Long-term effect of exposure to ambient air pollution on the risk of active tuberculosis

  • Author Footnotes
    1 These authors contributed equally to this work.
    Zhongqi Li
    Footnotes
    1 These authors contributed equally to this work.
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Author Footnotes
    1 These authors contributed equally to this work.
    Xuhua Mao
    Footnotes
    1 These authors contributed equally to this work.
    Affiliations
    Department of Clinical Laboratory, Yixing People’s Hospital, Wuxi, 214200, People’s Republic of China
    Search for articles by this author
  • Qiao Liu
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Huan Song
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Ye Ji
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Dian Xu
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Beibei Qiu
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Dan Tian
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Jianming Wang
    Correspondence
    Corresponding author at: Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, 101 Longmian Ave., Nanjing, 211166, People’s Republic of China.
    Affiliations
    Department of Epidemiology, Center for Global Health, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China

    Key Laboratory of Infectious Diseases, School of Public Health, Nanjing Medical University, Nanjing, 211166, People’s Republic of China
    Search for articles by this author
  • Author Footnotes
    1 These authors contributed equally to this work.
Open AccessPublished:July 30, 2019DOI:https://doi.org/10.1016/j.ijid.2019.07.027

      Highlights

      • Exposure to ambient PM2.5, PM10, SO2, and NO2 contributes to an increased risk of active tuberculosis.
      • The effect of ambient air PM10 and NO2 in the risk of tuberculosis remains significant in the multi-pollutant models.
      • This is the first DLNM-based time-series study exploring the long-term effect of ambient air pollution on the risk of tuberculosis.
      • Reducing the concentration of ambient air pollutants is likely to bring about health benefits in tuberculosis-endemic areas.

      Abstract

      Objectives

      To explore the long-term effect of exposure to ambient air pollution on the risk of active tuberculosis (TB).

      Methods

      We constructed a distributed-lag nonlinear model (DLNM) to evaluate the relatively long-term influence of outdoor PM2.5, PM10, SO2 and NO2 exposure on active TB risk in the city of Lianyungang in Jiangsu Province, China.

      Results

      There were 7,282 TB cases reported in the study area during 2014–2017, with annual median (interquartile range) concentrations of PM2.5, PM10, SO2 and NO2 at 45.86 (34.57–64.14) μg/m3, 85.43 (62.86–116.14) μg/m3, 22.00 (15.71–30.86) μg/m3 and 30.00 (23.29–38.57) μg/m3, respectively. The single-pollutant model showed that for each 10 μg/m3 increase in concentration, the cumulative relative risk of TB was 1.12 (lag 0–24 weeks, 95% CI: 1.03–1.22) for PM2.5 with reference to 35 μg/m3; 1.11 (lag 0–21 weeks, 95% CI: 1.06–1.17) for PM10 with reference to 70 μg/m3; 1.37 (lag 0–20 weeks, 95% CI: 1.16–1.62) for SO2 with reference to 60 μg/m3; and 1.29 (lag 0–22 weeks, 95% CI: 1.11–1.49) for NO2 with reference to 40 μg/m3. In the multipollutant model considering both PM10 and NO2, the association remained significant.

      Conclusions

      Our results revealed a potential association between outdoor exposure to PM2.5, PM10, SO2, and NO2 and active TB. Considering that people from developing countries continue to be exposed to both severe outdoor air pollution and high rates of latent TB infection, the association between worsening air pollution and active TB deserves further attention.

      Abbreviations:

      TB (tuberculosis), SO2 (sulfur dioxide), NO2 (nitrogen dioxide), DLNM (distributed-lag nonlinear model), RR (relative risk), CI (confidence interval), M. tb (Mycobacterium tuberculosis), WHO (World Health Organization), CVD (cardiovascular disorders), GDP (Gross Domestic Product), PGDP (per capita GDP), PD (population density), NDP (number of doctors per 10,000 people), GAM (generalized additive model), WAT (weekly average temperature), WAP (weekly average air pressure), WAS (weekly average wind speed), WAH (weekly average relative humidity), WAST (weekly average sunshine time), ns (Natural cubic spline), df (degree of freedom), PACF (partial autocorrelation function), TNF (tumor necrosis factor), IFN-γ (interferon-gamma)

