Differences in Anticholinergic Exposure Assessment Across Burden Scales and Exposure Measures in a Population-Based Study: Implications for Targeted Deprescribing in Older Adults

Anticholinergic burden scales and exposure measures can support medication review in older adults; however, different instruments may classify different groups of individuals as exposed. Population-level comparisons of burden scales and exposure measures that are feasible for administrative dispensing data remain limited. To compare six anticholinergic burden scales and exposure measures feasible for application to administrative dispensing data with respect to exposure classification, agreement, medication profiles and subgroup exposure patterns in older adults. This retrospective cross-sectional study was conducted in Western Australian residents aged ≥ 65 years during 2017–2018 using linked population-level administrative data. Anticholinergic exposure was calculated using three dose-based exposure measures, with standardised dose metrics and narrower medication lists and three broad-inclusion burden scales, with wider medication lists and graded potency ratings. Outcomes were classified as exposed (score > 0 within each scale or measure) and high exposure (scale- or measure-specific thresholds). Agreement was assessed using Cohen’s kappa, medication profiles were described and adjusted subgroup patterns were compared using logistic regression. Amongst 330,684 older adults, the proportion classified as exposed ranged from 13.1 to 53.6%, whilst the proportion classified as having high exposure ranged from 0.7 to 6.9%. Agreement in identifying exposed individuals was highest amongst dose-based exposure measures and lower between dose-based exposure measures and broad-inclusion burden scales. High-exposure overlap was limited, particularly for the modified Drug Burden Index. Dose-based exposure measures captured narrower medication profiles, dominated by tricyclic antidepressants, antipsychotics and urological antimuscarinics. Broad-inclusion burden scales captured a wider range of therapeutic classes, including medications with mild and moderate anticholinergic effects. Higher comorbidity was associated with greater exposure across all scales and measures, with the largest differences in the proportion classified as exposed observed for broad-inclusion burden scales. Scale or measure selection substantially influences who is classified as exposed, medication profiles and exposure patterns across subgroups. Anticholinergic burden scales and exposure measures are therefore not interchangeable and selection should reflect whether the intended purpose is population surveillance, comprehensive medication review or targeted deprescribing.