Short-term effect of ambient air pollutant change on the risk of tuberculosis outpatient visits: a time-series study in Fuyang, China

There is growing evidence that air pollution plays a role in TB, and most studies have been conducted in the core countries with inconsistent results. Few studies have comprehensively included the six common air pollutants, so they cannot consider whether various pollutants interact with each other. Our objectives were to investigate the association between short-term exposure to six common air pollutants and the risk of tuberculosis outpatient visits in Fuyang, China, 2015–2020. We combined the two models to explore the effects of exposure to six air pollutants on the risk of tuberculosis outpatient visits, including the Poisson generalized linear regression model and distributed lag non-linear model (DLNM). We performed stratified analyses for the season, type of cases, gender, and age. We used the lag-specific relative risks and cumulative relative risk obtained by increasing pollutant concentration by per 10 units to evaluate the connection between six air pollutants and TB; PM2.5 (RR = 1.0018, 95% CI: 1.0004–1.0032, delay of 12 days) and SO2 (RR = 1.0169, 95% CI: 1.0007–1.0333, lag 0–16 days) were 0.9549 (95% CI: 0.9389–0.9712, lag 0 day) and 0.8212 (95% CI: 0.7351–0.9173, 0–20-day lag). Stratified analyses showed that seasonal differences had a greater impact on TB, males were more likely to develop TB than females, older people were more likely to develop TB than younger people, and air pollution had a great impact on new cases. Exposure to O3, CO, PM10, PM2.5, and NO2 increases the risk of TB outpatient visits, except SO2 which reduces the risk. The incidence of TB has seasonal fluctuations. It is necessary for the government to establish a sound environmental monitoring and early warning system to strengthen the monitoring and emission management of pollutants in the atmosphere. Management, prevention, and treatment measures should be developed for high-risk groups (males and older people), reducing the risk of TB by reducing their specific behaviors and changing their lifestyle. We need to pay more attention to the impact of seasonal effects on TB to protect TB patients and avoid a shortage of medical resources, and it is necessary for the government to develop some seasonal preventive measures in the future.


