Objective: Estimate differences in the rate of decline in verbal memory by levels of perceived relationship quality among community-dwelling adults.
Participants: In the English Longitudinal Study of Ageing (ELSA), n = 10,109 participants aged 50-89 years were assessed at wave 1 (baseline: 2002-03) and followed-up over 16 years to wave 9 (2017-18).
Methods: Verbal memory was assessed biennially by tests of immediate and delayed word-recall. Positive and negative aspects of perceived relationship quality (social support and strain, respectively) were measured by relationship type (spouse; children; extended family members; friends). Random effects within-between (REWB) modelling was used to separate between- and within-person effects. Associations were estimated between levels of social support/strain and (i) baseline levels of memory (main effects), and (ii) the 2-year decline in memory (interaction with time).
Results: Longitudinal associations were most prominent for men, specific to relationship type, and showed between- rather than within-person effects. Among men, higher spousal strain was associated with faster decline in memory (βbetween-effect-by-time = -0.043; 95% CI: -0.084, -0.002; p = 0.039), whilst greater support from children was associated with slower decline (βbetween-effect-by-time = 0.020; 95% CI: 0.002, 0.039; p = 0.033). Men with higher levels of strain from friends had lower baseline memory (βbetween-effect = -0.382; 95% CI: -0.627, -0.137; p = 0.002) and showed faster decline (βbetween-effect-by-time = -0.047; 95% CI: -0.095, 0.000; p = 0.051).
Conclusions: Differences between persons in levels of social support and social strain were modestly associated with the rate of memory decline, especially among men. Our findings can inform future research studies and intervention strategies designed to maximise the potential of social relations to promote healthy cognitive ageing.