Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

The contribution of outdoor air pollution sources to premature mortality on a global scale


Assessment of the global burden of disease is based on epidemiological cohort studies that connect premature mortality to a wide range of causes1,2,3,4,5, including the long-term health impacts of ozone and fine particulate matter with a diameter smaller than 2.5 micrometres (PM2.5)3,4,5,6,7,8,9. It has proved difficult to quantify premature mortality related to air pollution, notably in regions where air quality is not monitored, and also because the toxicity of particles from various sources may vary10. Here we use a global atmospheric chemistry model to investigate the link between premature mortality and seven emission source categories in urban and rural environments. In accord with the global burden of disease for 2010 (ref. 5), we calculate that outdoor air pollution, mostly by PM2.5, leads to 3.3 (95 per cent confidence interval 1.61–4.81) million premature deaths per year worldwide, predominantly in Asia. We primarily assume that all particles are equally toxic5, but also include a sensitivity study that accounts for differential toxicity. We find that emissions from residential energy use such as heating and cooking, prevalent in India and China, have the largest impact on premature mortality globally, being even more dominant if carbonaceous particles are assumed to be most toxic. Whereas in much of the USA and in a few other countries emissions from traffic and power generation are important, in eastern USA, Europe, Russia and East Asia agricultural emissions make the largest relative contribution to PM2.5, with the estimate of overall health impact depending on assumptions regarding particle toxicity. Model projections based on a business-as-usual emission scenario indicate that the contribution of outdoor air pollution to premature mortality could double by 2050.


Air pollution is associated with many health impacts, including chronic obstructive pulmonary disease (COPD) linked to enhanced ozone (O3), and acute lower respiratory illness (ALRI), cerebrovascular disease (CEV), ischaemic heart disease (IHD), COPD and lung cancer (LC) linked to PM2.5 (ref. 8). Many previous studies have been based on air quality measurements, largely focusing on urban pollution3,4,11,12,13,14. Atmospheric chemistry and transport models have been used to account for other environments, including those for which no measurement data are available15,16,17,18,19,20,21,22.

Recently, enhanced resolution regional and global models and satellite data have been applied to improve estimates of PM2.5 and O3 concentrations and their impact on air quality19,20,21,22,23,24. Here we present results obtained with an atmospheric chemistry–general circulation model, applied at high resolution to compute global air quality changes, combined with population data, country-level health statistics and pollution exposure response functions (Methods). Our calculations of air pollution related mortality are based on the method of the global burden of disease (GBD) for 2010 (ref. 5), applying improved exposure response functions that more realistically account for health effects at very high PM2.5 concentrations compared to former assessments8. This is particularly relevant for some parts of the world where air pollution has increased nearly unabated and for future scenarios that project the continued growth of emissions. Following the GBD5 we also include desert dust (which is largely natural) with PM2.5; hence strictly speaking we assess the effects of atmospheric composition.

The air quality guidelines of the World Health Organization (WHO) and national regulatory policies are based on exposure response functions that rely on PM2.5 mass concentrations, implicitly treating all fine particles as equally toxic without regard to their source and chemical composition. However, expert elicitation suggests that carbonaceous particles are more toxic than crustal material, nitrates and sulfates10. A recent study25 finds that PM2.5 from coal combustion leads to increased mortality risk from cardiovascular disease and LC, but that the evidence is much weaker for other sources, whereas estimates using non-specific PM2.5 mass alone may underestimate the total effect of PM2.5 on mortality. Further, this study did not find support for mortality from biomass combustion and soil dust particles25. However, this and a subsequent report by the Health Effects Institute in the USA also note that there were only a limited number of cities in these investigations where these sources and components were likely to be measured consistently26,27. While the evidence for differential toxicity is far from conclusive, we conducted a secondary analysis assuming that carbonaceous PM2.5 is five times more toxic than inorganic particles, though maintaining the same overall health impact of PM2.5.

We have calculated premature mortality linked to CEV, COPD, IHD and LC for adults ≥30 years old, and ALRI for infants <5 years old (Table 1 and Extended Data Tables 1 and 2). Our estimate of the global PM2.5 related mortality in 2010 is 3.15 million people with a 95% confidence interval (CI95) of 1.52–4.60 million. The main causes are CEV (1.31 million) and IHD (1.08 million), and secondary causes are COPD (374 thousand), ALRI (230 thousand) and LC (161 thousand). Our global estimate of O3 related mortality by COPD is 142 (CI95: 90–208) thousand. Our total estimate of 3.30 (CI95: 1.61–4.81) million people in 2010 agrees closely with the GBD5. This is in addition to the estimated 3.54 million deaths per year caused by indoor air pollution due to use of solid fuels for cooking and heating5. Figure 1 shows the geographic distribution and demonstrates the locations of hotspots in China, India and many of the large urban centres.

Table 1 Premature mortality related to PM2.5 and O3 for the population <5 and ≥30 years old
Figure 1: Mortality linked to outdoor air pollution in 2010.

Units of mortality, deaths per area of 100 km × 100 km (colour coded). In the white areas, annual mean PM2.5 and O3 are below the concentration–response thresholds where no excess mortality is expected.

PowerPoint slide

Considering the global population of 6.8 billion in 2010, it follows that the mean per capita mortality attributable to air pollution is about 5 per 10,000 person-years. Of these 5 persons per 10,000 worldwide, about 2 die by CEV, 1.6 by IHD, 0.8 by COPD, 0.35 by ALRI and 0.25 by LC. The highest per capita mortality is found in the Western Pacific region, followed by the Eastern Mediterranean and Southeast Asia. The combination of high per capita mortality with high population density explains the (by far) highest number of deaths in the Western Pacific, China being the main contributor (1.36 million per year). Note that the mortality attributable to air pollution in China is approximately an order of magnitude higher than that attributable to Chinese road transport injuries and HIV/AIDS, and ranks among the top causes of death28. Southeast Asia has the second highest premature mortality, where India is the main contributor (0.65 million per year). The global mortality linked to air pollution is strongly influenced by these high numbers in Asia.

We determined the impacts of seven source categories by subtracting them one by one from the emissions in our model. These sensitivity calculations show the efficacy of individually controlling these sources. The 15 countries with highest premature mortality attributable to air pollution in 2010 are listed in Table 2 along with the contribution of each source category. Residential and commercial energy use (RCO) is the largest source category worldwide, contributing nearly one-third, and almost a factor of 2 more under the alternative assumption of differential toxicity. Note that this only refers to mortality by outdoor exposure to this source. Our estimate of 1.0 million deaths per year by RCO is in addition to the 3.54 million deaths per year due to indoor air pollution from essentially the same source5.

