To investigate the relationship between arterial oxygen partial pressure (PaO₂) and all-cause mortality of post–cardiac surgery patients.

A retrospective cohort study based on the Medical Information Mart in Intensive Care-III (MIMIC-III) database.

The MIMIC-III database is a dataset comprising the clinical data of patients admitted to intensive care units (ICUs) at the Beth Israel Deaconess Medical Center.

Patients who underwent cardiac surgery under extracorporeal circulation.

The patients were categorized into 4 groups based on PaO₂ level within 24 hours of admission to the ICU after surgery. The primary outcome was in-hospital all-cause mortality. Both linear and nonlinear relationships between PaO₂ and outcome were examined in the study.

During a 1-year follow-up, 296 patients (5.5%) experienced all-cause mortality, including 73 (1.4%) during hospitalization, 83 (1.6%) within 30 days, and 167 (3.1%) within 90 days. The restricted cubic spline regression model revealed a nonlinear relationship between PaO₂ and in-hospital mortality (P_overall = 0.004; P_nonlinearity = 0.012). Subgroup analyses and sensitivity analyses were consistent with the primary analysis.

The relationship between PaO₂ level and all-cause in-hospital mortality followed a J-shaped curve among patients following cardiac surgery. When PaO₂ > 196.1mmHg, higher levels of PaO₂ within 24 hours after cardiac surgery were associated with an increased risk of hospital mortality.