Loneliness and Cognitive Function in Older Adults: Longitudinal Analysis in 15 Countries

This study aims to evaluate the directionality of the association between loneliness and cognitive performance in older adults, accounting for confounding factors. Data were from 55,662 adults aged >= 50 years who participated in Waves 5-8 of the Survey of Health, Ageing and Retirement in Europe (SHARE). Loneliness was assessed with the Three-Item Loneliness Scale (TILS) and with a one-item direct question. Cognitive performance was assessed with four measures: verbal fluency, numeracy, immediate recall, and delayed recall. Age, sex, geographical area, educational attainment, partnership status, depressive symptoms, and previous chronic diseases at baseline were used as covariates. We analyzed the associations with three-wave random intercept cross-lagged panel models (RI-CLPM) and conducted age-stratified analysis among those younger versus older than 65 years. Full information maximum likelihood estimators were used to handle missing values in Waves 6-8 in the main analyses. We also conducted additional sensitivity analyses stratified by retirement status (retired vs. not) at baseline. At the within-person level, loneliness and cognitive performance were not associated with each other among those aged 50-64 years in the main time-lagged analysis. Among those aged >= 65 years, loneliness was associated with lower cognitive performance in the next wave in all four cognitive domains. In addition, lower verbal fluency predicted greater loneliness in the next waves among this age group. Similar patterns were found independently of retirement status at baseline. These results suggest that loneliness is a psychosocial risk factor for cognitive decline among older adults (>= 65 years).