Abstract
OBJECTIVE: This study aimed to examine the associations between working after retirement and three-year changes in cognitive function, focusing on differences by work type, work hours, and education level in Taiwan. METHODS: Data were drawn from the Taiwan Health and Retirement Study (THRS), a nationally representative survey of labor and civil service pension recipients aged 50-74 years. A total of 2,176 retirees with valid cognitive assessments across two waves (2015-2016 and 2018-2019) were included in the analysis. Cognitive function was measured by the Saint Louis University Mental Status (SLUMS) examination. Working after retirement was categorized along two axes: type (paid, unpaid, self-employment, vs. fully retired) and hours (full-time, part-time, vs. fully retired). Multiple linear regression models were used to examine associations with three-year cognitive change, with stratified analyses by education level. RESULTS: In the overall sample, neither classification of working after retirement was significantly associated with cognitive change. However, among retirees with lower educational attainment, those engaged in full-time work experienced a slower decline in cognition compared with those who were fully retired (β = 1.04, SE = 0.50, p = .036). No such association was observed among those with higher education. CONCLUSION: These findings highlight heterogeneity in the associations between working after retirement and cognitive aging, with potential benefits of full-time work concentrated among lower-educated retirees. However, the lack of consistent effects across sensitivity analyses indicates the need for cautious interpretation. Future research should employ longer follow-ups and causal inference methods to better clarify the mechanisms. Policies should recognize that maintaining cognitive health requires not only employment opportunities but also broader support for diverse forms of engagement, such as social participation, across the lifespan.