Education Differences in the Adverse Impact of PM2.5 on Incident Cognitive Impairment Among U.S. Older Adults

Abstract

Background: Air pollution is linked to worse cognitive function in older adults, but whether differences in this relationship exist by education, a key risk factor for cognitive decline, remains unknown.

Objective: To determine if the association between fine particulate matter air pollution (PM2.5) and incident cognitive impairment varies by level of education in two cohorts assessed a decade apart.

Methods: We used data on adults ages 60 and older from the nationally representative Health and Retirement Study (HRS) linked with tract-level annual average PM2.5. We used mixed-effects logistic regression models to examine education differences in the association between PM2.5 and incident cognitive impairment in two cohorts: 2004 (n = 9,970) and 2014 (n = 9,185). Cognitive impairment was determined with tests of memory and processing speed for self-respondents and proxy and interviewer assessments of cognitive functioning in non-self-respondents.

Results: PM2.5 was unrelated to incident cognitive impairment among those with 13 or more years of education, but the probability of impairment increased with greater concentrations of PM2.5 among those with 8 or fewer years of education. The interaction between education and PM2.5 was only found in 2004, possibly because PM2.5 concentrations were much lower in 2014.

Conclusion: Education is a key determinant of cognitive decline and impairment, and in higher pollution contexts may serve as a protective factor against the harms of air pollution on the aging brain. Additionally, because air pollution is ubiquitous, and particularly harmful to vulnerable populations, even small improvements in air quality may have large impacts on population health.

Keywords: Aging; air pollution; cognition; dementia; education; modifiable risk factors.

Fig. 1.

Distribution of PM_2.5 among older U.S. adults in the HRS in 2004 and 2014; μ = mean and s.d. = standard deviation.

Fig. 2.

Predicted probability of incident cognitive impairment by levels of PM_2.5 and education among U.S. older adults in HRS in 2004. Probabilities are predicted from a fully adjusted mixed-effects logistic regression (Table 2, Model 4) with all covariates held constant at their mean.