The association between pan-immune-inflammation value with mortality in critically ill patients with sepsis-associated acute kidney injury

Abstract

Background: Sepsis-associated acute kidney injury (SA-AKI) significantly impacts global health. Early identification of SA-AKI patients at inflammatory and immune risk, followed by timely interventions, is critical for improving outcomes. The pan-immune-inflammation value (PIV) reflects systemic inflammation and immune status. However, its prognostic value in SA-AKI remains unexplored. Methods: This retrospective cohort study analyzed SA-AKI patients in the MIMIC-IV database. Cox regression assessed the association between PIV and mortality, while restricted cubic spline (RCS) regression explored the relationship between PIV and 30-day and 365-day mortality. Results: A total of 2,473 SA-AKI patients in our study were categorized into PIV quartiles: T1 (≤ 214), T2 (214–679), T3 (679–2,039), and T4 (> 2,039). PIV showed a nonlinear association with mortality. Higher PIV quartiles were linked to increased mortality, with 30-day rates of 26%, 22%, 35%, and 41% (P < 0.001) and 365-day mortality rates of 34%, 31%, 46%, and 54% (P < 0.001). Adjusted hazard ratios (HR) for 30-day mortality across quartiles were 1.00 (reference), 1.04(0.82, 1.31), 1.54 (1.25, 1.9), and 1.62 (1.32, 1.98), respectively. For 365-day mortality, the HR and 95% CI were 1.00 (reference), 1.06 (0.87, 1.30), 1.58 (1.32, 1.90), and 1.70 (1.42, 2.03). After adding PIV to SOFA score, the integrated discrimination improvement (IDI) for 30-day mortality was 0.005, and the net reclassification improvement (NRI) was 0.103. For 365-day mortality, the IDI was 0.009, and the NRI was 0.124. Regarding the APACHE II score, the IDI for 30-day mortality was 0.003, and the NRI was 0.081. For 365-day mortality, the IDI was 0.006, and the NRI was 0.107. Conclusion: Elevated PIV independently predicts both short- and long-term adverse outcomes in SA-AKI patients. Incorporating PIV into established critical illness prediction models, such as SOFA and APACHE II, enhances their prognostic accuracy.