A Narrative Review of Medication Optimisation in Older Adults with Polypharmacy: From Structured Medication Review to AI-Enabled Medication Safety

Context: Polypharmacy, commonly defined as the concurrent use of five or more medicines, and excessive polypharmacy, often defined as the use of ten or more medicines, are highly prevalent among older adults with multimorbidity. Although appropriate polypharmacy may be necessary, polypharmacy can be associated with adverse drug reactions, drug–drug interactions, falls, hospitalisation, cognitive impairment, non-adherence, treatment burden and (preventable) harm during transitions of care. Methods: This narrative review examines medication optimisation in older adults, tracing its evolution from structured medication review, pharmacist-led interventions, explicit prescribing and deprescribing criteria to digital decision support, artificial intelligence (AI)-enabled pharmacovigilance and personalised medication-risk management. European implementation perspectives, with Belgium and Poland as illustrative examples, are used to examine how professional roles, reimbursement mechanisms, clinical tools and health-data infrastructure shape medication safety. Results: This review integrates geriatric medication optimisation, European implementation experience and AI-enabled pharmacovigilance within a framework linking risk detection, clinical interpretation and treatment optimisation. Current evidence is more consistent for improvements in prescribing quality and medication-related problems than for hard clinical outcomes such as mortality, falls or hospitalisation. AI-supported approaches may improve risk detection and prioritisation, but their clinical value depends on data quality, interoperability, explainability, external validation, human oversight and prospective evidence of patient benefit. Conclusions: Future geriatric medication safety should combine comprehensive geriatric assessment, explicit clinical criteria, patient-centred deprescribing, pharmacist-led multidisciplinary review, interoperable health data and accountable AI-enabled decision support.