Safety analysis of therapeutic drugs for breast cancer patients and construction of a predictive model for serious adverse drug reactions

Objectives To explore the characteristics of the occurrence of antineoplastic drug adverse reactions (ADRs) in breast cancer and to utilize a computerized tool to identify early warning indicators of potentially serious ADRs. Methods We conducted descriptive statistical analyses of the demographic features, medication use characteristics, and clinical manifestations of suspected ADRs in ADR-exposed patients using data from the Shaanxi Provincial Adverse Drug Reaction Monitoring Center, China, from 2017 to 2021. Using disproportionality methods (reporting odds ratio, proportional reporting ratio, and comprehensive standard method), the relationship between drugs and ADRs was measured. Finally, a web-based clinical prediction model for serious ADRs based on binary logistic regression was developed to estimate individual event probabilities numerically. Results We developed a new computer-mineable breast cancer-ADR system. In total, 1119 ADR reports were received between 2017 and 2021, with an increasing trend in the number. Antineoplastic medications of natural sources made up the greatest portion of the drug category (530, 38.10%) while targeted drugs’ part increased with time. The medicine with the greatest number of ADR cases was docetaxel. Bone marrow failure was the most reported ADR. The disproportionality methods produced 19 signals of disproportionate reporting, two signals of disproportionate reporting were unknown ADRs. The occurrence of serious ADRs was shown to be substantially correlated with gender, platinum drugs, and blood and lymphatic system disorders. The clinical prediction model for serious ADRs had above-moderate discriminatory power ( C -index was 0.775). Conclusions The number of ADRs to breast cancer antineoplastic drugs was constantly increasing, with docetaxel ranking first, with the majority of ADRs presenting as bone marrow suppression, nausea, and vomiting. Data mining identified 19 signals of disproportionate reporting.