669 - Early Human Milk (HM) Feeding vs. Necrotizing Enterocolitis and Mortality in Very Preterm (VPT) or Very Low Birthweight (VLBW) Infants

Publication Number: 1646.669

Eric B. Ortigoza, UT Southwestern Medical Center, Dallas, TX, United States; Sujir P. Nayak, UTSW -Dallas, Dallas, TX, United States; Dimitrios Angelis, University of Texas Southwestern Medical School, Dallas, TX, United States; Vonny Chong, University of Texas Southwestern Medical School, Dallas, TX, United States; Cheryl Lair, University of Texas Southwestern Medical School, Dallas, TX, United States; Audrey Edwards, Parkland Health, Dallas, TX, United States; Patti Burchfield, University of Texas Southwestern Medical School, Dallas, TX, United States; Steven Brown, Parkland Health, Dallas, TX, United States; Mambarambath A.. Jaleel, University of Texas Southwestern Medical School, Irving, TX, United States; Luc P. Brion, UT Southwestern Medical Center, FRISCO, TX, United States

Background: Necrotizing enterocolitis (NEC) results in 20-30% mortality and significant morbidity among survivors. It is most frequent among VPT and VLBW infants. Systematic reviews demonstrate no increase in NEC in infants randomized to early feeding; however, there was substantial variability in the timing of feeding initiation and NEC outcome.

Objective: To assess whether early (first 2 days of life) initiation of enteral feeding with HM is associated with less NEC Stage ≥IIa, mortality, and combined outcomes of death/NEC Stage ≥IIa and death/NEC Stage III.

Design/Methods: This is a retrospective cohort study of infants born VPT ( < 33 weeks' gestation) or VLBW ( < 1500 g) at Parkland Health and Hospital System from 2016-2023. We excluded infants with severe congenital anomalies, comfort care only status or no enteral feeds. To mitigate selection bias between infants who receive early HM vs. controls (late or no HM), propensity score (PS) analysis was performed using SAS procedure "causaltrt." A multivariable logistic regression model was used to predict the odds of receiving early HM, which was used as a PS for each infant. Stabilized weights [augmented inverse probability weights (AIPW)] were calculated from PS values to create a "pseudo-population" where variables associated with early HM were balanced between early HM and control groups. This reduced selection bias between early HM and controls. Outcomes (NEC, mortality, and death/NEC) were assessed by the average treatment effect (ATE), the bootstrap bias corrected ATE and the adjusted odds ratio (aOR) were obtained from SAS procedure "glimmix."

Results: 1959 infants were included in this study (Table 1). 96 (4.9%) developed NEC, 46 (2.3%) were NEC stage III and 61 (3.1%) died before discharge. PS matching in 1503 infants minimized differences in all variables associated with early HM and all other measured variables except Apgar score < 4 (Table 2). Early HM was associated with no significant changes in either NEC or mortality to discharge. However early HM was associated with less mortality after 7 days of age (P=0.04), death/NEC Stage ≥IIa (P=0.006) and death/NEC Stage III (P=0.03) (Table 3).

Conclusion(s): PS analysis showed that early HM feeding, compared with control, was not associated with less NEC but was associated with less late mortality, death/NEC Stage ≥IIa and death/NEC Stage III.

Table 1: Demographics and baseline variables


Table 2: Differences Between Propensity Score (PS)-Matched Fed Early HM vs. Late HM (Controls) Were Minimized vs. Before PS Matching
Values are either median (IQR) from Wilcoxon-Signed Rank test with P-values from t-test on ranks; or % with Chi-square. Age at starting feeds is by design different. AIPW, augmented inverse probability weighting (weight); CPAP, continuous positive airway pressure. Note: Among all variables independently associated with early HM feeding (epoch, maternal diabetes mellitus, cesarean delivery, maximum inhaled oxygen concentration, donor breast milk feeding), standardized weighted difference was <0.1 between early HM feeding and control groups.

Table 3: Results of Propensity Score Analysis (PS) of Early HM Starting Within 1st 2 Days
ATE, average treatment effect; OR, odds ratio; CI, confidence interval.
*One among: death, intraventricular/periventricular hemorrhage (IVH/PVH) grade III or IV, cystic periventricular leukomalacia (PVL), NEC Stage ≥IIa, proven sepsis, severe retinopathy of prematurity (ROP), severe bronchopulmonary dysplasia (BPD), surgical patent ductus arteriosus (PDA) ligation