The association between lung cancer incidence and ambient air pollution in China: A spatiotemporal analysis

Environmental Research 144 (2016) 60–65

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

The association between lung cancer incidence and ambient air pollution in China: A spatiotemporal analysis Yuming Guo a,n,1, Hongmei Zeng b,1, Rongshou Zheng b, Shanshan Li a, Adrian G. Barnett c, Siwei Zhang b, Xiaonong Zou b, Rachel Huxley d, Wanqing Chen b,nn, Gail Williams a a

Division of Epidemiology and Biostatistics, School of Public Health, University of Queensland, Brisbane, Australia National Office for Cancer Prevention and Control, National Cancer Center, Chinese Academy of Medical Sciences, Cancer Hospital, Beijing, China c School of Public Health and Social Work & Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Australia d School of Public Health, Curtin University, Perth, Australia b

art ic l e i nf o

a b s t r a c t

Article history: Received 11 September 2015 Received in revised form 2 November 2015 Accepted 3 November 2015

Background: China is experiencing more and more days of serious air pollution recently, and has the highest lung cancer burden in the world. Objectives: To examine the associations between lung cancer incidence and fine particles (PM2.5) and ozone in China. Methods: We used 75 communities’ data of lung cancer incidence from the National Cancer Registration of China from 1990 to 2009. The annual concentrations of fine particles (PM2.5) and ozone at 0.1°
0.1° spatial resolution were generated by combing remote sensing, global chemical transport models, and improvements in coverage of surface measurements. A spatial age-period-cohort model was used to examine the relative risks of lung cancer incidence associated with the air pollutants, after adjusting for impacts of age, period, and birth cohort, sex, and community type (rural and urban) as well as the spatial variation on lung cancer incidence. Results: The relative risks of lung cancer incidence related to a 10 mg/m3 increase in 2-year average PM2.5 were 1.055 (95% confidence interval (CI): 1.038, 1.072) for men, 1.149 (1.120, 1.178) for women, 1.060 (1.044, 1.075) for an urban communities, 1.037 (0.998, 1.078) for a rural population, 1.074 (1.052, 1.096) for people aged 30–65 years, and 1.111 (1.077, 1.146) for those aged over 75 years. Ozone also had a significant association with lung cancer incidence. Conclusions: The increased risks of lung cancer incidence were associated with PM2.5 and ozone air pollution. Control measures to reduce air pollution would likely lower the future incidence of lung cancer. & 2015 Elsevier Inc. All rights reserved.

Keywords: Air pollution Lung cancer incidence Fine particles Ozone Spatial age-period-cohort study

1. Introduction Lung cancer is now the most common cancer in the world, with the majority of the cases in developing countries (Ferlay et al., 2010; Jemal et al., 2011). China has the highest lung cancer burden in the world (Zhao et al., 2010). According to the latest Chinese cancer registration annual report, the world age-standardized incidence rate of lung cancer was 47.5 per 100,000 for men and 22.2 per 100,000 for women in 2009 (Chen et al., 2013), and these incidences are expected to rise (Chen et al., 2011). Determining the risk factors associated with this high burden is crucial for cancer prevention and control. The established risk n

Corresponding author. Corresponding author. E-mail addresses: [email protected] (Y. Guo), [email protected] (W. Chen). 1 These authors contributed equally to the study.

nn

http://dx.doi.org/10.1016/j.envres.2015.11.004 0013-9351/& 2015 Elsevier Inc. All rights reserved.

factors for lung cancer include smoking (Correa et al., 1983; Hackshaw et al., 1997; Hecht, 2002; Janerich et al., 1990) and air pollution (Cohen, 2000; Mumford et al., 1987; Pope et al., 2002; Vineis et al., 2004). In particular, ambient air pollution is the most widespread environmental carcinogen (Cohen, 2000; Vineis et al., 2004). Globally, it is estimated that 12.8% of lung cancer death can be attributed to exposure of the fine particulate matter air pollution alone (Evans et al., 2013; Fajersztajn et al., 2013). In 2010, an estimated 223,000 deaths from lung cancer worldwide were attributed to air pollution (Straif et al., 2013). With the rapid economic growth and increased urbanization of rural areas, China is experiencing very high concentrations of air pollutants (Brauer et al., 2012). The average concentration of fine particulate matter in densely populated regions of China can exceed 100 μg/m3 (Guo et al., 2013). However, studies on ambient air pollution and lung cancer have never been performed at the national level. In the present study, we investigated lung cancer

