Predicting Bloodstream Infection in High-risk Groups of Liver Failure Patients with Artificial Liver Support System.

ABSTRACT

Background: Liver failure is a rare, life-threatening disease that has a high mortality rate and affects many organ systems. Bloodstream bacterial infection has played a key role in liver failure patients with plasma exchange-centered artificial liver support systems, but the predicted risk factors of infection have not been fully understood. Objective: We aimed to predict bloodstream bacterial infection in high-risk groups of liver failure patients during a plasma exchange-centered artificial liver support system. Design: This was a prospective cohort study. Setting: This study was performed in Nanjing Drum Tower Hospital, The Affiliated Hospital of Nanjing University Medical School. Participants: 118 liver failure patients with plasma exchange-centered artificial liver support system therapy from Nanjing Drum Tower Hospital from November 2019 to November 2020 were selected. Interventions: We used a stepwise binary logistic regression model to select the optimal risk factors of infection with minimum Akaike information criterion, and the Nomogram prognostic model for bloodstream infection was constructed for visualization. Primary Outcome Measures: Risk factors of bloodstream infection (2) predictive accuracy of the constructed nomogram model. Results: Among the 118 liver failure patients, 22 (18.64%) were diagnosed with bloodstream bacterial infection. The univariable and multivariate logistic regression analyses suggested that culture level, glucocorticoids use, number of punctures, blood platelet counts, white blood cell counts, and indwelling catheter time were the sex predictors of bloodstream infection for liver failure patients during plasma exchange-centered artificial liver support system (P = .042, P = .013, P = .025, P = .003, P = .024 and P = .026). The nomogram predictive model was established with high prediction accuracy, of which the area under the curve was 0.935 (95% confidence interval 0.884-0.986), the sensitivity was 0.955, and the specificity was 0.854. Conclusion: The constructed nomogram prognostic model can recognize the risk factors and accurately predict bloodstream infection for liver failure patients during plasma exchange-centered artificial liver support system.