Previous studies have demonstrated an association between short- and long-term exposure to air pollution and adverse health outcomes.1–4 This evidence, however, is based on observational studies, which sometimes face challenges dealing with bias. For example, in cohort studies that provide evidence on health effects from long-term exposure to air pollution, researchers should adjust for potential confounders in the statistical model, such as smoking, diet, socioeconomic status, or location-related determinants of health.5 However, these potential confounders are sometimes not measured, which harms the interpretation of the findings from the observational studies.
To overcome these situations, a quasi-experimental study design is advocated,5 which examines the effect of decline in air pollution from regulatory actions6–9 or unplanned events (e.g., plant closure,10 strike,11 policy for heating,12 or political change such as reunification13), but the number of these studies is still limited. In addition, when conducting such quasi-experimental studies, appropriate counterfactual scenarios are needed14 (i.e., a population benefiting from the regulations should be compared with an external population in an area where specific regulation is not conducted). However, appropriate external groups are difficult to find.
In recent years, the Japanese Government has mandated a series of regulations to reduce levels of air pollution, including the Automobile NOx (nitrogen oxides)/PM (particulate matter) Act in 2002, which prohibits registration of noncompliant vehicles in metropolitan areas (including Tokyo or Osaka mentioned below; eTable 1; https://links.lww.com/EDE/B91). The Tokyo Metropolitan Government (TMG), the capital city of Japan, and the neighboring three prefectures (Saitama, Chiba, and Kanagawa) additionally introduced an ordinance in October 2003 requiring any diesel vehicles traveling in their area to conform to their standards for PM. Subsequently, the TMG and Saitama Prefecture further tightened these standards in April 2006. Conversely, the Osaka Prefectural Government, the second largest city in economy in Japan after the TMG, delayed the regulation and introduced a more relaxed ordinance in January 2009 requiring any vehicles departing from or arriving at their area to conform to their standards for NOx and PM.
Using these situations, we evaluated the effects of the local diesel emission control, which likely reduced PM2.5 (PM less than 2.5 µm in diameter) concentrations, on standardized mortality rates among residents in Tokyo. In the previous study,15 we evaluated the effect of enforcement in 2006 adjusting for the change in mortality rate in the rest of Japan as a reference population and showed apparent benefits on cerebrovascular disease mortality but not on other cause-specific mortality. In the present study, we expanded the study period and attempted to adopt a more appropriate reference population. When examining the hypothesis, we adjusted for the change in mortality rates among residents in Osaka, where further regulations were not taken in place until recent years, as a reference population.
The TMG is divided into three areas: 23 central urbanized wards, surrounding cities, and islands (Figure 1). We analyzed the data collected from residents in the 23 urbanized wards (population 8,489,653 according to the 2005 census) who died between October 2000 and September 2012 (N = 702,845). Because decline in PM2.5 was anticipated after the diesel regulation in the TMG in October 2003, we started the study period from 3 years before the introduction of the regulation (October 2000). We also focused on Osaka in the Osaka Prefectural Government as a city with late introduction of the regulation, and analyzed the data collected from residents in the city who died during the same study period (N = 287,022). Tokyo’s 23 wards and Osaka are the first and second most densely populous areas in Japan (population density in 2005 >10,000 persons per km2) and, as shown in eTable 2 (https://links.lww.com/EDE/B91), both are urbanized. In addition, the contributions of motor vehicles to PM2.5 are similar, e.g., local studies estimated that they were 11.6% in Tokyo in 2008 and 8%–14% in Osaka in 2009 except secondary particles.16,17
Air Pollutants Data
In the present study, we used concentrations of nitrogen dioxide (NO2) and PM2.5 as the main air pollution exposures because the regulation by the Japanese Government, the TMG, and Osaka Prefecture attempted to reduce levels of NOx or PM. We obtained daily mean concentrations at one general station (not affected by specific sources)18 in each study area that has measured PM2.5 from April 2001, but not from October 2000, owing to improved availability of data as of April 2001. A single station in Tokyo’s 23 wards is located about 12 km from the centroid of the 23 wards and a 28-km buffer from the station covers the entire area of the wards, while a station in Osaka is located about 3.6 km from the centroid of the city and a 15-km buffer from the station covers the entire area of the city. We obtained the daily air pollutant data from the Bureau of Environment of the TMG and the Research Institute of Environment, Agriculture and Fisheries, Osaka Prefecture, supplemented by the database of the National Institute of Environmental Studies.
