Ambient air pollution and incident dementia: exploration of relevant exposure windows

Background: As dementia has a decades-long preclinical phase, earlier air pollution exposures may be more etiologically relevant to dementia risk than more recent exposures. Methods: We estimated exposures at Atherosclerosis Risk in Communities (ARIC) study participant addresses to criteria air pollutants, PM components, and trace metals in several exposure windows (1990–1994, 1995–1999, 2000–2004, 2005–2009, and 1990–2009) using a chemical transport model with observation data fusing at two resolutions, 4 km (CTM-4k) and finest available resolution (CTM-FR), and to PM_2.5, ozone, and NO₂ using universal kriging with land-use regression and partial least squares regression (UK-LUR-PLSR). We estimated the association between each exposure/exposure window and incident dementia from ARIC Visit 5 (2011–2013) to Visit 7 (2018–2019). Results: During follow-up (mean: 6.1 years) of 5,621 participants (mean baseline age: 76 years), 828 (14.7 %) developed dementia. Analyses did not support an association between most air pollutant exposures in any exposure window and incident dementia. However, we observed stronger associations in later time periods for PM_2.5, with significant associations in the latest time period with one exposure estimation approach: (HR (95 % CI) per 1 ug/m³ higher 2005–2009 p.m._2.5 exposure for CTM-FR: 1.11 (0.99, 1.25); CTM-4k: 1.07 (0.89, 1.22), UK-LUR-PLSR: 1.13 (1.03, 1.24). We saw similar patterns for NO₂, elemental carbon, Ni, and V. Discussion: We found little evidence supporting the hypothesized greater etiologic relevance of earlier exposures on incident dementia. Spatial confounding or acceleration of pathologic processes may explain the stronger associations observed with later exposure windows.