Introduction
Tuberculosis (TB) is a chronic contagion caused by tuberculosis bacilli infection. Mycobacterium TB may invade all sorts of organs of the human body, basically, encroach on the lung, called pulmonary tuberculosis (PTB). Tuberculosis is a global public health problem that seriously endangers public health. Tuberculosis is one of the most dangerous infectious diseases, causing the world's top 10 deaths (Wang et al. 2021). There are about 10 million new cases and 1.5 million deaths caused by TB every year. China is a country with a high burden of tuberculosis, ranking second in the world. It is reported that in 2019, an estimated 8.9 million to 11 million people worldwide infected with TB, with China ranking third, accounting for 8.4% of the global total (WHO 2020). We know from previous studies that poor nutrition, weather, diabetes, smoking, weakened immune systems, HIV/ AIDS, air pollution, alcohol consumption, and many other factors all play a large or small role in the progression of TB (Kim 2014;Rivas-Santiago et al. 2015;Li et al. 2019).
With the development and deepening of epidemiological studies, more and more evidence show the influence of air pollutants on tuberculosis. A recent review summarized the links and potential mechanisms between respiratory bacterial infections and air pollution and described the association between various air pollutants and the occurrence and development of tuberculosis by analyzing existing epidemiological studies (Pompilio and Di Bonaventura 2020). Another literature review also showed that the pollutant with the most significant impact on TB was fine particulate matter PM 2.5 (< 2.5 µm); nitrogen dioxide (NO 2 ), fine particulate matter PM 10 (< 10 µm), and sulfur dioxide (SO 2 ) also showed a significant correlation (Popovic et al. 2019). A meta-analysis published in March of this year pooled 17 related studies and suggested that long-term exposure to PM 10 , SO 2 , or NO 2 was associated with higher TB incidence (Xiang et al. 2021). An ecological study conducted in the USA, China, and Canada has observed a correlation between coal consumption and tuberculosis. The decline in TB rates has coincided with a decline in household coal consumption, indicating the link between air pollution and TB (Tremblay 2007). However, the available results on the relationship between air pollution and TB risk are inconsistent. Two studies in Mexico have shown that exposure to airborne particles inhibits the body's immune response to the TB bacterium. PM 2.5 and PM 10 can hinder the human body to produce cytokines and antimicrobial peptides, impair the immunity of the human lung, and promote the occurrence and development of TB Torres et al. 2019). However, other studies indicated that short-term or long-term exposure to PM 2.5 or PM 10 was not associated with TB risk (Liu et al. 2018;Xu et al. 2019;Xiang et al. 2021). A study in Seoul, South Korea, has confirmed that the interquartile increase in SO 2 concentration was associated with a 7% increment in TB incidence (Hwang et al. 2014), but a recent study conducted in Hefei suggests that exposure to SO 2 is inversely associated with tuberculosis (Huang et al. 2020a, b, c). Some studies show that SO 2 can enter the cell membrane of microorganisms, interfere with the activities of enzymes and proteins, inhibit the growth of microorganisms, and play an antibacterial role. And SO 2 can regulate lung and cardiovascular function (Álvaro-Meca et al. 2016;Ge et al. 2017). Exposure to NO 2 and ozone (O 3 ) can affect the number of TB clinics (Lin et al. 2019;Huang et al. 2020a, b, c;Liu et al. 2021). Air pollutants such as ozone and nitrogen NO 2 mainly cause the increase of free radicals in the body, leading to oxidative stress and inflammation . A study in Taiwan showed that CO exposure showed the highest population attributable fraction; in a region with higher ambient air pollution, it is most likely (80% risk probability) that the contributions of CO exposure to the development of TB were 1.6-12.2% (Lin et al. 2019). Studies suggest that heme oxygenase is produced by carbon monoxide, which has an inhibitory effect on the body's immune response and inflammation (Yang et al. 2020). Some studies have found gender and demographic differences in the relationship between air pollutants and tuberculosis. For example, a study in Hefei, China (Huang et al. 2020a, b, c), showed that the effect of NO 2 exposure on the risk of outpatient tuberculosis visits remained statistically significant in male and young subgroups. In addition, the elderly are more susceptible to PM 2.5 . In one Chengdu study, SO 2 exposure was found to have a statistically significant increase in active TB cases only in the male group (Zhu et al. 2018). A time-series study in Jinan found that most air pollutants (PM 2.5 , SO 2 , O 3 , and CO) were significantly associated with an increased risk of TB in people younger than 60 years of age (Liu et al. 2018). In addition, although there is a lot of research in developed countries, still less research in China, so more research is urgently needed to assess the relationship between TB and air pollution in China. Fuyang is located in the northwest of Anhui Province. Because of the rapid development of the economy and the constant growth of population, Fuyang is facing a severe air pollution problem. In addition, Fuyang has a large population base and is the most populous city in Anhui Province.
The association between various air pollutants and TB has also been explored in time-series studies conducted in multiple Chinese cities (Chengdu, Hefei, and Wuhan) (Zhu et al. 2018;Huang et al. 2020a, b, c). However, few studies have comprehensively explored the effects of exposure to six pollutants, and the influence between contaminants cannot be entirely excluded by the multi-pollutant model. Therefore, we aim to explore the relationship between daily exposure to PM 10 (particulate matter < 10 μm in aerodynamic diameter), sulfur dioxide (SO 2 ), PM 2.5 (particulate matter < 2.5 μm in aerodynamic diameter), nitrogen dioxide (NO 2 ), ozone (O 3 ), and carbon monoxide (CO) and the risk of tuberculosis outpatient visits in Fuyang. Subgroup analyses for seasonal variation, type of cases, sex, and age were also performed to assess their correction effects.

Ethical statement
This study was approved by the Anhui Medical University Ethics Committee. All patient information included in the study was unidentified and anonymous.

