This study aims to explore the relationship between arterial partial pressure of oxygen (PaO₂) and in-hospital mortality in patients admitted for the first time with intracerebral hemorrhage, with emphasis on the effect of hyperoxia on mortality.

We screened the MIMIC IV database for 1985 patients with cerebral hemorrhage admitted for the first time between 2008 and 2019. Exclusions included cases with duplicate records, patients with tumors, aneurysms, traumatic cerebral hemorrhage, cerebral hemorrhage due to blood disorders, and cases missing PaO₂ values — 828 patients for final analysis. Based on PaO₂ levels within 24hours of admission, patients were divided into three groups: high (>150mmHg), medium (100–150mmHg), and low (<100mmHg). A logistic mixed-effects regression model was used to analyze the relationship between PaO₂ and hospital mortality.

The overall mortality rate during hospitalization was 37.1%. After adjusting for confounding factors, the mortality risk ratio (OR) of the low PaO₂ group and high PaO₂ group were 1.78 (95%CI: 1.04–3.03, P=0.034) and 2.09 (95%CI: 1.28–3.42, P=0.003), respectively. However, no significant associations were found between PaO₂levels and 28-, 60-, or 90-day mortality. Subgroup analysis showed an interaction between in-hospital mortality and PaO₂ level in patients with chronic lung disease (P=0.002). Sensitivity analysis showed that the relationship still existed after excluding extreme PaO₂ values.

Elevated arterial oxygen tension within the first 24hours of admission was independently associated with increased in-hospital mortality in patients with intracerebral hemorrhage.