Text

Background/Aim: Falls are a major cause of morbidity among older adults. While deprescribing fall-risk-increasing drugs (FRIDs) is recommended for secondary prevention, their high prevalence limits efficient targeting in primary prevention. Stratification by pharmacological FRID subclasses or drugs with fall-risk-increasing adverse reactions (FRIAR-drugs) may improve identification of high-risk patients, yet their added value in routine care is uncertain.

Therefore, the aim of this study is to compare pharmacological FRID classifications with a FRIAR-based framework for predicting fall-related hospitalisations and to identify drug subgroups that differentiate fall-risk among exposed older adults in a primary prevention context.

Methods: We conducted a retrospective cohort study including 120,867 adults aged ≥65 years prescribed at least one FRID or FRIAR-drug at baseline. Fall-related hospitalisations in 2019 were identified using ICD-10 codes. Multivariate logistic regression evaluated pharmacological FRID (sub)groups and FRIAR-drug classifications, adjusted for established general predictors. Model performance was assessed using internal validation and sensitivity analyses.

Results: During follow-up, 2,788 individuals (2.3%) experienced a fall-related hospitalisation. Pharmacological subgroups associated with higher risk included serotonin re-uptake inhibitors, older antiepileptics, and drugs with moderate anticholinergic properties. In contrast, the FRIAR-drug framework added little prognostic information beyond established FRID classifications. The final model demonstrated moderate discrimination (AUC 0.73) with stable predictive performance on internal validation.

Discussion/Conclusion: Pharmacological stratification of FRIDs allows clinically useful differentiation of fall-risk among exposed older adults, whereas FRIAR-based classification adds limited prognostic value. These findings support more selective prioritisation of medication review in primary-preventive fall strategies, although external validation and evaluation of decision-relevant risk stratification are required.