Urban-rural differences in health behaviors and cognition: Two Taiwanese older cohorts

Although urban-rural differences in dementia prevalence have been widely reported, little is known about whether vascular risk factors and health behaviors relate differently to cognitive function across urban and rural populations. This study investigated urban-rural differences in these factors and their associations with cognition.

We analyzed data on demographics, vascular risk factors, lifestyle factors, laboratory measures, and comprehensive neuropsychological assessments. Suboptimal cognition was defined as >1 SD below appropriate norms; predictors were examined using Firth's penalized likelihood logistic regression, followed by interaction analyses to assess region-specific associations.

Overall, 1,284 non-demented older adults were included, comprising 949 and 335 from the rural and urban cohort, respectively. Compared with the urban cohort, participants in the rural cohort were younger (70.1 ± 6.8 vs 71.3 ± 3.6 years, p < 0.001), more often male (49.6% vs 30.4%, p < 0.001), and less educated (4.0 ± 4.6 vs 14.2 ± 3.1 years, p < 0.001). In the rural cohort, higher systolic blood pressure (1.02, 95% confidence interval (CI) 1.01-1.04), and male sex (2.56, 1.49-4.44) were associated with suboptimal global cognition, whereas smoking cessation was protective (0.40, 0.20-0.79). Low physical activity (1.95, 1.35-2.80) further predicted suboptimal visuospatial function. No statistically significant associations were observed in the urban cohort after correction for multiple comparisons. Interaction analyses indicated that the associations between smoking cessation and global cognition (p = 0.013) and between low physical activity and visuospatial function (p = 0.033) differed significantly between the rural and urban cohorts, with stronger associations in the rural cohort.

Health behaviors showed stronger associations with cognitive function in rural than in urban cohort, warranting region-tailored prevention strategies. Urban-rural comparisons were based on two independent community cohorts; future studies using a single population-based cohort are needed to validate these findings.