The V-shaped association between the ratio of neutrophil counts to prognostic nutritional index and 30-, 60-, and 90-day mortality in elderly critically ill patients aged 65 and older with sepsis: a retrospective study based on the MIMIC database

Objective To investigate the association between the ratio of neutrophil counts to prognostic nutritional index (NPNR) and mortality risk in elderly patients with severe sepsis aged 65 and older. Methods This retrospective study utilized data from the MIMIC-IV (v2.2) database. We selected elderly patients (aged ≥65 years) with sepsis admitted to the ICU for the first time. After excluding patients with hematologic disorders, autoimmune diseases, or active malignancies, 13,176 patients were screened. Among them, 1,179 had complete data for NPNR calculation and were included. The primary endpoint was 30-day mortality, with secondary endpoints including 60-day and 90-day in-hospital and ICU mortality. We used Kaplan–Meier survival analysis, Cox proportional hazards models, restricted cubic spline (RCS) regression models, and comprehensive sensitivity analyses to assess the association between NPNR and mortality risk. Results NPNR was significantly associated with increased mortality risk. Kaplan–Meier survival curves demonstrated lower survival rates in extreme quartiles (Q1 and Q4) across all time frames ( p < 0.001). Cox regression showed significant associations for NPNR as a continuous variable and in higher quartiles. After adjustment for confounders including albumin infusion status, the V-shaped association persisted in RCS models ( p < 0.001). Sensitivity analyses using multiple imputation and subgroup stratification confirmed robustness. The excluded patients ( n = 11,997) showed comparable baseline severity to the included cohort. Conclusion NPNR is a significant predictor of mortality in elderly sepsis patients. Its V-shaped association reflects dual risks of immunonutritional depletion and hyperinflammation, highlighting the clinical utility of this composite marker.