636 - Timing and Type of First Enteral Feed and Relationship to Adverse In-Hospital Outcomes Among Very Preterm Newborns

Publication Number: 1613.636

Gabriel Byram, Pacific Northwest University of Health Sciences, Yakima, WA, United States; Olivia C. Brandon, University of Washington School of Medicine, Seattle, WA, United States; Rebecca Hoban, Seattle Children's, Seattle, WA, United States; Sandy Huang, University of Washington School of Medicine, Sammamish, WA, United States; Elizabeth Riffel, St. Joseph's Children's Hospital, Tacoma, WA, United States; John Haddock, St. Joseph Medical Center, Puyallup, WA, United States; Addysen Pulley, University of Washington School of Medicine, Puyallup, WA, United States; Sophia Rodriguez, University of Washington School of Medicine, Kirkland, WA, United States; Clara G.. McAdams, University of Washington School of Medicine, Tacoma, WA, United States; Ineeya Senthilnathan Kayal, University of Washington School of Medicine, Seattle, WA, United States; Virginia Higgins, University of Washington School of Medicine, Edina, MN, United States; Natalie Chapkis, Seattle Children's/ University of Washington, Seattle, WA, United States; Laura A.. Boyce, Seattle Children's, Tacoma, WA, United States; Jill Irvine, university of washington medical center, seattle, WA, United States; Robin G. LaGrandeur, University of Washington School of Medicine, Seattle, WA, United States; Sarah Kolnik, University of Washington - Seattle Children's Hospital, Seattle, WA, United States; Stephen E.. Welty, University of Washington School of Medicine, Tacoma, WV, United States; Krystle Perez, University of Washington School of Medicine, Snohomish, WA, United States; Katie M.. Strobel, University of Washington School of Medicine, Seattle, WA, United States; Gregory C. Valentine, University of Washington School of Medicine, Tacoma, WA, United States

Background: Very preterm (VP) neonates (born < 32 weeks' gestation) often receive their first enteral feed hours to days after birth, yet the clinical consequences of timing to first feed and first feed milk type, donor human milk (DHM) vs parent's own milk (POM), remain uncertain. Understanding early optimal feeding strategies among VP neonates is critical to reduce adverse health outcomes.

Objective: To examine how timing and type of first enteral human milk feed relate to comorbidities among VP neonates.

Design/Methods: We retrospectively studied neonates admitted to two level III/IV US neonatal intensive care units from 2017-2024, with complete fluid data, survival ≥3 days, and documented maternal and neonatal characteristics, including first feed time and type. Multiple logistic regression, adjusting for indicators of severity of illness, identified associations between time of first enteral feed and adverse in-hospital outcomes. Composite outcomes including mortality and transfer to outside hospitals were analyzed to address survivor bias. Interaction analyses assessed whether associations differed by first feed milk type.

Results: Among n = 378 VP neonates, those with first feed >24 hours after birth had lower GA, lower birthweight, lower 5-minute Apgar score, more first-week steroid/pressor/inotrope use, more POM as first feed, and were more frequently delivered by C-section than those fed within 24 hours (Table 1). Unadjusted and adjusted findings are shown in Figures 1 and 2, respectively. Time to first feed was not significantly associated with necrotizing enterocolitis/mortality, sepsis/mortality, or retinopathy of prematurity/mortality. However, each additional hour of delay in initiating enteral feeds was associated with 4.5% increased odds of bronchopulmonary dysplasia (BPD) alone (adjusted odds ratio [aOR] 1.045, 95% CI 1.002, 1.092, p = 0.044) and a trend toward increased BPD/mortality composite (aOR 1.038, 95% CI 0.996, 1.084, p = 0.084); Figure 2. First milk type was not associated with adjusted outcomes when accounting for time to first feed, and no significant interaction was observed between feed time and type (p > 0.2).

Conclusion(s): Each hour of delay in initiating enteral feeds was associated with higher odds of BPD, independent of first milk type and after adjusting for severity of illness and site of enrollment. Prospective clinical trials are needed to confirm these findings, which suggest that even modest delays in starting enteral feeds may increase the risk of developing BPD among VP neonates.

Table 1. Feed-Time Groups (≤12, >12-≤24, >24 h) - Maternal/Neonatal Characteristics.
p-values obtained via Fisher's Exact Test for categorical variables and via Kruskal-Wallis Rank Sum Test for continuous or ordinal variables. Categorical variables are shown as n/N (%). Continuous or ordinal variables are shown as median [interquartile range] (n). Differences in sample size across rows reflect exclusion of records with unknown values. To reduce influence of extreme outliers, time to first feed was Winsorized at 95th percentile.

Figure 1. Locally Estimated Scatterplot Smoothing (LOESS)-smoothed proportion of outcomes by time to first feed.
LOESS fits (blue curves) show the proportion of each outcome as a function of age (hours) at first enteral feed (Winsorized at the 95th percentile). Events/n by panel: A - necrotizing enterocolitis (NEC) or mortality (40/368); B - sepsis or mortality (43/364); C - retinopathy of prematurity (ROP) or mortality (92/325); D - bronchopulmonary dysplasia (BPD) or mortality (91/287). Mortality was defined as any mortality on or after 3 days in the hospital (all-cause mortality). NEC was defined as Modified Bell's stage II or higher. Sepsis was defined as culture-positive sepsis occurring on or after 3 days in the hospital. ROP was defined as any diagnosis of ROP in either eye (including stage I or higher). BPD was defined as oxygen or positive-pressure support at 36 weeks post-menstrual age. Gray bands are 95% pointwise confidence intervals. Hollow circles are individual data points.

Figure 2. Forest plot of adjusted odds ratios for first feed time and type.
Estimates obtained by multiple logistic regression. Associations in panel A - first feed time (hours) and B - parent's own milk (POM) as first milk type. 95% confidence intervals are solid black lines. Point fill encodes p-value: black = p < 0.05; white = p ≥ 0.05. Axes use logarithmic odds ratio scales (equal distances represent equal ratios). All models include: time of first enteral feed, type of milk in first feed, gestational age, small-for-gestational age (SGA; < 10th percentile per Fenton 2013), 5-minute Apgar score, mode of delivery, pressor, inotrope, or steroid use in first week, and site. Event count / n per model: mortality: 28/369; necrotizing enterocolitis (NEC) or mortality: 39/360; sepsis or mortality: 42/355; retinopathy of prematurity (ROP) or mortality: 89/318; bronchopulmonary dysplasia (BPD) or mortality: 89/281; BPD/mortality/transfer on oxygen (O2): 93/290; BPD: 61/253. Outcome definitions: mortality is all-cause mortality occurring on or after 3 days in the hospital; NEC is modified Bell's stage II or higher; sepsis is culture-positive sepsis occurring on or after 3 days of age; ROP is any diagnosis of ROP in either eye (including stage I or higher); BPD is oxygen or positive pressure ventilation requirement at 36 weeks post-menstrual age (PMA). To reduce survivor bias, outcomes are composited with mortality. Transfer on O2 defined as transfer to external healthcare system on continuous oxygen between ≥ 34 to < 36 weeks post-menstrual age, included as a sensitivity analysis to confirm the findings are not altered by transfers out prior to 36 weeks PMA for the BPD analyses.