      Keywords

      Introduction

      Worldwide, tuberculosis (TB) is the tenth leading cause of death (). One-third of the global population is infected with the mycobacterium tuberculosis (M.tb) pathogen, but only 10% of them eventually develop active TB (
      • Sgaragli G.
      • Frosini M.
      Human tuberculosis I. Epidemiology, diagnosis and pathogenetic mechanisms.
      ). Previous studies claimed that factors such as active and passive smoking, indoor air pollution, malnutrition or polymorphisms of immune-related genes were suggested to contribute to the increased risk of TB (
      • Ferrara G.
      • Murray M.
      • Winthrop K.
      • Centis R.
      • Sotgiu G.
      • Migliori G.B.
      • et al.
      Risk factors associated with pulmonary tuberculosis: smoking, diabetes and anti-TNFalpha drugs.
      ,
      • Lin H.-H.
      • Ezzati M.
      • Murray M.
      Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and meta-analysis.
      ,
      • Lonnroth K.
      • Jaramillo E.
      • Williams B.G.
      • Dye C.
      • Raviglione M.
      Drivers of tuberculosis epidemics: the role of risk factors and social determinants.
      ).
      Since the 1990s, global economic development has begun to accelerate, but the accompanying environmental pollution has become increasingly serious (
      • Chen B.
      • Kan H.
      Air pollution and population health: a global challenge.
      ). The WHO has estimated 2.4 million deaths due to air pollution-associated causes per year (
      • Sierra-Vargas M.P.
      • Teran L.M.
      Air pollution: impact and prevention.
      ). An increasing number of epidemiological studies have shown the adverse effects of outdoor exposure to air pollution on human health. A prospective analysis in nine European countries revealed a positive relation between exposures to particulate matter (PM) with aerodynamic diameter ≤10 μm (PM10) and ≤2.5 μm (PM2·5) and lung cancer risk (
      • Raaschou-Nielsen O.
      • Andersen Z.J.
      • Beelen R.
      • Samoli E.
      • Stafoggia M.
      • Weinmayr G.
      • et al.
      Air pollution and lung cancer incidence in 17 European cohorts: prospective analyses from the European Study of Cohorts for Air Pollution Effects (ESCAPE).
      ). In Hefei, China, Zhang et al. indicated that PM10, sulfur dioxide (SO2) and nitrogen dioxide (NO2) could significantly increase the cardiovascular disorders (CVD) mortality (
      • Zhang C.
      • Ding R.
      • Xiao C.
      • Xu Y.
      • Cheng H.
      • Zhu F.
      • et al.
      Association between air pollution and cardiovascular mortality in Hefei, China: a time-series analysis.
      ). A systematic review found that ambient air pollution exposure might contribute to the occurrence and development of chronic respiratory diseases (
      • Guan W.-J.
      • Zheng X.-Y.
      • Chung K.F.
      • Zhong N.-S.
      Impact of air pollution on the burden of chronic respiratory diseases in China: time for urgent action.
      ). Studies also indicated that outdoor exposure to air pollution might increase the risk of active TB. In Chengdu, China, Zhu et al. reported that people exposed to higher concentrations of PM10, SO2 and NO2 in the short term were more likely to develop TB (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ). A study in Taiwan revealed a possible link between PM2.5, CO, NO2 and TB (
      • Lai T.C.
      • Chiang C.Y.
      • Wu C.F.
      • Yang S.L.
      • Liu D.P.
      • Chan C.C.
      • et al.
      Ambient air pollution and risk of tuberculosis: a cohort study.
      ). Another study in North Carolina also suggested a potential correlation between particulate matter exposure and TB risk (
      • Smith G.S.
      • Schoenbach V.J.
      • Richardson D.B.
      • Gammon M.D.
      Particulate air pollution and susceptibility to the development of pulmonary tuberculosis disease in North Carolina: an ecological study.
      ). In South Korea, Hwang et al. found that male TB patients were more susceptible to incremental concentrations of SO2 than females (
      • Hwang S.S.
      • Kang S.
      • Lee J.Y.
      • Lee J.S.
      • Kim H.J.
      • Han S.K.
      • et al.
      Impact of outdoor air pollution on the incidence of tuberculosis in the Seoul metropolitan area, South Korea.
      ). Conversely, in a time-series analysis in Ningbo, China, Ge et al. reported that ambient SO2 exposure might be a protective factor for TB (
      • Ge E.
      • Fan M.
      • Qiu H.
      • Hu H.
      • Tian L.
      • Wang X.
      • et al.
      Ambient sulfur dioxide levels associated with reduced risk of initial outpatient visits for tuberculosis: a population based time series analysis.
      ). However, the above studies were mostly carried out in economically developed areas. Whether there is a similar relationship in economically underdeveloped regions requires further study.
      Nonlinear exposure-lag-response relationships have been demonstrated between outdoor air pollution and multiple health outcomes (
      • Chen F.
      • Deng Z.
      • Deng Y.
      • Qiao Z.
      • Lan L.
      • Meng Q.
      • et al.
      Attributable risk of ambient PM10 on daily mortality and years of life lost in Chengdu, China.
      ,
      • Guo Y.
      • Ma Y.
      • Zhang Y.
      • Huang S.
      • Wu Y.
      • Yu S.
      • et al.
      Time series analysis of ambient air pollution effects on daily mortality.
      ). The distributed-lag nonlinear model (DLNM) is an advanced time-series analysis method proposed by Gasparrini (
      • Gasparrini A.
      Distributed lag linear and non-linear models in R: the package dlnm.
      ,
      • Gasparrini A.
      • Armstrong B.
      • Kenward M.G.
      Distributed lag non-linear models.
      ), that has been widely used to explore the relationships between meteorological factors and health outcomes (
      • Buteau S.
      • Goldberg M.S.
      • Burnett R.T.
      • Gasparrini A.
      • Valois M.F.
      • Brophy J.M.
      • et al.
      Associations between ambient air pollution and daily mortality in a cohort of congestive heart failure: case-crossover and nested case-control analyses using a distributed lag nonlinear model.
      ,
      • Gasparrini A.
      Modeling exposure-lag-response associations with distributed lag non-linear models.
      ,
      • Neophytou A.M.
      • Picciotto S.
      • Brown D.M.
      • Gallagher L.E.
      • Checkoway H.
      • Eisen E.A.
      • et al.
      Exposure-lag-response in longitudinal studies: application of distributed-lag nonlinear models in an occupational cohort.
      ), but has seldom been applied to evaluate the relationship between outdoor exposure to air pollution and active TB risk. Thus, we performed a time-series study using DLNM to explore the correlation between a relatively long-term exposure to outdoor air pollution and active TB risk in a Chinese population.

      Materials and methods

       Study area and study subjects

      We selected the city of Lianyungang as the study site. It located in the northeastern region of Jiangsu Province, China, with an area of approximately 7.6 thousand square kilometers and 4.5 million permanent residents in 2017. Lianyungang is an economically underdeveloped city, where both the Gross Domestic Product (GDP) and per capita GDP (PGDP) rank 12th among the 13 cities in Jiangsu Province. TB cases reported from January 1, 2014, to December 31, 2017, were extracted through the Tuberculosis Management Information System. Diagnosis of active TB refers to the health industry standards of the People’s Republic of China- diagnosis for pulmonary tuberculosis. Patient demographic characteristics and clinical records were also extracted. We divided the study period into 209 weeks and calculated the per-week number of cases. Because January 1, 2014, was a Wednesday, the first week was only five days. This study was approved by the ethics committee of Nanjing Medical University.

       Air pollution and meteorological data

      We collected daily concentrations of ambient PM2.5, PM10, NO2 and SO2 from the Lianyungang Environmental Monitoring Terminal Station (https://www.aqistudy.cn/). CO and O3 were excluded because of missing values in the database. Simultaneous meteorological data, including average temperature (°C), pressure (hPa), wind speed (m/s), relative humidity (%), and sunshine time (h), were downloaded from the National Meteorological Data Sharing Center (http://data.cma.cn/). We calculated the weekly average air pollutants and meteorological factors for modeling.

       Socioeconomic indicators

      The socioeconomic indicators of Lianyungang including PGDP, population density (PD), and the number of doctors per 10,000 people (NDP), from 2014 to 2017, were collected from the Department of Statistics and entered into the model as covariates.