Study location
Fuyang (114° 52′-116° 49′ E, 32° 25′-34° 04′ N) is located in the southern part of the Huang-Huai-Hai Plain. It is a northwest city with the largest population in Anhui Province ( Fig. 1). By the end of 2019, the permanent resident population of Fuyang City was 8.259 million, an increase of 52,000 over the previous year, with a year-on-year increase of 0.63%. Fuyang City is located in the southern margin of the warm temperate zone, a warm temperate semi-humid monsoon climate. The four seasons are distinct, and the rainfall is moderate. Because Fuyang is near the Huaihe River in the south, and the north subtropical humid monsoon climate is in the south of the Huaihe River, the climate of Fuyang is characterized by a transition zone from the warm temperate zone to north subtropical zone (Anhui Provincial Bureau of Statistics of China, 2020). In recent years, due to population growth and economic development, Fuyang City has faced a severe air pollution problem.

Tuberculosis data
Tuberculosis is a class B infectious disease, one of the contagious diseases that must be reported in China. All case reports in this study were collected from the Tuberculosis Prevention and Control Institute of Anhui Chest Hospital and reported by Fuyang Center for Disease Control and Prevention. All cases in this study were tuberculosis of pulmonary. We collected admission information for all TB patients from January 1, 2015, to December 31, 2020, including age, gender, type of the cases, home address, and date of the current visit. In addition, because of the long incubation period of TB, typically between 4 and 8 weeks, the patient's time of infection is hard to confirm. Therefore, we included recurrent and new cases in this study (Huang et al. 2020a, b, c).

Statistical analysis
Descriptive statistical The demographic characteristics of tuberculosis were analyzed using simple descriptive statistical methods, including sex and age. A simple descriptive analysis of the distribution of air pollutant data and meteorological data was also carried out. The correlation among meteorological factors, air pollutants, and tuberculosis outpatient visits was described by the Spearman correlation coefficient and scatter plot.
Construction of the statistical model Based on previous biological findings that ambient air pollutants have a delayed effect on health outcomes, most epidemiological studies have found nonlinear exposure-response relationships (Chen et al. 2017;Guo et al. 2017). So, we used a distributed lag nonlinear model (DLNM) to examine the relationship between air pollutants and the risk of tuberculosis outpatient visits (Gasparrini 2014). In addition, because the numbers of daily reports of TB outpatients were generally considered to be rare events, we used a generalized linear model (GLM) based on the quasi-Poisson distribution to construct a function fitting the exposure-response and effects of lag (Huang et al. 2020a, b, c). First, to understand the influence of exposure to six air pollutants on the risk of tuberculosis outpatient visits, we used the function of natural cubic spline (7 degrees of freedom per year) to control for long-term trends and seasonal variations (Bhaskaran et al. 2013;Zhu et al. 2018). Meanwhile, the ns function with 3 degrees of freedom is used to control the mean temperature (MT) and relative humidity (RH) of meteorological factors (Chen et al. 2010(Chen et al. , 2017. We also controlled for the day of the week (Dow) and holiday effect. We classify the virtual variable holidays and assign values to holidays and non-holidays, 1 represents holidays, and 0 represents non-holidays.
Then, we inquired two-dimensional connection between air pollutants and the exposure, lag and response of tuberculosis outpatient visits by the cross-basis functions, with the initial maximum lag day set at 28 days (TB incubation period 28-56 days), using linear cubic spline function fitting exposed-response relationship, natural cubic spline function fitting lag-response relationship, because the purpose of this study was to investigate the effects of shortterm exposure. Therefore, we calculated different cumulative lag effects of different pollutants through a single-pollutant model, selected the lag days when the first peak or lowest value was reached, and obtained the QAIC (quasi-Poisson Akaike information criteria) values of different lag days (Supplementary Table S1). Finally, the best lag date was selected by QAIC (Gasparrini et al. 2010;Huang et al. 2020a, b, c). In order to solve the collinearity relationship between air pollutants and meteorological factors, we only added variables with correlation values less than 0.7 into the model (Liu et al. 2017). The basic model was as follows: At is the actual tuberculosis outpatient visits in t days, et is the expected tuberculosis outpatient visits t days, W T X is a cross-basis function, ns represents the natural cubic spline function, df 1 and df 2 represent the degrees of freedom of average temperature and relative humidity, respectively, and and are the regression coefficients of Dow and holiday respectively.
We regarded the level 1 concentration limits as a reference in the National Environmental Quality Standard (GB3059-2012), using RR (relative risk) estimates and 95% CI (confidence intervals) to represent the lag specificity and cumulative risk of TB outpatient visits with each 10-unit increase in six air pollutant concentrations. After a single-pollutant regression model, we modeled the effects of air pollution by gender (male, female), age (< 15 years old, 15-65 years old, and > 65 years old), case (new case and recurrent case), and season (cold and warm). Calculate the RR and 95% CI using the following formula: where Ê 1 and Ê 2 are the point estimates for the two subgroups and Ŝ E 1 and Ŝ E 2 are the corresponding standard errors, respectively (Huang et al. 2020a, b, c).
To test the robustness of the results, we also performed sensitivity analysis: (a) the multi-pollutant models were used to adjust for confounding factors among air pollutants; (b) different degrees of freedom were taken in the ns function of time variables (4-6 df); and (c) changing the degrees of freedom in the ns function of meteorological variables (4-7 df). All analyses were performed through the "splines," "dlnm," and "mgcv" packages in the R software (version 4.0.0). Table 1 provides summary statistics of the variables in this study, including TB case, atmospheric pollutants, and meteorology measures. The daily mean concentration of PM 2.5 was 55.18 μg/m 3 (0-276 μg/m 3 ), PM 10 was 85.89 μg/ m 3 (0-379 μg/m 3 ), SO 2 was 13.33 μg/m 3 (2-237 μg/m 3 ), CO was 0.77 mg/m 3 (2-2.9 mg/m 3 ), NO 2 was 32.09 μg/ m 3 (9-93 μg/m 3 ), and O 3 was 92.77 μg/m 3 (0-226 μg/m 3 ). From 2015 to 2020, there were 21,591 cases of tuberculosis, 6384 females and 15,207 males, 93.7% were new cases, and 6.3% were recurrent cases. Among them, 0.94% were younger than 15 years old, 30.26% were older than 65, and 68.8% were between 15 and 65 years old. The minimum and maximum daily mean temperatures were 7.8℃ and 33.5℃, respectively. The minimum value of relative humidity is 33%, and the maximum is 99%. Figure 2 shows the Spearman correlation coefficient and scatter plot. The distribution of air pollutant concentrations and tuberculosis hospitalizations from 2015 to 2020 are presented in Fig. 3 and Supplementary Fig. S2.

