Abstract: Objective To investigate the risk factors of drug-induced liver injury (DILI) induced by cefuroxime, and to construct the related prediction model.Methods Two hundred and seventy-one patients treated with cefuroxime sodium injection in our hospital from October 2019 to October 2020 were retrospectively selected through automatic monitoring system. The patients were divided into training group (n=203) and validation group (n=68) by cluster random grouping method. The risk factors of DILI caused by cefuroxime in training group were analyzed by univariate analysis and multivariate logistic regression analysis. The related nomogram prediction model was also established.Results Age ≥ 60 years (OR=3.497, 95%CI: 1.177~10.391), BMI ≥ 28 kg/m² (OR=3.333, 95%CI: 1.207~9.202), alcoholism (OR=3.399, 95%CI: 1.129~10.234) ), hypoalbuminemia (OR=3.272, 95%CI: 1.088~9.837) and dosage > 2.25 g/d (OR=9.045, 95%CI: 3.397~24.083) were independent risk factors of DILI caused by cefuroxime (P<0.05). A nomogram model was established to predict the DILI induced by cefuroxime based on the risk factors, and the model was also validated. The results showed that the C-index of training group and validation group were 0.774 and 0.758, respectively. The calibration curve was close to the standard curve. The area under the curve (AUC) of training group and validation group were 0.785 (95% CI: 0.0735-0.834) and 0.765 (95% CI: 0.707-0.779), indicating that the model had a good prediction ability.Conclusion There are many risk factors for DILI caused by cefuroxime. The nomogram model based on risk factors can accurately predict the incidence of DILI induced by cefuroxime.

Key words: Cefuroxime, Drug-induced liver injury, Automatic monitoring, Nomogram model, Model validation