Article
WITHDRAWN: A Comparative Study of Life-Years Lost Attributable to Air Particulate Matter in Asia-Pacific and European Countries in 2019
https://doi.org/10.21203/rs.3.rs-2332124/v1
This work is licensed under a CC BY 4.0 License
posted
You are reading this older preprint version
The full text of this preprint has been withdrawn by the authors while they make corrections to the work. Therefore, the authors do not wish this work to be cited as a reference. Questions should be directed to the corresponding author.
Editorial notes are used to provide important context regarding the topic of a preprint or to alert readers to potential issues concerning that preprint or a downstream publication associated with it. For more information on editorial notes, see our Editorial Policies.
Air particulate matter (PM) and its harmful effects on human health have attracted attention from global public health. Environmental risk factors, mainly air pollution, have been included in the program of the prevention and control of non-communicable diseases (NCDs) priorities by the United Nations since 2018. As pollution has been the fifth-ranking risk factor for death by NCDs globally[1–4], it is also in the Sustainable Development Goals (SDGs) target 3.9, by reducing the number of mortality caused by pollution by 2030[3]. Research on mortality and morbidity risk attributable to particulate air pollution exposure, both indoor and outdoor, shows that stroke, ischemic heart disease (IHD), chronic obstructive pulmonary disease (COPD) and tracheal, bronchus and lung cancer (TBL) are the most affected causes of death by this pollutant[5–7]. Based on the evidence, stroke and IHD were the two main causes of death attributable to PM, accounting for over 70 per cent of the total deaths, while COPD and TBL were the following causes of death due to the pollutant globally[2, 8].
Outdoor particulate matter (particles with a diameter of 2.5 \(\mu\)g/m3 or smaller, PM2.5), which severely impacts human health[9, 10], has been further studied than indoor pollution. Specially, since household air pollution (HAP) from solid fuels has declined in various high-income countries. However, traditional indoor cooking and heating are still found in a considerable amount in low- and middle-income countries[11, 12]. Low-income countries are likely to have higher use of solid fuels in houses than higher-income countries leading to health effects from HAP[13, 14], especially in African and Asian countries[15, 16].
According to the study of health impacts of air pollution exposure from 1990 to 2019 in 43 European countries, the improvement of the PM concentration correlated with a reduction in cause-specific morbidity and mortality attributable to the pollutant in the region. Moreover, the study stated that social and economic factors played a crucial role in health inequality due to PM2.5[17].By contrast, most Asia-Pacific (APAC) countries have been experiencing a rise in air pollution-related mortality and the highest levels of PM across the world[18, 19]. Additionally, there are few studies dealing with the association between socio-demographic indexes (SDI) -measured by income and education[20]- on air pollution-related mortality in APAC[21, 22]. Thus, we focus on the UN region of APAC, as a highly polluted region, and compare it to Europe, as an improved air region, additionally we study PM mortality by socio-demographic levels in both regions.
The metric used for this study is life-years lost (LYL), or the complement of life expectancy, and corresponds to years that the population is short from an age benchmark. Similar to life expectancy which is measured in years, LYL refers to the years lost due to premature death[23–25]. Furthermore, LYL can be decomposed into cause-specific premature deaths, for example LYL due to PM attributable mortality, and used for comparison between countries and regions. Since all is based on life tables, without the confounding effect of the population composition, comparisons of life expectancy and LYL can be done across the APAC and European countries.
This study aims to: 1) investigate life-years lost attributable to ambient PM and household air pollution by the most affected causes of death, including stroke, ischemic heart disease, chronic obstructive pulmonary disease and tracheal, bronchus and lung cancer; 2) compare cause-specific LYL due to PM in Asia-Pacific and European countries; and 3) explore LYL attributable to indoor and outdoor PM across different populations defined in terms of SDI.
