The prognostic nutritional index vs. MELD for short-term mortality in critically ill liver disease: a retrospective multi-cohort study.

Malnutrition, sarcopenia and frailty are common in advanced liver disease and predict poorer survival, encouraging simple blood-based indices when formal assessment is impractical. The most widely used, the Prognostic Nutritional Index (PNI), combines serum albumin and the peripheral lymphocyte count; yet both have been reported to be strongly influenced by systemic inflammation as well as by nutrition, and the index is seldom benchmarked against a disease-specific reference or against albumin alone. We evaluated whether the PNI provides clinically useful prognostic information for short-term mortality in critically ill patients with liver disease, using three independent cohorts: a North American intensive-care primary cohort (n = 3,653), a multicentre external-validation cohort (n = 922) and a single-center Chinese cohort in which hepatitis B was the leading cause of liver disease (n = 453). In each cohort we quantified discrimination of 28-day or in-hospital mortality, compared the PNI with serum albumin alone and with the Model for End-Stage Liver Disease (MELD), and tested whether it improved discrimination or reclassification beyond MELD. The PNI discriminated death only marginally better than chance (area under the curve 0.52-0.56) and no better than albumin alone, so its lymphocyte component added no independent signal; the weak per-standard-deviation association seen in the primary cohort was not significant in either validation cohort. MELD discriminated substantially better everywhere (0.67-0.70). The conventional PNI threshold classified 89-97% of patients as abnormal and was uninformative. Across three case-mixes the PNI behaved as a weak, non-specific prognostic discriminator, adding nothing to MELD, MELD-Na or MELD 3.0; in an exploratory paired subgroup, re-measurement on days 3-7 did not improve discrimination. This is consistent with its components tracking systemic inflammation and chronic physiologic reserve-an axis largely orthogonal to the acute organ failure that drives short-term mortality. Separating the index into its components and testing incremental value provides the comparative evidence this literature has lacked. The PNI was assessed as a discriminator of short-term mortality, not as a measure of nutritional status: nutritional risk is better identified with instruments validated for that purpose, and mortality risk with MELD.