**Background:** Nutritional support is critical for critically ill patients, particularly those with neurological injury who experience hypermetabolic responses. Malnutrition is associated with higher mortality (32% vs. 14%, P=0.018) and nosocomial infection rates (23.4% vs. 3.5%, P<0.001). However, optimal feeding timing and route in neurocritically ill patients remain unclear, and nutrition is often deprioritized in early management.
**Methods:** This retrospective, single-center observational study included patients admitted to the neurosurgical ICU at Samsung Medical Center from January 2013 to December 2019. Inclusion required ICU stay >5 days for neurocritical illness or postoperative management. Exclusion criteria included insufficient records, do-not-resuscitate orders, non-neurosurgical admissions, and transfers. Among 12,743 screened patients, 1,353 were analyzed. Early nutrition was defined as starting enteral or parenteral nutrition within 72 hours of ICU admission. The primary endpoint was in-hospital mortality; secondary endpoint was infectious complications. Propensity score matching (PSM) with nearest-neighbor matching (caliper 0.2 SD of logit) and propensity score weighting overlap weights (PSOW) were used to control selection bias. Standardized mean differences (SMD) <10% indicated adequate covariate balance. Multivariable logistic regression with stepwise variable selection and doubly robust estimation were performed.
**Key Results:** Early nutrition was initiated in 384 patients (28.4%): 152 (11.2%) received EEN and 232 (17.1%) received EPN. Mean age was 50.5±23.2 years; 52.3% were male. Malignancy (55.3%) and hypertension (34.5%) were the most common comorbidities; brain tumors (37.5%) and intracerebral hemorrhage (17.4%) were the leading ICU admission causes. In the unadjusted overall population, in-hospital mortality was significantly lower with early nutrition (14.1% vs. 33.1%, P<0.001). However, after PSM and PSOW adjustment, no significant difference in in-hospital mortality was observed between early and late nutrition groups (P=0.234 and P=0.094, respectively). Similarly, infectious complication rates did not differ significantly between early and late nutrition groups in any analysis (overall: 8.3% vs. 8.5%, P=0.999; adjusted: all P>0.05). When comparing EEN, EPN, and late nutrition, EEN patients had the lowest in-hospital mortality (5.9% vs. 19.4% for EPN vs. 33.1% for late nutrition, P<0.001) and lowest infectious complications (2.0% vs. 12.5% for EPN vs. 8.5% for late nutrition, P=0.001). In multivariable analyses across all three populations (overall, PSM-adjusted, PSOW-adjusted), early nutrition was not significantly associated with in-hospital mortality or infectious complications (all P>0.05), but EEN was significantly associated with both outcomes (all P<0.05).
**Clinical Implications:** This study suggests that early enteral nutrition specifically—rather than early nutrition in general—may reduce mortality and infectious complications in neurocritically ill patients. The findings highlight the importance of distinguishing between enteral and parenteral routes when evaluating early feeding strategies. EEN may preserve gastrointestinal integrity, prevent bacterial translocation, and support immune function during the hypermetabolic phase of brain injury. The lack of benefit from early nutrition overall may be attributable to confounding by EPN, which has been associated with worse outcomes. Limitations include the retrospective design, single-center setting, lack of standardized nutritional protocols, absence of calorie/protein intake data, and a high proportion of brain tumor patients (37.5%) that may limit generalizability. Prospective randomized trials are needed to confirm these findings and establish optimal feeding protocols for neurocritically ill patients.