Risk of new-onset dementia during the 2–4 years following COVID-19: A population-based cohort study of adults aged 50 and older

The long-term association between COVID-19 and new-onset dementia (NOD) remains uncertain, with most previous studies limited to the first two years after infection. Using data from the English Longitudinal Study of Ageing (ELSA), we examined whether early-pandemic COVID-19 infection in 2020 was associated with newly reported all-cause dementia approximately 2–4 years later among adults aged 50 years or older. Participants with early infection were compared with those with no recorded infection through 2023–2024. Associations with NOD were estimated using mean bias-reduced logistic regression. Crude NOD risks were similar between groups (1.1% vs 1.0%). After adjustment, the odds ratio (OR) was 2.74 (95% confidence interval [CI] 0.81–9.23); however, the confidence interval included the null and was compatible with both no association and a potentially substantial increase in risk. Findings were similar after restriction to non-hospitalised infection and in a sensitivity analysis using inverse-probability-of-censoring weighting. Overall, early-pandemic COVID-19 was not clearly associated with subsequent NOD. However, the small number of dementia events resulted in substantial uncertainty, meaning that an increased longer-term NOD risk could not be excluded. Larger studies with continued follow-up are needed, particularly to determine whether any excess risk is concentrated among older adults with pre-existing cognitive vulnerability.