The Ministry of Health, Labor, and Welfare in Japan provided electronic data on all deaths in Tokyo’s 23 wards and Osaka with causes of death during the study period, stripped of names and addresses. Underlying causes of death were coded according to the International Classification of Diseases, Tenth Revision (ICD-10). We counted daily numbers of deaths from all causes (nontrauma, A00-R99), cardiovascular disease (I10–I70), ischemic heart disease (I20–I25), cerebrovascular disease (I60–I69), pulmonary disease (J00–J99), lung cancer (C33–C34), and other causes (deaths excluding cardiovascular and pulmonary disease from nontrauma deaths). To compare mortality rates during the study period as well as between two study areas, we calculated daily age-standardized mortality rates. We first calculated daily all-cause and cause-specific mortality rates for each age category (5-year intervals up to 85 years, and over 85 years) in each area divided by daily population estimated by linear interpolation using age-specific census population counts (1995, 2000, 2005, and 2010) and number of population reported by each local Government. Then, we multiplied the daily mortality rates for each age category by the 2005 population census in each area as the weights to calculate the expected number of deaths. The overall sum of daily expected cases for each age category was divided by the total number of 2005 population census to calculate the daily standardized mortality rates. We did not adjust for sex.
We divided the study period (October 2000 to September 2012) into four intervals as follows because an ordinance was introduced in October 2003: October 2000 to September 2003 (reference period); October 2003 to September 2006; October 2006 to September 2009; and October 2009 to September 2012. We compared concentrations of pollutants with age-standardized mortality rates between the 3-year intervals in Tokyo’s 23 wards and Osaka.
To examine the changes in age-standardized mortality rates between the 3-year intervals in Tokyo’s 23 wards, we conducted interrupted-time series analysis7 using generalized Poisson regression models. We regressed the log of age-standardized mortality rates in Tokyo on the indicator of the 3-year intervals and estimated rate ratios of mortality for the other three intervals compared with the interval from October 2000 to September 2003 (reference period). We expressed percent changes in mortality by rate ratio minus one multiplied by 100.
First, to take into account the trends in mortality rates in the reference population (i.e., residents in Osaka), we weighted daily age-standardized mortality rates in Tokyo’s 23 wards with the daily trends in age-standardized mortality rates in Osaka. Specifically, we obtained weighted daily age-standardized mortality rates in Tokyo’s 23 wards using the following formula:
where rateTi is the age-standardized mortality rates in Tokyo’s 23 wards on dayi; rateO1Oct2000 is the age-standardized mortality rates in Osaka at the start of the study period (October 1, 2000); and rateOi is the age-standardized mortality rates in Osaka on dayi. By doing this, we could incorporate the daily trends in mortality in the reference population into the daily weighted standardized mortality rates in Tokyo’s 23 wards. We then adjusted for day of the week and public holidays as dummy variables, daily number of influenza patients obtained from the Tokyo Metropolitan Infectious Disease Surveillance Center as a continuous variable, and same-day average temperature and same-day relative humidity using a smoothing function. We obtained daily average temperature and relative humidity in Tokyo’s 23 wards from the Japan Meteorological Agency and used restricted cubic splines with five knots for temperature and three knots for humidity.
