Correlation between SHR and 28-day all-cause mortality risk in patients with spontaneous intracerebral hemorrhage

The stress hyperglycemia ratio (SHR) has recently been suggested as a dependable indication for predicting adverse outcomes in critically ill ICU patients. Previous studies have primarily focused on cardiovascular disorders include myocardial infarction and coronary artery disease, but the connection between SHR and prognosis in severely ill individuals with cerebrovascular disorders, particularly those with spontaneous intracerebral hemorrhage, remains unclear. This investigation aims to investigate the relationship between SHR and adverse outcomes in critically ill individuals experiencing spontaneous intracerebral hemorrhage.

This study analyzed information collected from critically ill patients in the MIMIC database, grouped by tertiles of SHR, to demonstrate the clinical characteristics of the different groups of patients. The primary outcome was 28-day all-cause mortality. Cox regression and Kaplan-Meier curves, with restricted cubic spline, were used to investigate the connection between SHR and risk of death in individuals with spontaneous cerebral hemorrhage. Additionally, subgroup analyses were conducted to investigate the effect of different baseline characteristics on the correlation between SHR and risk of death. Lastly, predictive models were constructed using machine learning algorithms to illustrate the predictive efficacy of SHR.

The study analyzed the data of 815 patients who met the inclusion criteria. The average age of all patients was 69.46 ± 15.05, and 432 patients were male (53.0%). Cox regression analysis demonstrated the connection of SHR with the risk of death as a continuous variable and a categorical variable. Subsequently, Kaplan-Meier curves were generated to illustrate the survival characteristics for the three cohorts of patients(P<0.001). Further restricted cubic splines showed a nonlinear connection between SHR and the risk of death; Higher SHR was significantly correlated with an elevated mortality risk in individuals diagnosed with cerebral hemorrhage (HR>1, p<0.05). In the subgroup analysis, higher SHR was consistently correlated with the mortality risk in the vast majority of subgroups. Finally, five machine learning models were built and ROC were plotted to demonstrate the predictive efficacy of SHR for poor prognosis.

SHR can be used as an noteworthy predictor of adverse outcome in patients with spontaneous cerebral hemorrhage. Higher SHR is strongly correlated with an elevated risk of unfavorable prognosis, particularly in non-diabetic patients. Additionally, it may help to risk stratify this specific population.