AUTHOR=Hu Xuyang , Liang Jiaru , Hao Wenjian , Zhou Jiaqi , Gao Yuling , Gong Xiaoyang , Liu Yong 

TITLE=Prognostic value of inflammatory markers for in-hospital mortality in intensive care patients with acute ischemic stroke: a retrospective observational study based on MIMIC-IV

JOURNAL=Frontiers in Neurology

VOLUME=Volume 14 - 2023

YEAR=2023

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2023.1174711

DOI=10.3389/fneur.2023.1174711

ISSN=1664-2295

ABSTRACT=Background: Acute ischemic stroke (AIS) is still the primary reason of death and disability worldwide. This study aimed to investigate the relationship between SII and in-hospital mortality in AIS intensive care patients and to investigate which inflammatory marker (SII, NLR, PLR, and bilirubin) was more effective in predicting AIS in-hospital mortality.
Methods: We selected patients older than 18 who were diagnosed with AIS on admission in the Medical Information Mart for Intensive Care-IV database. We collected their baseline characteristics,  including various clinical and laboratory data. We used the generalized additive model to examine the link between SII and in-hospital mortality in patients with AIS. Differences in in-hospital mortality between groups were summarized by Kaplan-Meier survival analysis and conducted log-rank tests. The predictive value of four inflammatory markers (SII, NLR, PLR, and bilirubin) in predicting in-hospital mortality in AIS patients was evaluated by receiver operating characteristic analysis.
Results: 463 suitable patients were included in our study, we yielded an in-hospital mortality rate of 12.31%. The results of the GAM analysis show the in-hospital mortality of AIS patients and SII was positively correlated but not linearly. High SII levels were related with a higher risk of in-hospital death, according to unadjusted Cox regression. We also found those with SII levels over 1232 (Q2 group) had a considerably higher chance of dying in the hospital than those with low levels of SII (Q1 group). Kaplan-Meier analysis demonstrated that patients with elevated SII levels had worse chances of surviving their hospital stays than those with low levels of SII, and these differences achieved statistical significance. According to the results of ROC analysis, the AUC of in-hospital mortality of AIS patients predicted by SII was 0.65, which discovered that SII had better discriminating than NLR, NLR, PLR, and bilirubin. 
Conclusions: The in-hospital mortality of AIS patients and SII are positively correlated but not linearly. High levels of SII are connected to worse prognosis in patients with AIS. SII has a modest level of discrimination for forecasting the in-hospital mortality. SII is slightly better than NLR and significantly better than PLR with bilirubin in predicting it.