Overall association between daily TB outpatient visits and air pollutants
As shown in Supplementary Fig. S1, from the overall exposure-response effect, it can be concluded that the exposure concentrations of PM 10 , PM 2.5 , NO 2 , O 3 , and CO have a positive relationship with the risk of tuberculosis. In contrast, SO 2 has a negative relationship with the risk of tuberculosis.

Association between daily TB outpatient visits and PM 10
In the single-pollutant model, the daily lag effect of increasing PM 10 concentration by 10 μg/m 3 reached statistical significance on the 14th day and lasted until the 20th day (RR = 1.001, 95% CI:

Association between daily TB outpatient visits and CO
As shown in the Fig. 4, when the lag time was 11 days, the single-day lag risk of the CO single-pollutant model reached statistical significance (RR = 1.0037, 95% CI: 1.0005-1.0068, lag 11 days). The cumulative lag effect of CO exposure showed a bimodal distribution. The first time was statistically significant with a lag of 0-1 days (RR = 1.013, 95% CI: 1.0001-1.0261, lag 0-1 day), and the second time was statistically significant with a lag of 0-13 days (RR = 1.0337, 95% CI: 1.0036-1.0647, lag 0-13 days). When stratified by sex, age, and season, the effect of CO exposure was still statistically significant in males, 15-65 years old, > 65 years old, new case, and cold quarter subgroups (Figs. 5, 6 and 7, and Supplementary  Table S4). Figure 4 shows that O 3 exposure immediately increased the risk of TB outpatient visits (RR = 1.049, 95% CI: 1.0012-1.0086, 0-day lag). The cumulative risk curve was bimodal and statistically significant after 0-4 and 0-9 days of lag. In the subgroup analysis, the effect of O 3 had a significant short-term effect in male, female, > 65 years old, 15-65 years old, new case, and cold quarter subgroup (Figs. 5,6,7,and Supplementary Table S5).