       Statistics analysis

      We used the Spearman correlation test to identify the correlations among four air pollutants and five meteorological factors. Considering the nonlinear exposure-lag-response relationship between outdoor exposure to air pollution and health (
      • Chen F.
      • Deng Z.
      • Deng Y.
      • Qiao Z.
      • Lan L.
      • Meng Q.
      • et al.
      Attributable risk of ambient PM10 on daily mortality and years of life lost in Chengdu, China.
      ,
      • Guo Y.
      • Ma Y.
      • Zhang Y.
      • Huang S.
      • Wu Y.
      • Yu S.
      • et al.
      Time series analysis of ambient air pollution effects on daily mortality.
      ), we adopted the DLNM to control the exposure-lag-response effect. The DLNM adds a lag dimension to the relationship between exposure and response and describes the variation of dependent variables in terms of both independent variable and lag dimensions. The core element of DLNM is the construction of a cross-basis function, which can be obtained by calculating the tensor product of functions. We constructed the model following the methods proposed by Gasparrini (
      • Gasparrini A.
      Distributed lag linear and non-linear models in R: the package dlnm.
      ,
      • Gasparrini A.
      • Armstrong B.
      • Kenward M.G.
      Distributed lag non-linear models.
      ). Since the weekly reported active TB cases are generally considered to be rare and nonindependent events, we used the generalized additive model (GAM) based on the quasi-Poisson distribution (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ).
      First, we constructed a basic model without air pollutants. We created a time variable of “week” (week = 1, 2, …, 209) to control the long-term patterns (including seasonal fluctuations) of the TB case reporting. Five meteorological factors, including weekly average temperature (WAT), weekly average air pressure (WAP), weekly average wind speed (WAS), weekly average relative humidity (WAH), weekly average sunshine time (WAST), and three socioeconomic indicators (PGDP, PD and NDP), were entered into the basic model as covariates. A natural cubic spline (ns) function was used to control the associations between the time-related variable of “week”, meteorological factors and active TB (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ). The degrees of freedom (df) of the ns were defined based on the principle of minimizing the sum of the absolute values of the partial autocorrelation function (PACF) of the residuals in the basic model (
      • Wood S.N.
      Generalized additive model: an introduction with R.
      ). Then, we modeled the exposure-response relationship through a linear function and the lag-response relationship through an ns function to build a cross-basis function for each air pollutant (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ). We hypothesized that an increased concentration of four ambient air pollutants could increase the risk of M. tb infection and progression to active TB. The progression from M. tb infection to clinically observed symptoms takes several months to years, and some latent infections even last for life. The average incubation period of TB is estimated to be three months. In addition, the diagnosis and reporting of TB also require times that can cause a delay. Thus, if the increased concentration of ambient air pollutants contributes to TB, extra time is required for an infected individual to develop active disease, be diagnosed, and finally be reported by a health care institution (
      • Murray M.
      • Oxlade O.
      • Lin H.H.
      Modeling social, environmental and biological determinants of tuberculosis.
      ,
      • You S.
      • Tong Y.W.
      • Neoh K.G.
      • Dai Y.
      • Wang C.H.
      On the association between outdoor PM2.5 concentration and the seasonality of tuberculosis for Beijing and Hong Kong.
      ). This type of lag was assumed to be over 2 months because the median time interval between infection and active disease was 7 weeks (
      • Naranbat N.
      • Nymadawa P.
      • Schopfer K.
      • Rieder H.L.
      Seasonality of tuberculosis in an Eastern-Asian country with an extreme continental climate.
      ). Leung et al. recommended that the maximum lag time was 6 months (
      • Leung C.C.
      • Yew W.W.
      • Chan T.Y.
      • Tam C.M.
      • Chan C.Y.
      • Chan C.K.
      • et al.
      Seasonal pattern of tuberculosis in Hong Kong.
      ). Here, we set the maximum lag time as 25 weeks.
      To avoid a multiple collinearity problem, we screened variables by setting the inclusion criteria of |r| <0.7 (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ). Since there was a high correlation between the WAT and WAP (r = −0.93), we removed the WAP from the basic model. We further added the constructed cross-basis function to the basic model. The final single-pollutant model was Yt=quasiPoissonμt, where μt=α+WXTη+nsweek,df1*4+nsWAT,df2+nsWAS,df3+nsWAH,df4+nsWAST,df5+β1PGDP+β2PD+β3NDP. Yt and μt were the observed and expected number of TB cases on week t, respectively; α was an intercept; WXTη was a cross-basis function of each air pollutant; df1 was the annual df of the time variable "week"; df2-df5 were the dfs of meteorological factors; β1-β3 were the coefficients of PGDP, PD and NDP, respectively. For PM2.5 and PM10, df1-df5 were defined as 1, 4, 5, 4, and 1, respectively. For SO2 and NO2, df1 was defined as 2, while df3-df5 were defined as 1, 4, and 1, respectively. Df2 was missing because the WAT was highly correlated with both SO2 (r = −0.79) and NO2 (r = −0.72), and the WAT was not involved in the single-pollutant model of SO2 and NO2.
      By referring to the annual average concentration threshold of each air pollutant specified by the National Ministry of Ecology and Environmental Protection (GB3095-2012), we defined the reference values as 35 μg/m3 for PM2.5, 70 μg/m3 for PM10, 60 μg/m3 for SO2 and 40 μg/m3 for NO2. We calculated the cumulative relative risk (RR) and 95% confidence interval (CI) to express the strength of association between every 10 μg/m3 increase in each air pollutant concentration and active TB risk (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ).
      We further assessed whether the effects of air pollutants were modified by gender or age. We calculated the 95% CI to identify the difference between subgroup effects using the following formula: Q1ˆ-Q2ˆ±1.96(SE1ˆ)2+(SE2ˆ)2. In this formula, Q1ˆ and Q2ˆ are the point estimates of the RRs, and SE1ˆ and SE2ˆ are their standard errors. If the 95% CI contained a value of zero, there was no evidence showing modification by gender or age (
      • Huang J.
      • Pan X.
      • Guo X.
      • Li G.
      Impacts of air pollution wave on years of life lost: a crucial way to communicate the health risks of air pollution to the public.
      ,
      • Schenker N.
      • Gentleman J.F.
      On judging the significance of differences by examining the overlap between confidence intervals.
      ). Since PM2.5 and SO2 were highly correlated with the other three air pollutants, we constructed the multipollutant model only for PM10 and NO2 in order to evaluate the stability of the effects.
      All analyses were performed with the “dlnm”, “mgcv” and “splines” packages in R software version 3.5.3 (https://www.r-project.org/). The significance level was set at 0.05.

      Results

       Characteristics of TB cases

      There were 7,282 active TB cases reported in the study area from 2014 to 2017, with an annual incidence of 34.39/105. As shown in Table 1, 74.43% of TB cases were males and 60.24% were younger than 60 years old. Most cases (82.00%) were manual workers and were of Han nationality (99.71%). Only 3.17% were migrant populations. Figure 1 shows the number of weekly reported TB cases from 2014 to 2017.
      Table 1Characteristics of newly reported active tuberculosis cases in Lianyungang, 2014–2017.
      CharacteristicsN (%)
      Gender
       Male5420 (74.43)
       Female1862 (25.57)
      Age group
       <60 years old4387 (60.24)
       ≥60 years old2895 (39.76)
      Occupation
       Students354 (4.86)
       Manual workers5971 (82.00)
       Office workers92 (1.26)
       Unemployed518 (7.11)
       Retired179 (2.46)
       Others168 (2.31)
      Race
       Han nationality7261 (99.71)
       Others21 (0.29)
      Household register
       Local residents7051 (96.83)
       Other cities residents231 (3.17)
      Type of TB
       Pulmonary TB7281 (99.99)
       Extrapulmonary TB1 (0.01)
      Figure 1
      Figure 1Weekly reported numbers of new TB cases and weekly average air pollution concentrations in Lianyungang, 2014–2017.
      TB: tuberculosis; PM2.5: particulate matter with aerodynamic diameter of ≤2.5 μm; PM10: particulate matter with aerodynamic diameter of ≤10 μm; SO2: sulfur dioxide; NO2: nitrogen dioxide.