Table 2 Top 15 ranked countries of premature mortality linked to outdoor air pollution in 2010

The next largest anthropogenic source category is agriculture (AGR), contributing one-fifth; however, this reduces significantly under the assumption of differential particle toxicity. The successive principal anthropogenic categories are power generation (PG), industry (IND), biomass burning (BB) and land traffic (TRA), and taken together they cause nearly one-third of all air pollution mortality. If carbonaceous particles are five times more toxic than sulfates and nitrates, these sources together account for one-quarter of the mortality. Natural sources make up for the remaining one-sixth of the total. However, if crustal material is five times less toxic than carbonaceous PM2.5 this reduces considerably. The most important source category in each region in 2010 is shown in Fig. 2.

Figure 2: Source categories responsible for the largest impact on mortality linked to outdoor air pollution in 2010.

Source categories (colour coded): IND, industry; TRA, land traffic; RCO, residential and commercial energy use (for example, heating, cooking); BB, biomass burning; PG, power generation; AGR, agriculture; and NAT, natural. In the white areas, annual mean PM2.5 is below the concentration–response threshold.

PowerPoint slide

RCO is foremost in the populous parts of Asia. It refers to small combustion sources, especially biofuel use (for heating and cooking), and also waste disposal and diesel generators. In China it contributes about 32%, in India, Bangladesh, Indonesia and Vietnam 50–60%, while in Nepal it is highest with nearly 70% (Extended Data Table 3). In western countries it is typically 5–10%, although in France and Poland it contributes about 15%. The contribution of this pollution source to mortality is sensitive to toxicity assumptions and large uncertainty related to IHD. Because of the comparatively large fraction of carbonaceous PM2.5, under our alternative calculations where these aerosols are five times more toxic, RCO increases from 31% to 59% of global air pollution mortality. If, on the other hand, we assume that RCO does not contribute to IHD mortality, this fraction decreases from 31% to 26% (Methods).

Agriculture (AGR) has a remarkably large impact on PM2.5, and is the leading source category in Europe, Russia, Turkey, Korea, Japan and the Eastern USA (Fig. 2). In many European countries, its contribution is 40% or higher. Agricultural releases of ammonia (NH3) from fertilizer use and domesticated animals affect air quality through several multiphase chemical pathways, forming ammonium sulphate and nitrate. Since NH3 abundance is often limiting in PM2.5 formation, reduction of its emissions can make an important contribution to air quality control29. As agricultural emissions mostly form inorganic PM2.5, the impact on mortality diminishes under the assumption that carbonaceous PM2.5 is five times more toxic.

Natural sources (NAT) contribute strongly to mortality, being dominant in northern Africa and the Middle East, and also a leading category in Central Asia (Table 2 and Fig. 2). Although we categorize airborne desert dust as natural, a fraction is anthropogenic due to the role of humans in desertification and agricultural practices30. The chronic health and mortality impacts associated with exposure to dust are more uncertain than those due to typical air pollution in industrialized countries where most of the epidemiological cohort studies have been carried out. If all fine particles are equally toxic, then natural sources are responsible for about one-sixth of air pollution mortality. If fine carbonaceous particles are five times more toxic than crustal material, then natural sources account for only about one-tenth of air pollution induced mortality.

Power generation (PG) by fossil fuel fired power plants is the third largest anthropogenic source category, being an important source of SO2 and NOx, which are converted to sulfate and nitrate in the atmosphere. It accounts for about one-seventh of population exposure to PM2.5 and O3. Power plant emissions are quite important in the USA (>30%) and in Russia, Korea and Turkey (roughly 20%). Emissions from power generation also have particularly large impacts on fine particle concentrations in the Middle East, but frequently these go unnoticed as they are masked by desert dust. The role of this source is sensitive to the assumed PM2.5 toxicity, reducing by a factor of 2 if sulfate and nitrate are five times less toxic than carbonaceous PM2.5.

Industry (IND) is among the smaller source categories, with a global fraction of about 7% (Table 2); nevertheless, it contributes about twice this percentage in most of the western world. It includes iron and steel, chemical, pulp and paper, food, solvent and other manufacturing sectors, oil refineries and fuel production. This source of air pollution is generally significant in industrialized countries and emerging economies, but rarely the leading cause of premature mortality. Under the differential toxicity assumption, its contribution to mortality would reduce by more than a factor of 2.

Our calculations suggest that land traffic (TRA) emissions are responsible for about one-fifth of mortality by ambient PM2.5 and O3 in Germany, the UK and the USA, while globally they account for about 5%. Because emissions of NOx are the dominant source of traffic-related PM2.5 in the form of nitrate, together with carbonaceous PM2.5, the results from our alternative calculations—assuming carbonaceous particles are five times more toxic than nitrates and other inorganics—also indicate a 5% contribution, globally. Note that this contribution is likely to be a lower limit as traffic also emits other pollutants that are not included or influential on PM2.5 (ref. 31) (Methods).

Biomass burning (BB) is also a relatively small source category with a global contribution of about 5%. Nevertheless, its areal range is large, for example in South America and Africa. It is the main source of air pollution in large parts of Canada, Siberia, Africa, South America and Australia. Because in many parts of these countries annual mean PM2.5 is below the concentration–response threshold (Methods), these areas are shown white in Fig. 2. Biomass burning is also widespread in southeastern Asia, although in populous parts of Vietnam and Indonesia (for example, Java) residential energy use is larger and therefore the leading category (Table 2).

In the Southern Hemisphere biomass burning is generally the leading contributor to PM2.5, with some exceptions. In Brazil it contributes about 70%, and in many African countries its impact can also be high, up to >90% in Angola. Note that the health impacts of PM2.5 from biomass burning are quite uncertain, especially the attribution of IHD related mortality, due to a dearth of epidemiological cohort studies in regions where this pollution source predominates (Methods). Our calculations suggest that it is responsible for between 5% (equal toxicity) and 8% (differential toxicity) of air pollution induced mortality.

To understand how the premature mortality attributable to air pollution may develop in the coming decades, we applied a business-as-usual (BaU) emission scenario for the years 2025 and 2050, assuming that only currently agreed legislation is implemented that will affect future emissions32. Thus air quality and emission standards are fixed. Results for 2050 are presented here, and for 2025 in Extended Data Fig. 2 and Extended Data Tables 4, 5. Under the BaU scenario, moderate though significant increases of premature mortality will occur in Europe and the Americas, to a large degree in urban areas. Large increases are projected in Southeast Asia and the Western Pacific, leading to a global growth of premature mortality to 6.6 (CI95: 3.4–9.3) million (+100%) in 2050 (Table 1). This compares to a negligible population increase of infants (<5 years old), and a substantial increase (+68%) among people ≥30 years old in 2050 (implying an ageing population). Globally, the per capita mortality is projected to increase from 5 per 10,000 person-year in 2010 to about 7 per 10,000 person-year in 2050. The mortality attributable to air pollution will continue to be dominated by Asia with an unchanged fraction of about 75%.

