Association Between Red Cell Distribution Width and In-Hospital Mortality in Patients With Chronic Kidney Disease: A Retrospective Cohort Study

Background Elevated red cell distribution width (RDW) independently predicts mortality in cardiovascular and chronic kidney disease (CKD) populations globally, but data from Indian hospitalized non-dialysis CKD patients, where sepsis and late presentations are prevalent, remain scarce. In this study, we evaluated RDW as a prognostic marker for in-hospital mortality in this setting. Methodology We conducted a retrospective, single-center cohort study of 176 adults (≥18 years) hospitalized with Kidney Disease: Improving Global Outcomes (KDIGO)-defined CKD (January 2022 to December 2024) at a tertiary Indian hospital after excluding those with acute kidney injury-on-CKD and stays <48 hours. Multivariable logistic regression assessed the RDW-mortality association, adjusted for sepsis, heart failure, coronary artery disease, and albumin. Discrimination was assessed by receiver operating characteristics (ROC)-area under the curve (AUC); bootstrapped confidence intervals (CIs) were used, given the modest event count (n = 43). Results In-hospital mortality was 24.4% (43/176). Non-survivors had significantly higher median RDW (16.1% vs. 14.2%, p < 0.001). RDW independently predicted mortality (adjusted odds ratio (OR) = 1.30 per 1% increase, 95% CI = 1.07-1.57), consistent with non-dialysis CKD studies. Discrimination was moderate for RDW alone (AUC = 0.696) and improved in the multivariable model (AUC = 0.774). An exploratory cutoff of 14.8% yielded a sensitivity of 78.6% and a specificity of 59.8%. Conclusions Elevated RDW is an independent predictor of in-hospital mortality in advanced hospitalized CKD in an Indian tertiary setting, consistent with global evidence. Despite moderate discrimination, its value lies in being a low-cost, universally available marker that can facilitate early risk stratification, particularly in resource-limited settings with high sepsis and malnutrition burden. However, prospective external validation is warranted to confirm these findings.