Adjusting for long-term secular trends in the mortality rates in the reference population when quantifying health benefits associated with air pollution controls is challenging and we used a sort of weighting scheme for doing this. Instead, previous studies7,19 adjusted for the secular trend by directly putting the mortality rates in the reference population as a covariate or by putting the secular trend as a covariate after smoothing the rates in the reference population, although the former approach may lead to biased estimates.19 In the sensitivity analyses, therefore, instead of weighting mortality rates in Tokyo’s 23 wards by mortality rates in Osaka, we directly adjusted for the daily age-standardized mortality rates in Osaka or we adjusted the trend in mortality rates in Osaka using lowess smoothing with windows of 30, 90, and 150 days. We chose the number of windows based on a simulation that showed an optimal window of 1 to 5 months to remove the extreme values but to leave the seasonal pattern.19
In further sensitivity analyses, we used restricted cubic splines with seven knots for temperature and five knots for humidity to check robustness of the degrees of freedom. We also changed the reference period (from 2000–2003 to 2003–2006), so as to control more completely for potential confounding by smoking, and we calculated changes in mortality rates comparing the periods from October 2006 to September 2009 and from October 2009 to September 2012 versus the period from October 2003 to September 2006. The rationale for the further analysis was to remove the potential effects of a downward trend in smoking. According to the Comprehensive Survey of Living Conditions conducted by the Ministry of Health, Labor, and Welfare, Japan,20 smoking prevalence (smoking daily or sometimes) in 23 Tokyo wards was 32.5% in 2001, 28.1% in 2004, 26.2% in 2007, and 21.2% in 2010, while that in Osaka was 28.8% in 2001, 28.9% in 2004, 27.8% in 2007, and 23.8% in 2010. Thus, a pattern of decline in smoking prevalence since 2001 was a little different between the Tokyo wards and Osaka but the prevalence in 2004 and the decline from 2004 to 2010 was similar.
All confidence intervals were calculated at the 95% level. Stata statistical software (Stata SE version 13; Stata Corp., College Station, TX) was used for all analyses.
The analyses of national data (unlinkable anonymized data) are considered to be exempt from the need for ethical review according to the Ethical Guidelines for Epidemiological Research in Japan.
Trends in air pollutants in Tokyo’s 23 wards and Osaka are shown in Figure 2 and monthly trends are shown in eFigure 1 (https://links.lww.com/EDE/B91). Decline in NO2 during the study period was similar in the two areas, while decline in PM2.5 was larger in Tokyo’s 23 wards (from 24.4 μg/m3 in the interval from October 2000 to September 2003 to 16.2 μg/m3 in the interval from October 2009 to September 2012) than in Osaka (22.7 to 18.9 μg/m3 during the same period) and the difference in the decline was 4.5 μg/m3 (eTable 3; https://links.lww.com/EDE/B91).
We show age-standardized mortality rates in Tokyo’s 23 wards and Osaka separated by 3-year intervals in Table 1. Declines in mortality rates compared with the first 3-year interval were larger in Tokyo’s 23 wards compared with Osaka except for mortality from other causes.
In the interrupted time-series analyses (Table 2), we observed declines in all-cause and cause-specific mortality in recent intervals in the Tokyo wards in the crude model. These remained when we weighted by daily age-standardized mortality rates in Osaka. In the fully-adjusted model, percent changes in mortality between the first 3-year interval (October 2000 to September 2003) and the last 3-year interval (October 2009 to September 2012) were −6.0% for all causes, −11% for cardiovascular disease, −10% for ischemic heart disease, −6.2% for cerebrovascular disease, −22% for pulmonary disease, and −4.9% for lung cancer, but we did not observe decline in mortality from other causes.
When we adjusted for the daily age-standardized mortality rates in Osaka directly or using lowess smoothing (Table 3), the results when we used lowess smoothing with a window of 150 days were close to the main findings. The percent changes in all-cause and cardiovascular mortality were the smallest when we used lowess smoothing with a window of 150 days, while those in respiratory and lung cancer mortality were the smallest when we directly adjusted the rates. However, the declines in mortality were consistently observed in recent intervals in Tokyo’s 23 wards in both analyses.
In additional sensitivity analysis, even if adjusted for temperature and humidity using different degrees of freedom, the results did not change substantially. When we changed the reference period in the sensitivity analysis, we still observed declines in all-cause and cause-specific mortality in Tokyo’s 23 wards (Table 4).