       Characteristics of air pollutants and meteorological factors

      The weekly average concentration of air pollutants is also shown in Figure 1. The range of weekly average concentrations was 15.29–156.71 μg/m3 for PM2.5, 29.43–234.60 μg/m3 for PM10, 4.43–79.60 μg/m3 for SO2, and 10.86–65.57 μg/m3 for NO2. The annual median (interquartile range) concentration of each air pollutant was 45.86 (34.57–64.14) μg/m3 for PM2.5, 85.43 μg/m3 (62.86–116.14) for PM10, 22.00 μg/m3 (15.71–30.86) for SO2 and 30.00 (23.29–38.57) μg/m3 for NO2. The information on five meteorological factors is also listed in Table 2.
      Table 2Weekly average air pollutant and meteorological indicators in Lianyungang, 2014–2017.
      VariablesMinimumQ25MedianMeanQ75Maximum
      Pollutant concentration
       PM2.5 (μg/m3)15.2934.5745.8652.3064.14156.71
       PM10 (μg/m3)29.4362.8685.4392.54116.14234.60
       SO2 (μg/m3)4.4315.7122.0024.8330.8679.60
       NO2 (μg/m3)10.8623.2930.0031.8438.5765.57
      Meteorological indicator
       Temperature (°C)−4.245.6416.1914.7922.9431.00
       Air pressure (hPa)1001.001009.001017.001017.001025.001034.00
       Wind speed (m/s)1.572.142.472.502.804.27
       Relative humidity (%)46.7166.0074.7173.9082.2992.00
       Sunshine time (h)0.914.936.306.207.6011.17
      Q25: 25% quartile; Q75: 75% quartile; PM2.5: particulate matter with aerodynamic diameter of ≤2.5 μm; PM10: particulate matter with aerodynamic diameter of ≤10 μm; SO2: sulfur dioxide; NO2: nitrogen dioxide.
      The four air pollutants were positively associated with each other (P < 0.01). PM2.5, PM10 and SO2 were positively associated with WAP and negatively associated with WAT and WAH (P < 0.01), while NO2 was positively associated with WAP (P < 0.01) and negatively associated with WAT, WAS, WAH and WAST (P < 0.05) (Table 3).
      Table 3Spearman rank correlation coefficients between weekly average air pollutant concentrations and meteorological factors in Lianyungang, 2014–2017.
      VariablesPM2.5PM10SO2NO2TemperatureAir pressureWind speedRelative humiditySunshine time
      PM2.510.92**0.79**0.74**−0.64**0.55**−0.09−0.45**−0.09
      PM1010.81**0.67**−0.59**0.48**0.03−0.51**0.04
      SO210.73**−0.79**0.72**−0.04−0.59**−0.04
      NO21−0.72**0.71**−0.30**−0.47**−0.14*
      Temperature1−0.93**0.030.52**0.30**
      Air pressure1−0.19**−0.48**−0.28**
      Wind speed1−0.130.10
      Relative humidity1−0.46**
      Sunshine time1
      PM2.5: particulate matter with aerodynamic diameter of ≤2.5 μm; PM10: particulate matter with aerodynamic diameter of ≤10 μm; SO2: sulfur dioxide; NO2: nitrogen dioxide; **: P < 0.01; *: P < 0.05.

       PM2.5 and TB

      As shown in Figure 2 (A1 and A2), for the single-pollutant model, the association between a 10 μg/m3 increase in PM2.5 concentration and active TB was significant (with reference to 35 μg/m3). The cumulative RR reached its maximum at a lag of 24 weeks (RR: 1.12, 95% CI: 1.03–1.22) (Table 4). After stratification by gender and age, the association remained significant in males (RR: 1.13, 95% CI: 1.03–1.24) and young groups (RR: 1.13, 95% CI: 1.02–1.24). The effect was not altered by gender or age (P > 0.05) (Table S1).
      Figure 2
      Figure 2The relative risk of a 10 μg/m3 increase in four air pollutants on active TB cases at different lag weeks, based on the single-pollutant model.
      RR: relative risk; CI: confidence interval; PM2.5: particulate matter with aerodynamic diameter of ≤2.5 μm; PM10: particulate matter with aerodynamic diameter of ≤10 μm; SO2: sulfur dioxide; NO2: nitrogen dioxide; TB: tuberculosis; Pooled RRs (A1, B1, C1 and D1); Cumulative RRs (A2, B2, C2 and D2).
      Table 4Cumulative RR and 95% CI for the association between a 10 μg/m3 increase in PM2.5 concentration and active TB cases at a lag of 0 to 24 weeks in Lianyungang, 2014–2017.
      VariablesSingle-pollutant model
      Adjusted for long-term trend, meteorological factors (temperature, wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).
      All cases1.12 (1.03–1.22)
      Sex
       Male1.13 (1.03–1.24)
       Female1.11 (0.97–1.26)
      Age group
       <60 years old1.13 (1.02–1.24)
       ≥60 years old1.11 (0.98–1.25)
      RR: relative risk; CI: confidence interval; PM2.5: particulate matter with aerodynamic diameter of ≤2.5 μm; TB: tuberculosis.
      a Adjusted for long-term trend, meteorological factors (temperature, wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).

       PM10 and TB

      As shown in Figure 2 (B1 and B2), for the single-pollutant model, the association between a 10 μg/m3 increase in PM10 concentration and active TB was significant (with reference to 70 μg/m3). The cumulative RR reached its maximum at a lag of 21 weeks (RR: 1.11, 95% CI: 1.06–1.17) (Table 5). After stratification by gender and age, the associations remained significant in each subgroup. In the multipollutant model, the association was still significant, with a cumulative RR of 1.07 (95% CI: 1.02–1.13) at a lag of 0 to 21 weeks (Table 5). The associations remained significant in males (1.07, 95% CI: 1.02–1.14), young (1.06, 95% CI: 1.00–1.13) and old groups (1.09, 95% CI: 1.01–1.17). The effect was not altered by gender or age (P > 0.05) (Table S2).
      Table 5Cumulative RR and 95% CI for the association between a 10 μg/m3 increase in PM10 concentration and active TB cases at a lag of 0 to 21 weeks in Lianyungang, 2014–2017.
      Single-pollutant model
      Adjusted for long-term trend, meteorological factors (temperature, wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).
      Multipollutant model
      Adjusted for long-term trend, meteorological factors (wind speed, relative humidity and sunshine time), socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people) and NO2.
      All cases1.11 (1.06–1.17)1.07 (1.02–1.13)
      Sex
       Male1.12 (1.06–1.18)1.07 (1.02–1.14)
       Female1.10 (1.02–1.18)1.08 (0.99–1.17)
      Age group
       <60 years old1.12 (1.06–1.18)1.06 (1.00–1.13)
       ≥60 years old1.10 (1.03–1.18)1.09 (1.01–1.17)
      RR: relative risk; CI: confidence interval; PM10: particulate matter with aerodynamic diameter of ≤10 μm; TB: tuberculosis.
      a Adjusted for long-term trend, meteorological factors (temperature, wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).
      b Adjusted for long-term trend, meteorological factors (wind speed, relative humidity and sunshine time), socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people) and NO2.

