Abstract: Objective To develop and validate a risk scoring tool for pulmonary infection in patients with acute-on-chronic liver failure（ACLF）. Methods 585 patients with ACLF in a tertiary Class A hospital from January 2009 to September 2022 were selected and assigned to the modeling group（411 cases） and the validation group（174 cases） in a 7 ∶ 3 ratio. We used univariate and multivariate logistic regression to analyze the risk factors for pulmonary infection in ACLF patients. We constructed a predictive model for pulmonary infection risk in ACLF patients using modeling group data,and validated the model using validation group data. The discriminability and calibration of the constructed model were evaluated using receiver operating characteristic curves and H-L test. Then we developed a risk scoring tool based on a risk prediction model and applied it in clinical practice. Results 585 patients with ACLF were included,and a total incidence of pulmonary infection was 36.41%. The incidence rates of pulmonary infection in modeling group and validation group were 37.71% and 33.33%,respectively. The results of multivariate logistic regression showed that fever,cough,concomitant infections,combined use of antibiotics≥2,femoral vein catheterization treatment,and Child-Pugh grading were risk factors for pulmonary infection in ACLF patients（P<0.05）. The area under the ROC curve of the prediction model is 0.949（0.928~0.970）,with a sensitivity of 96.1% and a specificity of 80.5%. The H-L test result was（χ²=9.850,P=0.276）. The results showed that the model has good prediction and calibration. The maximum value of the developed scoring tool‘s Youden index wss 0.766,with a critical score of 16. Based on the probability comparison chart of ACLF patients with pulmonary infection,the patients were divided into four levels：low,medium,medium high,and high risk. The corresponding sensitivity was 96.1%,the specificity was 80.5%,and the H-L test result was（χ²=6.005,P=0.650）. And the risk scoring tool was initially applied to clinical practice,with an accuracy of 86.8%. Conclusion The risk scoring tool for pulmonary infection has a high degree of discrimination and calibration,which can help medical staff quickly,conveniently,and early identify the risk of pulmonary infection in ACLF patients.

Key words: Acute-on-Chronic Liver Failure, Pulmonary Infection, Scoring Tool, Nursing Care