Prediction of Fungal Infection Development and Their Impact on Survival Using the NACSELD Cohort

Bacterial infections are associated with negative outcomes in cirrhosis but fungal infections are being increasingly recognized. The objective of this study is to define risk factors for fungal infection development and impact on 30-day survival.In a large, multi-center cirrhotic inpatient cohort, demographics, cirrhosis details, intensive care unit (ICU), organ failures/acute-on-chronic liver failure (ACLF), and 30-day survival were compared between patients without infections and with bacterial infections alone, with those with fungal infections. Variables associated with fungal infection development were determined using multi-variable regression. Ordinal variables (0=no infection, 1=community-acquired bacterial infection, 2=nosocomial bacterial, and 3=fungal infection) were input into a 30-day survival model.A total of 2,743 patients (1,691 no infection, 918 bacterial, and 134 fungal infections) were included. Patients with fungal infection, all of which were nosocomial, were more likely to be admitted with bacterial infections, on spontaneous bacterial peritonitis prophylaxis, and have diabetes and advanced cirrhosis. Bacterial infection types did not predict risk for fungal infections. Multi-variable analysis showed male gender to be protective, whereas diabetes, longer stay, ICU admission, acute kidney injury (AKI), and admission bacterial infection were associated with fungal infection development (area under the curve (AUC)=0.82). Fungal infections were associated with significantly higher ACLF, inpatient stay, ICU admission, and worse 30-day survival. The case fatality rate was 30% with most fungal infections but >50% for fungemia and fungal peritonitis. On a multi-variable analysis, age, AKI, model for end-stage liver disease, ICU admission, and ordinal infection variables impaired survival (P<0.0001, AUC=0.83).Fungal infections are associated with a poor 30-day survival in hospitalized cirrhotic patients compared with uninfected patients, and those with bacterial infections. Patients with diabetes, AKI, and those with an admission bacterial infection form a high-risk subgroup.