Y. Guo et al. / Environmental Research 144 (2016) 60–65

incidence in relation to long-term exposure to two ambient air pollutants, fine particulate matter (PM2.5) and ozone (O3), using population-based national cancer registration data of China.

2. Methods 2.1. Study design and participants The National Cancer Registration Center of China is responsible for the collection, evaluation and publication of cancer statistics from population-based cancer registries in China each year since 1970s. All data on cancer incidence are reported to populationbased cancer registries from hospitals, community health centers or other departments, including centers of township medical insurance and the New-type Rural Cooperative Medical System. Based on the integrity and quality of lung cancer data, a total of 75 cancer registries out of 104 (72%) from the national cancer database were selected from 1990 to 2009 in this study (Fig. 1). ICD10 (International Classification of Disease for Oncology, version 10) was used to classify lung cancer cases. The detailed information on each case including year and age at diagnosis, gender and community type (rural or urban area) was used. Population data was collected for each community and year from local statistics bureaus. We limited analyses to persons at least 30 years old, because few cases occurred below this age. We stratified the lung cancer incidence into 12 age groups (30–34 years, 35–39 years, 40–44 years, 80–84 years, and 85 þ years) for each community. 2.2. Exposure assessment We used data on annual mean PM2.5 and O3 for the years 1990 and 2005 from a previous study (Brauer et al., 2012), which estimated the concentration of global air pollution to assess the global burden of disease attributable to outdoor air pollution. In brief, Ambient PM2.5 and O3 exposures for the Earth’s entire human

61

population were estimated, which allowed the inclusion of populations in smaller cities and rural areas to examine the disease burden related to air pollution. Remote sensing, global chemicaltransport models, and improvements in coverage of surface measurements were combined to estimate the global estimates of annual average ambient concentrations of PM2.5 and O3 at 0.1°
0.1° spatial resolution for the years 1990 and 2005. We spatially matched our study communities with the global air pollution data using latitude and longitude for the years 1990 and 2005. We then predicted the annual concentrations of PM2.5 and O3 during the years 1990–2009 for each community using a linear regression model, because the data for air pollution is only available for the years 1990 and 2005. Lastly we linked the lung cancer incidence data and predicted annual concentrations of PM2.5 and O3 during the years 1990– 2009 for each of the 75 communities. 2.3. Statistical analysis Age-period-cohort models can separate the effects of age from the effects of risk factors in relation to calendar time and birth cohort effects (Robertson and Boyle, 1998). The incidence of lung cancer increases with age, and there were substantial birth cohort effect and period effect (Eilstein et al., 2008; Mdzinarishvili et al., 2009). In this study, we therefore included separate variables for age, period, and cohort effects, and gradually extended the model to include information on air pollution. Thus we examined the association between air pollution and lung cancer incidence after controlling for the effects of age, period, and birth cohort. As there might be spatial cluster in lung cancer incidence, we also included a spatial term in the analyses to control for the spatial distribution of lung cancer incidence (Mdzinarishvili et al., 2010). We used an over-dispersed Poisson regression model to fit the spatial age-period-cohort model:

Lung cancer rate/100,000

0

500

1000 1500 km

32.1−71.1 71.1−110 ● 110−148 ● 148−187 ● 187−225 225−264 264−303 303−380 380−419

Fig. 1. The location of the 75 study communities and standardized lung cancer incidence rate for people aged 4 30 years in urban (red colour) and rural (purple colour) China, during 1990–2009. The rate was standardized by world Segi's population 1985. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

