The relationship between pan immune inflammatory index (PIV) and 28 day mortality in patients with severe urinary sepsis: a retrospective study of a multinational dual cohort study.

Background: Urosepsis is a critical condition originating from urinary tract infections, characterized by rapid progression and high mortality. The Pan-immune-inflammation Value (PIV), a novel composite index reflecting integrated immune and inflammatory status, has shown prognostic value in various critically ill and septic populations. However, its independent prognostic value in urosepsis remains inadequately explored. This study therefore aims to systematically evaluate the ability of PIV to predict 28-day adverse outcomes in patients with urosepsis. Materials and methods: This study employed a retrospective dual-cohort design. The internal training cohort comprised adult ICU patients with urosepsis from the MIMIC-IV database, while the external validation cohort was derived from the electronic medical record system of Anshun Municipal People's Hospital. To investigate the association between the PIV and short-term adverse outcomes in urosepsis patients, we utilized a range of statistical methods, including multivariable Cox regression, restricted cubic spline (RCS) analysis, subgroup analysis, and Kaplan-Meier survival curves. To develop a parsimonious predictive model, the internal cohort was randomly split into training and testing sets at a 7:3 ratio. Within the training set, an ensemble machine learning strategy-incorporating the Boruta algorithm, LASSO-Cox regression, random forest (RF), gradient boosting (GBDT), and support vector machine (SVM)-was applied to identify key predictive variables from serological tests, comorbidities, demographic characteristics, and vital signs. Based on the selected features, prognostic models were constructed using multivariable Cox regression in the training, testing, and external validation sets, respectively. The discriminatory power of these models against traditional disease severity scores was assessed using receiver operating characteristic (ROC) curves, with the area under the curve (AUC) quantifying predictive performance. Result: A total of 1,686 patients with severe urosepsis were included in this study. In the fully adjusted model, both continuous PIV and PIV quartiles were independently associated with 28-day adverse outcomes. For 28-day ICU mortality, each unit increase in continuous PIV was associated with a 76.4% higher risk (HR 1.764, 95% CI 1.340-2.323, Conclusion: This study confirms PIV as an independent predictor of 28-day mortality risk in patients with urosepsis across two cohorts. The prediction model incorporating PIV and four other clinical variables exhibited good discrimination and calibration, with prognostic performance superior to that of traditional disease severity scores. Future prospective, multicenter studies involving diverse geographic and ethnic populations are needed to further validate the generalizability of this model, thereby facilitating precise risk stratification for patients with urosepsis.