       SO2 and TB

      As shown in Figure 2 (C1 and C2), for the single-pollutant model, the association between a 10 μg/m3 increase in SO2 concentration and active TB was significant (with reference to 60 μg/m3). The cumulative RR reached a maximum at a lag of 20 weeks (RR: 1.37, 95% CI: 1.16–1.62) (Table 6). After stratification by gender and age, the associations remained significant in each subgroup. The effect was not altered by gender or age (P > 0.05) (Table S3).
      Table 6Cumulative RR and 95% CI for the association between a 10 μg/m3 increase in SO2 concentration and active TB cases at a lag of 0 to 20 weeks in Lianyungang, 2014–2017.
      Single-pollutant model
      Adjusted for long-term trend, meteorological factors (wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).
      All cases1.37 (1.16–1.62)
      Sex
       Male1.39 (1.16–1.66)
       Female1.33 (1.03–1.72)
      Age group
       <60 years old1.36 (1.13–1.63)
       ≥60 years old1.37 (1.08–1.74)
      RR: relative risk; CI: confidence interval; SO2: sulfur dioxide; TB: tuberculosis.
      a Adjusted for long-term trend, meteorological factors (wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).

       NO2 and TB

      As shown in Figure 2 (D1 and D2), for the single-pollutant model, the association between a 10 μg/m3 increase in NO2 concentration and active TB was significant (with reference to 40 μg/m3). The cumulative RR reached its maximum at a lag of 22 weeks (RR: 1.29, 95% CI: 1.11–1.49) (Table 7). After stratification by gender and age, the association remained significant in each subgroup. In the multipollutant model, the association was still significant, with a cumulative RR of 1.30 (95% CI: 1.11–1.53) at a lag of 0 to 22 weeks (Table 7). After stratification by gender and age, the associations remained significant in each subgroup. The effect was not altered by gender or age (P > 0.05) (Table S4).
      Table 7Cumulative RR and 95% CI for the association between a 10 μg/m3 increase in NO2 concentration and active TB cases at a lag of 0 to 22 weeks in Lianyungang, 2014–2017.
      Single-pollutant model
      Adjusted for long-term trend, meteorological factors (wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).
      Multipollutant model
      Based on a single-pollutant model, adjusted with PM10.
      All cases1.29 (1.11–1.49)1.30 (1.11–1.53)
      Sex
       Male1.28 (1.09–1.50)1.29 (1.08–1.54)
       Female1.32 (1.06–1.65)1.35 (1.05–1.72)
      Age group
       <60 years old1.30 (1.11–1.52)1.30 (1.09–1.55)
       ≥60 years old1.26 (1.02–1.55)1.35 (1.08–1.69)
      RR: relative risk; CI: confidence interval; NO2: nitrogen dioxide; TB: tuberculosis.
      a Adjusted for long-term trend, meteorological factors (wind speed, relative humidity and sunshine time) and socioeconomic indicators (per capita gross domestic product, population density and the number of doctors per 10,000 people).
      b Based on a single-pollutant model, adjusted with PM10.

