Estimation of Mortality and Hospital Admissions Attributed to Criteria Air Pollutants in Tehran Metropolis , Iran ( 2013 – 2016 )

Health impact assessments are useful for governmental authorities and decision-makers to determine the need for action and address potential public health problems arising from exposure to air pollution. The present study was conducted to assess the short-term health impacts of ambient air pollution in Tehran using the AirQ 2.2.3 model for March 2013–March 2016. Hourly concentrations of PM10, PM2.5, O3, NO2 and SO2 were acquired from the Department of Environment (DOE) and Tehran Air Quality Control Company (TAQCC). Air pollution data was validated according to the USEPA criteria, and only valid monitoring stations for each of the three years were entered to the AirQ 2.2.3 model. The pollutant concentrations were lower in the March 2015-March 2016 period compared to the previous years. The three-year average (± standard deviation) of PM10 and PM2.5 concentrations were 80.21 (± 34.21) and 39.17 (± 17.26) μg m, respectively. The three-year averages (± standard deviation) for ozone (O3), nitrogen dioxide (NO2), and sulphur dioxide (SO2) were 54.88 (± 24.15), 103.97 (± 25.88) and 39.84 (± 11.17) μg m, respectively. The total estimated number of deaths attributed to PM10, PM2.5, O3, NO2 and SO2 over these three years were 4192, 4336, 1363, 2830, and 1216, respectively. The health impacts attributed to all pollutants except for PM10 were estimated to decrease in 2016, compared to the prior years. However, the air quality in Tehran still poses significant risks to public health. In conclusion, urgent efforts are needed such as mandating the replacing of old and poorly functioning vehicles from the roadways in order to reduce the health burden that air pollution is currently imposing on this city.


INTRODUCTION
The World Health Organization (WHO) reported that in 2012 about 7 million deaths could be attributed to exposure to air pollution, of which 3.7 million could be the result of exposure to ambient air pollution (WHO, 2014).Epidemiological studies have shown positive associations between air pollutants such as particulate matter of aerodynamic diameter less than 10 µm (PM 10 ), nitrogen dioxide (NO 2 ), sulphur dioxide (SO 2 ), ozone (O 3 ) and carbon monoxide (CO) and adverse health outcomes (Goldberg et al., 2013;Crouse et al., 2015;Fischer et al., 2015;Liu et al., 2015;Wong et al., 2015;Yorifuji et al., 2016).Many studies have been conducted to assess the health impacts of air pollution in various cities of the world (Boldo et al., 2006;Fattore et al., 2011;Orru et al., 2012;Allen et al., 2013), including Iranian cities (Naddafi et al., 2012a;Gholampour et al., 2014;Ghozikali et al., 2016;Mohammadi et al., 2016).These results could be of interest to authorities and decision-makers (Fattore et al., 2011;Likhvar et al., 2015).
The World Bank reported that the economic cost of air pollution has increased significantly from 1990 to 2013, in which the total welfare losses due to premature deaths from exposure to air pollution increased by 94 percent.The welfare losses due to air pollution in the Middle East and North Africa in 2013 were reported to be 154 billion USD, an increase of 108% compared to 1990.In addition, the costs due to exposure to PM 2.5 in the Middle East and North Africa have increased from 62 to 141 billion USD in the same period (World Bank, 2016).Due to the contribution of air pollution to decreased public health and its economic impact on the country, quantification of possible mortality and hospitalizations is useful in assessing the likely damages.