62

Y. Guo et al. / Environmental Research 144 (2016) 60–65

Log (μt, i ) = α + β Age + S(Period) + λCohort + ηGender + υLocation + S(latitude, longitude) + offset(log(population))

(1)

where t is the year of the observation; i is the community of the lung cancer incidence; Yt,i is the observed yearly incidence counts on year t at community i; α is the intercept; Age is a categorical variable of age group for lung cancer incidence,and β is vector of coefficients for Age. S(.) is a natural cubic spline and Period is the year of lung cancer incidence. Cohort is a categorical variable representing birth cohort (year of birth), and λ is vector of coefficients. Gender represents the gender category and η is the vector

of corresponding coefficients. Location is a binary variable that is 1 for urban areas and 0 for rural areas. A flexible spatial term “latitude, longitude” smoothed using a natural cubic spline was used to control for spatial variation in lung cancer incidence. The degrees of freedom for spline functions were chosen using the Akaike information criterion for quasi-Poisson models (Q-AIC). The log of the population size for each age group for each community was modelled as the offset. Model (1) was used to control for the effects of age, period, birth cohort, and gender, as well as the spatial variation on lung cancer incidence. To examine the association between air pollutants and lung cancer incidence, we added 2-year average PM2.5 and 2-year average O3 separately to Model (1). We used a linear

PM2.5 350

PM2.5 (ug/m3)

300 250 200 150 100 50

O3

O3 (ppb)

70 60 50 40

Fig. 2. The spatial distribution of mean concentrations of modelled PM2.5 (mg/m3) and O3 (ppb) in China during the study period.

Y. Guo et al. / Environmental Research 144 (2016) 60–65

term for PM2.5 and O3, as our initial analyses showed the associations between PM2.5 and O3 and lung cancer incidence were linear. We also examined the effects of air pollutants on lung cancer incidence separately for men, women, rural residents, urban residents, people aged 30–65 years, people aged 65–75 years, and those aged over 75 years. A series of sensitivity analyses were conducted to check the robustness of our results. We removed the period and spatial term respectively from the spatial age-period-cohort model, to examine whether we underestimated the associations between air pollutants and the lung cancer incidence. To consider the uncertainties of the predicted annual concentrations of air pollutants, for each community we simulated 10,000 data set for PM2.5 and O3 respectively by adding random residuals (mean ¼0 with standard deviation ¼5). We rerun the spatial age-period-cohort model 10,000 times for PM2.5 and O3 respectively using the simulated data set. We calculated the mean of the effect estimates and confidence intervals.

3. Results

Table 1 The relative risks of lung cancer incidence associated with an increase of 10 mg/m3 in PM2.5 and an increase of 10 ppb in O3 in difference groups in China, during 1990– 2009. Group

PM2.5

All Male Female Urban Rural Age 30–65 Age 65–75 Age475

1.074 1.055 1.149 1.060 1.037 1.074 1.101 1.111

0.0

0.12

−0.5

0.04

−0.12

−2.0

−0.2 60

80

PM2.5 (ug/m3)

100

(1.079, 1.095) (1.082, 1.102) (1.065, 1.094) (1.075, 1.092) (0.980, 1.028) (1.071, 1.096) (1.105, 1.133) (1.081, 1.115)

−0.04

−1.5

40

1.087 1.092 1.080 1.083 1.004 1.083 1.119 1.098

To the best of our knowledge, this study is the first to assess the

0.2

20

(1.060, 1.089) (1.038, 1.072) (1.120, 1.178) (1.044, 1.075) (0.998, 1.078) (1.052, 1.096) (1.076, 1.126) (1.077, 1.146)

