Identification, reliability, and validity of drug–drug interaction checkers in chronic diseases: A systematic review

Drug-drug interactions (DDIs) pose major risks in chronic disease management where polypharmacy is common. Although digital DDI tools are used widely to aid healthcare professionals, tools rely on theoretical interactions. This limitation questions their reliability and validity for clinical relevancy. Hence, this systematic review identified available DDI tools and evaluated their reliability and validity in chronic disease management. Following PRISMA guidelines, MEDLINE, Embase, Scopus, Web of Science, Google Scholar, ACM, and IEEE Xplore were searched for studies published between January 2015 and April 2025. Methodological quality, reliability, and validity were assessed using tailored versions of QUADAS-2 and QUADAS-C through agreement and performance metrics. Forty-four studies met inclusion criteria. Micromedex, Drugs.com®, Lexicomp™, Medscape®, Epocrates®, and UpToDate® were most frequently assessed (median: three tools per study). Inter-rater reliability was generally low, with limited agreement between DDI tools (Fleiss' κ -0.155 to 0.483; Cohen's κ -0.065 to 0.726; Gwet's AC1 0.12 to 0.46). Negative inter-rater reliability statistics indicate disagreement not due to chance, suggesting oppositional classification for some DDIs. Two studies used Kendall's W and Kruskal-Wallis tests reported significant differences (P < 0.05). Seventeen studies examined consensus, seven using clinicians, drug labels, or textbooks as reference standards. High risks of bias were noted in medication selection, reference quality, and index comparability. Overall, DDI tools showed limited reliability and limited evidence of validity against consistent clinical outcome-anchor standards, with poor consensus even when used collectively. Standardised evaluation frameworks and greater algorithm transparency are needed to enhance accuracy and harmonisation across tools.