The urban population is expected to grow relatively rapidly from 3.6 billion in 2010 to 5.2 billion in 2050, and combined with increasing air pollution concentrations the health impacts will escalate. Our estimate of urban premature mortality by outdoor air pollution in 2010 is 2.0 million, increasing to 4.3 million in 2050, representing 60% of the global total in 2010 and 65% in 2050. Urban population growth is responsible for part of this change, but the levels of air pollution in urban areas are also projected to grow rapidly. This is evident from our finding that the per capita mortality attributable to air pollution in 2010 is about 50% higher in urban than in rural environments. Under the BaU scenario this difference is expected to increase to nearly 90% in 2050.

Recently, much emphasis has been placed on rapidly emerging megacities (Methods). We calculate that 17 megacities and conurbations in Asia rank among the top 30 in terms of premature mortality worldwide, the leading one being the Pearl River Delta. When viewed instead from the perspective of individual risk, Tianjin and Beijing rank highest (Extended Data Table 6). While the per capita mortality attributable to air pollution is already extraordinary in Chinese megacities, according to the BaU scenario it will become even higher in Chinese and also Indian megacities by 2050. The combined premature mortality in the 30 largest conurbations accounts for about 7% of the worldwide burden of air pollution, indicating the relevance of all urban areas.

Our results suggest that if the projected increase in mortality attributable to air pollution is to be avoided, intensive air quality control measures will be needed, particularly in South and East Asia. The poorly characterized uncertainty about the relative toxicity of various classes of particles such as sulfates, nitrates, organics, crustal materials, black carbon, and especially smoke from biomass combustion, limits unambiguous attribution of sources. Nevertheless, our study suggests that emissions from residential energy use should be considered in air pollution control strategies and, if all fine particles are equally toxic, the reduction of agricultural emissions would improve air quality. An improvement in the efficacy of air pollution controls requires a better understanding of the relative toxicity of particles from various emissions sources.


Model and emissions

We used the global ECHAM5/MESSy atmospheric chemistry (EMAC)–general circulation model at a spatial resolution of T106L31, that is, with a spherical spectral truncation of T106, which corresponds to a quadratic Gaussian grid of approximately 1.1° × 1.1° latitude × longitude (110 km at the Equator), with 31 vertical hybrid terrain-following and pressure levels up to 10 hPa in the lower stratosphere. The core atmospheric model is the 5th generation European Centre Hamburg (ECHAM5, version 5.3.01) general circulation model33. EMAC includes sub-models that represent tropospheric and stratospheric processes and their interaction with oceans, land and human influences34,35,36. It uses the Modular Earth Submodel System (MESSy, v.1.09) to link submodels that describe emissions, atmospheric chemistry, aerosol and deposition processes; the results have been tested against in situ and remote sensing observations37,38,39,40,41,42,43,44,45,46,47,48,49.

Following up on Lelieveld et al.21, who focused on the year 2005, we present results for the years 2010, 2025 and 2050, applying monthly varying emission data from Doering et al.50, also used by Pozzer et al.32. The data are from the Emission Database for Global Atmospheric Research (EDGAR), prepared by the Joint Research Centre of the European Commission in Ispra (Italy) at a resolution of 0.1° latitude and longitude50,51. For the year 2010 we performed sensitivity calculations in which seven main emission categories have been removed one by one to compute the impact of these sources and to estimate their contributions to air quality control and related mortality. We first calculated the apportionment of source categories to the total PM2.5 and O3 concentrations and then applied the computed fractions to the total mortalities attributable to air pollution.

The categories are: (1) ‘Natural’ (NAT), mostly desert dust but locally also sea salt and dimethyl sulphide derived sulphate, some nitrate and ammonium from natural sources, volcanic sulphur emissions and organics released by the vegetation; (2) ‘Industry’ (IND), including iron and steel, chemical, pulp and paper, food, solvent and other manufacturing sectors, oil refineries and fuel production; (3) ‘Land transport’ (TRA), that is, road and non-road transport on land; (4) ‘Residential and commercial energy use’ (RCO), referring to local and commercial energy use from small combustion sources for space heating and cooking, including diesel generators and biofuel use; (5) ‘Power generation’ (PG), that is, public energy production by fossil fuel fired power plants; 6) ‘Biomass burning’ (BB), that is, tropical forest fires and deforestation, savanna and shrub fires, middle and high latitude forest and grassland fires, and agricultural waste burning; and (7) ‘Agriculture’ (AGR), dominated by ammonia emissions associated with the use of fertilizers and domesticated animals. Not included in these categories are air traffic and shipping. We find that the removal of individual source categories leads to a near-linear response in the modelled contributions to mortality, indicated by the small scaling corrections needed (about 10%) to add up to 100% in the country level contributions, that is, in Table 2 and Extended Data Table 3.

The BaU scenarios for 2025 and 2050 assume that energy and food consumption are largely determined by population growth and economic development, which in turn drive air pollution sources based on current legislation and technology32,50,51. This represents a pessimistic, but plausible future prospect. Comparable to Shindell et al.52, and different from the Representative Concentration Pathways of the Intergovernmental Panel on Climate Change53, the BaU scenario differentiates between air pollution and climate change mitigation measures, as the latter typically require relatively long-term and structural societal changes. The scenarios used here are based on projections for energy and fuel computed by the Prospective Outlook for the Long-term Energy System (POLES) model51,54 and for agriculture, land-use and waste projections by the Integrated Model to Assess the Global Environment (IMAGE)55.

The population development in the BaU scenario is consistent with our mortality calculations, as described below, projecting 9 billion people in 2050. For additional details we refer to Pozzer et al.32 and references therein. While BaU projections should not be conceived as ‘predictions’, especially for 2050, they represent the current trajectory into the future and may be considered a worst-case scenario, to explore what can be expected if air quality policies and health care remain as they are today. Note that these results are not sensitive to differential toxicity assumptions as the total mortality induced by PM2.5 is not affected, only the attribution to source categories. For the future scenarios we used the baseline mortalities for 2010. Hence the implicit assumption is that smoking habits, diets and health care remain unchanged.

The model meteorology has been forced by pre-calculated sea surface temperatures and ice coverage based on a 10-year climatology (2000–2009) adopted from the AMIP-II database56,57. The model was applied in atmospheric chemistry-transport mode by switching the coupling between radiation and atmospheric chemistry off, so that atmospheric composition changes do not influence the model dynamics32. This is justified considering that air quality projections are primarily driven by emissions rather than climate change58,59, even though natural sources, biomass burning and deposition processes can be influenced by climatic conditions20,59,60,61,62. For example, Fang et al.62 project a 4% climate change effect for PM2.5 related mortality and less than 1% for O3 related mortality by the end of the 21st century.