4. Discussion

0.5

−1.0

O3

increase of 10 mg/m3 in PM2.5 and an increase of 10 ppb in O3 were associated with relative risks of 1.074 (95% confidence interval (CI): 1.060, 1.089) and 1.087 (1.079, 1.095) in lung cancer incidence, respectively. Women, urban residents, and the elderly had higher relative risks of lung cancer associated with exposure to PM2.5 than men, rural residents, and the young, respectively. Urban residents and the elderly had higher relative risks of lung cancer associated with exposure to O3 than rural residents and the young, respectively. Within both men and women, urban residents had a higher risk of lung cancer associated with exposure to PM2.5 and O3 than those living in rural areas (Table 2). Women had similar raised relative risks of lung cancer related to exposure to PM2.5 across all ages, but elderly men had higher risks than young men. We found similar patterns for O3. We further examined the relative risks of lung cancer incidence associated with PM2.5 and O3 by geographical area (Table 3). In general, the effects estimates of lung cancer associated with air pollutants were higher in urban area than those in rural area, except for people aged 30–65 years with exposure to PM2.5. The elderly urban resident had a greater risk of lung cancer related to exposure to PM2.5 and O3 than the young urban resident. When we removed the period and spatial terms from the models, respectively, the effect estimates were slightly increased but not significantly. When rerun the models using the simulated data set with random residuals, the mean effect estimates were not changed, but the confidence intervals were slightly wide.

Log (RR)

Log (RR)

There were 368,762 lung cancer cases, including 247,533 (67%) men and 312678 (85%) cases living in urban area. Table S1 shows the rates of lung cancer incidence in China from 1990 to 2009. The crude rates of lung cancer incidence increased significantly from 1990 to 2009 for all groups. However, the standardized rates changed slightly during the study period for all groups. Men had higher crude and standardized rates than women. There was spatial variation in the standardized lung cancer incidence rate across China (Fig. 1). Lung cancer rates increased with age across different periods and birth cohorts, while the rates unchanged along the period (Supplemental material, Fig. S1). However, lung cancer cases increased along the period, and early birth cohorts had high lung cancer cases (Supplemental material, Fig. S2) The spatial distribution of PM2.5 and O3 in China is shown in Fig. 2. For PM2.5, the central-east and northwest of China were highly polluted. O3 levels were high in the north and southwest of China. Concentrations of both PM2.5 and O3 were higher in urban areas than in rural areas (Supplemental material, Table S2). Fig. 3 shows the associations between lung cancer incidence and exposure to 2-year average of PM2.5 and O3 in China during 1990–2009. Generally, both PM2.5 and O3 had linear effects on lung cancer incidence, which means the higher concentrations of PM2.5 and O3, the higher risk of lung cancer incidence in China. The associations between PM2.5 and O3 and lung cancer incidence in different groups are shown in Table 1. A two-year

63

40

50

60

70

O3 (ppb)

Fig. 3. : The associations between PM2.5, O3 and lung cancer incidence in China during 1990–2009, using spatial age-period-cohort design. A natural cubic spline with 4 degrees of freedom was used for PM2.5 and O3, respectively.

64

Y. Guo et al. / Environmental Research 144 (2016) 60–65

Table 2 The relative risks of lung cancer incidence associated with an increase of 10 mg/m3 in PM2.5 and an increase of 10 ppb in O3 in the subgroups of women and men in China, during 1990–2009. Group

Urban Rural Age 30–65 Age 65–75 Age4 75

O3

PM2.5 Women

Men

1.128 1.061 1.176 1.175 1.163

1.042 1.033 1.038 1.102 1.084

(1.097, 1.159) (0.985, 1.143) (1.132, 1.222) (1.125, 1.226) (1.101, 1.228)

(1.024, 1.061) (0.988, 1.080) (1.013, 1.063) (1.073, 1.132) (1.044, 1.126)