To study the health impacts of air pollution in APAC versus European countries, cause-specific LYL in 2019 are presented. Figure 1 shows that the average life-years lost due to ambient particulate matter (PM2.5) in the Asia-Pacific region is 1.65 years, that is 1.24 years higher than the average in Europe (0.41 years lost). In APAC (Fig. 1A), the highest number of years lost was observed in the Syrian Arab Republic with almost three years lost, whereas New Zealand and Australia had the lowest number of years lost in the region (< 0.1 years). This was the case even when the level of pollution in Syrian Arab Republic was about 30 \(\mu\)g/m3 lower than in Qatar, India and Nepal. In Europe, North Macedonia had the largest number of years lost, while Sweden showed the lowest number in 2019. This was also in concordance with the level of ambient particulate matter concentration, which was high in North Macedonia and low in Sweden (Fig. 1B). The positive relationship between LYL due to ambient particulate matter and the level of PM concentration was high in both regions, but higher in Europe (Pearson correlation of r = 0.927, significant at p < 0.0001) than for APAC (r = 0.705, p < 0.0001, see Figure A1 in Supplementary Material, SM).
[Figure 1A and 1B about here]
The average LYL attributable to household air pollution in APAC amounted to 0.57 years, which was over 50 percent higher than in European countries (~ 0.01 years lost). Various countries in the Asia- Pacific, particularly the Pacific Islands, such as Solomon Islands, Vanuatu, Kiribati and Papua New Guinea, had a high level of LYL due to HAP, when compared to other Asian countries. Also, LYL due to HAP for these countries were 2.5 times higher than LYL due to PM2.5 in the APAC region (Fig. 2A). Similar to the average levels in the two regions, the HAP concentration for European countries was lower than for APAC. There was a small number of countries in Europe having years of life lost due to HAP, mainly in the Balkan region (Albania, Bosnia and Herzegovina, North Macedonia and Montenegro), while most European countries have zero LYL, particularly several Nordic countries, including Finland, Sweden and Iceland, showed zero LYL due to HAP (Fig. 2B). As observed in Fig. 2, in both Europe and APAC, countries showed positive correlation between the proportion of HAP and the number of LYL due to HAP (r = 0.952, p < 0.0001 for EU, and r = 0.856, p < 0.0001 for APAC, see Figure A1 in SM).
[Figure 2A and 2B about here]
Additionally, LYL by IHD (for PM2.5 it varied from 35.06%-53.05% among APAC and from 37.05%-53.48% among European countries, while for HAP 31.32%-77.15% and 36.92%-66.25% were observed across countries in APAC and Europe respectively) and stroke (19.50%-36.34% and 18.28%-41.44% for PM2.5 in APAC and Europe, and 0.00%-42.14% and 0.00%-50.00% for HAP in APAC and Europe), in both continents, were the majority of causes of premature death, with three quarters, in comparison to COPD (4.21%-11.73% and 2.73%-7.77% for PM2.5 in APAC and Europe, and 0.00%-12.95% and 0.00%-3.98% for HAP in APAC and Europe), and TBL (3.36%-11.07% and 11.44%-21.46% for PM2.5 in APAC and Europe, and 0.00%-10.54% and 0.00%-19.66% for HAP in APAC and Europe). However, APAC experienced more premature deaths by COPD due to the two types of particulate matter, PM2.5 and HAP, than in European countries.
Cause-specific life-years lost attributable to particulate matter pollution across the socio-demographics.