We evaluated the effects of the local diesel emission control ordinance on standardized mortality rates among residents in Tokyo’s 23 wards, compared with the change in mortality rates in a reference population (residents in Osaka), which introduced such a regulation later. We found that the decline in PM2.5 and the improvement in age-standardized mortality rates related to cardiorespiratory disease and lung cancer were greater in Tokyo’s 23 wards compared with Osaka during the study period.
Until now, several quasi-experimental studies examined the effect of decline in air pollution from regulatory actions6–9 or unplanned events10–13 and our study is similar to the former types of the studies, in particular those examining the impact of traffic-related regulations. Some European studies evaluated the impact of registrations to reduce traffic congestions or volumes and showed some apparent benefits on years of life gain21,22 or hospital admissions for bronchitis.23 Several studies also examined the benefit of regulations including traffic-related ones implemented during Olympics9,24–26 and international games.27 Although the benefit related to the Atlanta Olympics was equivocal,24,26 changes in air pollution levels during the Beijing Olympics were associated with changes in some biomarkers related with systematic inflammation, thrombosis, and pulmonary inflammation.9,25 The present study is in line with the findings from these previous studies.
In our previous study,15 we evaluated the effect of enforcement in 2006 during the period from April 2003 to December 2008 directly adjusting for the change in mortality rate in the rest of Japan as a reference population and showed some benefits on cerebrovascular disease mortality but not on all or other cause-specific mortality. In the present study, we expanded the study period (from October 2000 to September 2012) to evaluate the introduction of the regulation in 2003, adopted a more appropriate reference population, and controlled more appropriately the secular trends in mortality, which would make the benefit of the control more apparent.
The observed larger decline in PM2.5 in Tokyo’s 23 wards compared with Osaka may have been driven by the diesel emission control in the TMG and neighboring three prefectures, which is a long-term intervention that has lasted for several years, as opposed to short-term regulations such as those introduced during the Olympic Games.9,26 According to the TMG,28 diesel vehicles that did not meet the standards were required to be replaced with new compliant vehicles or have a device fitted to reduce the emission of PM. New vehicles had a 7-year grace period to meet the obligation. All noncompliant vehicles are prohibited from traveling in the Tokyo area and the TMG regularly inspects diesel vehicles on the roads to monitor compliance, which has resulted in high compliance (over 97%).29 In addition, the TMG and the Saitama Prefecture enforced the standard for PM emission in April 2006 (eTable 1; https://links.lww.com/EDE/B91). Such a regulation in the TMG and the neighboring prefectures should have led to a larger but step-by-step decline in PM2.5 in Tokyo’s 23 wards compared with Osaka. But, as mentioned, the regulation is a long-term one and not so abrupt (e.g., a 7-year grace period for new vehicles), it may have taken some time to observe a lager decline in PM2.5 in Tokyo’s 23 wards (Figure 2).
An effectiveness of the control was evaluated by a local study which showed that elemental carbon, a component of PM2.5 and often used as a traffic-exposure surrogate,30 emitted from diesel vehicles decreased by 44% after the introduction of the regulation.31 Indeed, concentrations as well as percentages of elemental carbon in PM2.5 have decreased in Tokyo’s 23 wards during the study period (eFigure 2; https://links.lww.com/EDE/B91), e.g., concentrations of elemental carbon in the station where we obtained NO2 and PM2.5 data for this study were 6.6 µg/m3 in 2001 and 1.14 µg/m3 in 2012.32,33 Moreover, the TMG estimated that net PM emissions by motor vehicles decreased from 3,180 tons in 2000 to 710 tons in 2010,34,35 and the control should have played a major role in reducing the net PM emissions by motor vehicles, possibly in combination with a national regulation (Automobile NOx/PM Act) introduced in 2002.
According to a committee on PM2.5 established by the TMG, in addition to the diesel emission control, waste incinerators, and volatile organic compounds emission controls also contributed to a decline in PM2.5 from 2001 to 2008 in Tokyo.36 These controls, however, also have been implemented in Osaka based on a local regulation and these controls, as well as the national regulation (Automobile NOx/PM Act) on vehicles, may have contributed to a decline, but smaller than that seen in Tokyo, in Osaka’s PM2.5 levels.