BACKGROUND
The AirQ 2.2.3 model has been developed by the WHO European Centre for Environment and Health, Bilthoven Division, and estimates the health effects of air pollutants such as mortality and hospital admission among people within a specific area and over a given time period (Fattore et al., 2011).
Multiple studies have been conducted to estimate the health impacts of various air pollutants using the AirQ 2.2.3 model.Kermani et al. (2016a) investigated PM 2.5 concentrations and its health effect in Tehran during 2005-2014.The annual mean concentrations of PM 2.5 during these 10 years were 34.92, 30.29, 33.09, 41.40, 38.38, 47.02, 47.31, 40.75, 43.26 and 36.15,respectively.All the annual concentrations were higher than the WHO's guideline.In addition, the total attributable deaths during the whole period were estimated to be 20015 cases (Kermani et al., 2016a).In another study in Tehran, the number of cardiovascular and respiratory hospital admissions due to exposure to PM 10 during 2005-2014 was estimated using the AirQ 2.2.3 model.The total cases of cardiovascular and respiratory hospital admissions in the whole period were reported 20990 and 54352, respectively (Kermani et al., 2016b).Naddafi et al. (2012a) assessed the health impacts of PM 10 , NO 2 , SO 2 , and O 3 in Tehran during 2010.The number of all-cause, cardiovascular, and respiratory deaths, and cardiovascular, respiratory, and chronic obstruction pulmonary disease hospital admissions, and acute myocardial infarctions were estimated.They concluded that the magnitude of the health impacts estimated for Tehran underscores the need for urgent action to reduce the health burden of air pollution (Naddafi et al., 2012a).
Another study estimated health outcomes for cardiovascular and respiratory mortality attributable to O 3 and NO 2 using the AirQ program.The results showed that the total cumulative number of mortalities attributed to NO 2 and O 3 were 1593 and 946, respectively, that represents about 2.66% and 1.58% of total non-accidental mortality in Tehran (Kermani et al., 2016c).In general, few studies exist about the health impacts of particulate air pollution, especially PM 2.5 .However, the majority of these studies have a serious weakness in methodology because they used the old WHO criteria for validating monitoring stations.
The numbers of hospital admissions and mortality related to NO 2 in five Iranian cities (Mashhad, Tabriz, Shiraz, Isfahan and Arak) in 2011-2012 were estimated using the AirQ model.The highest numbers of estimated adverse health outcomes were in Mashhad and Isfahan.These values likely result from the increasing number of vehicles with related traffic, fuel usage, and high levels of temporary and permanent population in religious and tourist sites (Asl et al., 2015).Miri et al. (2016) estimated the mortality and morbidity resulting from exposure to ambient air pollution in Mashhad metropolis using AirQ model.The attributable proportion of total mortality values attributed to exposure to PM 10 , PM 2.5 , SO 2 , NO 2 and O 3 were, respectively, 4.24%, 4.57%, 0.99%, 2.21%, 2.08%, and 1.61% of the total deaths occurring in the year of study (Miri et al., 2016).Goudarzi et al. (2015) estimated the all-cause and respiratory mortality attributable to PM 10 in Ahvaz city during 2009.Their analysis predicted 1165 all-cause deaths and 115 respiratory deaths annually for each 10 µg m -3 increase in PM 10 .They concluded that the high number of deaths could be the result of higher PM 10 average or because of sustained highconcentration days in Ahvaz (Goudarzi et al., 2015).There are other studies about the application of AirQ 2.2.3 for a variety of air pollutants in different Iranian cities (Goudarzi, 2014;Geravandi et al., 2015;Nourmoradi et al., 2015;Ghozikali et al., 2016;Mohammadi et al., 2016;Nourmoradi et al., 2016;Khaniabadi et al., 2017).However, there has not been a recent analysis of the impacts of each of the criteria pollutants on health outcomes in Tehran, Iran's most populous city.
The present study was conducted to assess the shortterm health impacts of outdoor air pollution in Tehran using the AirQ model for the March 2013-March 2016 period.

Location and Time
Tehran is the capital of Iran located at latitude and longitude of 35° 41'N, 51° 25'E.It has about 9 million inhabitants and 3 million personal vehicles (Shahbazi, 2015).

