Personalised estimation of exposure to ambient air pollution and application in a longitudinal cohort analysis of cognitive function in London-dwelling older adults.

Abstract

Background: Accurate estimates of personal exposure to ambient air pollution are difficult to obtain and epidemiological studies generally rely on residence-based estimates, averaged spatially and temporally, derived from monitoring networks or models. Few epidemiological studies have compared the associated health effects of personal exposure and residence-based estimates.

Objective: To evaluate the association between exposure to air pollution and cognitive function using exposure estimates taking mobility and location into account.

Methods: Residence-based dispersion model estimates of ambient NO₂, PM₁₀ and PM_2.5 were assigned to 768 London-dwelling participants of the English Longitudinal Study of Ageing. The London Hybrid Exposure Model was implemented to adjust estimates per pollutant to reflect the estimated time-activity patterns of each participant based on age and residential location. Single pollutant linear mixed-effects models were fit for both exposure assessment methods to investigate the associations between assigned pollutant concentrations and cognitive function over a follow-up period of up to 15 years.

Results: Increased long-term exposures to residence-based ambient NO₂ (IQR: 11.10 µg/m³), PM₁₀ (2.35 µg/m³), and PM_2.5 (2.50 µg/m³) were associated with decreases of -0.10 [95% CI: -0.20, 0.00], -0.07 [-0.11, -0.02] and -0.14 [-0.21, -0.06], respectively, in composite memory score. Similar decreases were observed for executive function scores (-0.38 [-0.58, -0.18], -0.11 [-0.20, -0.02] and -0.14 [-0.29, 0.01], respectively). When applying personalised exposure estimates, which were substantially lower, similar decreases were observed for composite memory score per IQR, but a consistent pattern of slightly more adverse effects with executive function score was evident.

Impact statement: The present study constructed a framework through which time-activity information derived from a representative sample could be applied to estimates of ambient air pollution concentrations assigned to individuals in epidemiological cohort studies, with the intention of adjusting commonly used residence-based estimates to reflect population mobility and time spent in various microenvironments. Estimates of exposure were markedly lower when incorporating time-activity, likely because people in European populations spend a large proportion of their time indoors, where their exposure to ambient air pollution may be reduced through infiltration, which is not taken into account in residence-based ambient estimates. Further work into such methods could provide insights into the efficacy of personalising exposure estimates.