Estimating the Effects of Hypothetical Ambient PM2.5 Interventions on the Risk of Dementia Using the Parametric g-Formula in the UK Biobank Cohort

Abstract

Background: Emerging evidence identifies ambient particulate matter (PM) with an aerodynamic diameter $\le 2.5\mu m$ (${\mathrm{PM}}_{2.5}$) as a modifiable risk factor for dementia, but the potential health benefits gained by enacting regulations that reduce ${\mathrm{PM}}_{2.5}$ remain unclear.

Objectives: Our aim was to estimate the total effects of hypothetical ambient ${\mathrm{PM}}_{2.5}$ interventions starting in late life on the risk of dementia in a cohort using the parametric g-formula.

Methods: We used data from 291,495 participants in the UK Biobank cohort who were free of dementia and $\ge 55$ y of age at baseline (2010). We estimated the total effects of hypothetical ambient ${\mathrm{PM}}_{2.5}$ interventions (achieving annual average standards of $12{\mu g/m}^3$, $10{\mu g/m}^3$, and $9{\mu g/m}^3$) from 2010 to 2019 on the risk of dementia by calculating the difference between the estimated 10-y risk of dementia under a specified hypothetical intervention and the risk under no intervention using the parametric g-formula.

Results: In comparison with no intervention, the estimated 10-y risk difference of dementia was $-0.54$ per 1,000 population [95% confidence interval (CI): $-1.00$, $-0.10$], $-1.36$ per 1,000 population (95% CI: $-2.44$, $-0.25$), $-1.92$ per 1,000 population (95% CI: $-3.39$, $-0.33$), with ${\mathrm{PM}}_{2.5}$ interventions achieving annual average standards of $12{\mu g/m}^3$, $10{\mu g/m}^3$, and $9{\mu g/m}^3$, respectively.

Discussion: The estimated 10-y risk of dementia decreased if the individual ambient ${\mathrm{PM}}_{2.5}$ exposure was reduced due to more stringent ${\mathrm{PM}}_{2.5}$ standards in late life in comparison with the natural course without intervention on ambient ${\mathrm{PM}}_{2.5}$ exposure. Our findings, obtained using the parametric g-formula-a causal inference method that can directly evaluate the impact of hypothetical interventions-suggest that policies reducing ambient ${\mathrm{PM}}_{2.5}$ pollution may lower the risk of dementia among UK Biobank participants who would experience more stringent ambient ${\mathrm{PM}}_{2.5}$ standards in late life. https://doi.org/10.1289/EHP14723.

Figure 1.

Estimated 10-y risk difference (95% CI) of dementia between different threshold interventions on ambient ${\mathrm{PM}}_{2.5}$ and no intervention for the main analysis and sensitivity analyses. The risk difference is the absolute difference in the 10-y risk when comparing the hypothetical ${\mathrm{PM}}_{2.5}$ intervention with no intervention on ${\mathrm{PM}}_{2.5}$. Note: In the main analysis, we used the parametric g-formula to estimate the 10-y risk of dementia adjusting for baseline covariates: sex, baseline age, educational attainment, 9-y moving average (2001–2009) of annual mean temperature ($°\mathrm{C}$), 9-y moving average (2001–2009) of population density (${\text{person}/\mathrm{m}}^2$), 9-y moving average (2001–2009) of GDHI per head index, and 9-y moving average (2001–2009) of ${\mathrm{PM}}_{2.5}$ ($\mathrm{\mu }\mathrm{g}/\mathrm{m}³$); and time-varying covariates: annual mean temperature ($°\mathrm{C}$), annual population density ($\text{person}/\mathrm{m}²$), annual GDHI per head index, and annual mean ${\mathrm{PM}}_{2.5}$ ($\mathrm{\mu }\mathrm{g}/\mathrm{m}³$). The sensitivity analyses are shown in green below the main analyses, with differing factors described in the figure. Numeric results presented in Table S4. Note: CI, confidence interval; GDHI, gross disposable household income; ${\mathrm{PM}}_{2.5}$, particulate matter with an aerodynamic diameter $\le 2.5\mathrm{\mu }\mathrm{m}$.