Fig. 1

Data source and study design. Using an integrated healthcare database in Shenzhen, China, we attempted to conduct a population-based retrospective cohort study. We selected 176 150 community-dwelling older adults aged ≥65 years who attended a health check-up and screened negative for cognitive impairment between 2018 and 2022 from the geriatric health management records. A 2-year “look-back” period from 2016 to 2017 was set to extract disease history. We identified osteoporosis, osteoporotic fractures, and dementia cases through linking to outpatient and inpatient medical records, and death registration records. Information on covariates was obtained from various sources, including resident health records and electronic medical records. The first health check-up with cognitive function tested during recruitment was set as the baseline date. Participants were then split into two groups according to whether they had a prior osteoporosis diagnosis: the osteoporosis group (i.e., exposure group) and the control group. Follow-up for incident dementia began at the baseline date for both groups and continued until the earliest dementia diagnosis, death, loss to follow-up, or end of follow-up (June 30th, 2023). ^*Incident osteoporosis during the follow-up period was treated as a time-varying exposure in a sensitivity analysis