The relationship between elderly nutritional risk index and short-term all-cause mortality in critically ill patients with cerebral injury: a retrospective cohort study from two cohorts

Zirong GaoXun MoYi GeYanhui LuShanshan Yu^*

• Affiliated Jinyang Hospital of Guizhou Medical University, Guiyang, China

The Geriatric Nutritional Risk Index (GNRI) is a simple and objective tool for assessing the risk of malnutrition, with potential practicality in Critically ill patients with cerebral injury (CIPCI). However, the current evidence is still limited.Method: Patients diagnosed with cerebral injury were retrospectively retrieved from the Medical Information Marketplace for Intensive Care (MIMIC-IV) and the Affiliated Jinyang Hospital of Guizhou Medical University. Various statistical methods, including restricted cubic spline regression (RCS) regression, multivariate logistic regression, subgroup analyses, and KM survival curves, were used to examine the association between the GNRI index and short-term adverse outcomes in CIPCI. Finally, we used LASSO-COX regression and multifactorial COX regression to develop risk prediction models and assessed the validity of the risk models using subject work characteristics (ROC) curves, area under the ROC (AUC).The final 1,244 CIPCIs were included in the analysis, of which the 28-day ICU/in-hospital mortality rates were 23.5% and 27.18%, respectively. In the fully adjusted model, continuous GNRI and GNRI category were significantly associated with 28-day ICU/in-hospital mortality, with HRs of, respectively, continuous GNRI: 0.97 (0.96-0.99)/0.97 (0.96-0.99) and [moderate nutritional risk vs. no nutritional risk:1.66 (1.01-2.74) and 2.03 (1.26-3.28), higher nutritional risk vs no nutritional risk: 1.79(1.08-2.97)/2.11 (1.30-3.42)]. Meanwhile, the RCS showed an inverse linear association between continuous GNRI and both 28-day ICU/hospitalization mortality, and this association remained consistent across most subgroups. In addition, the results of the ROC curves showed that the risk model constructed in this study could identify highly-risk CIPCI more effectively than the traditional critical care score. Finally, all results have been further confirmed in the external queue.This study identifies that GNRI is negatively associated with short-term mortality in critically ill patients with cerebral injury and provides a simple and effective tool for risk stratification, allowing clinicians to identify at-risk individuals and provide timely intervention.