In addition to the larger decline in PM2.5 in Tokyo’s 23 wards, residents in the area experienced larger improvements in the age-standardized mortality rates compared with Osaka. When conducting quasi-experimental studies to quantify health benefits associated with air pollution controls, appropriate counterfactual scenarios are needed14 (i.e., a population benefiting from the regulations should be compared with an external population in an area where specific regulation is not conducted). In addition, the appropriate external population should be comparable with the regulation-affected population.19 However, an appropriate external population is difficult to find. The reference population in the present study, Osaka, adopted such a regulation more recently (in 2009) and is very urbanized, and so met the above-mentioned assumptions, making the results more robust.
The most important concern is the differential downward trend in smoking prevalence between Tokyo’s 23 wards and Osaka. We therefore adopted a different reference period (October 2003 to September 2006) in the sensitivity analysis, but the results remained robust. In addition, Tokyo’s 23 wards benefited from a larger decline in cardiovascular and respiratory disease mortality than that in lung cancer mortality (which is sometimes used as an indicator of cumulative exposure to smoking37). Therefore, the differential trends in smoking prevalence between the two areas could not explain all of the declines in mortality in Tokyo.
In addition, differential exposure to health care, the health care system, or health-related policies might explain the results. However, Japan has a universal health insurance system that covers all of its citizens so access to health care or the health care system would not explain the different declines in mortality. In addition, health-related policies introduced during the study period were implemented nationwide (e.g., Health Promotion Act in 2003 or Medical Care Act updated in 2006); thus, health-related policies would not explain the different declines in mortality.
Although we attempted to adopt more appropriate reference population (i.e., Osaka), there are still discrepancies in some characteristics (e.g., area, population size, proportion of secondary industry workers, mean per capita income) between Tokyo’s 23 wards and Osaka (eTable 2; https://links.lww.com/EDE/B91). These may somehow contribute to baseline differences in air pollutants concentrations and mortality between the areas (Figure 2 and Table 1). However, because characteristics related to health (i.e., number of hospitals, clinics, and medical doctors per person) are similar and, as mentioned, the health care system (including access to health care or health-related policy) is the same between the areas, these characteristic differences may not affect the different declines in mortality between the areas materially. If it were true that these differences affected the differential declines, we might have observed larger declines even in mortality from other causes in Tokyo’s 23 wards. Selecting an appropriate reference population is challenging, but adjusting the change in mortality rates among residents in Osaka as a reference population would be appropriate.
Even though we can find the appropriate reference population, adjusting for long-term secular trends in the mortality rates in the reference population is also challenging. We used a sort of weighting scheme in the main analysis and we also adjusted for the daily age-standardized mortality rates in Osaka directly or using lowess smoothing in sensitivity analyses. The main findings were close to the findings when we used lowess smoothing with a window of 150 days, which is considered to produce less bias;19 thus, we could control the secular trends in mortality appropriately.
However, we only adjusted for secular trends in mortality in the reference population and we did not conduct separate analysis of a comparison population, unlike other intervention studies8,19 (i.e., doing the same analyses for Tokyo in an additional comparison population living in another city unaffected by the control). Such an analysis may provide an additional assessment of whether the effects observed in Tokyo can be attributed to the policy or to other factors happening at the same time.
A limitation is that we obtained air pollutants from only one station in each area and assumed that all of the residents were exposed to the same concentration without considering the spatial distribution of exposure. Although large spatial differences for NO2 are demonstrated, PM2.5 is known to be more spatially stable than other pollutants;30 thus, the measurements should reflect urban concentrations. Because PM2.5 arises not only from traffic sources, we show the results on elemental carbon in eFigure 2 (https://links.lww.com/EDE/B91), which still shows the decline during the study period.
Following institution of a local diesel emission control policy in Tokyo, we demonstrated a larger decline in PM2.5 and improvement in mortality compared with the reference population group.
We appreciate Dr. Wataru Sakamoto for his advice on statistical analyses and also appreciate the valuable support from Ms. Saori Irie.
