Abstract / Summary
Abstract Introduction The systemic immune-inflammation index (SII) has been proposed as a simple prognostic biomarker in COVID-19, but its performance relative to routinely available inflammatory markers and its incremental value beyond clinical predictors remain uncertain. We evaluated the association, discriminative performance, and incremental prognostic value of admission SII for in-hospital mortality in patients hospitalized with COVID-19. Methods We conducted a retrospective multicenter study including adult patients hospitalized for confirmed moderate-to-severe SARS-CoV-2 infection after Emergency Department (ED) admission between March 1 and April 30, 2020, across six hospitals in northeastern France. The association between SII and in-hospital mortality was assessed using multivariable logistic regression with alternative SII parameterizations. Discrimination was compared with the neutrophil-to-lymphocyte ratio (NLR), platelet-to-lymphocyte ratio (PLR), lymphocyte count, and C-reactive protein (CRP). Incremental prognostic value beyond a clinical model was assessed using changes in discrimination, reclassification metrics, and decision curve analysis. Results Among 1,035 patients, 139 (13.4%) died during hospitalization. SII dichotomized at the ROC-derived threshold of 1,723 was independently associated with mortality (adjusted OR = 1.54, 95%CI: 1.03–2.30; p = 0.037). However, this association was not significant when SII was analyzed continuously (adjusted OR = 1.04 per 1,000 units; p = 0.194), after log-transformation (adjusted OR = 1.19; p = 0.096), or across tertiles (p for trend = 0.096). SII showed weak discrimination for mortality (AUC = 0.563, 95%CI: 0.509–0.618) and performed significantly worse than NLR (AUC = 0.615; p < 0.001). Adding SII to the clinical model did not significantly improve discrimination (AUC = 0.7363 vs. 0.7402; p = 0.487), and decision curve analysis showed no meaningful increase in net benefit. Conclusion SII showed limited standalone prognostic performance and provided no meaningful incremental value beyond clinical predictors. These findings support moving beyond isolated inflammatory indices toward integrated multimodal approaches for risk stratification in COVID-19.