Although our model resolution does not resolve small-scale heterogeneities in the urban environment, a comparison with satellite and ground-based remote sensing observations indicates that this is not critical. The exposure response functions used to calculate mortalities are based on annual mean concentrations for which these heterogeneities largely average out. This is illustrated by Extended Data Fig. 3, which compares a simulation for the year 2010 with ground-based AERONET remote sensing data of aerosol optical depth (AOD) ( Since our model approximates though not replicates meteorological conditions for the year 2010, and local flows near the AERONET stations cannot be captured, substantial scatter around the ideal 1:1 comparison is expected. The comparison shows that the model mean error and bias are small (the latter absent for the annual mean), and the correlation good. We have also performed a comparison between MODIS (satellite) and AERONET data of AOD, leading to similar spread and correlations, the latter also increasing through averaging (not shown).

The primary differences in the relationships between emissions and exposures for ground level sources, such as traffic, in comparison with elevated sources, such as power plants, have been accounted for in our model43. The relative impacts of secondary particles (such as sulfates and nitrates) from these sources are expected to be realistically simulated. On the other hand, models such as ours cannot capture the fine structure of near-source gradients in ultrafine PM along transportation corridors. Because of this our estimates of the relative impacts of urban traffic and urban sources of primary fine particles may be biased downward, though only to the extent that ultrafine PM is in fact responsible for the mortality seen in cohort studies. As discussed above, the relative toxicity of various constituents of ambient PM2.5 has not been well established. Our sense is that the sensitivity study, allowing for carbonaceous particles to be five times as toxic as sulfates, nitrates and crustal material, is adequate to cover any potential differences in the relationships between emissions, exposure and differential toxicity of traffic related PM2.5.

To investigate if our model reproduces urban concentration increments of PM2.5 and O3, that is, comparing the urban background with the rural environment, we compare our results with recent case studies63,64,65,66,67. For Paris and London our model computes urban PM2.5 increments of 18% and 2%, respectively, consistent with the measurements and highly resolved model calculations. Our model calculations suggest that the leading sources of PM2.5 in Paris are residential energy use, agriculture and traffic. Agricultural emissions (NH3/NH4+) are transported from the rural environment and contribute to PM2.5 in the city. For London we calculate that PM2.5 is most strongly influenced by agriculture, traffic and power generation. The limited contribution by land traffic and the importance of atmospheric transport for air quality in London have been corroborated by observational analysis63. For Beijing we calculate an urban PM2.5 increment of 5%, consistent with the conclusion by Zhang et al.67 that regional sources are crucial contributors to PM2.5. They estimate the contribution by traffic and waste incineration at 4%; our results suggest that traffic alone contributes 3% in this city and residential energy use 47%, which we find to be representative of China (Table 2).

Our model calculations indicate that these relatively small urban increments for PM2.5 are typical for many, though not all, cities. For example, for Johannesburg (including Pretoria) we find +41% and for the Pearl River area +62%, and in both conurbations residential energy use is the leading source of PM2.5. For O3 we find generally small and negative urban increments due to titration of O3 by local traffic emissions (in Paris −7% and in London −5%). Negative urban increments due to NO by traffic of a few per cent (comparing weekend with weekdays) have also been documented for American cities68. For Chicago, New York, Los Angeles and Atlanta we find negative O3 increments of 1–5% due to traffic and power generation.

Sample size

No statistical methods were used to predetermine sample size.

Exposure response functions

The premature mortality attributable to PM2.5 and O3 has been calculated by applying the EMAC model for the present (2010) and projected future (2025, 2050) concentrations. We combined the results with epidemiological exposure response functions by employing the following relationship to estimate the excess (that is, premature) mortality:

ΔMort is a function of the baseline mortality rate due to a particular disease category yo for countries and/or regions estimated by the World Health Organization69 (the regions and strata are listed in the Extended Data Table 1). The term (RR − 1)/RR is the attributable fraction and RR is the relative risk. The disease specific baseline mortality rates have been obtained from the WHO Health Statistics and Health Information System. The value of RR is calculated for the different disease categories attributed to PM2.5 and O3 for the population below 5 years of age (ALRI) and 30 years and older (IHD, CEV, COPD, LC) using exposure response functions from the 2010 GBD analysis of the WHO (and described below).

The population (Pop) data for regions, countries and urban areas have been obtained from the NASA Socioeconomic Data and Applications Center (SEDAC), hosted by the Columbia University Center for International Earth Science Information Network (CIESIN), available at a resolution of 2.5′ × 2.5′ (about 5 km × 5 km) (, and projections by the United Nations Department of Economic and Social Affairs/Population Division70 ( Urban areas are defined by applying a population density threshold of 400 individuals per km2, while for megacities and major conurbations the threshold is 2,000 individuals per km2. We note that the resolution of our atmospheric model, about 1° latitude/longitude, is coarser than that of the population data, and our model does not resolve details of the urban environment. However, our anthropogenic emission data are aggregated from a resolution of 10 km to that of the model grid, accounting for relevant details such as altitude dependence (for example, stack emissions and hot plume rise effects)43.

Lelieveld et al.21 (henceforth L2013) derived the relative risk RR from the following exposure response function:

The term X represents the model calculated annual mean concentration of PM2.5 or O3. The value of Xo is the threshold concentration below which no additional risk is assumed (concentration–response threshold). The parameter b is the concentration response coefficient. However, it has been argued that this expression is based on epidemiological cohort studies in the USA and Europe where annual mean PM2.5 concentrations are typically below 30 μg m−3, which may not be representative for countries where air pollution levels can be much higher, for example in South and East Asia. This is particularly relevant for our BaU scenario. Therefore, here we have used the revised exposure response function of Burnett et al.8 who also included epidemiological data from the exposure to second-hand smoke, indoor air pollution and active smoking to account for high PM2.5 concentrations, and tested eight different expressions. The best fit to the data was found for the following relationship, which was also used by Lim et al.5 for the GBD for the year 2010:

The RR functions were derived by Burnett et al.8. We applied this model for the different categories, represented by their figures 1 and 2, shown to be superior to other forms previously used in burden assessments. We also adopted the upper and lower bounds, likewise shown in these figures, representing the 95% confidence intervals (CI95). The latter were derived based on Monte Carlo simulations, leading to 1,000 sets of coefficients and exposure response functions from which the upper and lower bounds were calculated.

Following Burnett et al.8 and Lim et al.5 we combine all aerosol types, hence including natural particulates such as desert dust. Note that by using PM2.5 mass, we do not distinguish the possibly different toxicity of various kinds of particles. This information is not available from epidemiological cohort studies, but could potentially substantially affect both our overall estimates of mortality and the geographical patterns. This is addressed by sensitivity calculations presented in the main text, Table 2 and Extended Data Fig. 1. For COPD related to O3 we applied the exposure response function by Ostro et al.3:

where b is 0.1521 and Xo the average of the range 33.3–41.9 p.p.b.v. O3 indicated by Lim et al.5, that is, 37.6 p.p.b.v. Previously we used model calculated pre-industrial O3 concentrations to estimate XXo (ref. 21), leading to about 20% higher estimates for mortality by ‘respiratory disease’ related solely to O3 compared to the current estimate for COPD due to both PM2.5 and O3.

