Abstract: Objective To establish a visual nomogram with predicting value for infection in patients with liver failure. Methods The patients with liver failure admitted to our hospital from January 2014 to December 2019 were collected and they were divided into infected group and non-infected group. The clinical data of patients were collected and analyzed by univariate and multivariate logistic regression analysis. Based on the screened independent risk factors, R language software was used to visualize the logistic regression (LR) model to obtain the nomogram, and the distinguishing ability and consistency ability of the model was evaluated through internal verification. Results Univariate analysis showed that the average ages of infected group and non-infected group were (44±13) and (49±14) respectively, (t=-3.273, P=0.001). There were 59 and 19 patients with liver cirrhosis in infected group (39.3%) and non-infected group (17.6%) respectively, χ²=13.46, P<0.001; There were 105 (70%) and 49 (48%) patients with hepatic encephalopathy in infected group and non-infected group respectively, χ²=4.341, P=0.037; There were 87 (58%) and 20 (19.6%) patients with ascites in infected group and non-infected group, χ²=36.63, P<0.001. There were 39 (26%) and 4 (3.92%) patients with hepatorenal syndrome in infected group and non-infected group, χ²=20.91, P<0.001; APACHEII score in infected group and non-infected group were (19.8±5.30) and (16.6±5.24) respectively, t=-4.8, P<0.001. The white blood cell (WBC) count in infected group and non-infected group were 9.56 (6.37-14.6) ×10⁹/L and 7.25 (5.77-10.7) ×10⁹/L, Z=-2.807, P<0.001. The alanine aminotransferase level in infected group and non-infected group were 309 (64-773) U/L and 554 (182-1039) U/L, Z=-2.577, P=0.01. The albumin level in infected group and non-infected group were (31.3±5.2) g/L and (33.1±5.0) g/L, t=2.657, P=0.008. The cholinesterase level in infected group and non-infected group were 2916 (2196-4123) U/L and 3966 (2840-4883) U/L, Z=-3.767, P<0.001. The serum potassium of the infected group and non-infected group were 3.98 (3.51-4.34) mmol/Land 4.27 (3.72-4.54) mmol/L, t=-2.675, P=0.007. The serum ions of infected group and non-infected group were (26.9±11.2) mmol/L and (31.7±12.3) mmol/L, t=-3.196, P=0.002. The creatinine of infected group and non-infected group were 71 (52-108) μmol/L and 57.5 (45.8-75.3) μmol/L, Z=-3.676, P<0.001. There were statistically significant differences in above indexes between the 2 groups. Multivariate logistic regression analysis showed that ascites (OR=4.90, 95%CI: 2.51~9.56), hepatic encephalopathy stage 4 (OR=8.41, 95%CI: 2.18 ~ 32.4), hepatorenal syndrome (OR=3.77, 95%CI: 1.18-12.1) and APACHEII score (OR=1.08, 95%CI: 1.02-1.15) were independent risk factors for infection in patients with liver failure (P<0.05). The consistency index (C-index) of the nomogram model predicting infection in patients with liver failure was 0.803. The calibration curve showed that the average absolute error between the predicted result and the actual result of the nomogram model was 0.038, and the receiver operator characteristic (ROC) curve shows that the area under the curve was 0.812 (95%CI: 0.760 ~ 0.864). Conclusion Ascites, hepatic encephalopathy, hepatorenal syndrome and higher APACHEII scores are independent risk factors for infection in patients with liver failure. The nomogram model based on these 4 risk factors can accurately predict the probability of infection in patients with liver failure.

Key words: Liver failure, Infection, Nomogram