associations between ambient PM2.5 and O3 and lung cancer in China. A large data set of 75 communities spread across the country during 1990–2009 was used in present study. We developed the spatial age-period-cohort mode which fitted the data very well (Supplemental material, Fig. S3). There were statistically significant associations between PM2.5 and O3 and lung cancer in China. In general, the effects of air pollution on lung cancer were greater in women, elderly and urban people than those in men, the young, and rural people, respectively. The present study found significant associations between ambient air pollution and lung cancer in China. Many studies in USA and Europe have reported that ambient air pollutants including particulate matter and ozone were associated with the risk for lung cancer incidence or deaths (Cohen, 2000; Pope et al., 2002; Raaschou-Nielsen et al., 2013; Vineis et al., 2004). Studies showed significant associations between risk for lung cancer incidence and PM2.5 with relative risks of 1.18 (95% CI 0.96, 1.46) per 5 μg/m3 in Europe (Raaschou-Nielsen et al., 2013) and in Italy 1.05 (1.01, 1.10) per 10 μg/m3 (Cesaroni et al., 2013). Recently, the International Agency for Research on Cancer (IARC) has concluded that outdoor air pollution causes lung cancer (Straif et al., 2013). However, no study has reported the relationship between ambient PM2.5 and O3 and lung cancer in China. A cohort study showed that an increase of 10 μg/m3 of total suspended particle, sulphur dioxide, and nitrogen oxides corresponded to 1.1% (95% CI:  0.1%, 2.3%), 4.2% (95% CI: 2.3%, 6.2%), and 2.7% (95% CI:  0.9%, 6.5%) increase of lung cancer mortality in China (Cao et al., 2011). One interesting finding is that the relative risks of lung cancer incidence associated with PM2.5 were higher in women than in men, but not significant for O3. Importantly, in China, smoking was more prevalent among men (63%) than women (4%) (Yang et al., 1999). Several epidemiological studies have reported that the significant association of air pollution and lung cancer mainly exist in non-smokers (Raaschou-Nielsen et al., 2011) and never-smokers (Beelen et al., 2008; Raaschou-Nielsen et al., 2010). The potential biological mechanism for women being more sensitive to outdoor air pollution than men might be that women have smaller lung size and airway diameter. This might increase women's airway reactivity and exacerbate particulate deposition (Bennett et al., 1996). However, further individual level studies collecting clinic, behavioural, and demographic data may help elucidate why

Women

Men

1.076 0.963 1.094 1.093 1.095

1.089 1.015 1.080 1.135 1.096

(1.061, 1.092) (0.919, 1.009) (1.071, 1.117) (1.068, 1.118) (1.067, 1.125)

(1.078, 1.099) (0.986, 1.044) (1.065, 1.096) (1.118, 1.153) (1.075, 1.117)

women are likely to be the more vulnerable than men. People living in urban areas are more likely to develop lung cancer than those living in rural area. This might be caused by the more serious air pollution in urban area than rural area (Li et al., 2014). Our data shows that people living urban cities have higher exposures to air pollution than those living in rural areas (Table S2). In addition, urban air pollution combines toxic effects from many constituents, leading to more severe effects on human health (Yue et al., 2013). The elderly were more vulnerable to air pollution associated with lung cancer than the young. The elderly are more sensitive than younger people to the air pollution both in terms of physiology and behaviour. It is acknowledged that many physiological regulatory functions weaken with age (Collins, 1987). This means their bodies are less able to compensate for the impacts of air pollutants, which might increase risks of health events. In addition, the elderly usually spends more time outdoors than the young because of work necessities, which many result in that the elderly experiences greater exposure to the outdoor air pollution than the young (Tong et al., 2015). Many studies have found that the relative risks of morality or morbidity associated with air pollutants were higher in elderly than the young in China (Chen et al., 2012a, 2012b). This study has several strengths. To the best of our knowledge, this is the first study to examine the associations between PM2.5 and O3 and lung cancer incidence in China. We used national lung cancer registration data at 75 communities with varying air pollution exposure levels. The exposure assessment of air pollution used a standardized protocol from the global burden disease study (Brauer et al., 2012). The spatial age-period-cohort model was used, which produced a high predictive ability. Our study overcomes several limitations of previous cohort studies, for example, broad national coverage and lung cancer incidence as the outcome. There are also some limitations. We did not control for the trend of smoking rate, as smoking data at this detailed spatial level were not available. We only had air pollution data for the years 1990 and 2005, and interpolated the data to 1990–2009. This might underestimate the effect estimates of air pollution on lung cancer, because random measurement error in exposure will bias the effect estimates towards the null (Hutcheon et al., 2010). However, our sensitivity analyses showed that the effect estimates

