Association of Neutrophil Percentage-to-Albumin Ratio With Mortality in ICU Patients With Acute Pancreatitis: A Retrospective Study.

OBJECTIVE: To investigate whether the neutrophil percentage-to-albumin ratio (NPAR) predicts 28- and 365-day mortality in ICU patients with acute pancreatitis (AP) and improves the performance of existing ICU scoring systems. METHODS: We extracted data on 437 ICU AP patients from the MIMIC-IV 3.0 database. Selection bias was assessed by comparing included and excluded patients. The optimal NPAR cutoff was determined by maximally selected rank statistics. Kaplan-Meier and Cox regression assessed the NPAR-mortality association, with 200-iteration bootstrap resampling for internal validation. Time-dependent AUC, continuous net reclassification improvement (NRI), and integrated discrimination improvement (IDI) evaluated whether adding NPAR to SAPS II, SOFA, and SIRS improved prediction. RESULTS: The optimal NPAR cutoff was 34.6. Included patients had more severe disease than excluded patients (SOFA 6.75 vs. 5.32; 28-day mortality 17.4% vs. 11.8%). High NPAR (≥ 34.6) was significantly associated with lower survival (log-rank p < 0.001) and independently predicted 28-day mortality (HR = 2.043, 95% CI 1.235-3.380, p = 0.005) and 365-day mortality (HR = 1.749, 95% CI 1.174-2.606, p = 0.006). After bootstrap validation, optimism-corrected HRs were 1.820 and 1.560, with corrected AUCs of 0.765 and 0.695, respectively. Adding NPAR to SIRS significantly improved AUC (28-day: 0.613 vs. 0.529, p = 0.011; 365-day: 0.580 vs. 0.507, p = 0.004), NRI, and IDI (all p < 0.05), whereas adding NPAR to SAPS II or SOFA yielded no significant improvement (all p > 0.05). CONCLUSION: NPAR ≥ 34.6 is an independent predictor of 28- and 365-day mortality in ICU AP patients, with robust internal validation. Combining NPAR with SIRS significantly enhances mortality prediction.