Mortality Attributable to Ambient Air Pollution: A Review of Global Estimates

Abstract

Since the publication of the first epidemiological study to establish the connection between long-term exposure to atmospheric pollution and effects on human health, major efforts have been dedicated to estimate the attributable mortality burden, especially in the context of the Global Burden of Disease (GBD). In this work, we review the estimates of excess mortality attributable to outdoor air pollution at the global scale, by comparing studies available in the literature. We find large differences between the estimates, which are related to the exposure response functions as well as the number of health outcomes included in the calculations, aspects where further improvements are necessary. Furthermore, we show that despite the considerable advancements in our understanding of health impacts of air pollution and the consequent improvement in the accuracy of the global estimates, their precision has not increased in the last decades. We offer recommendations for future measurements and research directions, which will help to improve our understanding and quantification of air pollution-health relationships.

Keywords: global estimates; health impact; outdoor air pollution.

Figure 1

Graphical representation of estimates of global excess mortality attributable to atmospheric fine particulate (PM_2.5) air pollution. The symbols show the average estimates, while error bars represent the 95% confidence levels. The color code denotes the ERF used in the study (yellow: LI, green: IER, orange: GEMM, blue: MRBRT, gray: LI with β from Vodonos et al. [2018], red: FUSION). The dashed lines represent calculations for all‐cause mortality. The circles are estimates from the GBD. The x‐axis indicates the publication year. Data are from Table 1.

Figure 2

Colors and symbols as Figure 1. The x‐axis indicates the years for which the estimates were derived.

Figure 3

Graphical representation of estimates of global excess mortality attributable to ozone (O₃) pollution. The symbols are the average estimates, while error bars represent the 95% confidence level. The color code denotes the cause of death used in the study (yellow: all respiratory diseases, green: COPD). All studies used a LI ERF, with the exception of the dashed ones, which adopted an LL ERF. The circles are estimates from the GBD. The x‐axis indicates the publication year. Data are from Table 2.

Figure 4

Normalized Standard Error (NSE) in (%) for PM_2.5 mortality estimated by the studies included in this review. The x‐axis shows the publication year. Colors and symbols are the same as in Figure 1.

Figure 5

Two aerosol distributions with the same PM_2.5 mass. The blue curve represents the number density of an aerosol distribution, the green curve represents the mass density of the distribution in the same panel. The PM_2.5 is estimated as the integral of the green curve between zero and the 2.5 μm, the last limit depicted with the red line. The median for the number and mass are also presented. Please note the different vertical axis for the green and blue curves. The aerosol density is identical in both distributions. The total mass of PM_2.5 is 32 μgm⁻³, equivalent to the global population weighted annual average exposure for the year 2016 (Brauer et al., ; van Donkelaar et al., 2021).