To investigate the relation between socio-demographic index and different causes of death due to PM2.5 and HAP, countries were categorised into three groups, namely low, medium, and high SDI countries. Figure 3A top row shows results for APAC, there was no significant difference in LYL due to PM2.5 by stroke and IHD (Kruskal Wallis H test of H = 5.11, p = 0.078 for stroke, H = 5.45, p = 0.066 for IHD). However, significant differences between SDI groups for COPD (H = 10.04, p = 0.007) and TBL (H = 9.07, p = 0.011) are observed, although the latter was greater for the high as opposed to the low SDI. The analysis of LYL by COPD and TBL with Bonferroni test to compare cause-specific LYL attributable to PM2.5 in different SDI groups. As a result, there was a significant difference for SDI groups in LYL by COPD, t=-3.17, p = 0.005, with low SDI countries (\(\stackrel{-}{X}\)=0.12) having higher average LYL due to PM2.5 by this cause of death than high SDI countries (\(\stackrel{-}{X}\)=0.05). On the contrary, low SDI countries (\(\stackrel{-}{X}\)=0.03) had LYL in TBL lower than medium (\(\stackrel{-}{X}\)=0.07, t=-2.52, p = 0.035), and high SDI countries (\(\stackrel{-}{X}\)=0.06, t = 2.80, p = 0.015). Given the small values of the TBL results, any conclusion should be considered with caution.
Figure 3A bottom row shows that in terms of LYL attributable to HAP, the lowest SDI group was significantly different from the other groups (H = 39.35 for stroke, H = 43.23 for IHD, H = 41.82 for COPD and H = 34.81 for TBL, where p < 0.001 for all four causes). The group of countries with lower SDI had significantly larger LYL due to HAP than groups of countries with higher SDI for stroke (\(\stackrel{-}{X}\)=0.82, \(\stackrel{-}{X}\)=0.13 and \(\stackrel{-}{X}\)=0.02 for low, medium, and high SDI countries, respectively) and IHD (\(\stackrel{-}{X}\)=0.88, \(\stackrel{-}{X}\)=0.15 and \(\stackrel{-}{X}\)=0.02 for low, medium, and high SDI countries, respectively). Furthermore, the lowest SDI countries had more than tenfold and fivefold higher LYL than medium SDI countries for COPD (t=4.91, p<0.001) and TBL (t = 4.44, p < 0.001), and had over 138 times and 17 times higher LYL than the highest SDI countries for COPD (t=-6.27, p<0.001) and TBL (t=-5.74, p < 0.001). Consequently, Asia-Pacific countries with lower SDI countries have higher LYL attributable to PM2.5 and HAP than countries with higher SDI countries.
Similar results to those in APAC, were found for European countries with higher LYL among the lowest SDI countries when compared to the medium and high SDI countries. Significant differences between SDI groups due to PM2.5 for stroke (H = 18.50, p < 0.0001) and IHD (H = 18.35, p < 0.0001) are observed, while there was no difference in LYL by COPD and TBL; moreover, the lowest SDI countries had significantly higher LYL from the other groups (H = 21.79 for stroke, H = 22.49 for IHD, H = 14.50 for COPD and H = 15.62 for TBL, where p < 0.001 for all four causes). However, the disparity between LYL for household air pollution by SDI, are very small in Europe compared to the high gaps seen in APAC (see Fig. 3B).
[Figure 3A and 3B about here]
Air pollution, both indoor and outdoor, threatening human health and shortening life expectancy is a major global environmental issue. The impacts of particulate air pollution on mortality and morbidity have been studied in several countries, not only in highly polluted regions but also in countries with lower concentration[13, 21, 26, 27].
This research showed that the average life-years lost due to ambient particulate matter and household air pollution in APAC was higher than in European countries. This finding has been observed in previous results showing that the highest mortality and morbidity attributable to PM2.5 globally were recorded in Asia between 1990 and 2019, while Europe had relatively low PM2.5-related mortality over the same period[7].