For detailed discussion of uncertainties and sensitivity calculations that address the shape of exposure response functions, we refer to earlier work5,8,21,22 and references therein. L2013 estimated statistical uncertainties by propagating the quantified (random) errors of all parameters in the exposure response functions. They found that the CI95 of estimated mortality attributable to air pollution in Europe, North and South America, South and East Asia are within 40%, whereas they are 100–170% in Africa and the Middle East. Our results are very close to the GBD, which substantiates the estimates by Lim et al.5 and provides consistency with the most recent estimates for 2010, serving as a basis for our investigations.

We emphasize that the confidence intervals described here, and those reported by Lim et al.5, reflect only the statistical uncertainty of the parameters used in the concentration–response functions. It is known that the uncertainty in interpretation of epidemiological results can be dominated by other model or epistemic uncertainties, such as those having to do with the control of confounders. Sources of uncertainty have been summarized by Kinney et al.71, who underscore the need to determine the differential toxicity of specific component species within the complex mixture of particulate matter. Our sensitivity calculations (Table 2 and Extended Data Fig. 1) corroborate that this can have significant influence, especially in areas where carbonaceous compounds contribute strongly to PM2.5.

We emphasize the dearth of studies that link PM2.5 from biomass combustion emissions—rich in carbonaceous particles—to IHD. Expert judgment studies on the toxicity of particulate matter have reported uncertainties much larger than those suggested by analysis of parameter uncertainty alone10,72. Although the CI95 intervals provided above include a larger range of parameters and uncertainties than these earlier studies, they should be viewed as lower bounds on the true uncertainty in estimates of the health effects of PM2.5 exposure, especially PM2.5 from biomass burning and biofuel use. If we consider the possibility that biomass burning (BB, including agricultural waste burning) and residential energy use (RCO, dominated by biofuel use) do not contribute to mortality by IHD, the total mortality attributable to air pollution would decrease from 3.3 to 3.0 million per year (Extended Data Table 7). The largest effect is found in Southeast Asia where biomass combustion (RCO and BB) is a main source of air pollution. While the global contribution by residential energy use, as presented in Table 2, would decrease from 31% to 26%, and of biomass burning from 5% to 4% (the other categories increase proportionally), the ranking of the different sources and hence our conclusions remain unchanged, as RCO and BB would still be the largest and smallest source category, respectively.

Issues such as the shape of the concentration–response functions and the existence and specific levels of concentration–response thresholds have been discussed by the experts10,71,72. These have been accounted for by Burnett et al.8, however, uncertainty related to the differences in central estimates given by various cohort studies is not reflected in the estimates of parameter uncertainty by Lim et al.5. This problem has grown more substantial recently as the results from new cohort studies have become available73. Furthermore, uncertainty about the relative toxicity of different constituents of PM2.5 remains. Since the current study underscores that the sources of mortality attributable to PM2.5 can differ strongly between different regions (Fig. 2), this aspect merits greater attention in future.

Comparison to previous work

We estimate the combined (PM2.5 and O3 related) global mortality attributable to air pollution in 2010 at 3.3 million. Our global estimate for PM2.5 related mortality of 3.15 million per year is close to that of 3.22 million per year in the GBD study for 2010 (ref. 4). However, it is substantially higher than the recent multi-model study of Silva et al.20 for the year 2000, being 2.1 million per year. The difference can be explained by the focus of Silva et al.20 on anthropogenic pollution in 2000, whereas our study and the GBD account for emission increases between 2000 and 2010 and also include natural sources.

Our global estimate of O3 related mortality by COPD in 2010 is 142,000, substantially lower than the estimates of Anenberg et al.18, 700,000 deaths in 2000; L2013, 773,000 in 2005; and Silva et al.20, 470,000 deaths in 2000; but quite close to the GBD estimate of 152,000 deaths in 2010. Much of the difference between our results (and those from the 2010 GBD) and previous work is explained by the fact that we attribute COPD to both O3 and PM2.5. When our results for COPD from both O3 and PM2.5 are combined, our overall estimate of COPD mortality from air pollution agrees with the above-mentioned studies within about 25–30%. The remaining differences are largely due to the use of a concentration response threshold, Xo, in our new work, which substantially reduces mortality estimates. Anenberg et al.18 and L2013 did not apply a threshold but computed the natural background based on preindustrial emissions. In these analyses the calculated ambient concentrations are typically lower than Xo. For example, the global average O3 ambient concentration at the surface in our pre-industrial simulation is 19 p.p.b.v. The global mortality estimate for 2010 presented here is 10% higher than that of L2013 for 2005. This is primarily due to the fact that we also account for natural sources in the present work. If we subtract the natural fraction, our estimate of mortality attributable to anthropogenic air pollution for 2010 is 9% lower than that of L2013, mostly related to the new exposure response functions applied here.

Our calculations suggest that natural sources contribute relatively strongly to mortality attributable to air pollution (18%), about 600,000 per year, which is to a large degree caused by airborne desert dust. Recently we reported a global dust-related mortality rate of about 400,000 per year, substantially lower than the present estimate22. While here we follow the GBD methodology5, it is likely to yield an upper limit. Instead of the annual mean dust concentrations Giannadaki et al.22 used the median concentrations, motivated by the intermittent nature of dust events. Their sensitivity calculations indicate that had they used the mean concentration instead, their estimate of global dust-related mortality would have increased from 402,000 per year to 622,000 per year. Finally, if we assume that carbonaceous aerosols are five times more toxic than other compounds, including dust particles, the contribution by natural sources would decrease from about 600,000 per year (18%) to 360,000 per year (11%).