Data Collection
Hourly concentrations of PM 10 , PM 2.5 , O 3 , NO 2 and SO 2 from 21 March 2013 to 19 March 2016 (three years in the Persian calendar) were acquired from the Department of Environment (DOE), and Tehran Air Quality Control Company (TAQCC).The locations of monitoring stations are illustrated in Fig. 1.Relative risk values were adopted from several European meta-analyses that have been used in previous studies regarding health impact quantification in Iran (WHO, 2000;Anderson et al., 2004;Gryparis et al., 2004;Samoli et al., 2006).The number of deaths for all mortality was obtained from National Organization for Civil Registration of Iran.Baseline incidence (BI) for nonaccidental mortality was calculated using the ratio of accidental deaths given by Ministry of Health and Medical Education of Iran (Khosravi, 2016).In addition, BI values for cardiovascular and respiratory mortality were calculated by multiplying the reported non-accidental mortality value by the ratio of cardiovascular and respiratory deaths given by Ministry of Health and Medical Education.BI values for cause-specific hospital admission were taken as the default values in AirQ 2.

Data Validation
The initial filtering of the data included the deletion of zero and negative data.The concentrations of the gaseous pollutants were converted to µg m -3 (Boguski, 2006).Then, EPA's criteria for data completeness requirements was used to determine valid stations for entering the model (USEPA, 2015).After assessment of the data from each station, only qualified data sets were selected.Finally, 24-hour average values were calculated for PM 10 , PM 2.5 , and SO 2 .In addition, eight-hour moving average and one-hour average were calculated for O 3 and NO 2 , respectively.Only valid stations common in all the three years were selected to enter the model, and the others were excluded.The overall 24-h city-wide averages were calculated from the 24-h averages of all of the included stations on each day.The frequency of days within the AirQ-defined concentration ranges was calculated for the model.

AirQ Software 2.2.3
WHO has developed and released AirQ 2.2.3 to assess the short-and long-term health effects of ambient air pollution.This program calculated the attributable proportion, the attributable cases, the attributable cases per 100,000 persons, and the proportion of cases in each concentration range.Epidemiological studies are the sources of the relative risk values and the concentration-response functions used in the model.The required model inputs include the annual and seasonal mean and maximum, the 98 th percentile, the number of days within each specified concentration range, demographical information, baseline incidence, and relative risk values.

The Concentration of Pollutants
The descriptive statistics of air pollutant concentrations including PM 10 , PM 2.5 , O 3 , NO 2 , and SO 2 are illustrated in Fig. 2. The average (± SD), maximum, 98th percentile, and seasonal mean (± SD) and maximum values of PM 10 , PM 2.5 , O 3 , NO 2 , and SO 2 are presented in Table 1.The three-year mean values (± standard deviation) of PM 10 and PM 2.5 were 80.21 (± 34.21) and 39.17 (± 17.26) µg m -3 , respectively.By comparing the mean concentrations from Table 1, it can be seen that the PM 10 concentrations have decreased during the three-year period.In addition, PM 2.5 concentrations have decreased over these three years.Naddafi et al. (2012a) reported that the average concentration of PM 10 in Tehran during 2010 was 90.6 µg m -3 .The annual average of PM 10 values were 1.3 and 4.5 times higher than the world-wide average (71 µg m -3 ) for 2011 (WHO, 2011), and the WHO guideline values (Naddafi et al., 2012a), respectively.The high PM concentrations in Tehran are accounted by TAQCC.They indicated that there are about 3 million personal vehicles in Tehran, of which 25% are more than 10 years old and 75% have emissions with Euro-2 standard and less (Shahbazi, 2015).About 70% of particulate matter in Tehran during 2015 was emitted from mobile sources (Ahadi, 2016).In addition, dust from Middle Eastern dust storms affects the particulate concentrations in Tehran (Sowlat et al., 2012;Sowlat et al., 2013).
The three-year mean values (± standard deviation) for O 3 , NO 2 and SO 2 were 54.88 (± 24.15), 103.97 (± 25.88) and 39.84 (± 11.17) µg m -3 , respectively.The average concentration of O 3 in the third year decreased substantially compared to the other years by 36% and 24% reduction from year 2 and year 1, respectively.The mean NO 2 concentration during the third year was lower than the similar periods in previous years.Mobile sources, energy production, and domestic sector are responsible for about 46%, 24%, and 23% of total NO x emission in Tehran (Ahadi, 2016).Also the results indicated that the March 2015-March 2016 year had a lower SO 2 concentration compared to the first (24% lower) and second year (9% lower).The dominant source of emitted SO x in ambient air of Tehran is the energy production sector, which reflects the use of fossil fuels (Ahadi, 2016).Naddafi et al. (2012a) reported that the average concentration of O 3 , NO 2 , and SO 2 were in Tehran in 2010 were 68.82, 85.00 and 89.16 µg m -3 , respectively (Naddafi et al., 2012a).Since AirQ model estimates the health effects based on the number of days in concentration ranges, the related charts (Figs.S1 and S2) are provided in Supplementary file.

