Retinal artery occlusion does not act as an independent marker of upcoming dementia: results from a Danish 20-year cohort study

Purpose Retinal artery occlusion (RAO) is a vision threatening disease associated with cerebral vascular dysfunction, which may reflect initial signs of cerebral pathology. Early detection of patients in risk of dementia could allow for preventative treatment. Hence, this study aimed to investigate RAO as an independent biomarker of incident dementia.Methods This study was a nationwide, 20-year longitudinal cohort study in Denmark with inclusion from 1998 to 2020 and follow up until the end of 2022. We identified 2 205 159 individuals aged 65 or older through the Danish national health registers and monitored RAO (exposure) and dementia (outcome) status. We calculated incidence rate and performed a Cox regression analysis with hazard ratio (HR) and 95% confidence interval (CI) for RAO as a marker of dementia in a crude, a semi-adjusted (age and sex), and a fully adjusted model (furthermore adjusted for marital status and systemic comorbidity.)Results We identified 8 863 individuals with RAO. Incidence rates were higher among exposed compared to unexposed individuals (12.28 and 8.18 per 1000 person-years at risk, respectively). Individuals with RAO were more likely to be male and older at inclusion, to have hypertension, dyslipidaemia, cardiovascular disease, chronic kidney disease, and diabetes (p < 0.001). RAO was not associated with all-cause dementia in the crude analysis (HR 1.07 CI [1.00-1.17]) or in the fully adjusted analysis (HR 0.98 CI [0.91-1.06].Conclusion Although individuals with RAO had a higher incidence of dementia compared to unexposed individuals, these associations were lost when confounders were taken into account.

Automated summary

This study investigates whether retinal artery occlusion (RAO) can be an independent biomarker for incident dementia. A nationwide cohort study in Denmark showed that individuals with RAO had a higher incidence of dementia. However, this association disappeared when confounding factors were taken into account.