  1. 1

    Murray, C. & Lopez, A. D. The Global Burden of Disease: A Comprehensive Assessment of Mortality and Disability from Diseases, Injuries, and Risk Factors in 1990 and Projected in 2020 (Harvard Univ. Press, 1996)

    Google Scholar 

  2. 2

    Ezzati, M. et al. Selected major risk factors and global and regional burden of disease. Lancet 360, 1347–1360 (2002)

    Article  PubMed  Google Scholar 

  3. 3

    Ostro, B. Outdoor Air Pollution: Assessing the Environmental Burden of Disease at National and Local Levels (World Health Organization Environmental Burden of Disease Series No. 5, WHO, Geneva, 2004)

    Google Scholar 

  4. 4

    Cohen, A. J. et al. The global burden of disease due to outdoor air pollution. J. Toxicol. Environ. Health A 68, 1301–1307 (2005)

    Article  CAS  PubMed  Google Scholar 

  5. 5

    Lim, S. S. et al. A comparative risk assessment of burden of disease and injury attributable to 67 risk factors and risk factor clusters in 21 regions, 1990–2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet 380, 2224–2260 (2012); correction 381, 628 (2013)

    Article  PubMed  PubMed Central  Google Scholar 

  6. 6

    Pope, C. A., III & Dockery, D. W. Health effects of fine particulate air pollution: lines that connect. J. Air Waste Manag. Assoc. 56, 709–742 (2006)

    Article  CAS  PubMed  Google Scholar 

  7. 7

    Beelen, R. et al. Effects of long-term exposure to air pollution on natural-cause mortality: an analysis of 22 European cohorts within the multicentre ESCAPE project. Lancet 383, 785–795 (2014)

    Article  CAS  PubMed  Google Scholar 

  8. 8

    Burnett, R. T. et al. An integrated risk function for estimating the Global Burden of Disease attributable to ambient fine particulate matter exposure. Environ. Health Perspect. 122, 397–403 (2014)

    Article  PubMed  PubMed Central  Google Scholar 

  9. 9

    Jerrett, M. et al. Long-term ozone exposure and mortality. N. Engl. J. Med. 360, 1085–1095 (2009)

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. 10

    Tuomisto, J. T., Wilson, A., Evans, J. S. & Tainio, M. Uncertainty in mortality response to airborne fine particulate matter: combining European air pollution experts. Reliab. Eng. Syst. Saf. 93, 732–744 (2008)

    Article  Google Scholar 

  11. 11

    Pope, C. A., III et al. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. J. Am. Med. Assoc. 287, 1132–1141 (2002)

    Article  CAS  Google Scholar 

  12. 12

    Prüss-Üstün, A., Bonjour, S. & Corvalan, C. The impact of the environment on health by country: a meta-synthesis. Environ. Health 7, (2008)

  13. 13

    Russell, A. G. & Brunekreef, B. A focus on particulate matter and health. Environ. Sci. Technol. 43, 4620–4625 (2009)

    Article  ADS  CAS  PubMed  Google Scholar 

  14. 14

    Gurjar, B. R. et al. Human health risks in megacities due to air pollution. Atmos. Environ. 44, 4606–4613 (2010)

    Article  ADS  CAS  Google Scholar 

  15. 15

    West, J. J., Fiore, A. M., Horowitz, L. W. & Mauzerall, D. L. Global health benefits of mitigating ozone pollution with methane emission controls. Proc. Natl Acad. Sci. USA 103, 3988–3993 (2006)

    Article  ADS  CAS  PubMed  Google Scholar 

  16. 16

    Duncan, B. N. et al. The influence of European pollution on ozone in the Near East and northern Africa. Atmos. Chem. Phys. 8, 2267–2283 (2008)

    Article  ADS  CAS  Google Scholar 

  17. 17

    Liu, J., Mauzerall, D. L. & Horowitz, L. W. Evaluating inter-continental transport of fine aerosols: (2) Global health impact. Atmos. Environ. 43, 4339–4347 (2009)

    Article  ADS  CAS  Google Scholar 

  18. 18

    Anenberg, S. C., Horowitz, L. W., Tong, D. Q. & West, J. J. An estimate of the global burden of anthropogenic ozone and fine particulate matter on premature human mortality using atmospheric modeling. Environ. Health Perspect. 118, 1189–1195 (2010)

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. 19

    Fann, N. et al. Estimating the national public health burden associated with exposure to ambient PM2.5 and ozone. Risk Anal. 32, 81–95 (2012)

    Article  PubMed  Google Scholar 

  20. 20

    Silva, R. A. et al. Global premature mortality due to anthropogenic outdoor air pollution and the contribution of past climate change. Environ. Res. Lett. 8, (2013)

    Article  ADS  CAS  Google Scholar 

  21. 21

    Lelieveld, J., Barlas, C., Giannadaki, D. & Pozzer, A. Model calculated global, regional and megacity premature mortality due to air pollution by ozone and fine particulate matter. Atmos. Chem. Phys. 13, 7023–7037 (2013)

    Article  ADS  CAS  Google Scholar 

  22. 22

    Giannadaki, D., Pozzer, A. & Lelieveld, J. Modeled global effects of airborne desert dust on air quality and premature mortality. Atmos. Chem. Phys. 14, 957–968 (2014)

    Article  ADS  CAS  Google Scholar 

  23. 23

    van Donkelaar, A. et al. Global estimates of ambient fine particulate matter concentrations from satellite-based aerosol optical depth: development and application. Environ. Health Perspect. 118, 847–855 (2010)

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  24. 24

    Brauer, M. et al. Exposure assessment for estimation of the Global Burden of Disease attributable to outdoor air pollution. Environ. Sci. Technol. 46, 652–660 (2012)

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

  25. 25

    Thurston, G. D. et al. in National Particle Component Toxicity (NPACT) Initiative: Integrated Epidemiologic and Toxicologic Studies of the Health Effects of Particulate Matter Components (eds Lippmann, M. et al.) 127–166 (Health Effects Institute Research Report 177, Boston, 2013)

    Google Scholar 

  26. 26

    Lippmann M., et al. (eds) National Particle Component Toxicity (NPACT) Initiative: Integrated Epidemiologic and Toxicologic Studies of the Health Effects of Particulate Matter Components (Health Effects Institute Research Report 177, Boston, 2013)

    Google Scholar 

  27. 27

    Vedal, S. et al. National Particle Component Toxicity (NPACT) Initiative: Report on Cardiovascular Effects (Health Effects Institute Research Report 178, Boston, 2013)

    Google Scholar 

  28. 28

    Yang, G. et al. Rapid health transition in China, 1990–2010: findings from the Global Burden of Disease Study 2010. Lancet 381, 1987–2015 (2013)

    Article  PubMed  Google Scholar 

  29. 29

    Megaritis, A. G., Fountoukis, C., Charalampidis, P. E., Pilinis, C. & Pandis, S. N. Response of fine particulate matter concentrations to changes of emissions and temperature in Europe. Atmos. Chem. Phys. 13, 3423–3443 (2013)

    Article  ADS  CAS  Google Scholar 

  30. 30

    Ginoux, P., Prospero, J. M., Gill, T. E., Hsu, N. C. & Zhao, M. Global-scale attribution of anthropogenic and natural dust sources and their emission rates based on MODIS Deep Blue aerosol products. Rev. Geophys. 50, RG3005 (2012)

    Article  ADS  Google Scholar 

  31. 31

    Tager, I. et al. Traffic-related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects (Health Effects Institute Special Report 17, Boston, 2010)

    Google Scholar 

  32. 32

    Pozzer, A. et al. Effects of business-as-usual anthropogenic emissions on air quality. Atmos. Chem. Phys. 12, 6915–6937 (2012)