Health Impacts
The number of total, cardiovascular and respiratory mortality and cardiovascular and respiratory hospitalization attributed to short-term exposure to PM 10 in concentrations above 10 µg m -3 is presented in Table 2.The total number of deaths attributed to PM 10 over these three years was 4192.The PM 10 health effects in March 2015-March 2016    (Ahadi, 2016).This can be due to the use of valid monitoring stations' datasets in this study.Naddafi et al. (2012a) conducted a similar study on the health impacts of Tehran's ambient air pollutants (PM 10 , O 3 , NO 2 and SO 2 ) for concentrations above 10 µg m -3 in 2010 using AirQ 2.2.3.The results indicated that the total mortality attributable to PM 10 , O 3 , NO 2 and SO 2 was 2194, 819, 1050 and 1458, respectively.The cardiovascular mortality attributable to PM 10 , O 3 , NO 2 and SO 2 was 1367, 574, 591 and 1202, respectively.The excess respiratory deaths attributable to PM 10 , O 3 and SO 2 were 402, 299 and 310, respectively.They also reported that PM 10 caused 2580 and 6677 cases of cardiovascular and respiratory hospital admissions, respectively.The number of hospital admissions attributed to O 3 , NO 2 and SO 2 was reported 424, 247 and 298, respectively.Furthermore, about 305 and 556 excess cases of acute myocardial infarction were attributed to NO 2 and SO 2 concentrations, respectively (Naddafi et al., 2012a).Air pollution in Mashhad, the second largest metropolis in Iran was estimated to cause total mortality due to PM 10 , PM 2.5 , O 3 , NO 2 and SO about 557, 600, 130, 290 and 274 cases per year, respectively.PM 2.5 showed the highest attributable deaths (Miri et al., 2016), mainly due to the higher relative risk value and a high PM 2.5 /PM 10 ratio (0.49).Boldo et al. (2006) assessed the health impacts of longterm exposure to PM 2.5 in 23 European cities.About 16,926 premature deaths from all causes, including 11,612 cardiopulmonary deaths and 1901 lung-cancer deaths, could be prevented annually if long-term exposure to PM 2.5 levels were reduced to 15 µg m -3 in each city (Boldo et al., 2006).Fattore et al. (2011) estimated short-and long-term effects of air pollution in two municipalities in an industrialized area of Northern Italy using AirQ 2.2.3.They reported that in case of short-term exposure, PM 2.5 showed the highest health impact on the 24,000 inhabitants, causing an excess of eight deaths out of 177 annually.O 3 and NO 2 each caused about three excess cases of total mortality.The results on long-term effects showed 433, 180, and 72 years of life lost (YLL) for mortality for all-cause, cardiopulmonary diseases and lung cancer, respectively in a year (Fattore et al., 2011).In a study in 13 Italian cities with about 13 million inhabitants, the health impacts of PM 10 and O 3 were estimated.The authors reported that chronic exposure to PM 10 in adults older than 30 years caused about 8220 excess non-accidental deaths in concentrations above the reference value of 20 µg m -3 .In addition, 1372 deaths are attributed to short-term exposure to PM 10 .Furthermore, a total of 516 premature deaths from all causes (0.6% of total acute mortality), excluding accidents are attributable every year to O 3 (Martuzzi, 2006).
WHO has estimated that the total mortality attributed to PM 2.5 concentrations in Iran were 26267 during 2014.In addition, the years of lost life (YLLs) and disability adjusted life years (DALY) were 703207 and 726027 years, respectively.However, that study had serious limitations such as considering a constant value of concentration for all the cities and populations (WHO, 2016).The World Bank announced that the total deaths due to air pollution in Iran in 1990 and 2013 were 17035 and 21680 cases, respectively.The related total welfare losses and the contribution in GDP were estimated to be 30.6 billion USD and 2.48% in 2013, respectively.Furthermore, the total lost labor output and its contribution in Iran's GDP were 1471 billion USD and 0.12%, respectively (World Bank, 2016).
About 85% of the total mass of pollutants are emitted from mobile sources (Shahbazi, 2015).Thus, improvement in policies, regulations, traffic plans, fuel substitution, etc., and consequence reduction in emitted pollutants can show a significant effect on decreasing health impacts and economic costs of air pollution in Tehran.

