Multivariate risk assessment and analysis of brain injury in high-risk infants in neonatal intensive care unit.

BACKGROUND

To explore risk factors, establish, and validate a risk prediction model for brain injury in high-risk infants in neonatal intensive care units (NICU).

METHODS

A case-control study was conducted involving 819 high-risk infants admitted to the Second People's Hospital of Wuhu City from March 2019 to March 2023. Participants were divided into a modeling group (n = 655) and a validation group (n = 164) at an 8:2 ratio. In the modeling cohort, brain injury was diagnosed by cranial ultrasound or magnetic resonance imaging (MRI). Intraventricular hemorrhage (IVH) was graded using the Papile classification and Volpe classification, and hypoxic-ischemic encephalopathy (HIE) was staged according to the Sarnat staging system. The modeling group was further divided into a brain injury group (n = 131) and a non-brain injury group (n = 524) based on the presence or absence of brain injury. Clinical data, maternal pregnancy status, and laboratory indicators were monitored and recorded. Risk factors for brain injury were analyzed through univariate and logistic regression analyses. Independent risk factors identified in the modeling group were used to establish a predictive model. A nomogram was developed to visualize the prediction model using R-4.0.2 software. The model's performance was evaluated and validated using discrimination and calibration assessments in both groups.

RESULTS

Meconium-stained amniotic fluid (MSAF), neonatal respiratory distress syndrome (NRDS), and pregnancy-associated anemia were independent risk factors for brain injury (P < 0.05). The 5-minute Apgar score served as a protective factor against brain injury (P = 0.01). Additionally, low umbilical artery PH and low base excess (BE) were associated with a higher brain injury (P < 0.05). Furthermore, higher serum levels of brain-derived neurotrophic factor (BDNF) and 25(OH)D were protective factors (P < 0.05). The predictive model showed an area under the ROC curve (AUC) of 0.935, indicating high accuracy.

CONCLUSION

Meconium-stained amniotic fluid (MSAF), NRDS, pregnancy-associated anemia, and low umbilical artery PH, and low BE are independent risk factors for brain injury. A higher 5-minute Apgar score and elevated serum levels of BDNF and 25(OH)D are protective factors. The risk prediction model established on the basis of Papile, Volpe, and Sarnat standardized criteria shows favorable performance and can be used for early risk stratification of brain injury in high-risk NICU infants.