Polypharmacy Thresholds that Best Predict Emergency Room Visits and Mortality in Older Adults: A Population-Based...

BackgroundPolypharmacy is prevalent among older adults. Identifying informative medication thresholds could enhance clinical decision-making and healthcare planning.ObjectivesTo measure the predictive capacity of medication count and polypharmacy thresholds on (1) frequent emergency visits and (2) mortality, in community-dwelling older adults in Quebec, Canada; To measure the predictive capacity of inappropriateness criteria (potentially inappropriate medications, anticholinergic burden level, and drug-drug interactions) on these same outcomes.MethodsWe conducted a population-based study using the Quebec Integrated Chronic Disease Surveillance System. Using multivariable logistic regression models, we assessed the predictive capacity [area under the curve (AUC)] of medication use (medication count, polypharmacy thresholds, number of inappropriateness criteria) on frequent emergency visits (≥3) and mortality from April 2, 2022, to March 31, 2023. Findings were interpreted from two perspectives: clinical (with models including individuals' clinical characteristics, such as chronic diseases) and public health (with models excluding clinical characteristics to mimic polypharmacy measures usually used in surveillance).ResultsMedication count remained the most informative predictor in the clinical perspective analyses, with AUC = 0.739 for frequent emergency visits and AUC = 0.804 for one-year mortality, while medication thresholds lacked discrimination. Public health perspective analyses resulted in a bell-shaped predictive pattern peaking at eight medications for both frequent emergency visits (AUC = 0.690) and one-year mortality (AUC = 0.764). Inappropriateness criteria did not outperform medication count or polypharmacy thresholds.ConclusionFrom a clinical perspective, medication count is more informative than polypharmacy thresholds. From a public health perspective, a threshold of eight medications may serve as a useful polypharmacy measure.