CONCLUSIONS
Short-term health impacts of Tehran's air pollutants (PM 10 , PM 2.5 , O 3 , NO 2 , and SO 2 ) in concentrations above 10 µg m -3 were assessed using AirQ 2.2.3 model in the period of 21 March 2013 to 19 March 2016.Only monitoring stations with all three years of valid data were included in the study.The results showed that the health impacts of PM 2.5 , O 3 , NO 2 and SO 2 have decreased in March 2015-March 2016 in comparison to the previous year.There were likely to be significant public health impacts from exposure to high concentrations of air pollutants in Tehran, and urgent efforts are required such as mandating the removal of old and poorly functioning vehicles from the roadways in order to reduce the burden of air pollution is currently placing on this city.
Three one-year periods were considered in this study.The first period is from 21 March 2013 to 20 March 2014.The second period is from 21 March 2014 to 20 March 2015.Also, the third period is between 21 March 2015 and 19 March 2016.
2.3.The populations of Tehran during these three periods were obtained from Statistical Centre of Iran.The city population in March 2013-March 2014, March 2014-March 2015, and March 2015-March 2016 periods were 8209730, 8652820, and 8866500, respectively.

Fig. 1 .
Fig. 1.Location of ambient air monitoring stations of Tehran.

Table 2 .
(Forouzanfar et al., 2016)utable to short-term exposure to air pollutants above 10 µg m -3 , Tehran(2013)(2014)(2015)(2016). to its previous years.This is due to the increase in Population, which has neutralized the effect of reduction in PM 10 concentration.In case of PM 2.5 , the attributable total number of deaths is 4336 cases.The total mortality decreased constantly from 2013 to 2016.The mortality in March 2015-March 2016 showed a reduction of 9% in comparison to March 2013-March 2014.Particulate matter has been implicated as the seventh leading risk factor for premature death and disability in Iran(Forouzanfar et al., 2016).The total number of deaths attributable to O 3 for the 3year period was 1363.Total mortality in the March 2015-March 2016 period decreased 38% and 28% compared to the March 2013-March 2014 and March 2014-March 2015 periods, respectively.The same pattern can be detected in other related health effects of O 3 concentrations.The total cases of mortality attributable to NO 2 were 2830.The increase in health impacts of NO 2 is related to the population increase in March 2014-March 2015 comparing to its previous year.The total number of deaths attributable to SO 2 was 1216 cases.The SO 2 's health effects have decreased constantly from 2013 to 2016.For total mortality, PM 2.5 and PM 10 have shown the highest health impacts in the March 2013-March 2014, March 2014-March 2015, and March 2015-March 2016 periods.In addition, PM 10 was responsible for most of the cardiovascular and respiratory deaths in each of the three years.For COPD hospital admissions, O 3 and NO 2 have the highest values for March 2013-March 2014, March 2014-March 2015, and March 2015-March 2016.The health impacts attributed to all pollutants except for PM 10 have decreased for March 2015-March 2016 compared to the previous year.The reduction in concentration and health effects of air pollutants may be different from national reports regarding Tehran's air pollution