The concentration level, both for PM2.5 and HAP, was highly correlated to LYL attributable to pollution. In simple terms, for countries with high pollution the average life years are shorten more than for low polluted countries. However, we found great heterogeneity in LYL attributable to PM2.5 and HAP across subregions. The highest number of LYL due to PM2.5 was found in Western Asia, in countries like the Syrian Arab Republic, Iraq, and Oman, while New Zealand and Australia, had the lowest number of years lost in the region. The emergence of particles smaller than 2.5 \(\mu\)g results from gasoline and diesel fuel engines[28]. Studies on air pollution in Western Asia found that most countries in this subregion increased the use of gas and diesel vehicles, industrial emissions, outdoor cooking and smoking habits in public spaces over the few decades[29, 30], resulting in high mortality due to the pollutant. For example, the Syrian Arab Republic had the highest LYL due to PM2.5 in the region, originating from high levels of outdoor air pollution, high smoking rates in public areas, and cooking outside[29]. In contrast, most high-income countries such as Australia, New Zealand and Japan reduced the use of these sources resulting in decreased number of mortality due to PM[30, 31]. Various Pacific Islands countries, such as Solomon Islands, Vanuatu, Kiribati and Papua New Guinea, had a high LYL due to HAP compared to other Asian countries. One of the significant health risks in the Solomon Islands was caused by indoor air pollution from wood fire ovens, which are the primary source of energy for cooking in households in the Islands[32]. On the contrary, Australia and New Zealand showed the lowest LYL, about zero lost years, due to this pollutant. The introduction of electric cooking, the use of rangehoods and having a separate space for kitchens in various high-income countries translated in a decline in mortality due to decreased emissions of air pollutants from solid fuel use[33, 34].
On the other hand, the highest LYL due to PM2.5 and HAP in European countries was observed in the Balkan Peninsula (North Macedonia, Bosnia and Herzegovina, Serbia and Albania), whereas LYL was low in the Nordic region (Sweden, Finland and Iceland). Differences in the region are emphasized in previous research, with Eastern European and Balkan nations, compared to the rest of Europe, still use conventional energy sources, owing to the reduction in solid fuel use in home cooking and heating[17, 35–37]. However, cooking with solid fuels at home was less than 5% across high-income European countries from 1990 to 2010[34]. In Europe, the dominant sources of air pollution are industrial activities, home fuel consumption, and transportation[38].
As a summary of cause-specific LYL attributable to PM2.5 and HAP, LYL by ischemic heart disease (IHD) and stroke, in both continents, played a major role in causes of premature death compared to chronic obstructive pulmonary disease (COPD), tracheal, bronchus and lung cancer (TBL) and other causes of death. Previous studies on the health impacts of particulate matter exposure concluded that IHD and stroke were the top two causes of mortality due to PM[39–41]. However, LYL by COPD due to both pollutants were shown more in APAC than in European countries. Traditional cooking from solid fuels is found in Asian population (Asian Street food), particularly in low- and middle-income countries. Studies have suggested that ingredients, cooking oil and cooking techniques are correlated with levels of indoor and outdoor air pollution; most Asian foods are cooked by traditionally stir-frying and deep-frying indoors and outdoors, with high temperature and oil leading to higher emissions than Western or European food[42–46].
In addition, great variability in causes of death attributable to air pollution across socio-demographic levels were observed in both continents. This analysis outlined significant differences between SDI groups for LYL by COPD and TBL due to PM2.5 in APAC. LYL in TBL for low SDI in the Asia-Pacific countries were lower than medium and high SDI countries, while the lowest SDI countries had more LYL by COPD than the highest SDI countries. One possible explanation of why low SDI countries had less LYL to PM2.5 for TBL in 2019 than high SDI countries is related to the development of lung cancer screening which it was particularly pronounced in countries with higher SDI where screening was first available [47, 48]. Also, premature death due to TBL may be confounded by smoking habits, which result in higher LYL for cancers in countries with higher smoking prevalence, than in middle and low-income countries where the population cannot afford the tobacco products with high levels of nicotine and toxicants[49]. For COPD, given the high-cost treatments and a lack of medical equipment, low-income countries may have relatively limited access COPD medications and diagnostic tests[43].
