Serum Selenium Concentration Is Not Associated with Incident Cognitive Impairment in a Cohort of General US Adults: Findings From the REGARDS Study

OBJECTIVES: To examine serum selenium concentrations in relation to the incidence of cognitive impairment in apparently healthy US adults. METHODS: A random sub-cohort (N = 2,666) was drawn from the Reasons for Geographic and Racial Differences in Stroke (REGARDS) study consisting of Blacks and Whites, aged 45 or older across the United States. After excluding those with baseline cognitive impairment and stroke, and missing information on serum selenium, a total of 2,136 participants (mean age at baseline = 63 years) remained in the analyses. Serum selenium was assessed at baseline and was categorized into quartiles. The Six-item screener (SIS) was used to characterize global cognitive function, and the Enhanced Cognitive Battery (ECB), consisting of 4 assessments: Word List Learning, Delayed Recall, Animal Fluency, and letter F Fluency, was used to examine participants’ domains of cognitive function. Multivariable-adjusted odds ratio (ORs) and the corresponding 95% confidence intervals (CIs) were estimated using logistic regression models. Because 474 participants did not have the ECB score, we performed multiple imputation to account for the missingness. Restricted cubic spline analysis was performed to assess the potential non-linear associations. RESULTS: After adjusting for potential confounders, serum selenium was not associated with cognitive impairment. The multivariable-adjusted ORs (95% CI) from quintile 2 (Q2) to quintile 4 (Q4), with quintile 1 (Q1) as the referent were 1.07 (0.67, 1.71), 1.10 (0.68, 1.77), and 0.79 (0.49, 1.29); P(linear )= 0.44 for SIS and 0.91 (0.38, 2.15), 0.53 (0.19, 1.47), and 1.16 (0.50, 2.68); P(linear )= 0.67 for ECB. In the sensitivity analysis, which was conducted using an imputed dataset, a similar pattern of ORs was observed for ECB (Q2 vs. Q1 to Q4 vs. Q1): 0.98 (0.44, 2.18), 0.75 (0.27, 2.10), and 1.35 (0.58, 3.15); P(linear )= 0.49]. Additionally, results from the spline analysis indicated that selenium was not associated with either outcome in a non-linear fashion. CONCLUSIONS: Findings from this study do not support serum selenium as a predictor of incident cognitive impairment among US adults. FUNDING SOURCES: National Institutes of Health.