Development and validation of a risk prediction model for cefoperazone/sulbactam-related coagulation disorders in Chinese inpatients – A case-control study

Background and aim: Long-term use of cefoperazone/sulbactam (CPZ/SAM) may lead to vitamin K-dependent coagulation dysfunction which is significantly associated with increased risks of major bleeding and death. However, there is a lack of prediction models estimating the risk for this adverse drug reaction. This study aimed at developing and internally validating a risk prediction model for coagulation-related adverse reactions in Chinese inpatients treated with CPZ/SAM. Methods: A case-control study was conducted in 11,092 adult inpatients admitted to our hospital from 2020 to 2021. We selected patients reporting coagulation-related adverse reactions through the Adverse Drug Events Active Surveillance and Assessment System-2 developed by our team. The final predictors were obtained by using the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm. The prediction model was visualized as a static nomogram and a web-based dynamic nomogram. Internal validation and calibration were performed using bootstrap enhanced procedure with 1000 resampling. Results: We found 215 coagulation-related adverse reactions (5.1%) in 4212 patients treated with CPZ/SAM. A final population of 208 cases and 624 controls was considered for model development and validation. Malnutrition, renal insufficiency, parenteral nutrition, longer treatment duration and red blood cell count were identified as final predictors. There is a nonlinear relationship between treatment duration and coagulation disorder. The model shows good discrimination and calibration, with the validated area under the receiver operating characteristic curve (AUROC)being 0.739 (0.699-0.779), p for Hosmer-lemeshow more than 0.05 and Brier score being 0.152. The decision curve analysis reveals that the net benefit of the model is higher, compared with the hypothesis that coagulopathy occurs in all patients with medication or no coagulopathy occurs. Conclusions: The nomogram model quantifies the risk for coagulation-related adverse reactions in Chinese inpatients treated with CPZ/SAM, supporting individual assessment and interventions to mitigate the risk.