Enhancing the usability of the 2023 beers criteria: A structured framework for geriatric prescribing safety

The American Geriatrics Society (AGS) Beers Criteria is one of the most widely used tools for identifying potentially inappropriate medications (PIMs) in older adults worldwide. However, its narrative format, ambiguous definitions, and lack of machine readability hinder integration with electronic health records (EHRs), clinical decision support systems, and real-world data analysis. Therefore, this study aimed to develop a structured, programmable, and multidimensional evaluation framework to transform the 2023 AGS Beers Criteria into computable logic components, thereby supporting automated PIM identification and clinical application.

We systematically reviewed 87 PIM-related statements from Tables 2-6 of the 2023 AGS Beers Criteria. A multidisciplinary team deconstructed each criterion into structured logical components and categorized them into the following five main domains: within-prescription, outside-prescription, recommendation/strength, key issues, and interpretative information. Twenty-four subcategories were identified, including dose, route, laboratory data, co-medications, grouping issues, and ambiguous expressions.

Only 8.0% of the 2023 AGS Beers Criteria could be evaluated using drug names alone, while 92.0% required multi-factorial analysis. We identified 404 computable elements: 30.2% could be assessed using prescription-only data, 25.2% required external data such as diagnoses or laboratory data, and 44.6% involved interpretive or structural complexity.

This novel framework addresses key barriers to the programmatic implementation of the 2023 AGS Beers Criteria. It enables artificial intelligence-supported PIM identification, facilitates EHR integration, and establishes a scalable infrastructure for geriatric pharmacovigilance, ultimately improving medication safety for older adults.