1. Brook RD, Rajagopalan S, Pope CA III, et al.; American Heart Association Council on Epidemiology and Prevention, Council on the Kidney in Cardiovascular Disease, and Council on Nutrition, Physical Activity and Metabolism. Particulate matter air pollution and cardiovascular disease: An update to the scientific statement from the American Heart Association. Circulation. 2010;121:2331–2378.
2. Hoek G, Krishnan RM, Beelen R, et al. Long-term air pollution exposure and cardio- respiratory mortality: a review. Environ Health. 2013;12:43.
3. Atkinson RW, Kang S, Anderson HR, Mills IC, Walton HA. Epidemiological time series studies of PM2.5 and daily mortality and hospital admissions: a systematic review and meta-analysis. Thorax. 2014;69:660–665.
4. World Health Organization. Review of evidence on health aspects of air pollution-REVIHAAP Project: final technical report Denmark: 2013.WHO Regional Office for Europe.
5. Dominici F, Greenstone M, Sunstein CR. Science and regulation. Particulate matter matters. Science. 2014;344:257–259.
6. Chay K, Dobkin C, Greenstone M. The Clean Air Act of 1970 and adult mortality. J Risk Uncertainty. 2003;27:279–300.
7. Clancy L, Goodman P, Sinclair H, Dockery DW. Effect of air-pollution control on death rates in Dublin, Ireland: an intervention study. Lancet. 2002;360:1210–1214.
8. Johnston FH, Hanigan IC, Henderson SB, Morgan GG. Evaluation of interventions to reduce air pollution from biomass smoke on mortality in Launceston, Australia: retrospective analysis of daily mortality, 1994-2007. BMJ. 2013;346:e8446.
9. Rich DQ, Kipen HM, Huang W, et al. Association between changes in air pollution levels during the Beijing Olympics and biomarkers of inflammation and thrombosis in healthy young adults. JAMA. 2012;307:2068–2078.
10. Pope CA III, Schwartz J, Ransom MR. Daily mortality and PM10 pollution in Utah Valley. Arch Environ Health. 1992;47:211–217.
11. Pope CA III, Rodermund DL, Gee MM. Mortality effects of a copper smelter strike and reduced ambient sulfate particulate matter air pollution. Environ Health Perspect. 2007;115:679–683.
12. Chen Y, Ebenstein A, Greenstone M, Li H. Evidence on the impact of sustained exposure to air pollution on life expectancy from China’s Huai River policy. Proc Natl Acad Sci U S A. 2013;110:12936–12941.
13. Peters A, Breitner S, Cyrys J, et al. The influence of improved air quality on mortality risks in Erfurt, Germany. Res Rep Health Eff Inst. 2009:5–77; discussion 79–90.
14. van Erp AM, Kelly FJ, Demerjian KL, Pope CA, Cohen AJ. Progress in research to assess the effectiveness of air quality interventions towards improving public health. Air Qual Atmos Health. 2012;5:217–230.
15. Yorifuji T, Kawachi I, Kaneda M, Takao S, Kashima S, Doi H. Diesel vehicle emission and death rates in Tokyo, Japan: a natural experiment. Sci Total Environ. 2011;409:3620–3627.
16. Research Institute of Environment Agriculture and Fisheries Osaka Prefecture. [Summary of investigations on PM2.5 in the FY 2009] (in Japanese). Available at: http://www.kannousuiken-osaka.or.jp/kankyo/gijutsu/pm25/h21_overview.html
. Accessed December 2, 2015.
17. Miyoshi T, Akiyama K, Ueno H, Yokota H, Ishii K, Uchida Y, Higuchi Y, Ito Y. [Estimation of the source contribution of PM2.5 by CMB] (in Japanese). Annual Report. 2011;Tokyo: Tokyo Metropolitan Research Institute for Environmental Protection, 16–22.
18. Kashima S, Yorifuji T, Tsuda T, Doi H. Application of land use regression to regulatory air quality data in Japan. Sci Total Environ. 2009;407:3055–3062.
19. Dockery DW, Rich DQ, Goodman PG, et al. Effect of air pollution control on mortality and hospital admissions in Ireland. Res Rep Health Eff Inst. 2013:3–109.
