Analysis of consensus among drug interaction databases with regard to combinations of psychotropics

• Published in: Basic & clinical pharmacology & toxicology Vol. 126; no. 2; pp. 126 - 132
• Main Authors: Schjøtt, Jan, Schjøtt, Pernille, Assmus, Jörg
• Format: Journal Article
• Language: English
• Published: England Wiley Subscription Services, Inc 01.02.2020
• Subjects: Artificial intelligence; classification; consensus; Decision support systems; Drug interaction; Drug interactions; Nursing homes; Pharmacodynamics; Pharmacokinetics; Pharmacology; psychotropics; consensus; databases; drug interactions; classification; psychotropics
• ISSN: 1742-7835; 1742-7843
• DOI: 10.1111/bcpt.13312

Drug interaction databases are important tools in today’s clinical decision support. However, there is great variation with regard to classification and presentation of interactions among databases. The present study aimed to investigate consensus among databases with regard to combinations of psychotropics. A database integrated in Norwegian computerised clinical decision support systems and three international recommended subscription databases were compared. Combinations of psychotropics (two or more) prescribed to patients 65 years or older on a single day from three nursing homes in Bergen, Norway 16 years apart (2000 and 2016) were studied. The databases were compared in a common analysis with the following questions: interaction (no, not contraindicated or contraindicated), type (pharmacodynamic or pharmacokinetic), the total number of interactions, and the first ranked interaction among several in each patient. Consensus among the four drug interaction databases was associated with pharmacokinetic interactions involving mainly older psychotropics in the common analysis. The qualities that best characterised interactions with consensus was primarily the evidence including a description of manageability. There was a surprising lack of consensus with regard to contraindicated interactions, even when older psychotropics were involved. Lack of consensus decreased with the number of psychotropics in the combinations. This was mainly because the highest ranked interactions in the respective databases involved different drugs. We propose evidence and manageability as core factors when ranking and presenting interactions in clinical decision support.