Document resource
Background and Importance Direct oral anticoagulants (DOACs) use is widespread and expected to grow with generic entry and lower prices, making knowledge of drug–drug interactions essential. In clinical practice, only those deemed clinically relevant are of concern, as they directly affect efficacy and safety. Several drug–interaction databases are available, but their consistency in classifying clinically relevant interactions has not been systematically assessed.Aim and Objectives To identify the clinically relevant interactions with DOACs reported by four interaction databases and to evaluate the level of agreement among them.Material and Methods A cross-sectional observational study was performed using Lexicomp, Medinteract, Micromedex, and Drugs.com. All reported interactions for apixaban, dabigatran, edoxaban, and rivaroxaban were retrieved. Severity ratings were dichotomised as clinically relevant (X/D, major, contraindicated/major) or not clinically relevant (C, moderate; B, minor; A, undetermined/unknown). Interactions classified as clinically relevant in at least one database were analysed. Overall agreement was defined as the average pairwise percent agreement among the four databases; full consensus required unanimous classification as clinically relevant. Agreement was assessed using Gwet’s AC, with Fleiss’ kappa as sensitivity analysis. Interpretation followed Altman’s thresholds: ≤0.20 poor, 0.21–0.40 fair, 0.41–0.60 moderate, 0.61–0.80 good, ≥0.81 excellent. Analyses were conducted with Stata 17.Results A total of 1,088 interactions were considered clinically relevant by at least one database: dabigatran 314 (28.9%), edoxaban 271 (24.9%), rivaroxaban 258 (23.7%), and apixaban 245 (22.5%). Overall agreement was low (percent agreement 49.1%; 95% CI 48%–50%). Full consensus across the four databases was achieved in only 69 interactions (6.3%; 95% CI 4.9%–7.9%). Gwet’s AC was 0.0218 (95% CI 0.006–0.037; p=0.006). Fleiss’ kappa was negative, indicating disagreement beyond chance. According to Altman’s thresholds, concordance was poor.Conclusion and Relevance Many DOAC interactions are potentially clinically relevant. Major databases show poor agreement, with full concordance in only 6.3% of cases. This inconsistency challenges pharmacists, who should consult more than one source to optimise safety and avoid both undertreatment and unwarranted restrictions.Conflict of Interest No conflict of interest