Early 48-h glucose-to-lymphocyte ratio trajectories and 28-day mortality in sepsis-associated acute kidney injury: a retrospective landmark secondary analysis using MIMIC-IV

Abstract Sepsis-associated acute kidney injury (SA-AKI) is a common and clinically important phenotype of critical illness, yet early risk assessment still relies mainly on static variables. We evaluated whether early 48-h glucose-to-lymphocyte ratio (GLR) trajectories were associated with 28-day mortality in critically ill patients with SA-AKI. This retrospective landmark secondary analysis used a curated SA-AKI cohort derived from MIMIC-IV. Adults who survived beyond 48 h, stayed in the intensive care unit for at least 24 h, had no pre-existing renal disease or malignancy, and had paired glucose and lymphocyte measurements in the 0–24-h and 24–48-h windows were included. Patients were classified into four trajectory groups according to baseline GLR and early GLR change. Associations with 28-day mortality were examined using Kaplan–Meier analysis, multivariable Cox models, restricted cubic splines, multiple imputation, measurement-frequency summaries, alternative exposure definitions, and model-performance comparisons. Among 1,179 included patients, 241 (20.4%) died within 28 days after the 48-h landmark. In the primary multiple-imputation fully adjusted model, both the high baseline + decreasing (pooled HR 1.53; P = 0.049) and high baseline + increasing trajectories (pooled HR 1.83; P = 0.012) were associated with higher mortality. In the complete-case sensitivity analysis, the high baseline + increasing trajectory remained associated with higher mortality (HR 1.90, 95% CI 1.14–3.17; P = 0.014), whereas the high baseline + decreasing trajectory was attenuated (HR 1.49, 95% CI 0.93–2.38; P = 0.094). Baseline GLR analyzed continuously remained associated with mortality, while absolute and percentage GLR change alone were not. Baseline GLR carried the most robust prognostic signal, while early GLR trajectory provided modest additional context, particularly among patients with high baseline GLR. Early dynamic metabolic-immune monitoring may therefore help refine, but should not replace, baseline risk assessment in SA-AKI.