Table 3 The relative risks of lung cancer incidence associated with an increase of 10 mg/m3 in PM2.5 and an increase of 10 ppb in O3 in the subgroups of rural and urban areas in China, during 1990–2009. Group

Age 30–65 Age 65–75 Age4 75

PM2.5

O3

Rural

Urban

Rural

Urban

1.042 (0.986, 1.102) 0.992 (0.929, 1.058) 1.071 (0.976, 1.176)

1.042 (0.974, 1.115) 1.037 (0.965, 1.116) 1.148 (1.039, 1.267)

1.001 (0.961, 1.043) 0.967 (0.923, 1.012) 0.879 (0.826, 0.934)

1.049 (1.032, 1.067) 1.071 (1.054, 1.088) 1.079 (1.060, 1.098)

Y. Guo et al. / Environmental Research 144 (2016) 60–65

did not change when we added random residuals to the annual concentrations. This study used community-level incidence and air pollution exposure, but not individual level information on exposure and outcome. community-level exposure may introduce exposure bias which might make the our effect estimates smaller than the real associations (Hutcheon et al., 2010). However, this study is still useful and important, especially in the context that China has dearth of studies evaluating long-term exposure effects of air pollution on lung cancer.

5. Conclusions The findings suggest that lung cancer is associated with ambient air pollution in China. Air pollution is a serious problem in China, and on the basis of our findings, decrease in concentrations of air pollution can be expected to greatly reduce the future number of lung cancer cases in China.

Conflict of interests The authors have declared that no competing interests exist.

Ethical approval This study was approved by the University of Queensland’s behavior and social sciences ethical review committee (#2013000739).

Acknowledgements We gratefully acknowledged the cooperation of all the population-based cancer registries in providing cancer statistics, data collection, sorting, verification and database creation. YG is supported by the University of Queensland Research Fellowship. The study was funded by Hope Run Malathon Fund (Cancer Institute & Hospital, Chinese Academy of Medical Sciences, LC2011Y41), and the Australia National Health and Medical Research Council (#APP1030259). No funding bodies had any role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2015.11. 004.

References Beelen, R., et al., 2008. Long-term exposure to traffic-related air pollution and lung cancer risk. Epidemiology 19, 702–710. Bennett, W.D., et al., 1996. Variability of fine particle deposition in healthy adults: effect of age and gender. Am. J. Respir. Crit. Care Med. 153, 1641–1647. Brauer, M., et al., 2012. Exposure assessment for estimation of the global burden of