      Discussion

      In this time-series study performed in a northeastern region of Jiangsu Province, China, we observed a potential correlation between relatively long-term outdoor exposure to PM2.5, PM10, SO2 and NO2, and active TB. Ambient PM10 and NO2 remained significant in the multipollutant models, and the association was not altered in subgroups of different genders and ages. To date, this is the first DLNM-based time-series study exploring the role of relatively long-term outdoor air pollution on active TB risk.
      A study in Chengdu, China also reported exposure to ambient PM10, NO2 and SO2 in relation to increased morbidity of TB (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ). However, the maximum lag time in their model was 28 days for PM10 and 5 days for SO2 and NO2, which could only estimate the short-term effects. Progression from infection with M.tb to the clinically observable immunological response usually takes a few months or even several years, and the diagnosis and reporting of TB also require some time; this lag time should be included in the time-series model (
      • Murray M.
      • Oxlade O.
      • Lin H.H.
      Modeling social, environmental and biological determinants of tuberculosis.
      ,
      • You S.
      • Tong Y.W.
      • Neoh K.G.
      • Dai Y.
      • Wang C.H.
      On the association between outdoor PM2.5 concentration and the seasonality of tuberculosis for Beijing and Hong Kong.
      ). The commonly used lag time in TB modeling was two months though sometimes, up to six months was used (
      • Leung C.C.
      • Yew W.W.
      • Chan T.Y.
      • Tam C.M.
      • Chan C.Y.
      • Chan C.K.
      • et al.
      Seasonal pattern of tuberculosis in Hong Kong.
      ). Therefore, we set the maximum lag time as 25 weeks, which was rational and could be applied to evaluate a relatively long-term effect caused by air pollution exposure.
      The particle diameter of PM2.5 is fine enough to reach the pulmonary alveoli along the respiratory tract. PM2.5 and some toxic compounds attached to it may contribute to a variety of diseases (
      • Chen B.
      • Kan H.
      Air pollution and population health: a global challenge.
      ,
      • Guan W.-J.
      • Zheng X.-Y.
      • Chung K.F.
      • Zhong N.-S.
      Impact of air pollution on the burden of chronic respiratory diseases in China: time for urgent action.
      ,
      • Raaschou-Nielsen O.
      • Andersen Z.J.
      • Beelen R.
      • Samoli E.
      • Stafoggia M.
      • Weinmayr G.
      • et al.
      Air pollution and lung cancer incidence in 17 European cohorts: prospective analyses from the European Study of Cohorts for Air Pollution Effects (ESCAPE).
      ). In a systematic review, the pollutant most frequently associated with TB outcomes was PM2.5, where 4 out of 6 studies demonstrated a significant association (
      • Popovic I.
      • Soares Magalhaes R.J.
      • Ge E.
      • Marks G.B.
      • Dong G.H.
      • Wei X.
      • et al.
      A systematic literature review and critical appraisal of epidemiological studies on outdoor air pollution and tuberculosis outcomes.
      ). However, previous studies have shown inconsistent conclusions on the relationship between ambient PM10 and SO2 exposure and active TB risk (
      • Popovic I.
      • Soares Magalhaes R.J.
      • Ge E.
      • Marks G.B.
      • Dong G.H.
      • Wei X.
      • et al.
      A systematic literature review and critical appraisal of epidemiological studies on outdoor air pollution and tuberculosis outcomes.
      ).
      A study in Taipei found that increased concentration of PM10 increased the incidence of culture-positive TB, which is consistent with our results (
      • Chen K.Y.
      • Chuang K.J.
      • Liu H.C.
      • Lee K.Y.
      • Feng P.H.
      • Su C.L.
      • et al.
      Particulate matter is associated with sputum culture conversion in patients with culture-positive tuberculosis.
      ). However, the effect of PM10 did not appear to be associated with TB in other studies in nothern California (
      • Smith G.S.
      • Van Den Eeden S.K.
      • Garcia C.
      • Shan J.
      • Baxter R.
      • Herring A.H.
      • et al.
      Air pollution and pulmonary tuberculosis: a nested case-control study among members of a Northern California health plan.
      ) and South Korea (
      • Hwang S.S.
      • Kang S.
      • Lee J.Y.
      • Lee J.S.
      • Kim H.J.
      • Han S.K.
      • et al.
      Impact of outdoor air pollution on the incidence of tuberculosis in the Seoul metropolitan area, South Korea.
      ). This inconsistency may be partly attributed to a disparity in air pollutant levels. For example, the median concentration of PM10 was 20.6 μg/m3 in nothern California and 62.8 μg/m3 in South Korea, which were lower than the value of our study at 85.4 μg/m3. In addition, other factors such as meteorological conditions and socioeconomic levels may also differ in ways that impact the exposure-response effect (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ).
      SO2, a dominant air pollutant, is released into the atmosphere through the combustion of fossil fuels for energy conversion. A study in South Korea reported that an increased concentration of SO2 might increase the morbidity of TB in males but not in females (
      • Hwang S.S.
      • Kang S.
      • Lee J.Y.
      • Lee J.S.
      • Kim H.J.
      • Han S.K.
      • et al.
      Impact of outdoor air pollution on the incidence of tuberculosis in the Seoul metropolitan area, South Korea.
      ). Conversely, a study in Ningbo, China, found that exposure to SO2 might function as a protective factor against TB (
      • Ge E.
      • Fan M.
      • Qiu H.
      • Hu H.
      • Tian L.
      • Wang X.
      • et al.
      Ambient sulfur dioxide levels associated with reduced risk of initial outpatient visits for tuberculosis: a population based time series analysis.
      ). This counterintuitive protective effects may be attributed to the antimicrobial properties of SO2. It is noteworthy that these studies explored the effects of SO2 on TB in the context of short-term exposure.
      Since inhaling high concentrations of NO2 can irritate the respiratory tract, people living in environments with high concentrations of NO2 may be more susceptible to respiratory infections. An increased concentration of NO2 contributed to the increased morbidity of active TB in Taiwan (
      • Lai T.C.
      • Chiang C.Y.
      • Wu C.F.
      • Yang S.L.
      • Liu D.P.
      • Chan C.C.
      • et al.
      Ambient air pollution and risk of tuberculosis: a cohort study.
      ) and northern California (
      • Smith G.S.
      • Van Den Eeden S.K.
      • Garcia C.
      • Shan J.
      • Baxter R.
      • Herring A.H.
      • et al.
      Air pollution and pulmonary tuberculosis: a nested case-control study among members of a Northern California health plan.
      ). Our results also indicated a positive link between NO2 exposure and TB risk. The cumulative RR reached its maximum at a lag of 22 weeks, and this association was not altered by gender or age.
      The potential role of outdoor exposure to air pollution in active TB may be explained by the following reasons. First, the mucous lining of the human nasal and respiratory tract surface is the first line of defense in combating M.tb infection by stopping it from entering the pulmonary alveoli. Exposure to ambient air pollutants may weaken the clearance of respiratory secretions, resulting in M.tb escaping the first line of host defense (
      • Zhu S.
      • Xia L.
      • Wu J.
      • Chen S.
      • Chen F.
      • Zeng F.
      • et al.
      Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
      ). Second, particulate matter, particularly PM2.5, can reach the pulmonary alveoli and impair the defense function of alveolar macrophages (
      • Ni L.
      • Chuang C.C.
      • Zuo L.
      Fine particulate matter in acute exacerbation of COPD.
      ). Additionally, both NO2 and SO2 are easily soluble in water, which makes them able to damage the upper respiratory tract mucosa. Third, during the early phase of infection, alveolar macrophages can inhibit the proliferation of M.tb by phagocytosis and the formation of granulomas in the lungs (
      • Palanisamy G.S.
      • Kirk N.M.
      • Ackart D.F.
      • Shanley C.A.
      • Orme I.M.
      • Basaraba R.J.
      Evidence for oxidative stress and defective antioxidant response in guinea pigs with tuberculosis.
      ), while SO2 can affect the macrophage function and the ability of alveolar clearance and mucociliary transport (
      • Hwang S.S.
      • Kang S.
      • Lee J.Y.
      • Lee J.S.
      • Kim H.J.
      • Han S.K.
      • et al.
      Impact of outdoor air pollution on the incidence of tuberculosis in the Seoul metropolitan area, South Korea.
      ). Exposure to SO2 at a concentration of 12.5 ppm in vitro can cause the death of 62% of alveolar macrophages within 30 minutes (
      • Kienast K.
      • Muller-Quernheim J.
      • Knorst M.
      • Schlegel J.
      • Mang C.
      • Ferlinz R.
      Realistic in vitro study of oxygen radical liberation by alveolar macrophages and mononuclear cells of the peripheral blood after short-term exposure to SO2.
      ). Fourth, continuous exposure to SO2 and NO2 might decrease the production of tumor necrosis factor (TNF)-α and interferon-gamma (IFN-γ) (
      • Saito Y.
      • Azuma A.
      • Kudo S.
      • Takizawa H.
      • Sugawara I.
      Effects of diesel exhaust on murine alveolar macrophages and a macrophage cell line.
      ,
      • Saito Y.
      • Azuma A.
      • Kudo S.
      • Takizawa H.
      • Sugawara I.
      Long-term inhalation of diesel exhaust affects cytokine expression in murine lung tissues: comparison between low- and high-dose diesel exhaust exposure.
      ), which are essential for combating M.tb infection (
      • Dutta N.K.
      • Karakousis P.C.
      Latent tuberculosis infection: myths, models, and molecular mechanisms.
      ,
      • Flad H.D.
      • Gercken J.
      • Hubner L.
      • Schluter C.
      • Pryjma J.
      • Ernst M.
      Cytokines in mycobacterial infections: in vitro and ex vivo studies.
      ,
      • Fremond C.
      • Allie N.
      • Dambuza I.
      • Grivennikov S.I.
      • Yeremeev V.
      • Quesniaux V.F.
      • et al.
      Membrane TNF confers protection to acute mycobacterial infection.
      ).
      Our study has several limitations. First, our data on ambient pollutant concentrations came from local environmental monitoring terminal stations and might not provide precise exposure estimates for the entire city. Second, we assumed that all the inhabitants of this study were exposed to similar weekly average levels of ambient air pollutants. There may exist an ecological fallacy, and the misclassification of exposure is inevitable. Due to missing data, other air pollutants such as CO and O3 were not included in our study.

