Fig. 2

Flowchart of study population selection. In the full cohort analysis, we first selected older adults aged ≥65 years who attended a cognitive screening between 2018 and 2022 (n = 190 880). We excluded individuals with dementia diagnosis or dementia-related medication dispensation (donepezil, rivastigmine, galantamine, memantine) before baseline (n = 724), individuals with suspected cognitive impairment at baseline (n = 4 084), who left the city without any follow-up records (n = 3 565), without any inpatient or outpatient visit during study period (n = 4 967), with data missing on covariates (n = 998), died or diagnosed with dementia within 6 months following baseline date (n = 392). Eventually, 176 150 eligible participants were included in the final analyses. In the matched cohort analysis, 9 530 older adults with osteoporosis were successfully matched with 43 913 controls by propensity score matching