65

disease attributable to outdoor air pollution. Environ. Sci. Technol. 46, 652–660. Cao, J., et al., 2011. Association between long-term exposure to outdoor air pollution and mortality in China: a cohort study. J. Hazard. Mater. 186, 1594–1600. Cesaroni, G., et al., 2013. Long-term exposure to urban air pollution and mortality in a cohort of more than a million adults in Rome. Environ. Health Perspect. 121, 324–331. Chen, R., et al., 2012a. Association of particulate air pollution with daily mortality the China air pollution and health effects study. Am. J. Epidemiol. 175, 1173–1181. Chen, R.J., et al., 2012b. Short-term exposure to sulfur dioxide and daily mortality in 17 Chinese cities: the China air pollution and health effects study (CAPES). Environ. Res. 118, 101–106. Chen, W., et al., 2013. Report of incidence and mortality in China cancer registries, 2009. Chin. J. Cancer Res. 25, 10. Chen, W.Q., et al., 2011. Bayesian age-period-cohort prediction of lung cancer incidence in China. Thorac. Cancer 2, 149–155. Cohen, A.J., 2000. Outdoor air pollution and lung cancer. Environ. health Perspect. 108 (Suppl 4), S743–S750. Collins, K.J., 1987. Effects of cold on old people. Br. J. Hosp. Med. 38, 506. Correa, P., et al., 1983. Passive smoking and lung cancer. Lancet 322, 595–597. Eilstein, D., et al., 2008. Lung cancer mortality in France. Trend analysis and projection between 1975 and 2012, using a Bayesian age-period-cohort model. Lung Cancer 59, 282–290. Evans, J., et al., 2013. Estimates of global mortality attributable to particulate air pollution using satellite imagery. Environ. Res. 120, 33–42. Fajersztajn, L., et al., 2013. Air pollution: a potentially modifiable risk factor for lung cancer. Nat. Rev. Cancer 13, 674–678. Ferlay, J., et al., 2010. Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008. Int. J. Cancer 127, 2893–2917. Guo, Y., et al., 2013. The burden of air pollution on years of life lost in Beijing, China, 2004–08: retrospective regression analysis of daily deaths. BMJ 347, f7139. Hackshaw, A.K., et al., 1997. The accumulated evidence on lung cancer and environmental tobacco smoke. BMJ 315, 980–988. Hecht, S.S., 2002. Cigarette smoking and lung cancer: chemical mechanisms and approaches to prevention. Lancet Oncol. 3, 461–469. Hutcheon, J.A., et al., 2010. Random measurement error and regression dilution bias. BMJ 340, c2289. Janerich, D.T., et al., 1990. Lung cancer and exposure to tobacco smoke in the household. New. Engl. J. Med. 323, 632–636. Jemal, A., et al., 2011. Global cancer statistics. CA: Cancer J. Clin. 61, 69–90. Li, W., et al., 2014. Distribution of atmospheric particulate matter (PM) in rural field, rural village and urban areas of northern China. Environ. Pollut. 185, 134–140. Mdzinarishvili, T., et al., 2009. A novel approach for analysis of the log-linear ageperiod-cohort model: application to lung cancer incidence. Cancer Inform. 7, 271–280. Mdzinarishvili, T., et al., 2010. Estimation of hazard functions in the log-linear ageperiod-cohort model: application to lung cancer risk associated with geographical area. Cancer Inform. 9, 67–78. Mumford, J.L., et al., 1987. Lung cancer and indoor air pollution in Xuan Wei, China. Science 235, 217–220. Pope III, C.A., et al., 2002. Lung cancer, cardiopulmonary mortality, and long-term exposure to fine particulate air pollution. JAMA: J. Am. Med. Assoc. 287, 1132–1141. Raaschou-Nielsen, O., et al., 2010. Air pollution from traffic and risk for lung cancer in three Danish cohorts. Cancer Epidemiol. Biomark. Prev.: Publ. Am. Assoc. Cancer Res. 19, 1284–1291, Cosponsored by the American Society of Preventive Oncology. Raaschou-Nielsen, O., et al., 2011. Lung cancer incidence and long-term exposure to air pollution from traffic. Environ. Health Perspect. 119, 860–865. Raaschou-Nielsen, O., et al., 2013. Air pollution and lung cancer incidence in 17 European cohorts: prospective analyses from the European Study of Cohorts for Air Pollution Effects (ESCAPE). Lancet Oncol. 14, 813–822. Robertson, C., Boyle, P., 1998. Age–period–cohort analysis of chronic disease rates. I: modelling approach. Stat. Med. 17, 1305–1323. Straif, K., et al., 2013. IARC Scientific Publication No. 161: Air Pollution and Cancer. International Agency for Research on Cance, Geneva. Tong, L., et al., 2015. The association between air pollutants and morbidity for diabetes and liver diseases modified by sexes, ages, and seasons in Tianjin, China. Environ. Sci. Pollut. Res. Int. 22, 1215–1219. Vineis, P., et al., 2004. Outdoor air pollution and lung cancer: recent epidemiologic evidence. Int. J. Cancer J. Int. Cancer 111, 647–652. Yang, G.H., et al., 1999. Smoking in China – findings of the 1996 National Prevalence Survey. JAMA – J. Am. Med. Assoc. 282, 1247–1253. Yue, D., et al., 2013. Comparison of particle number size distributions and new particle formation between the urban and rural sites in the PRD region, China. Atmos. Environ. 76, 181–188. Zhao, P., et al., 2010. Cancer trends in China. Jpn. J. Clin. Oncol. 40, 281–285.