Enteral feeding timing and clinical outcomes in neonates with hypoxic-ischemic encephalopathy undergoing therapeutic hypothermia: a retrospective cohort study highlighting the challenge of confounding by indication

Objective To analyze real-world feeding decisions in neonates with hypoxic-ischemic encephalopathy (HIE) undergoing therapeutic hypothermia (TH), assess potential confounding by indication, and clarify how this bias may affect the interpretation of the association between feeding timing and clinical outcomes. Methods We conducted a single-center retrospective cohort study involving 94 consecutive neonates with moderate to severe HIE who received standardized TH at the Children's Hospital of Chongqing Medical University from March 2024 to March 2025. These neonates were assigned to either the early enteral feeding (EEF) group (initiated during TH/rewarming, n = 48) or the delayed enteral feeding (DEF) group (initiated post-TH/rewarming, n = 46). The primary outcome was gastrointestinal adverse events (GIAEs). We used multivariate logistic regression to adjust for confounders and performed subgroup analyses to explore feeding safety across different disease severity levels. Results The DEF group had significantly higher rates of invasive ventilation (28.3% vs. 6.3%), hemodynamic support (82.6% vs. 29.2%), and peripherally inserted central catheter (PICC) placement (50.0% vs. 14.6%) (all P < 0.001), suggesting confounding by indication. No significant differences existed between the two groups in GIAEs (22.9% vs. 21.7%, P = 0.891), feeding intolerance (18.8% vs. 17.4%, P = 0.864), or necrotizing enterocolitis (NEC) (4.2% vs. 4.3%, P > 0.999). After adjustment, EEF was not independently associated with GIAEs (OR = 0.75, 95% CI: 0.23–2.44, P = 0.636), but it shortened time to feeding initiation and parenteral nutrition duration, and reduced hospitalization costs (all P < 0.05). In the severe HIE subgroup ( n = 11), NEC incidence was 18.2%, higher than the overall cohort (4.3%, P = 0.042), though no intergroup difference was seen due to the small sample size. Conclusions Confounding by indication may have contributed to the observed associations between early feeding and outcomes in prior studies. In clinically stable neonates with moderate HIE, early enteral feeding (≤20 mL/kg/day, with gradual advancement <20 mL/kg/day) was associated with shorter parenteral nutrition duration and reduced hospitalization costs. However, these findings should be interpreted with caution given the potential for residual confounding, and future prospective studies are needed for confirmation.