Evaluation of drug-drug interaction checkers along clinically relevant adverse drug events in oncology and hematology pediatric patients

Background and Objective: Pediatric patients are vulnerable to drug-drug interactions (DDIs), studies on Pediatric patient are generally limited particularly in Hemato-Oncology patients, the aim of this study is to evaluate the performance of DDIs checkers, along with most frequent DDIs and its prevalence. Materials and Methods: A retrospective observational study was conducted at the Yeditepe Hospital's Oncology Center from 2019 to 2023. Two different tools, Drugs.com and Lexicomp were utilized, and the agreement between these tools was assessed using the Kappa value. Tools' sensitivity, specificity, Positive Predictive Value (PPV), and Negative Predictive Value (NPV) was determined to evaluate the accuracy of these tools compared to Stockley's interaction book. The Spearman correlation test to compare the length of hospitalization with exposure to DDIs. Statistical significance was defined as a p-value of < 0.05. Logistic regression was performed to identify variables associated with the patients' final status and length of hospitalization (LOS). The consistency of ChatGPT was assessed using the Intraclass Correlation Coefficient (ICC). the study design was approved by an ethics committee. Statistical analysis was carried out using SPSS 25. Results: A total of 155 patients were initially screened, with 80 met the inclusion criteria. Drugs.com and Lexicomp identified 752 and 473 unique drug-drug interactions (DDIs), respectively, resulting in an average DDI prevalence of 78%. Spearman was 0.79 and 0.85 for Lexicomp and Drugs.com, respectively. The Mann-Whitney showed a significant difference in DDI exposure between surviving and deceased patients. The performance of Lexicomp and Drugs.com revealed sensitivity values 0.5 - 0.77, specificity values 0.5 - 0.36, PPV values 0.31 - 0.36, and NPV values 0.31 - 0.77. The accuracy for both tools was 0.5, and Kappa value was 0.575. ICC for ChatGPT's consistency, showed a poor result with a Cronbach's Alpha of 0.630. Conclusion: The study revealed that pediatric patients with hematologic oncology diseases are at a high risk of drug-drug interactions (DDIs). The accuracy of the tools used in this study was found to be moderate, and there was poor agreement between these tools. Keywords: Drug interaction checker, Adverse effect, DDIs, Online software, Oncology