Screening for Potential Drug–Drug Interactions in Patients Receiving Anticoagulant Therapy: A Comparison of Three Drug Interaction Databases for Consistency in Severity Rating, Evidence Classification, and Clinical Management

Background and Objectives: For patients on anticoagulants, the risk of possible drug–drug interactions (pDDIs) is particularly higher due to complex polypharmacy. Clinical decision-making is largely guided by drug interaction databases (DIDs); however, inconsistencies in programming may compromise therapeutic safety and effectiveness. The current study is designed to assess and contrast the consistency of severity rating, evidence classification, and clinical management of pDDIs across three DIDs. Materials and Methods: The study was conducted using real patient data from the outpatient medicine and cardiology department of a public hospital in the United Arab Emirates. Prescriptions containing anticoagulants were evaluated using three databases for pDDI screening: Micromedex, Lexicomp, and Drugs.com. Consensus was assessed using Fleiss’ kappa, and correlations between variables were evaluated using Spearman’s rank correlation coefficient, using a threshold of p < 0.05 to assess statistical significance. Results: A total of 130 prescriptions were analyzed, and 3237 pDDIs involving 1143 interaction pairs were retrieved. Of these, 107 pDDI pairs were consistently identified across all three databases. Significant inter-database variability was observed in the severity classification and management recommendations of pDDIs across the three databases. Regarding evidence classification, both Micromedex and Lexicomp rated most interactions with fair evidence, while Drugs.com provided no evidence ratings. Although some correlations were observed—particularly between Lexicomp’s and Drugs.com—overall agreement across databases was slight to fair (p < 0.05). Conclusions: Marked inconsistencies across the databases were identified in the classification and categorization of pDDIs and their associated parameters. Category-wise agreement analysis provides more meaningful insights beyond overall agreement by revealing clinically relevant concordance and divergence among databases.