      Conclusions

      Our results indicate that exposure to ambient PM2.5 (lag 0–24 weeks), PM10 (lag 0–21 weeks), SO2 (lag 0–20 weeks) and NO2 (lag 0–22 weeks) can increase the risk of active TB. Reducing the concentration of ambient air pollutants is likely to bring about health benefits in TB-endemic areas.

      Author contributions

      ZL, XM and JW conceived and designed the study. ZL, XM, QL, HS, YJ, DX, BQ and DT were involved in the data analysis and collection. ZL and XM conducted the data analysis and interpretation. ZL, XM and JW drafted and wrote the article and all authors provided critical revisions. All authors approved the final version of the article.

      Funding

      The study was supported by the National Key R&D Program of China ( 2017YFC0907000 ), National Natural Science Foundation of China ( 81973103 ), Qing Lan Project of Jiangsu Province (2019), and Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.

      Competing interests

      The authors declare no competing financial interest.

      Patient consent for publication

      Not required.

      Data sharing statement

      All data generated or analyzed during this study are included in this published article.

      Ethical approval

      This study was approved by the ethics committee of Nanjing Medical University.

      Appendix A. Supplementary data

      The following is Supplementary data to this article:

      References

        • Buteau S.
        • Goldberg M.S.
        • Burnett R.T.
        • Gasparrini A.
        • Valois M.F.
        • Brophy J.M.
        • et al.
        Associations between ambient air pollution and daily mortality in a cohort of congestive heart failure: case-crossover and nested case-control analyses using a distributed lag nonlinear model.
        Environ Int. 2018; 113 (Available from:): 313-324https://doi.org/10.1016/j.envint.2018.01.003
        • Chen B.
        • Kan H.
        Air pollution and population health: a global challenge.
        Environ Health Prev Med. 2008; 13 (Available from:): 94-101https://doi.org/10.1007/s12199-007-0018-5
        • Chen F.
        • Deng Z.
        • Deng Y.
        • Qiao Z.
        • Lan L.
        • Meng Q.
        • et al.
        Attributable risk of ambient PM10 on daily mortality and years of life lost in Chengdu, China.
        Sci Total Environ. 2017; 581–582 (Available from:): 426-433https://doi.org/10.1016/j.scitotenv.2016.12.151
        • Chen K.Y.
        • Chuang K.J.
        • Liu H.C.
        • Lee K.Y.
        • Feng P.H.
        • Su C.L.
        • et al.
        Particulate matter is associated with sputum culture conversion in patients with culture-positive tuberculosis.
        Ther Clin Risk Manag. 2016; 12 (Available from:): 41-46https://doi.org/10.2147/tcrm.s92927
        • Dutta N.K.
        • Karakousis P.C.
        Latent tuberculosis infection: myths, models, and molecular mechanisms.
        Microbiol Mol Biol Rev. 2014; 78 (Available from:): 343-371https://doi.org/10.1128/MMBR.00010-14
        • Ferrara G.
        • Murray M.
        • Winthrop K.
        • Centis R.
        • Sotgiu G.
        • Migliori G.B.
        • et al.
        Risk factors associated with pulmonary tuberculosis: smoking, diabetes and anti-TNFalpha drugs.
        Curr Opin Pulm Med. 2012; 18 (Available from:): 233-240https://doi.org/10.1097/MCP.0b013e328351f9d6
        • Flad H.D.
        • Gercken J.
        • Hubner L.
        • Schluter C.
        • Pryjma J.
        • Ernst M.
        Cytokines in mycobacterial infections: in vitro and ex vivo studies.
        Arch Immunol Ther Exp. 1995; 43: 153-158
        • Fremond C.
        • Allie N.
        • Dambuza I.
        • Grivennikov S.I.
        • Yeremeev V.
        • Quesniaux V.F.
        • et al.
        Membrane TNF confers protection to acute mycobacterial infection.
        Respir Res. 2005; 6 (Available from:): 136https://doi.org/10.1186/1465-9921-6-136
        • Gasparrini A.
        Distributed lag linear and non-linear models in R: the package dlnm.
        J Stat Softw. 2011; 43: 1-20
        • Gasparrini A.
        Modeling exposure-lag-response associations with distributed lag non-linear models.
        Stat Med. 2014; 33 (Available from:): 881-899https://doi.org/10.1002/sim.5963
        • Gasparrini A.
        • Armstrong B.
        • Kenward M.G.
        Distributed lag non-linear models.
        Stat Med. 2010; 29 (Available from:): 2224-2234https://doi.org/10.1002/sim.3940
        • Ge E.
        • Fan M.
        • Qiu H.
        • Hu H.
        • Tian L.
        • Wang X.
        • et al.
        Ambient sulfur dioxide levels associated with reduced risk of initial outpatient visits for tuberculosis: a population based time series analysis.
        Environ Pollut. 2017; 228 (Available from:): 408-415https://doi.org/10.1016/j.envpol.2017.05.051
        • Guan W.-J.
        • Zheng X.-Y.
        • Chung K.F.
        • Zhong N.-S.
        Impact of air pollution on the burden of chronic respiratory diseases in China: time for urgent action.
        Lancet. 2016; 388 (Available from:): 1939-1951https://doi.org/10.1016/s0140-6736(16)31597-5
        • Guo Y.
        • Ma Y.
        • Zhang Y.
        • Huang S.
        • Wu Y.
        • Yu S.
        • et al.
        Time series analysis of ambient air pollution effects on daily mortality.
        Environ Sci Pollut Res Int. 2017; 24 (Available from:): 20261-20272https://doi.org/10.1007/s11356-017-9502-7
        • Huang J.
        • Pan X.
        • Guo X.
        • Li G.
        Impacts of air pollution wave on years of life lost: a crucial way to communicate the health risks of air pollution to the public.
        Environ Int. 2018; 113 (Available from:): 42-49https://doi.org/10.1016/j.envint.2018.01.022
        • Hwang S.S.
        • Kang S.
        • Lee J.Y.
        • Lee J.S.
        • Kim H.J.
        • Han S.K.
        • et al.
        Impact of outdoor air pollution on the incidence of tuberculosis in the Seoul metropolitan area, South Korea.
        Korean J Intern Med. 2014; 29 (Available from:): 183-190https://doi.org/10.3904/kjim.2014.29.2.183
        • Kienast K.
        • Muller-Quernheim J.
        • Knorst M.
        • Schlegel J.
        • Mang C.
        • Ferlinz R.
        Realistic in vitro study of oxygen radical liberation by alveolar macrophages and mononuclear cells of the peripheral blood after short-term exposure to SO2.
        Pneumologie (Stuttgart, Germany). 1993; 47: 60-65