Association between daily TB outpatient visits and SO 2
Different from other pollutants, when an increase of 10 μg/ m 3 in SO 2 concentration, the risk of TB outpatient visits was reduced. In the single-pollutant model, the singleday lag risk of SO 2 reached statistical significance at 0 day (RR = 0.9549, 95% CI: 0.9389-0.9712, lag 0 day). The cumulative lag risk reached its lowest at 20 days (RR = 0.8212, 95% CI: 0.7351-0.9173, 0-20-day lag). In the subgroup analysis, all subgroups except the < 15 years old and recurrent case had statistical significance (Figs. 5,6,7,and Supplementary Table S6).

Sensitivity analysis
Sensitivity analysis: In the multi-pollutant model, we changed the degree of freedom of the time variable ns (4-6df) and the meteorological variable ns (4-7df), the principal results showed no significant change (Table 2 and   Supplementary Table S8-25), as we can see from Table 2, other pollutants were added based on each single pollutant model, PM2.5, PM10, CO, NO2 and O3 still increased the risk of tuberculosis, while SO2 still showed a protective effect. The results show that the model has good performance, and the calculation results are reliable.

Discussion
In this study, we used time-series analysis to evaluate the short-term impact of air pollution exposure on the risk of tuberculosis outpatient visits in Fuyang from 2015 to 2020. We found that exposure to O 3 , CO, PM 10 , PM 2.5 , and NO 2 increases the risk of TB outpatient visits, and SO 2 reduces the risk. In addition, subgroup analysis showed that the pollutants had a more significant impact on men and the elderly. The multi-pollutant model results showed that the effects of PM 2.5 , CO, SO 2 , and NO 2 on the TB visit risk remained stable after adjustment, while the effects on PM 10 and O 3 were reduced to some degree.
The findings of this study, which are supported by several previous studies, suggest that tuberculosis is affected by PM 2.5 concentrations (Jassal et al. 2013;Smith et al. 2014;Liu et al. 2017). Two recent time-series studies in the Chinese cities of Jinan and Hefei also confirmed the risk of PM 2.5 on tuberculosis (Liu et al. 2018;Huang et al. 2020a, b, c). Several potential mechanisms can be used to explain the harm of PM 2.5 on TB: (1) When PM 2.5 exposure is increased, the immune function of the human respiratory system will be altered or impaired. Exposure may harm lung immunity by inducing nitrosation stressors and oxidation (Nel 2005). Therefore, in combination with the results of this study and previous studies, it can be seen that PM 2.5 can damage the immune response of the lungs and accelerate the progress of tuberculosis. We can see that from some epidemiological studies, PM 10 , has a similar effect and reaction mechanism to PM 2.5 (Rivas-Santiago et al. 2015; Kim et al. 2020;Pompilio and Di Bonaventura 2020;Xiang et al. 2021). PM 2.5 and PM 10 are so small in diameter that they can pass through the physical barrier of the respiratory tract. The aerosol nuclei containing Mycobacterium tuberculosis were colonized deep in lung tissue with PM 2.5 and PM 10 as carriers. This is also a risk factor for TB infection (Yang et al. 2020).

