Estimates of worldwide deaths associated with exposure to fine particles in atmospheric pollution provide some surprising results. The findings will guide future research and act as a wake-up call for policymakers. See Letter p.367
In this issue, Lelieveld et al.1 (page 367) estimate the number of worldwide deaths each year caused by seven sources of air pollution. To do this, they used advanced global atmospheric-chemistry models, detailed country-level population and health data, and integrated exposure–response (IER) functions — statistical models that describe how mortality varies with exposure to fine particulate air pollution. The atmospheric-chemistry model allowed the researchers to attribute air pollution and premature deaths in different regions to emissions associated with various sectors of the economy.
More than 3.2 million deaths per year have been attributed2 to exposure to outdoor particulate matter known as PM2.5 — particles less than 2.5 micrometres in diameter, which can penetrate deep into the lungs and cause a wide range of health problems. Many parts of the United States and Europe have seen substantial improvements in air quality over recent decades as a result of regulatory interventions, and growing evidence3,4 suggests that these improvements benefit public health. But other regions, particularly countries in Asia with vast populations, continue to have poor air quality5 (Fig. 1), with the emissions of several key pollutants expected to increase in the future6. The overlap of high pollution and large populations takes a huge toll on public health, but little is known about the pollution sources that are responsible for premature deaths.
Enter Lelieveld and colleagues. The authors' results are surprising and potentially important for protecting public health globally. First, they estimate that ambient PM2.5 from commercial and residential energy sources contributes the most to premature deaths worldwide. These sources include solid fuel such as coal and biomass used for heating and cooking, local waste disposal and diesel generators. Such sources account for 32% of the premature deaths in China and 50–70% of those in India and other Asian nations.
The IER functions7 that the authors used pool epidemiological exposure–response information for mortality associated with exposure to outdoor particles, emissions from biomass burning, and tobacco smoke (both from active smoking and second-hand exposure). For deaths attributable to stroke and cardiovascular disease, the IER curve is steeper at low exposures (implying that the mortality effects of increases in PM2.5 are greater at lower particulate levels), but generally flattens at higher exposures. Large uncertainties in the IER for PM2.5 occur in the exposure range of approximately 30–100 micrograms per cubic metre (ref. 7), because no information for cardiovascular mortality due to outdoor PM2.5 is available, and because only a few studies of second-hand smoke exposure exist. A caveat to Lelieveld and colleagues' estimates of premature deaths from commercial and residential energy sources in Asian countries is that they fall mostly in these areas of high uncertainty.
Studies of the effects of biomass burning on cardiovascular disease or stroke at any level of exposure are also lacking8. Furthermore, the largest study so far to examine how sources of fine-particle air pollution affect heart-disease mortality9 found no effects for ambient PM2.5 from biomass burning in the United States. Nevertheless, as the authors point out, even if it is assumed that biomass burning and commercial and residential energy use do not contribute to mortality associated with heart disease, such energy use remains the largest factor for global mortality associated with air pollution overall, even though the total number of deaths declines.
Lelieveld and colleagues' next major finding is that agricultural sources are the second-largest contributor to global mortality from PM2.5 — releases of ammonia from livestock and fertilizers lead to atmospheric formation of ammonium nitrate and sulfate particles. Agricultural sources are the leading source of mortality in the eastern United States, Russia, Turkey, Korea, Japan and Europe, contributing to more than 40% of the deaths in many European countries.
This finding assumes that ammonium nitrate and sulfate have the same toxicity as other constituents of the atmospheric particle mixture. Some epidemiological studies10,11 do indeed report adverse effects from these particles, but many toxicological data indicate that they have little biological potency at ambient levels10. The contradictory evidence for ammonium sulfate probably arises because these particles are often mixed with metals and other toxic components from coal or industrial sources11. It could therefore be that Lelieveld et al. overestimate the effects of particles from agricultural sources. The finding is highly valuable, however, because agriculture has generally not been seen as a major source of air pollution or premature death, and because it suggests that much more attention needs to be paid to agricultural sources, by both scientists and policymakers.
Third, the researchers find that traffic-related pollution accounts for about 20% of the deaths from PM2.5 in the United States, the United Kingdom and Germany, but only 5% globally. The spatial resolution of their global assessment (which considers sub-areas of approximately 110 × 110 km) cannot capture the effects of finer-scale variation of traffic pollution. Other studies10,12 have found that variation in pollution about 50–500 metres from the roadside correlates with mortality. Mounting evidence10 also points to heightened effects on health and mortality from the components and reaction products of traffic emissions compared with other emission sources. Thus, the effects from traffic might be underestimated by Lelieveld and colleagues. But the findings send out two crucial messages: traffic emissions remain a major source of premature death in Western countries even after extensive regulatory action, and the rapid rate of growth in traffic in many regions may well lead to increased pollution and more premature deaths in the near future.
Finally, the authors project a doubling of mortality from air pollution by 2050 on the basis of projected rates of increase in pollution and population levels. This projection should sound alarm bells for public-health agencies around the world. It also raises the question of which sources should be reduced in different regions. The answer depends on how much trust we put in the IER curve. Because the steep part of the curve is at lower levels of ambient PM2.5, large benefits can accrue from relatively small reductions in air pollution in cleaner regions, whereas the flatness of the curve at high levels necessitates large reductions in the polluted areas of Asia to achieve major health benefits13.
Lelieveld and colleagues' findings suggest that about 1 million lives could be saved every year by reducing ambient exposure to pollution. A further 3.54 million lives per year could be saved by lowering indoor exposures from similar sources2, mainly through changes in commercial and residential energy use. Incentivizing the use of cleaner fuels or of electricity for local energy needs would reduce mortality from both indoor and ambient PM2.5 exposure and should be a priority in Asia and other regions that rely on solid fuels. For many parts of the world, more research is needed if we are to understand the impacts of agricultural practices on air pollution and mortality, and especially to determine the toxicity of ammonium nitrate and sulfate emanating from this source. And in countries that already have low ambient levels of pollution, sizeable benefits can still be achieved by reducing emissions from fossil-fuel power plants and traffic.
About this article
Effects of environmental policy on public risk perceptions of haze in Tianjin City: A difference-in-differences analysis
Renewable and Sustainable Energy Reviews (2019)