Résumé
Loneliness is a risk factor for late-life dementia. There is less consistent evidence of its association with cognitive performance. This study examined the replicability of the association between loneliness and overall and domain-specific cognitive function and informant-rated cognitive decline in cohorts from seven countries: the United States, England, India, China, South Africa, Mexico, and Chile.
Data were from the Harmonized Cognitive Assessment Protocol administered in seven population-based studies (total N > 20,000). Participants reported their loneliness, completed a battery of cognitive tests, and nominated a knowledgeable informant to rate their cognitive decline. Random-effect meta-analyses were used to summarize the associations from each cohort.
Loneliness was associated with poor overall cognitive performance and informant-rated cognitive decline controlling for sociodemographic factors (meta-analytic correlation for overall cognition = -.10 [95% CI = -.13, -.06] and informant-rated decline = .16 [95% CI = .14, .17]). Despite some heterogeneity, the associations were significant across samples from Africa, Asia, Europe, North, Central, and South America. The meta-analysis also indicated an association with specific cognitive domains: episodic memory, speed-attention, visuospatial abilities, numeric reasoning, and verbal fluency. The associations were attenuated but persisted when depressive symptoms were added as a covariate. Depression, cognitive impairment, and sociodemographic factors did not consistently moderate the associations across samples.
Loneliness is associated with poor performance across multiple domains of cognition and observer-rated cognitive decline, associations that replicated across diverse world regions and cultures.