Fig. 4
Lag-specific relative risks (95% CI) and cumulative risks (95% CI) of per 10-unit increase in the daily concentrations of air pollution on TB outpatient visits on different lag days, according to the singlepollutant model; TB, tuberculosis; PM2.5, particulate matter < 2.5 μm in aerodynamic diameter; PM10, particulate matter < 10 μm in aerodynamic diameter; CO, carbon monoxide; O3, ozone; SO2, sulfur dioxide; NO2, nitrogen dioxide ◂ We observed a protective effect of sulfur dioxide. Similar protective effects have been reported in cities such as Madrid in Spain and Ningbo and Hefei in China (Álvaro-Meca et al. 2016;Ge et al. 2017;Huang et al. 2020a, b, c). This protective effect of SO 2 is thought to be due to the antibacterial properties of SO 2 . The ability of SO 2 to block enzyme activity in microbial cell membranes may account for the antimicrobial properties (Ge et al. 2017). An in vitro study of SO 2 also demonstrated its inhibitory effect on the growth of Mycobacterium tuberculosis (Malwal et al. 2012). One study reported that male mice inhaled small amounts of SO 2 which can increase levels of inflammatory cytokines in lung tissue and inhibit the proliferation of Mycobacterium tuberculosis (Meng et al. 2005). However, some related studies have obtained the opposite result of our conclusion Fig. 5 Lag-specific relative risks (95% CI) of per 10-unit increase in the daily concentrations of air pollution on TB outpatient visits on different lag days, according to the single-pollutant model stratified by age; TB, tuberculosis; PM2.5, particulate matter < 2.5 μm in aero-dynamic diameter; PM10, particulate matter < 10 μm in aerodynamic diameter; CO, carbon monoxide; O3, ozone; SO2, sulfur dioxide; NO2, nitrogen dioxide Fig. 6 Lag-specific relative risks (95% CI) of per 10-unit increase in the daily concentrations of air pollution on TB outpatient visits on different lag days, according to the single-pollutant model stratified by gender and season; TB, tuberculosis; PM2.5, particulate matter < 2.5 μm in aerodynamic diameter; PM10, particulate matter < 10 μm in aerodynamic diameter; CO, carbon monoxide; O3, ozone; SO2, sulfur dioxide; NO2, nitrogen dioxide Lag(days) 30666 Environ Sci Pollut Res (2022) 29:30656-30672 that SO 2 is a risk factor (Liu et al. 2020. Therefore, it is necessary to study further the mechanism of action of SO 2 in the human body to provide a more comprehensive and scientific explanation. With regard to O 3 , we observed a positive correlation with the risk of TB. The lag effect was observed to be statistically significant at 0 day, indicating an acute effect of O 3 on the human body. The independent effect of O 3 on increased TB risk also was observed in a multi-city model study in Shandong Province, China . Laboratory evidence has shown that O 3 increases the incidence of airway inflammation, impairs lung function, and affects lung gas exchange (Smith et al. 2016). However, a nested case-control study in Northern California showed that O 3 exposure above the lowest quintile resulted in a reduced risk of tuberculosis. A time-series study from Hefei, China, also concluded that O 3 reduces the risk of tuberculosis outpatient visits (Smith et al. 2016;Huang et al. 2020a, b, c). Due to the differences in study results and the unclear biological mechanisms of O 3 associated with TB infection, further studies are recommended to explain the effect of O 3 on TB risk.

