Prognostic Nutritional Index as a Predictor of Mortality in 101,616 Patients Undergoing Hemodialysis

Abstract

High mortality in dialysis patients is linked to malnutrition and inflammation. Prognostic nutritional index (PNI), calculated from serum albumin level and total lymphocyte count, has been developed as a prognostic marker for cancer patients. We investigated the clinical utility of PNI in predicting mortality in patients undergoing hemodialysis. Thus, 101,616 patients who initiated hemodialysis in United States dialysis centers between 2007 and 2011 were included in this retrospective cohort study. Using the Cox regression model, we assessed the relationship between PNI and mortality. Further, the predictive value of PNI for one-year mortality was compared with that of its constituent using area under the receiver operating characteristic curve, net reclassification improvement, and integrated discrimination improvement. Higher PNI quartiles were incrementally associated with lower mortality; in patients with PNI values of 39.5–<43.1, 43.1–<46.6, and ≥46.6 (reference: PNI < 39.5), case-mix adjusted hazard ratios (95% confidence intervals) were 0.66 (0.64, 0.68), 0.49 (0.48, 0.51), and 0.36 (0.34, 0.37), respectively. PNI predicted mortality better than serum albumin level or total lymphocyte count alone. In the subgroup analysis, PNI performed well in predicting mortality in patients aged < 65 years. Our results indicate that PNI is a simple and practical prognostic marker in patients undergoing hemodialysis.