Skip to main content
. 2021 Jan 28;6(2):e106–e115. doi: 10.1016/S2468-2667(20)30258-9

Table 4.

Sex differences in 13-year cognitive decline, stratified by birth cohort and level of education

Base model
Memory
Low education
Birth cohort
1930–38 −0·04 (−0·11 to 0·04)
1939–45 −0·12 (−0·21 to −0·04)
1946–55 −0·17 (−0·24 to −0·09)
p value for sex difference by birth cohort 0·064
High education
Birth cohort
1930–38 0·02 (−0·11 to 0·14)
1939–45 −0·07 (−0·17 to 0·02)
1946–55 −0·08 (−0·16 to −0·00)
p value for sex difference by birth cohort 0·41
Fluency
Low education
Birth cohort
1930–38 −0·10 (−0·25 to 0·06)
1939–45 0·00 (−0·08 to 0·09)
1946–55 0·08 (−0·16 to 0·32)
p value for sex difference by birth cohort 0·29
High education
Birth cohort
1930–38 0·03 (−0·18 to 0·24)
1939–45 −0·13 (−0·24 to −0·03)
1946–55 0·16 (−0·11 to 0·44)
p value for sex difference by birth cohort 0·11

Data are mean difference between sexes in SDs (95% CI), or p values. Positive SD values indicate slower cognitive decline in men; negative values indicate slower cognitive decline in women. Low education was defined as qualifications below A-level; high education was A-level qualifications or higher. Results are shown for the reference category: participants aged 60 years. Base models included sex, sex by age, age2, age3, birth cohort, sex by birth cohort, birth cohort by age, sex by birth cohort by age, ethnicity, and practice effect. Memory models additionally included birth cohort by age3 and lower-order interactions. Fluency models additionally included practice effect by sex and birth cohort by sex by age2 and lower-order interactions.