        • Lai T.C.
        • Chiang C.Y.
        • Wu C.F.
        • Yang S.L.
        • Liu D.P.
        • Chan C.C.
        • et al.
        Ambient air pollution and risk of tuberculosis: a cohort study.
        Occup Environ Med. 2016; 73 (Available from:): 56-61https://doi.org/10.1136/oemed-2015-102995
        • Leung C.C.
        • Yew W.W.
        • Chan T.Y.
        • Tam C.M.
        • Chan C.Y.
        • Chan C.K.
        • et al.
        Seasonal pattern of tuberculosis in Hong Kong.
        Int J Epidemiol. 2005; 34 (Available from:): 924-930https://doi.org/10.1093/ije/dyi080
        • Lin H.-H.
        • Ezzati M.
        • Murray M.
        Tobacco smoke, indoor air pollution and tuberculosis: a systematic review and meta-analysis.
        PLoS Med. 2007; 4 (Available from:): e20https://doi.org/10.1371/journal.pmed.0040020
        • Lonnroth K.
        • Jaramillo E.
        • Williams B.G.
        • Dye C.
        • Raviglione M.
        Drivers of tuberculosis epidemics: the role of risk factors and social determinants.
        Soc Sci Med. 2009; 68 (Available from:): 2240-2246https://doi.org/10.1016/j.socscimed.2009.03.041
        • Murray M.
        • Oxlade O.
        • Lin H.H.
        Modeling social, environmental and biological determinants of tuberculosis.
        Int J Tuberc Lung Dis. 2011; 15 Suppl. 2 (Available from:): 64-70https://doi.org/10.5588/ijtld.10.0535
        • Naranbat N.
        • Nymadawa P.
        • Schopfer K.
        • Rieder H.L.
        Seasonality of tuberculosis in an Eastern-Asian country with an extreme continental climate.
        Eur Respir J. 2009; 34 (Available from:): 921-925https://doi.org/10.1183/09031936.00035309
        • Neophytou A.M.
        • Picciotto S.
        • Brown D.M.
        • Gallagher L.E.
        • Checkoway H.
        • Eisen E.A.
        • et al.
        Exposure-lag-response in longitudinal studies: application of distributed-lag nonlinear models in an occupational cohort.
        Am J Epidemiol. 2018; 187 (Available from:): 1539-1548https://doi.org/10.1093/aje/kwy019
        • Ni L.
        • Chuang C.C.
        • Zuo L.
        Fine particulate matter in acute exacerbation of COPD.
        Front Physiol. 2015; 6 (Available from:): 294https://doi.org/10.3389/fphys.2015.00294
        • Palanisamy G.S.
        • Kirk N.M.
        • Ackart D.F.
        • Shanley C.A.
        • Orme I.M.
        • Basaraba R.J.
        Evidence for oxidative stress and defective antioxidant response in guinea pigs with tuberculosis.
        PLoS One. 2011; 6 (Available from:)e26254https://doi.org/10.1371/journal.pone.0026254
        • Popovic I.
        • Soares Magalhaes R.J.
        • Ge E.
        • Marks G.B.
        • Dong G.H.
        • Wei X.
        • et al.
        A systematic literature review and critical appraisal of epidemiological studies on outdoor air pollution and tuberculosis outcomes.
        Environ Res. 2019; 170 (Available from:): 33-45https://doi.org/10.1016/j.envres.2018.12.011
        • Raaschou-Nielsen O.
        • Andersen Z.J.
        • Beelen R.
        • Samoli E.
        • Stafoggia M.
        • Weinmayr G.
        • et al.
        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. 2013; 14 (Available from:): 813-822https://doi.org/10.1016/s1470-2045(13)70279-1
        • Saito Y.
        • Azuma A.
        • Kudo S.
        • Takizawa H.
        • Sugawara I.
        Effects of diesel exhaust on murine alveolar macrophages and a macrophage cell line.
        Exp Lung Res. 2002; 28: 201-217
        • Saito Y.
        • Azuma A.
        • Kudo S.
        • Takizawa H.
        • Sugawara I.
        Long-term inhalation of diesel exhaust affects cytokine expression in murine lung tissues: comparison between low- and high-dose diesel exhaust exposure.
        Exp Lung Res. 2002; 28 (Available from:): 493-506https://doi.org/10.1080/01902140290096764
        • Schenker N.
        • Gentleman J.F.
        On judging the significance of differences by examining the overlap between confidence intervals.
        Am Stat. 2001; 55 (Available from:): 182-186https://doi.org/10.1198/000313001317097960
        • Sgaragli G.
        • Frosini M.
        Human tuberculosis I. Epidemiology, diagnosis and pathogenetic mechanisms.
        Curr Med Chem. 2016; 23: 2836-2873
        • Sierra-Vargas M.P.
        • Teran L.M.
        Air pollution: impact and prevention.
        Respirology. 2012; 17 (Available from:): 1031-1038https://doi.org/10.1111/j.1440-1843.2012.02213.x
        • Smith G.S.
        • Schoenbach V.J.
        • Richardson D.B.
        • Gammon M.D.
        Particulate air pollution and susceptibility to the development of pulmonary tuberculosis disease in North Carolina: an ecological study.
        Int J Environ Health Res. 2014; 24 (Available from:): 103-112https://doi.org/10.1080/09603123.2013.800959
        • Smith G.S.
        • Van Den Eeden S.K.
        • Garcia C.
        • Shan J.
        • Baxter R.
        • Herring A.H.
        • et al.
        Air pollution and pulmonary tuberculosis: a nested case-control study among members of a Northern California health plan.
        Environ Health Perspect. 2016; 124 (Available from:): 761-768https://doi.org/10.1289/ehp.1408166
        • WHO
        Global tuberculosis report 2018.
        2018 (. [Accessed 18 April 2019])
        • Wood S.N.
        Generalized additive model: an introduction with R.
        2th ed. Chapman and Hall/CRC, London2017
        • You S.
        • Tong Y.W.
        • Neoh K.G.
        • Dai Y.
        • Wang C.H.
        On the association between outdoor PM2.5 concentration and the seasonality of tuberculosis for Beijing and Hong Kong.
        Environ Pollut. 2016; 218 (Available from:): 1170-1179https://doi.org/10.1016/j.envpol.2016.08.071
        • Zhang C.
        • Ding R.
        • Xiao C.
        • Xu Y.
        • Cheng H.
        • Zhu F.
        • et al.
        Association between air pollution and cardiovascular mortality in Hefei, China: a time-series analysis.
        Environ Pollut. 2017; 229 (Available from:): 790-797https://doi.org/10.1016/j.envpol.2017.06.022
        • Zhu S.
        • Xia L.
        • Wu J.
        • Chen S.
        • Chen F.
        • Zeng F.
        • et al.
        Ambient air pollutants are associated with newly diagnosed tuberculosis: a time-series study in Chengdu, China.
        Sci Total Environ. 2018; 631–632 (Available from:): 47-55https://doi.org/10.1016/j.scitotenv.2018.03.017