Nutritional predictors of one-year survival in hemodialysis patients: insights from a cohort analysis.

BACKGROUND

Protein-energy wasting is linked to adverse outcomes in dialysis. The independent prognostic value of pragmatic nutritional indicators under strong age confounding warrants clarification.

METHODS

Retrospective single-center cohort of adults on thrice-weekly high-flux hemodialysis.

PRIMARY ENDPOINT: all-cause 1-year mortality. Primary analysis used multivariable logistic regression on the full cohort (N = 416) with age (continuous) and sex adjustment. Prespecified sensitivity analyses included age stratification (< 75 vs ≥ 75 years) and an age × GNRI interaction. Matched comparisons (± 1-year nearest neighbor, 1:1) are provided in the Supplement. Calibration (Brier score, calibration plot), discrimination (AUC), and incremental value metrics (ΔAUC, IDI/NRI) were computed.

RESULTS

In the full cohort, deaths at 1 year were 59/416 (14.2%). In multivariable models, GNRI was inversely associated with mortality (OR per 1-point 0.97; 95% CI 0.94-0.99; p = 0.010), while age increased risk (OR per year 1.06; 95% CI 1.03-1.09; p < 0.001). The AUC improved from 0.706 (age + sex) to 0.714 (age + sex + GNRI); IDI 0.030 and continuous NRI 0.301 indicated incremental predictive gain. The model showed reasonable calibration (Brier ≈ 0.103). In age-stratified analyses, the GNRI association was significant in subjects ≥ 75 years (OR 0.94; 95% CI 0.91-0.97; p < 0.001) but not in < 75 years, with a significant age × GNRI interaction (p = 0.011).

CONCLUSIONS

Under routine data availability, GNRI provides independent, clinically implementable prognostic information-especially in older hemodialysis patients-supporting its use in pragmatic risk stratification.