RESUMO
PURPOSE: Detecting potential drug interactions can lead to early interventions that protect patients from serious drug-related problems. The aim of this study was to evaluate the agreement among the lists of warfarin interactions provided by five information sources. METHODS: The lists of warfarin interactions and the corresponding severity ratings and documentation levels presented by the three compendia and by the World Health Organization (WHO) Model Formulary were all compared, and each list was compared to that provided on the package insert of Marevan, a brand of warfarin. The compendia used were: Drug Interaction Facts, Drug Interactions: Analysis and Management and DRUG-REAX. A kappa coefficient was used to calculate the agreement among the sources. RESULTS: A total of 537 interactions were listed. Only 13 (2.4%) were common to the five sources. The global Fleiss' kappa coefficient was -0.0080, which indicated poor agreement. Eleven warfarin interactions appeared only in the Marevan package insert. Importantly, 243 interactions (45.3% of the total) were deemed significant in at least one compendium. Only two warfarin interactions were reported as critical by all three compendia and by WHO. The most critical interactions cited by the compendia were missing from the package insert. CONCLUSIONS: Poor agreement was found among five sources listing warfarin interactions. Potentially severe clinical consequences might occur due to these discrepant recommendations. Finally, the lack of standard terminology and clinical guidance, as well as the possible inaccuracy of severity ratings and documentation might contribute to heterogeneous procedures in clinical practice.