Intraindividual variability in cognitive function in older people in the English Longitudinal Study of Ageing: modelling risk factors using a mixed-effects beta-binomial model

Cognitive functioning in older age has a huge impact on quality of life and physical and mental health. Whilst most research in cognition in older age has focussed on mean levels, there is some evidence that individuals with cognitive functioning that varies a lot around this may have different risk factors and outcomes to those with less variable functioning. Existing approaches to investigate such intraindividual variability (IIV) typically involve deriving a summary statistic for each person from residual error around a fitted mean. However, such methods ignore sampling variability, prohibit the exploration of associations with time-varying factors, and are biased by floor and ceiling effects. To address this, we fitted a mixed-effects location scale beta-binomial model to estimate average per-trial probability and IIV in a word recall test with the English Longitudinal Study of Aging (ELSA). After adjusting for mean performance, in an analysis of 9,873 individuals observed across 7 (mean: 3.4) waves we found IIV to be greater: at older ages; with lower education; in females; with more difficulties with activities of daily living; in later cohorts; and when interviewers recorded issues which may have affected the tests. Our study identifies groups with more varying cognitive performance, which has implications for their daily functioning and care. Further work is needed to identify the impact of this for future health outcomes.