The red cell distribution width to albumin ratio as a novel biomarker for predicting short-term mortality in severe pulmonary sepsis: a retrospective study with dual-cohort validation

Background: The Red Cell Distribution Width to Albumin Ratio (RAR) is a biomarker that reflects a patient's nutritional status, inflammatory response, and oxidative stress, showing significant potential in critical care medicine. To investigate its prognostic value, we conducted a retrospective study using a dual-cohort design to assess the association between RAR and short-term (28-day) mortality in patients with pulmonary sepsis. Materials and methods: We retrospectively identified patients with sepsis secondary to pulmonary infections from the Binzhou Medical University Hospital medical records and the Medical Information Mart for Intensive Care (MIMIC-IV) database. To examine the association between RAR and short-term adverse outcomes in these patients, we employed several statistical methods, including Kaplan-Meier survival curves, multivariable Cox regression, and restricted cubic spline (RCS) analysis. Subsequently, we applied machine learning algorithms-namely the Boruta algorithm, LASSO-COX regression, and Random Forests-to identify the most predictive features. These features were then used to develop a final multivariable Cox regression model for risk prediction. The performance of this predictive model was evaluated using receiver operating characteristic (ROC) curve analysis. Result: The final analysis included 6,065 patients with pulmonary sepsis. The 28-day ICU and in-hospital mortality rates were 20.50 and 19.30%, respectively. In the fully adjusted multivariable model, a higher RAR was significantly associated with increased 28-day ICU and in-hospital mortality, whether treated as a continuous or categorical variable. For each unit increase in the continuous RAR score, the hazard ratios (HR) for 28-day ICU and in-hospital mortality were 1.52 (95% CI: 1.28-1.80) and 1.30 (95% CI: 1.09-1.55), respectively. Similarly, when compared to the low RAR group, the high RAR group had hazard ratios of 1.45 (95% CI: 1.23-1.70) and 1.29 (95% CI: 1.09-1.52) for the two outcomes. The restricted cubic spline (RCS) analysis revealed a positive dose-response relationship between RAR levels and short-term adverse outcomes. Furthermore, the risk prediction model incorporating RAR and eight other independent predictors demonstrated superior performance in identifying high-risk patients compared to traditional critical illness scoring systems, as shown by receiver operating characteristic (ROC) analysis. All findings were consistently validated in the external cohort. Conclusion: In conclusion, our study demonstrates a significant inverse association between the RAR and short-term survival in patients with severe pulmonary sepsis. The RAR-based scoring system we developed shows promise as a practical adjunct tool for clinical risk assessment. Prospective validation is warranted to confirm its utility in improving risk stratification for this patient population.