20. Ministry of Health Labour and Welfare in Japan. [Comprehensive Survey of Living Conditions] (in Japanese). Available at: http://www.mhlw.go.jp/toukei/list/20–21.html
. Accessed June 2, 2015.
21. Johansson C, Burman L, Forsberg B. The effects of congestions tax on air quality and health. Atmos Environ. 2009;43:4843–4854.
22. Cesaroni G, Boogaard H, Jonkers S, et al. Health benefits of traffic-related air pollution reduction in different socioeconomic groups: the effect of low-emission zoning in Rome. Occup Environ Med. 2012;69:133–139.
23. Tonne C, Beevers S, Kelly FJ, Jarup L, Wilkinson P, Armstrong B. An approach for estimating the health effects of changes over time in air pollution: an illustration using cardio-respiratory hospital admissions in London. Occup Environ Med. 2010;67:422–427.
24. Friedman MS, Powell KE, Hutwagner L, Graham LM, Teague WG. Impact of changes in transportation and commuting behaviors during the 1996 Summer Olympic Games in Atlanta on air quality and childhood asthma. JAMA. 2001;285:897–905.
25. Huang W, Wang G, Lu SE, et al. Inflammatory and oxidative stress responses of healthy young adults to changes in air quality during the Beijing Olympics. Am J Respir Crit Care Med. 2012;186:1150–1159.
26. Peel JL, Klein M, Flanders WD, Mulholland JA, Tolbert PE. Impact of improved air quality during the 1996 Summer Olympic Games in Atlanta on multiple cardiovascular and respiratory outcomes. Res Rep Health Eff Inst. 2010:3–23; discussion 25–33.
27. Lee JT, Son JY, Cho YS. Benefits of mitigated ambient air quality due to transportation control on childhood asthma hospitalization during the 2002 summer Asian games in Busan, Korea. J Air Waste Manag Assoc. 2007;57:968–973.
28. Bureau of Environment of TMG. [Outline of regulation of vehicles in Tokyo Metroplotan Government] (in Japanese). Available at: https://www.kankyo.metro.tokyo.jp/vehicle/air_pollution/attachement/aramasi23.pdf
. Accessed June 8, 2015.
29. Bureau of Environment of TMG. [Status of regulation of noncompliant diesel vehicle] (in Japanese). Available at: http://www.kankyo.metro.tokyo.jp/vehicle/air_pollution/attachement/h27torishimari.pdf
. Accessed June 8, 2015.
30. HEI Panel on the Health Effects of Traffic-Related Air Pollution. Traffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. HEI Special Report 17. 2010.Boston, MA: Health Effects Institute.
31. Ueno H, Ishii K, Hoshi J, Kurita K, Amano S, Mitsuya R, Kinoshita T, Nakagawa T, Ihara M. [Evaluation of the Diesel Vehicle Emission Control Regulation by a Road Tunnel Study] (in Japanese). Annual Report. 2004:Tokyo: Tokyo Metropolitan Research Institute for Environmental Protection; 27–32.
32. Kanto Regional Air Quality Prevention Consortium. [Report on joint investigation on suspended particulate matter FY 2012] (in Japanese). 2014.
33. Bureau of Environment of TMG. [Summary of results on air pollution measurement] (in Japanese). 2005.
34. Bureau of Environment of TMG. [Future Policy to Reduce the Total Amount of Nitrogen Oxides and Particulate Matter Emitted from Motor Vehicles in Tokyo Metropolitan Government] (in Japanese). 2004.
35. Bureau of Environment of TMG. [Future Policy to Reduce the Total Amount of Nitrogen Oxides and Particulate Matter Emitted from Motor Vehicles in Tokyo Metropolitan Government] (in Japanese). 2013.
36. Committee on PM2.5 of Tokyo Metropolitan Government. [Report by Committee on PM2.5 of Tokyo Metropolitan Government] (in Japanese). 2011.Tokyo: Tokyo Metropolitan Government.
37. Pope CA 3rd, Ezzati M, Dockery DW. Fine-particulate air pollution and life expectancy in the United States. N Engl J Med. 2009;360:376–386.