Fig. 7
Lag-specific relative risks (95% CI) of per 10-unit increase in the daily concentrations of air pollution on TB outpatient visits on different lag days, according to the single-pollutant model stratified by the type of cases; TB, tuberculosis; PM2.5, particulate matter < 2.5 μm in aerodynamic diameter; PM10, particulate matter < 10 μm in aerodynamic diameter; CO, carbon monoxide; O3, ozone; SO2, sulfur dioxide; NO2, nitrogen dioxide The results of this study suggested that CO exposure increases the risk of TB outpatient visits. Similar protective effects have been reported in northern California. Those exposed to CO in the highest quintile had a 50% increase in TB prevalence compared with those exposed to the lowest quintile (Smith et al. 2016). Two studies in Wuhan and  (Park et al. 2003;Tremblay 2007). An experiment in mice found that exposure to diesel exhaust, including CO, reduced mRNA expression of several pro-inflammatory cytokines and cell signaling enzymes that control Mycobacterium tuberculosis infection (Hiramatsu et al. 2005).
Our study showed a positive correlation between NO 2 and TB. Two time-series studies in Chengdu and Hefei also reached similar conclusions (Zhu et al. 2018;Huang et al. 2020a, b, c). A study in Taiwan has shown that NO 2 is associated with tuberculosis which contributed 1.2-9.8% to developing TB (Lin et al. 2019). From a biochemical point of view, NO 2 is an irritating gas that can corrode and damage the alveoli and lower respiratory tract, and NO 2 is a component of photochemical smog and has acute toxic effects on the lungs (Wang et al. 2014;Smith et al. 2016). These may be the reasons for NO 2 's influence on TB outpatient visit risk.
In the subgroup analysis, we found that pollutant exposure had a more significant impact on the male group, and all pollutants except NO 2 were statistically significant. This may be due to men's greater exposure to outdoor environments and lifestyle factors such as smoking and drinking. In addition, there are substantial differences in hormonal status and airway physiology between males and females, which may also be one of the reasons (Sopori et al. 1998). In the age subgroup, the effects of pollutants on older age groups were greater. The possible reason is that with age, the body's immune defenses decline, and the body is more likely to be exposed to pollutants than younger people Huang et al. 2020a, b, c). Interestingly, contrary results were observed for PM 2.5 and PM 10 in the cold and warm seasons. In the cold season group, the risk of tuberculosis diagnosis was still increased. In contrast in the warm season group, PM 2.5 and PM 10 reached statistical significance when they lagged by 2 days and 1 day, respectively, reducing the risk of tuberculosis outpatient visits. Some possible reasons can explain this phenomenon: (1) The temporal distribution of particulate matter pollutants showed noticeable seasonal differences, with high in the cold season and low in the warm season. (2) During periods of high ambient air pollutants, particularly particulate matter exposure at high concentrations visible to the naked eye, humans prefer to stay indoors, leading to increased exposure and further increasing the risk of tuberculosis (Huang et al. 2020a, b, c). But the actual mechanism remains unclear, and more research is needed to determine the cause. In the subgroup analysis of case types, it was obvious that air pollution had a great impact on the new cases, and no significant effect was observed except for NO 2 in the recurrent case. The possible reason is that those who have had TB in the past have higher self-protection awareness, higher use of masks and lower exposure to air pollution than those who have never had TB.
Our study has several strengths. First, we explored for the first time the relationship between air pollution exposure and tuberculosis outpatient visits in Fuyang City, which is the most populous city in Anhui Province and has a good representation. Second, the impact of six contaminants on TB visit risk was thoroughly explored, and the sensitivity analysis showed that the results were robust and reliable. However, there are still some shortcomings. First of all, this study adopted the average daily concentration of fixed stations as the exposure concentration of air pollution, without excluding the influence of spatial heterogeneity of air pollution. Second, the ecological fallacy is inevitable. For example, we did not consider other risk factors, such as clusters of cases. Third, this study was limited to one area in Fuyang, which means that our findings should be carefully extended to other sites.

Conclusion
The results showed that six air pollutants all had an impact on the risk of tuberculosis outpatient visits. PM 10 , PM 2.5 , NO 2 , O 3 , and CO increased tuberculosis outpatient visits, while SO 2 decreased the number of tuberculosis outpatient visits. Men, the elderly, and new cases are more likely to be affected by air pollution, and the effects of the season are more pronounced than age or sex. The government should strengthen the monitoring of pollutants and reduce the discharge of pollutants. Protecting susceptible groups during the high-incidence season is of great significance to reduce the incidence of TB. Therefore, we should give more comprehensive prevention and treatment to high-risk groups. Meanwhile, more measures should be taken to against the effects of seasonal changes. and drafted the manuscript, and Xiao-Hong Kan and Xiu-Jun Zhang revised the manuscript. Cheng-Yang Hu, Kai Huang, Kun Ding, Xiao-Jing Yang, Xin Cheng, Kang-Di Zhang, Wen-Jie Yu, Jie Wang, Yong-Zhong Zhang, and Zhen-tao Ding contributed in collecting the data. Xin-Qiang Wang and Ying-Qing Li contributed equally. All authors read and approved the final manuscript.
Funding This study was supported by the Anhui Medical University (2019xkj019), Anhui Provincial Natural Science Foundation (2008085MH063), National Key Project for Infectious Disease (2018ZX10722301-001-004), and Major National Science and Technology Projects during the 12th Five-Year Plan period (2013ZX10003008-001-003).

Data availability
The datasets analyzed during the current study are available from the corresponding author on reasonable request.

Declarations
Ethics approval Not applicable.

Competing interests
The authors declare no competing interests.