On the contrary, SDI groups significantly differed in LYL due to PM2.5 by stroke and IHD for European countries. This aligns with the studies in Europe that lower SDI groups had higher disability-adjusted life years due to PM2.5 for stroke and IHD than countries with higher SDI[17]. Although circulatory system diseases are the primary causes of mortality in most high-income European countries, the number of deaths from this disease has been declining leading to less mortality for these diseases than in lower-income countries[50]. Enhanced lung cancer screening and inexpensive medical treatments in TBL, also reduced this mortality in European low SDI countries until all premature deaths in low SDI nations were hardly different from high SDI countries[48].
According to several studies, people with lower SDI (e.g. lesser education, not being white-collar employees and lower income) are more vulnerable to air pollution-related mortality[21, 51]. One of the conceivable reasons is that coal combustion, agricultural burning, traditional cooking method and growth in the number of industries and motor vehicles in lower SDI nations have contributed to an increase in PM. Besides, the development of medical care and public health services is limited in countries with lower SDI, potentially leading to higher mortality due to environmental risk factors[47, 48].
Limitations of this study should be acknowledged. Firstly, this study was limited by a lack of death counts attributable to PM for ages between 0 and 24. Thus, the findings only showed the impact of air pollution on adults and elderly population up to age 95. Secondly, the impact from global warming, indirectly caused by air pollution, on population health have been shown in several studies[14, 16]; however, the GBD estimations do not include those. Third, the level of PM concentration should be estimated from burning for any purpose, including cooking and heating, but the level of HAP data from GBD were quantified as the fraction of the population in each country that cooks with solid fuels. Fourth, the results reported here are averages at the national level. Research of air pollution-related mortality at the subnational level and comparative research between rural-urban areas should be further assessed. Finally, only four causes of death were included in this study, but other causes of death, such as chronic kidney disease, diabetes and other cardiovascular diseases[7, 13], should be further investigated.
In summary, this study explores and compares cause-specific LYL attributable to PM in APAC and European countries in 2019. LYL attributable to ambient particulate matter and household air pollution from solid fuels in APAC were significantly higher than in Europe. Stroke and IHD were the top two cause-specific LYL due to the pollutants in the two regions. As a summary analysis of associations between socio-demographic levels and LYL due to air pollution, APAC and European countries in the lower SDI had more premature deaths due to PM2.5 and HAP than countries with higher SDI. Air pollution affects health globally, however, low SDI populations are more heavily affected than other groups, adding to the global health inequity. Understanding the negative impact of particle air pollution on human health is critical for improving environmental health policies and for introducing more effective interventions for protecting public health around the world.
The data for this study were obtained from the Institute for Health Metrics and Evaluation (IHME) and Health Effects Institute (HEI). We used the age and cause-specific number of death attributable and non-attributable to particulate matter, the number of populations, the age-specific probability of death and the socio-demographic index (SDI) in 2019 for the analysis from the Global Burden of Disease (GBD) Results Tool[52]. The estimated level of particulate matter in 2019 by country was obtained from the State of Global Air[53]. As this study used data from public sources and did not have personal information, ethics approval from our respective Institutional Review Board (IRB) was not required.
The study analysed 68 Asia-Pacific countries and 43 European countries. The continent classification by United Nations (UN)[54] was used. However, some Pacific Island nations in APAC were not included in the analysis because of unavailable information in the database (a country list is detailed in Table A1 in SM).
