Related Experiment Video
Updated: Aug 23, 2026
A Bedside, Single Burr Hole Approach to Multimodality Monitoring in Severe Brain Injury
Published on: March 26, 2019
Early Postoperative Inflammatory Indices Predict Long-Term Neuro-ICU Stay in Patients Requiring Prolonged Care After
Biwu Wu1,2,3, Lichao Wei1,2,3, Haoyue Yuan1,2,3
1Department of Neurosurgical Intensive Care Unit, Huashan Hospital, Fudan University, Shanghai, China.
Background:
Long-term stay (LTS) in the neurosurgical intensive care unit (Neuro-ICU) after supratentorial brain tumor resection disproportionately consumes healthcare resources, but early risk stratification tools remain lacking. We evaluated five early postoperative inflammatory indices (neutrophil-to-lymphocyte ratio [NLR], platelet-to-lymphocyte ratio [PLR], lymphocyte-to-monocyte ratio [LMR], systemic immune-inflammation index [SII], and systemic inflammation response index [SIRI]) to develop an integrated prediction model.
Methods:
This retrospective study included patients requiring > 48-h Neuro-ICU care after elective supratentorial tumor resection. Patients were temporally partitioned into derivation (n = 432) and validation (n = 146) cohorts. Inflammatory indices were measured within 24 h postoperatively. Following least absolute shrinkage and selection operator (LASSO) feature selection, five machine learning classifiers were compared. Model performance was assessed for discrimination, calibration, and clinical utility, alongside subgroup analyses and a series of sensitivity analyses. A nomogram and web-based risk calculator were developed.
Results:
LTS occurred in 47.5% and 49.3% of the derivation and validation cohorts, respectively. LTS patients showed higher early postoperative SIRI and PLR and lower LMR, whereas preoperative indices did not differ. All three were independent predictors of LTS. The integrated model combined these indices with preoperative Karnofsky Performance Status, postoperative steroid use, and retained endotracheal intubation at blood sampling. It yielded an AUC of 0.845 in the derivation cohort and 0.816 in the validation cohort, compared with 0.715 for the clinical baseline model (p < 0.001; IDI 0.212; cNRI 0.981). These indices maintained independent predictive value, and model performance remained stable across subgroup and sensitivity analyses.
Conclusion:
Early postoperative SIRI, PLR, and LMR independently predict LTS in patients requiring prolonged care (> 48 h) after supratentorial brain tumor resection. An integrated model with these indices and early clinical variables provides reliable risk stratification within 24 postoperative hours.
