Answers are grounded in this paper and cite their sources.
Longitudinal cohort study uncovers higher baseline cognition but mixed rates of cognitive change with internet use in older adults, suggesting digital engagement alone does not guarantee protection.
Rapid global population aging has driven a continuous rise in the burden of cognitive decline and dementia, imposing substantial health, caregiving, and economic pressures worldwide. In the digital era, internet use has emerged as a novel source of social participation and cognitive stimulation for older adults. However, systematic, cross-national evidence regarding its association with both baseline cognitive levels and the rate of cognitive decline remains limited, particularly concerning how these associations vary across distinct cognitive trajectories. We analyzed 172,433 person-wave observations from four longitudinal aging cohorts in China, the United States, England, and Europe. Cohort-standardized cognitive z-scores were constructed, and Group-Based Trajectory Modeling was used to summarize cognitive change within a harmonized three-class specification. Independently estimated four-class cubic trajectory models were used to assess sensitivity to the prespecified class count. Generalized Estimating Equations and rate-of-change models estimated associations of internet use with cognitive level and longitudinal change. Four-way decomposition was used as an exploratory analysis of components involving life satisfaction and depressive symptoms. Internet use was associated with higher cognitive levels in all cohorts (cohort-specific estimates, 0.15–0.29 standard deviations; random-effects summary, 0.24 standard deviations; I²=96.5%). Because heterogeneity was very high, cohort-specific estimates were treated as primary. The largest trajectory-stratified estimate was observed in the low-and-declining group in ELSA (0.80 standard deviations), whereas the corresponding CHARLS estimate was small and nonsignificant. Under the alternative four-class specification, 15 of 16 class-specific internet-use coefficients were positive and statistically significant at p < 0.05; the only negative estimate was small and nonsignificant in the lowest CHARLS class. Internet use × time estimates remained heterogeneous. Life satisfaction and depressive symptoms explained little of the observed association; approximately 89% or more was allocated to the controlled direct-effect component, which should be interpreted as association not explained by the measured mediators rather than proof of an unmediated causal pathway. Internet use was consistently associated with higher cognitive levels, but its association with cognitive change was heterogeneous across countries and trajectory groups. The findings do not support a uniformly protective interpretation: faster decline was observed among internet users in CHARLS and in the low-and-declining groups in HRS and SHARE. Because exposure definitions, cognitive measures, and the harmonized trajectory specification differ in important ways across cohorts, the results should be interpreted as observational, cohort-specific associations. Intervention studies are needed before digital inclusion can be expected to improve cognitive outcomes. Not applicable.
No takes yet. Share an insight, caveat, or question.
Guo et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: