Background Previous studies have shown an excess risk of Alzheimer's disease and related dementias among women. Education is thought to have a causal association with dementia onset. We aimed to investigate the role of education in influencing sex differences in cognitive ageing.
MethodsWe analysed data from two prospective cohort studies in the UK; the English Longitudinal Study of Ageing (ELSA) and the Whitehall II study, to assess sex differences in cognitive performance and cognitive decline by birth cohort (birth year 1930-38, 1939-45, or 1946-55), before and after adjustment for education, and by high and low education level. Memory was assessed using immediate recall, for which data were available from all waves of the ELSA (2002-14) and Whitehall II (1997-2015) studies. Fluency was assessed using a semantic fluency test based on an animal naming task, with data available from all waves of the Whitehall II study and waves one to five (2002-10) and wave seven (2014) of the ELSA study. Cognitive scores were standardised separately in each study based on the mean and SD of the corresponding test among participants aged 50-59 years with secondary education. Findings 15 924 participants were included from the two studies. In pooled analyses, women had better memory scores than men in all birth cohorts, irrespective of adjustment for education (eg, at age 60 years, birth cohort 1930-38, mean difference between sexes [male scores minus female scores] -0•25 SDs [95% CI -0•32 to -0•19] after adjustment for education), and in both education level groups. Memory decline was faster in men than in women (at age 60 years, birth cohort 1946-55, mean difference in 13-year change -0•15 SDs [-0•20 to -0•09]; after adjustment for education -0•14 SDs [-0•20 to -0•08]). Men had better fluency scores than women in earlier birth cohorts and in the low education group (at age 60 years, birth cohort 1930-38, mean difference 0•20 SDs [95% CI 0•05 to 0•36]); but women had better fluency scores than men in later birth cohorts and in the high education group (at age 60 years, birth cohort 1946-55, mean difference -0•17 SDs [-0•24 to -0•10]). No sex differences were observed for fluency decline.Interpretation Our findings suggest that decreasing disparities between sexes in education, due to secular increases in educational opportunities, could attenuate sex differences in dementia risk and cognitive decline in the future.