The data of the age-specific and cause-specific number of death attributable to particulate matter (PM) were estimated by IHME, which divides the death counts due to PM into two categories, ambient particulate matter (PM2.5) and household air pollution from solid fuels (HAP). These two types of pollution were used in the calculation for the study. In brief, IHME estimated exposure to PM using an annual concentration of pollution from satellite remote sensing, simulations of chemical transport model, ground station, land-use data, and estimation of population size. Additional details of data sources and methodology have been presented elsewhere[55]. From this source and due to the availability of data, we used the cause-specific mortality data attributable to pollution by 5-year age groups from age 25 to 95 in 2019 for each country in APAC and Europe[55]. The four most affected causes of death by particulate matter (including their International Classification of Diseases 10th codes): stroke (ICD10: G45-G46.8, I60-I63.9, I65-I66.9, I67.0-I67.3, I67.5-I67.6,I68.1-I68.2, I69.0-I69.3), ischemic heart disease (IHD, ICD10: I20-I25.9), chronic obstructive pulmonary disease (COPD, ICD10: J41-J44.9) and tracheal, bronchus and lung cancer (TBL, ICD10: C33-C34.9, D02.1-D02.3, D14.2-D14.3, D38.1) were included in this study. The total number of deaths for all causes of death non-attributable to PM2.5 and HAP were also used for LYL calculations. Additionally, we used the number of populations to calculate the average LYL due to pollution in the whole region.
As the component of the LYL analysis requires life table data for each country, the age-specific probabilities of death were used for calculating abridged life tables from age 25 to 95 in 2019 for each country, matching the cause-specific death data. As such, life expectancy and life years lost between those ages were attained.
This study focuses on two types of air pollution: ambient particulate matter and household air pollution from solid fuels in 2019. The data of annual mean concentrations of these two types of particulate matter were reported in the State of Global Air by HEI[56].
Socio-demographic index (SDI) data were obtained from IHME. Countries were divided into three SDI groups: low (SDI\(<\)0.561 for APAC and SDI\(<\)0.779 for Europe), medium (SDI=0.561–0.754 for APAC and SDI=0.779–0.868 for Europe) and high (SDI\(>\)0.754 for APAC and SDI\(>\)0.868 for Europe) SDI countries using interquartile range.
Life-years lost (LYL) is the complement of life expectancy and measures the number of years lost in the population before a fixed age, which is 95 years for this study. As the available age-specific data from IHME was for ages 25 to 95, life expectancy in this analysis was calculated in that range using standard demographic techniques[57], and LYL was calculated as the complement to life expectancy (life expectancy + LYL = 70 = 95 − 25). Moreover, the cause-specific LYL attributable to PM2.5 and HAP by affected causes of death: stroke, IHD, COPD and TBL was used to compare their impact on population health in this study.
The calculation of cause-specific life-years lost (LYL) attributable to particulate matter pollution can be summarised as:
where \({}_{70}{e}_{25}\) is the life expectancy between ages 25 and 95, and\({{}_{70}ə}_{25}^{}\left(j\right)\) is the number of life years lost by causes of death \(j\) from ages 25 to 95. Details on the calculation procedures for the LYL have been presented elsewhere[19, 23, 25].
Differences in the cause-specific LYL attributable to particulate matter between three groups of socio-demographic index (SDI) countries
The confidence interval (CI) of the average LYL due to PM2.5 and HAP was calculated using an alpha value of 0.05 or Z-value at 95%[58]. Due to the assumptions for one-way ANOVA were not met, the Kruskal-Wallis H test by ranks was used to determine the significant difference in LYL attributable to PM2.5 and HAP among three groups of SDI countries[59].
A comparison analysis by H test and Post Hoc test used a significant value at 0.05. If the Kruskal-Wallis test shows a significant difference in LYL between three groups of SDI countries, multiple comparison tests with Bonferroni test[60] were added to determine statistical significance between each pair of subgroups. All results in this study were calculated and presented by using the R Statistical software version 1.3.109. Details of the additional calculations are provided in the supplementary information.
Data availability
The datasets used for analysing life-years lost attributable to ambient particulate matter and household air pollution, and socio-demographic index for each country are publicly available through the Global Burden of Disease Study 2019 (GBD 2019) repository at https://ghdx.healthdata.org/gbd-2019. The datasets regarding the estimated level of ambient particulate matter and household air pollution in 2019 by country were obtained from the State of Global Air 2020 which can be downloaded at https://www.stateofglobalair.org/data/. Other data generated or analysed during this study are included in this published article and its supplementary information files.
No competing interests reported.
posted
You are reading this older preprint version