LAUSR

Background: Long-term protective associations proposed between previous complex occupational tasks and cognitive functioning in later life point to work roles contributing to cognitive reserve. Objective: To examine occupational complexity involving data, people, and things in relation to the level of, and rate of change in, cognitive functioning. Methods: Participants were 1,290 members of the Australian Longitudinal Study of Ageing and initially aged 65-102 years (mean = 79). Information about main lifetime occupation was collected retrospectively. Cognition was assessed 4 times over a 13-year interval. Results: In multilevel models adjusted for demographics, medical conditions, and depressive symptoms, higher complexity involving data was associated with faster speed (β = 0.73, p < 0.001), better memory (β = 0.32, p < 0.05), and mental status (β = 0.40, p < 0.001) at baseline. These associations remained statistically reliable after adjusting for complexity with people and things, sedentary and heavy physical work, retirement age, and leisure activity. Complexity with things was associated with slower speed (β = -0.50, p < 0.001) and poorer mental status (β = -0.26, p < 0.01) and was not explained by other variables. There were no associations of occupational complexity with rates of cognitive decline over time. Conclusion: Older individuals retired from occupations characterized by higher complexity with data maintain their cognitive advantage over those with lower complexity into older adulthood, although without additional moderation of this advantage in terms of less postretirement cognitive decline. Complexity of work with things confers a negative relation to cognition whilst also not affecting postretirement cognitive change. Although the relative contributions of occupation or other early life influences for cognition remain to be established, it nevertheless may be beneficial to promote workplace design strategies and interventions that incorporate complex activities, particularly tasks involving data.