Longitudinal modeling of red blood cell distribution width dynamics and mortality risk in critically Ill patients with sepsis-associated acute kidney injury

Abstract

Background Sepsis-associated acute kidney injury (SA-AKI) is a critical condition with high mortality. Red cell distribution width (RDW) has emerged as a potential dynamic biomarker, but longitudinal RDW changes in SA-AKI remain underexplored. Methods This retrospective cohort study analyzed adult SA-AKI patients from the MIMIC-IV database (2008–2022). Group-Based Trajectory Modeling (GBTM) identified distinct longitudinal RDW patterns. Primary outcome was 28-day all-cause mortality. Secondary outcomes included 90-day mortality, continuous renal replacement therapy (CRRT) requirement, and ICU length of stay. Multivariable Cox models assessed associations. adjustment for any transfusion and any major hemorrhage, and an exclusion analysis in which all transfused patients were removed. Results Among 6,694 patients (mean age 65.5 years, 57.7% male), 28-day mortality was 22.5%. Four RDW trajectory groups were identified: Stable Low (27.8%), Gradual Increase (38.5%), Continuous Increase (27.6%), and Rapid Increase (6.1%). The Rapid Increase group demonstrated highest disease severity scores and poorest laboratory profiles. Compared to the Stable Low group, the Rapid Increase group had significantly elevated 28-day mortality risk after full adjustment (HR 4.27, P < 0.001), with consistent patterns for 90-day mortality and resource use. Associations remained consistent across subgroup analyses, multiple-imputation datasets, and the relaxed-inclusion cohort, and were minimally altered after adjusting for transfusion/hemorrhage or excluding transfused patients. Conclusions Dynamic RDW trajectories were independently associated with adverse outcomes in SA-AKI patients. Across extensive sensitivity analyses, these trajectories functioned as dynamic prognostic indicators without implying causality, supporting their use for risk stratification and individualized patient care.