is 4. 60%, and average inter-assay coefficient of variation is 6. 00%.

Stress Regulation: Buccal cell DNAm of NR3C1 exon 1F and HSD11B2 promoter. Buccal samples are collected by study team staff before feeding to avoid milk contamination. Samples are collected using the DNA/RNA Shield Collection Tube with Swab (Zymo Research, Inc., Irvine, CA) and stored at −70°C until analysis.

Genomic DNA is extracted from the samples using a Maxwell ® RSC Buccal Swab DNA kit (Promega, Maddison, WI) implemented on an automated Maxwell RSC48 device. The DNAm for NR3C1 exon 1F and HSD11B2 promoter is assessed using methods including deep amplicon sequencing from bisulfite-converted DNA using next-generation sequencing as previously described ( McFadden et al. , 2017 ).

Briefly, DNA is treated with bisulfite for methylation analysis using an EZ DNA Methylation-Lightning Kit (D5030; Zymo Research, Inc., Irvine, CA) per manufacturer’s instructions. CpG sites of interest are targeted using PCR amplification of the bisulfite-treated DNA, followed by sequencing of the amplicons on an Illumina MiSeq sequencer.

Preparation of converted DNA for high-throughput amplicon sequencing is performed in two PCR steps in a protocol termed “targeted amplicon sequencing (TAS)” as previously described ( Naqib et al. , 2018 ). Primer sets, as previously described ( Griffith et al.

, 2025 ; Griffith et al. , 2024 ; Lester et al. , 2015 ; Liu et al.

, 2020 ), target the NR3C1 exon 1F and HSD11B2 promoter, with amplicons covering 33 and 30 CpG sites, respectively. Bisulfite conversion controls are included with each sequencing run. Raw paired-end reads are processed using PEAR ( Zhang et al.

, 2014 ) to merge overlapping read pairs, and only the merged reads will be used for downstream analyses. Trimming is performed using Cutadapt to remove low-quality bases and adapter sequences. FastQC is used to assess quality of raw and merged reads.

The trimmed, merged reads are then aligned to the human reference genome (hg38) using Bismark ( Krueger & Andrews, 2011 ). Prior to alignment, the hg38 reference genome is preprocessed using Bismark’s genome preparation tool to create a bisulfite-converted index of the genome. Methylation data are extracted and summarized by methylation percent and overall counts of methylated and unmethylated reads per CpG site.

Neurodevelopment: NeoNatal Neurobehavioral Scale (NNNS-II). Neurodevelopment is assessed using the NNNS-II ( Lester et al. , 2015 ; Zhao et al.

, 2022 ). The NNNS-II is a standardized neurobehavioral scale that assesses neurological integrity and behavioral function of infants who are 32 to 48 weeks PMA. The NNNS-II has 115 items organized into 13 domains: attention, handling, self-regulation, habituation, arousal, excitability, lethargy, hypertonicity, hypotonicity, nonoptimal reflexes, asymmetric reflexes, quality of movement, and stress/abstinence.

The NNNS-II provides a summary score for each domain. High-risk neurodevelopment is characterized by higher scores in handling, arousal, excitability, lethargy, hypertonicity, hypotonicity, nonoptimal reflexes, asymmetric reflexes, and stress/abstinence, along with lower score in quality of movement, attention, self-regulation, habituation. The NNNS-II has strong internal consistency (Cronbach’s α = .

87 - . 90) ( Fink et al. , 2012 ; Lester & Tronick, 2004 ).

Concurrent validity is supported by the correlation between the NNNS-II scores and brain volumes in white matter, basal ganglia, and total brain tissue ( Brown et al. , 2009 ; Parikh et al. , 2022 ).

Reliability is acceptable (Cronbach’s α = . 56 - . 85) ( Lester & Tronick, 2004 ).

Predictive reliability is supported by correlation between the NNNS-II scores and motor outcomes at 18 months ( Miller-Loncar et al. , 2005 ) and 24 months, and later development of cerebral palsy ( Stephens et al. , 2010 ).

High-risk NNNS-II profiles are associated with short- and long-term outcomes, i.e., behavior problems and low IQ up to 4½ years of age ( Liu et al. , 2010 ). Test-retest reliability is shown by correlations in the NNNS-II scores ( r = .

30 - . 44) across 34, 40, and 44 weeks GA ( Lester & Tronick, 2004 ). A study team staff, blinded to group assignment, conducts and video records the NNNS-II assessments at the bedside.

The study team staff and PI randomly re-rate 10% the NNNS-II video recordings to assess intra-rater and inter-rater reliability. Neurodevelopment: Ages and Stages Questionnaire (ASQ-3). Neurodevelopment is also assessed by the ASQ-3 ( Shuffrey et al.

, 2022 ). The ASQ-3 is a validated and standardized developmental screening instrument that assesses developmental progress in children from 1 month to 3 years of age and is based on parental reports. The 2-month ASQ-3 assesses 3 domains: communication, gross motor, and personal-social skills.

