Standardizing and Comparing Management Recommendations for Potential Drug-Drug Interactions Across Different Interaction Checkers

Background: Potential drug-drug interactions (pDDIs) are frequent in clinical care, particularly among older patients. Accurate identification and management of pDDIs are essential for patient safety. Prescribers often rely on interaction checkers (ICs) to screen for pDDIs. However, these tools may provide inconsistent recommendations, potentially leading to suboptimal clinical decisions. Objective: This study aimed to develop a standardized approach for classifying and prioritizing pDDIs based on the clinical relevance of their management recommendations and to compare how these pDDIs are categorized across ICs. Methods: A scale was developed through a structured iterative process to classify pDDIs into four management categories (high priority, intermediate priority, low priority, data unavailable), based on the management recommendations extracted from six ICs. This scale was applied to 218 real-world pDDIs identified from 1923 patients, and the agreement was primarily assessed using Gwet's AC1. Main results: Overall agreement among the ICs was moderate (Gwet's AC1 = 0.44; 95% CI 0.39-0.50), with values ranging from 0.58 (0.51, 0.65) to 0.38 (0.31, 0.44) in leave-one-out analyses. The agreement was higher in binary analyses dichotomizing the scale into high- and intermediate-priority versus low-priority pDDIs (AC1 = 0.72; 95% CI 0.65-0.79), and in the classification of high-priority versus all other pDDIs (AC1 = 0.62; 95% CI 0.54-0.69). Conclusion: This study developed and tested a structured approach to systematically compare pDDI management across ICs and prioritize clinically relevant interactions. Its application revealed a generally limited agreement between ICs, pointing to the need for harmonized approaches and further studies to support more consistent, evidence-aligned pDDI management.