Supplementary Material for: Rural Versus Urban Residence in Adulthood and Incident Cognitive Impairment

Introduction: Rural versus urban living is a social determinant of cognitive health. We estimated the association of rural vs urban residence in the United States (US) with incident cognitive impairment (ICI) and assessed effect heterogeneity by sociodemographic, behavioral, and clinical factors. Methods: REGARDS is a population-based prospective observational cohort of 30,239 adults, 57% female, 36% Black, aged 45+, sampled from 48 contiguous states in the US in 2003-2007. We analyzed 20,878 participants who at baseline were cognitively intact with no history of stroke and had ICI assessed on average 9.4 years later. We classified participants’ home addresses at baseline as urban (population ≥50,000), large rural (10,000-49,999), or small rural (≤9,999) by Rural-Urban Commuting Area codes. We defined ICI as ≥1.5 SD below the mean on at least 2 of the following tests: word list learning, word list delayed recall, and animal naming. Results: Participants’ home addresses were 79.8% urban, 11.7% large rural, and 8.5% small rural. ICI occurred in 1,658 participants (7.9%). ICI occurred in 1,658 participants (7.9%). Small rural residents had higher odds of ICI than urban residents, adjusted for age, sex, race, region, and education (OR = 1.34 [95% CI: 1.10, 1.64]), and after further adjustment for income, health behaviors, and clinical characteristics (OR = 1.24 [95% CI: 1.02, 1.53]). Former smoking versus never, non-drinking versus light alcohol drinking, no exercise versus >4 times/wk, CES-D depressive symptom score of 2 versus 0, and fair versus excellent self-rated health had stronger associations with ICI in small rural areas than in urban areas. For example, in urban areas, lack of exercise was not associated with ICI (OR = 0.90 [95% CI: 0.77, 1.06]); however, lack of exercise combined with small rural residence was associated with 1.45 times the odds of ICI compared with >4 bouts of exercise/wk in urban areas (95% CI: 1.03, 2.03). Overall, large rural residence was not associated with ICI; however, black race, hypertension, and depressive symptoms had somewhat weaker associations with ICI, and heavy alcohol drinking a stronger association with ICI, in large rural areas than in urban areas. Conclusion: Small rural residence was associated with ICI among US adults. Further research to better understand why rural residents are at higher risk for developing ICI and mechanisms to ameliorate that risk will support efforts to advance rural public health.