Background This study aims to explore the association between the ROX index and clinical outcomes in patients with sepsis. Methods Data were extracted from the Medical Information Mart for Intensive Care IV (MIMIC-IV) database, including adult sepsis patients admitted to the intensive care unit (ICU). The primary outcome was 28-day mortality, while secondary outcomes included 7-day and 14-day mortality, ICU mortality, ICU length of stay (LOS), and hospital LOS. Restricted cubic spline models and Cox regression models were used to assess the associations between the ROX Index and clinical outcomes. Results A total of 23,502 sepsis patients were included, who were stratified into high (ROX ≥ 6.46) and low ROX (ROX 6.46) strata using a data-derived threshold survival analysis. In the unadjusted model, the high ROX stratum exhibited a significantly lower risk of 28-day mortality (HR = 0.33, 95% CI: 0.31–0.35, P 0.001), with consistent findings after adjustment for age, sex, and SOFA score respectively. Similar trends were observed for ICU, 7-day, and 14-day mortality. Restricted cubic spline analysis revealed a nonlinear “L”-shaped association, with the 28-day mortality risk decreasing until reaching a plateau at a ROX index of approximately 10.50. Additionally, the shortest ICU and hospital LOS were observed at ROX thresholds of 9.56 (4.87 days) and 9.29 (12.05 days), respectively. The ROX index showed a moderate predictive accuracy for 28-day mortality (AUC = 0.63, 95% CI: 0.62–0.64), outperforming the SOFA score (AUC = 0.59, 95% CI: 0.58–0.60). Subgroup analyses confirmed consistent associations across demographic and clinical subgroups (overall HR = 0.33, 95% CI: 0.31–0.35), with significant interaction effects observed in gender, patients with septic shock, heart failure, diabetes, chronic lung disease, and those not receiving non-invasive ventilation (all P 0.05). Conclusion In patients with sepsis, a higher ROX index is associated with significantly lower mortality rates and shorter ICU and hospital stays. However, the ROX index demonstrated moderate predictive accuracy.
Xiao‐Jun Ji (Tue,) studied this question.