    Article  ADS  CAS  Google Scholar 

  33. 33

    Roeckner, E. et al. Sensitivity of simulated climate to horizontal and vertical resolution in the ECHAM5 atmosphere model. J. Clim. 19, 3771–3791 (2006)

    Article  ADS  Google Scholar 

  34. 34

    Jöckel, P. et al. Technical Note: The Modular Earth Submodel System (MESSy) – a new approach towards earth system modeling. Atmos. Chem. Phys. 5, 433–444 (2005)

    Article  ADS  Google Scholar 

  35. 35

    Jöckel, P. et al. The atmospheric chemistry general circulation model ECHAM5/MESSy: Consistent simulation of ozone from the surface to the mesosphere. Atmos. Chem. Phys. 6, 5067–5104 (2006)

    Article  ADS  Google Scholar 

  36. 36

    Pozzer, A., Jöckel, P., Kern, B. & Haak, H. The atmosphere-ocean general circulation model EMAC-MPIOM. Geosci. Model Dev. 4, 771–784 (2011)

    Article  ADS  Google Scholar 

  37. 37

    Sander, R., Kerkweg, A., Jöckel, P. & Lelieveld, J. Technical note: The new comprehensive atmospheric chemistry module MECCA. Atmos. Chem. Phys. 5, 445–450 (2005)

    Article  ADS  CAS  Google Scholar 

  38. 38

    Kerkweg, A. et al. Technical Note: An implementation of the dry removal processes DRY DEPosition and SEDImentation in the Modular Earth Submodel System (MESSy). Atmos. Chem. Phys. 6, 4617–4632 (2006)

    Article  ADS  CAS  Google Scholar 

  39. 39

    Tost, H. et al. Technical note: A new comprehensive SCAVenging submodel for global atmospheric chemistry modeling. Atmos. Chem. Phys. 6, 565–574 (2006)

    Article  ADS  CAS  Google Scholar 

  40. 40

    Tost, H. et al. Global cloud and precipitation chemistry and wet deposition: tropospheric model simulations with ECHAM5/MESSy1. Atmos. Chem. Phys. 7, 2733–2757 (2007)

    Article  ADS  CAS  Google Scholar 

  41. 41

    Pozzer, A. et al. Technical Note: The MESSy-submodel AIRSEA calculating the air-sea exchange of chemical species. Atmos. Chem. Phys. 6, 5435–5444 (2006)

    Article  ADS  CAS  Google Scholar 

  42. 42

    Pozzer, A. et al. Simulating organic species with the global atmospheric chemistry general circulation model ECHAM5/MESSy1: a comparison of model results with observations. Atmos. Chem. Phys. 7, 2527–2550 (2007)

    Article  ADS  CAS  Google Scholar 

  43. 43

    Pozzer, A., Jöckel, P. & van Aardenne, J. The influence of the vertical distribution of emissions on tropospheric chemistry. Atmos. Chem. Phys. 9, 9417–9432 (2009)

    Article  ADS  CAS  Google Scholar 

  44. 44

    Pozzer, A. et al. Distributions and regional budgets of aerosols and their precursors simulated with the EMAC chemistry-climate model. Atmos. Chem. Phys. 12, 961–987 (2012)

    Article  ADS  CAS  Google Scholar 

  45. 45

    Astitha, M. et al. Parameterization of dust emissions in the global atmospheric chemistry-climate model EMAC: impact of nudging and soil properties. Atmos. Chem. Phys. 12, 11057–11083 (2012)

    Article  ADS  CAS  Google Scholar 

  46. 46

    Pringle, K. J. et al. Description and evaluation of GMXe: A new aerosol submodel for global simulations (v1). Geosci. Model Dev. 3, 391–412 (2010)

    Article  ADS  Google Scholar 

  47. 47

    Pringle, K. J. et al. Global distribution of the effective aerosol hygroscopicity parameter for CCN activation. Atmos. Chem. Phys. 10, 5241–5255 (2010)

    Article  ADS  CAS  Google Scholar 

  48. 48

    de Meij, A. et al. EMAC model evaluation and analysis of atmospheric aerosol properties and distribution. Atmos. Res. 114-115, 38–69 (2012)

    Article  CAS  Google Scholar 

  49. 49

    Christoudias, T. & Lelieveld, J. Modelling the global atmospheric transport and deposition of radionuclides from the Fukushima Dai-ichi nuclear accident. Atmos. Chem. Phys. 13, 1425–1438 (2013)

    Article  ADS  CAS  Google Scholar 

  50. 50

    Doering, U., Janssens-Maenhout, G., van Aardenne, J. & Pagliari, V. Climate Change and Impact Research in the Mediterranean Environment: Scenarios of Future Climate Change. JRC Tech. Note 62957 (Joint Research Centre, Ispra, 2010)

    Google Scholar 

  51. 51

    Van Aardenne, J. et al. Climate and Air Quality Impacts of Combined Climate Change and Air Pollution Policy Scenarios. JRC Sci. Tech. Rep. 61281 (Joint Research Centre, Ispra, 2010)

    Google Scholar 

  52. 52

    Shindell, D. et al. Simultaneously mitigating near-term climate change and improving human health and food security. Science 335, 183–189 (2012)

    Article  ADS  CAS  PubMed  Google Scholar 

  53. 53

    Stocker, T. F. et al. (eds) Climate Change 2013: The Physical Science Basis (Cambridge Univ. Press, 2013)

    Google Scholar 

  54. 54

    Russ, P., Wiesenthal, T., van Regenmorter, D. & Ciscar, J. C. Global Climate Policy Scenarios for 2030 and Beyond. Analysis of Greenhouse Gas Emission Reduction Pathway Scenarios with the POLES and GEM-E3 models. JRC Ref. Rep. EUR 23032 EN, (Joint Research Centre, Ispra, 2007)

    Google Scholar 

  55. 55

    Bouwman A. F., Kram T., Klein Goldewijk K., eds. Integrated Modelling of Global Environmental change. An Overview of IMAGE 2.4 (Netherlands Environmental Assessment Agency (MNP), Bilthoven, 2006)

    Google Scholar 

  56. 56

    Taylor, K., Williamson, D. & Zwiers, F. The Sea Surface Temperature and Sea Ice Concentration Boundary Conditions for AMIP II Simulations. PCMDI Tech. Rep. 60 (Program for Climate Model Diagnosis and Intercomparison, Lawrence Livermore National Laboratory, Livermore, California, 2000)

    Google Scholar 

  57. 57

    Hurrell, J. et al. A new sea surface temperature and sea ice boundary dataset for the Community Atmosphere Model. J. Clim. 21, 5145–5153 (2008)

    Article  ADS  Google Scholar 

  58. 58

    Jacob, D. J. & Winner, D. A. Effect of climate change on air quality. Atmos. Environ. 43, 51–63 (2009)