ASQ-3 yields a summary score for each domain. Higher ASQ-3 scores indicate more optimal neurodevelopment. It has strong concurrent validity ( r = .

85), 2-week test-retest reliability ( r = . 75 - . 82), inter-rater reliability ( r = .

43 - . 69), and internal consistency ( α = . 51 - .

87) ( Shuffrey et al. , 2022 ). The sensitivity of ASQ-3 as a screener for for developmental delays ranges from 21.

0 to 84. 2, while the specificity ranges from . 76 to .

91 ( Lipkin & Macias, 2020 ; Sheldrick et al. , 2020 ). The ASQ-3 is completed with the infant’s primary caregiver during a phone interview.

Oral Feeding Skills: Early Feeding Skill Assessment (EFS). Oral feeding skills are assessed by the EFS ( Thoyre et al. , 2018 ).

The EFS has 22 items within 5 subscales: respiratory regulation, oral-motor functioning, swallowing coordination, engagement, and physiologic stability. The EFS provides a summary score for each subscale and a total summary score. Higher EFS summary scores indicate more competent oral feeding skills.

Inter-rater reliability is good (Cronbach’s α = . 81). Construct validity is shown by correlations among EFS scores, GA and PMA.

A study team staff member, blinded to the group assignment, conducts and video records the EFS assessments at the bedside. The study team staff and PI randomly re-rate 10% of the EFS video recordings separately to assess intra-rater and inter-rater reliability. The clinical research nurse feeds the infant per standard protocol (i.e., 20 – 30 minutes, side-lying, swaddling, pacing) during the EFS assessments.

Oral Feeding Skills: Neonatal Feeding Assessment Tool (NeoEAT). Oral feeding skills are also assessed by the NeoEAT-Bottle-feeding ( Pados et al. , 2018 ).

The NeoEAT-Bottle-feeding has 64 items within 5 subscales: regulation, energy and physiologic stability, gastrointestinal tract function, sensory responsiveness, and compelling symptoms of problematic feeding. The NeoEAT-Bottle-feeding yields a summary score for each subscale and a total summary score. Lower NeoEAT-Bottle-feeding scores indicate more competent oral feeding skills.

Internal consistency (α = . 92) and test-retest reliability ( r = . 90) are excellent.

Construct validity is established with the Infant Gastroesophageal Reflux Questionnaire ( r = . 74) and the Infant Gastrointestinal Symptoms Questionnaire ( r = . 64).

Known-groups validity is supported by lower scores in healthy infants than ones with feeding problems. NeoEAT is completed with the infant’s primary caregiver during a phone interview. Covariates and Descriptive Measures.

Demographics includes infant sex, race, and ethnicity. Birth history includes GA (weeks), birthweight (g), delivery method (vaginal/C-section), Apgar score (1- and 5-minute), small for GA (yes/no), intra-uterine growth restriction (yes/no), premature rupture of membranes (yes/no), antenatal steroid use (yes/no), and medical diagnoses. Medical severity is assessed by the Neonatal Medical Index Classification (NMI) ( Korner et al.

, 1993 ). The NMI classifies infants’ medical condition as ranging from 1 for infants born weighing ≥ 1000 g and without major complications to 5 for infants born weighing < 1000 g and with very serious complications.

Hospital progression includes daily oral feeding intake, daily type of milk, duration of exclusive tube feedings (number of days infants received only gavage feedings, no oral feeding transition), duration of tube feeding (number of days infants receive gavage feedings, including oral feeding transition), duration of transition from first to full oral feeding, PMA (weeks) at oral initiation, full oral feeding, and discharge, duration of invasive ventilation, weight at discharge, type of feeding at discharge, type of milk at discharge, and length of NICU stay.

Early Life Stress is assessed daily using the modified Neonatal Infant Stressor Scale (NISS) ( Cong et al. , 2017 ; Newnham et al. , 2009 ) during NICU stay.

The modified NISS includes 47 acute and 23 chronic stress events. Stress intensity is rated from 1 (little stressful) to 4 (extremely stressful). Predictive and construct validity are reported for higher NISS scores and brain structural changes ( Smith et al.

, 2011 ), and poorer neurodevelopment ( Cong et al. , 2017 ; Gorzilio et al. , 2015 ).

Data from the electronic medical record will be retrospectively extracted for procedures and events corresponding to NISS-defined acute and chronic stress events. Acute stress events are summarized in frequency of occurrence, while chronic stress events are summarized in number of hours. Figure 2 .

After informed consent is obtained, infants are randomized into: Group 1 (M-MILK) and Group 2 (control). All data collection is conducted in the same manner and at the planned time points for both groups. We collect data for the NISS daily from the electronic medical record.

At baseline (T0), between day of life 2 to 4, we collect buccal and saliva samples. At oral feeding initiation (T1), we collect buccal and saliva samples, and the EFS assessment. At 36 weeks PMA (T2), we collect buccal and saliva samples, and the NNNS-II and EFS assessments.

At 2 months CA (T3), we