A predictive model for sepsis-associated liver injury incorporating serum 25(OH)D, bile acids, APRI, and age

Background: Sepsis-associated liver injury (SALI) is a severe complication with a poor prognosis. We aimed to identify early indicators and construct a preliminary predictive model for SALI. Methods: We conducted a retrospective analysis of 173 septic patients admitted to the intensive care unit from 2021 to 2024. Factors independently associated with SALI were identified using multivariable logistic regression, and a predictive model was constructed. Model performance was assessed using receiver operating characteristic (ROC) curves, calibration curve analysis, decision curve analysis, and clinical impact curve analysis. Results: Low serum 25-hydroxyvitamin D (25[OH]D) levels (adjusted odds ratio [AOR] 0.86), elevated total bile acids (TBA) (AOR 1.02), and an AST-to-platelet ratio index (APRI) >0.848 (AOR 9.81) were independently associated with SALI, while age >60 years showed an inverse association (AOR 0.35). A predictive model incorporating these variables achieved an area under the ROC curve (AUC) of 0.92 (95% CI 0.88–0.96), with favourable calibration and clinical net benefit in internal validation. Conclusions: Low serum 25(OH)D, elevated TBA, and APRI >0.85 were independently associated with SALI, whereas age >60 years was preliminarily found to be inversely associated with SALI. The model incorporating these variables exhibited favourable performance in internal validation. These findings are preliminary and require external prospective validation to confirm their generalizability.