    Article  ADS  CAS  Google Scholar 

  59. 59

    Pye, H. O. T. et al. Effect of changes in climate and emissions on future sulfate-nitrate-ammonium aerosol levels in the United States. J. Geophys. Res. 114 D01205, (2009)

    Article  ADS  CAS  Google Scholar 

  60. 60

    Hedegaard, G. B., Christensen, J. H. & Brandt, J. The relative importance of impacts from climate change vs. emissions change on air pollution levels in the 21st century. Atmos. Chem. Phys. 13, 3569–3585 (2013)

    Article  ADS  CAS  Google Scholar 

  61. 61

    Naik, V. et al. Preindustrial to present-day changes in tropospheric hydroxyl radical and methane lifetime from the Atmospheric Chemistry and Climate Model Intercomparison Project (ACCMIP). Atmos. Chem. Phys. 13, 5277–5298 (2013)

    Article  ADS  CAS  Google Scholar 

  62. 62

    Fang, Y. et al. Impacts of 21st century climate change on global air pollution-related premature mortality. Clim. Change 121, 239–253 (2013)

    Article  ADS  CAS  Google Scholar 

  63. 63

    Jones, A. M., Yin, J. & Harrison, R. M. The weekday–weekend difference and the estimation of the non-vehicle contributions to the urban increment of airborne particulate matter. Atmos. Environ. 42, 4467–4479 (2008)

    Article  ADS  CAS  Google Scholar 

  64. 64

    Harrison, R. M., Laxen, D., Moorcroft, S. & Laxen, K. Processes affecting concentrations of fine particulate matter (PM2.5) in the UK atmosphere. Atmos. Environ. 46, 115–124 (2012)

    Article  ADS  CAS  Google Scholar 

  65. 65

    Moussiopoulos, N. et al. An approach for determining urban concentration increments. Int. J. Environ. Pollut. 50, 376–385 (2012)

    Article  CAS  Google Scholar 

  66. 66

    Timmermans, R. M. A. et al. Quantification of the urban air pollution increment and its dependency on the use of down-scaled and bottom-up city emission inventories. Urban Clim. 6, 44–62 (2013)

    Article  Google Scholar 

  67. 67

    Zhang, R. et al. Chemical characterization and source apportionment of PM2.5 in Beijing: seasonal perspective. Atmos. Chem. Phys. 13, 7053-7074 (2013); Atmos. Chem. Phys. 14, 175 (2014)

    Article  ADS  CAS  Google Scholar 

  68. 68

    Blanchard, C. L., Tanenbaum, S. & Lawson, D. R. Differences between weekday and weekend air pollutant levels in Atlanta; Baltimore; Chicago; Dallas–Fort Worth; Denver; Houston; New York; Phoenix; Washington, DC; and surrounding areas. J. Air Waste Manag. Assoc. 58, 1598–1615 (2008)

    Article  CAS  PubMed  Google Scholar 

  69. 69

    World Health Organization. World Health Organization Statistical Information System (WHOSIS), Detailed Data Files of the WHO Mortality Database (WHO, Geneva, 2012)

  70. 70

    United Nations Department of Economic and Social Affairs/Population Division. World Population Prospects: the 2004 Revision. E.05.XIII.12 (United Nations, 2005)

  71. 71

    Kinney, P. L. et al. On the use of expert judgment to characterize uncertainties in the health benefits of regulatory controls of particulate matter. Environ. Sci. Policy 13, 434–443 (2010)

    Article  CAS  Google Scholar 

  72. 72

    Roman, H. A. et al. Expert judgment assessment of the mortality impact of changes in ambient fine particulate matter in the U.S. Environ. Sci. Technol. 42, 2268–2274 (2008)

    Article  ADS  CAS  PubMed  Google Scholar 

  73. 73

    Cao, J. et al. Association between long-term exposure to outdoor air pollution and mortality in China: A cohort study. J. Hazard. Mater. 186, 1594–1600 (2011)

    Article  CAS  PubMed  Google Scholar 

Download references


We are grateful to the EDGAR team of the Joint Research Centre in Ispra, Italy, for the emission data. We acknowledge support from the Distinguished Scientist Fellowship Program at the King Saud University, Riyadh. The research leading to these results has received funding from the European Research Council under the European Union’s Seventh Framework Programme (FP7/2007-2013)/ERC grant agreement no. 226144.

Author information




J.L., A.P. and M.F. planned the research, A.P. performed the model calculations, J.L., A.P., D.G. and J.S.E. analysed the results, and J.L. and J.S.E. wrote the paper. All authors contributed to the manuscript.

Corresponding author

Correspondence to J. Lelieveld.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Extended data figures and tables

Extended Data Figure 1 Source categories responsible for the largest impact on mortality linked to outdoor air pollution in 2010 from a sensitivity calculation with carbonaceous aerosol having a five times larger impact than inorganic and crustal compounds.

IND, industry; TRA, land traffic; RCO, residential energy use (for example, heating, cooking); BB, biomass burning; PG, power generation; AGR, agriculture; and NAT, natural.

Extended Data Figure 2 Increase in mortality linked to outdoor air pollution from 2010 to 2050 (business-as-usual scenario).

Units (colour coded), deaths per area of 100 km × 100 km. In the white areas, no additional mortality is projected.

Extended Data Figure 3 Comparison of EMAC model calculated aerosol optical depth (AOD) with AERONET observations, using all available measurements worldwide in the year 2010.

Although the comparison with individual data points shows a large scatter (left panel), the bias is small (MBE), and time averaging improves the agreement. The middle panel shows a comparison of the monthly means, and the right panel the annual means (that is, showing individual stations) for which the mean error (root mean square error, RMSE) is smallest, the correlation highest and the bias absent. The long-dashed line indicates absolute agreement, the bold short-dashed lines agreement within a factor of two and the short-dashed lines agreement within a factor of ten.

Extended Data Table 1 WHO regions, mortality strata, child and adult mortality characteristics, and the countries and territories included
Extended Data Table 2 Premature mortality related to PM2.5 and O3 in 2010
Extended Data Table 3 Premature mortality by PM2.5 and O3 related diseases in 2010 in countries where it exceeds 9,000 individuals per year (<5 and ≥30 years old)
Extended Data Table 4 Premature mortality related to PM2.5 and O3 in 2025
Extended Data Table 5 Premature mortality related to PM2.5 and O3 in 2050
Extended Data Table 6 Population and premature mortality (deaths per year) related to PM2.5 and O3 in the most polluted megacities and conurbations in 2010, 2025 and 2050
Extended Data Table 7 Premature mortality related to PM2.5 and O3 for the population aged <5 years and ≥30 years

Related audio

PowerPoint slides

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Lelieveld, J., Evans, J., Fnais, M. et al. The contribution of outdoor air pollution sources to premature mortality on a global scale. Nature 525, 367–371 (2015).

Download citation

Further reading


By submitting a comment you agree to abide by our Terms and Community Guidelines. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing