Prognostic nutritional index at discharge as a practical bedside tool for long-term all-cause mortality risk in pneumonia-induced sepsis survivors.

Abstract

Existing research has predominantly focused on short-term outcomes of sepsis during intensive care unit (ICU) admission, with limited evidence on long-term all-cause mortality. This multicenter ambispective longitudinal cohort study aimed to evaluate the prognostic value of the prognostic nutritional index (PNI), a composite marker reflecting nutritional and immune-inflammatory status, for long-term all-cause mortality in pneumonia-induced sepsis survivors. A total of 461 pneumonia-induced sepsis survivors with recurrence-free status for at least 3 months post-discharge were retrospectively enrolled and prospectively followed for a median of 36 months. Participants were stratified into tertiles based on discharge PNI scores. The association between PNI and long-term all-cause mortality was assessed, and its prognostic performance was compared with established indicators, including the Controlling Nutritional Status (CONUT) score, modified Glasgow Prognostic Score (mGPS), blood urea nitrogen-to-albumin ratio (BAR), neutrophil-to-lymphocyte ratio (NLR), and platelet-to-lymphocyte ratio (PLR). Restricted cubic spline (RCS) analysis revealed an inverted L-shaped, dose-dependent negative association between PNI and long-term all-cause mortality risk ( p for overall < 0.001). Multivariate Cox regression analysis showed that participants in the lowest PNI tertile had a 54.8% higher mortality risk than those in the highest tertile ( p for trend = 0.043). Subgroup analyses confirmed consistent prognostic value across diverse populations (all interaction p -values > 0.05). Kaplan-Meier analysis demonstrated that lower PNI levels were significantly associated with higher cumulative mortality (log-rank, p < 0.001). Multivariate linear regression further indicated that each one-tertile increase in PNI corresponded to a quantifiable reduction [calculated as |β| × tertile interval (5.9)] in related nutritional and inflammatory markers, in model III, including a reduction of 1.0585 in COUNT score (β = -0.1794), 0.1398 in mGPS (β = -0.0237), 0.0378 in BAR (β = -0.0064), 45.6619 in PLR (β = -7.7393), and 2.8951 in NLR (β = -0.4907) (all p < 0.001). Receiver operating characteristic (ROC) curve analysis identified a PNI clinical diagnostic cutoff of 39.00 for long-term all-cause mortality, with an area under the curve (AUC) of 0.672 ( p < 0.001) and a sensitivity of 0.64, outperforming the established indicators. These findings suggest that PNI may serve as a practical bedside tool for estimating long-term post-hospital all-cause mortality among survivors of pneumonia-induced sepsis, supporting personalized clinical decision-making and routine care as an auxiliary prognostic indicator.