Abstract
OBJECTIVE: The prevalence of meropenem-related thrombocytopenia has risen in tandem with the growing utilization of meropenem in clinical settings. Consequently, we aimed to develop a risk prediction model for meropenem-related thrombocytopenia in patients with pulmonary infection and to enhance the safety for the clinical administration of meropenem. METHODS: A retrospective case-control study was conducted to collect data. The training group consisted of patients who were treated with meropenem for pulmonary infection at a tertiary A hospital in Shiyan from January 2018 to December 2021. The external validation group was formed from patients at another tertiary A hospital from January 2019 to January 2020. Multivariable logistic regression analysis was employed to investigate the risk factors linked to meropenem-related thrombocytopenia. Subsequently, these factors were utilized to develop a nomogram for predicting meropenem-related thrombocytopenia. The Bootstrap method was conducted to internally validate the nomogram model. The discrimination, calibration, and clinical effectiveness of the model were assessed through the receiver operating characteristic (ROC) curve, calibration curve, and decision curve analysis (DCA). RESULTS: A total of 625 patients were included in the training group. Among them, 73 patients experienced meropenem-related thrombocytopenia. Multivariate logistic regression analysis revealed that hypertension, baseline platelet count, and combined with cephalosporins or penicillins were independent risk factors for meropenem-related thrombocytopenia in patients with pulmonary infection. A risk prediction model was developed based on these four factors. The model demonstrated an AUC (area under the ROC curve) of 0.774 (95%CI: 0.718 ~ 0.829), with a sensitivity of 0.685 and a specificity of 0.737. The optimal critical value was determined to be 0.137. Internal validation yielded an AUC of 0.761, while external validation resulted in an AUC of 0.750 (95%CI: 0.702 ~ 0.799). The calibration diagram indicated a high level of agreement between the predicting and actual probabilities. Furthermore, the DCA demonstrated that the model had significant clinical benefit and practical value. CONCLUSION: A risk prediction model based on hypertension, baseline platelet count, combined with cephalosporins or penicillins could effectively predict the occurrence of meropenem-related thrombocytopenia in patients with pulmonary infection.