Spatiotemporal trends and drivers of pulmonary tuberculosis incidence in China in the past two decades

Spatiotemporal trends and drivers of pulmonary tuberculosis incidence in China in the past two decades | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Spatiotemporal trends and drivers of pulmonary tuberculosis incidence in China in the past two decades Chen-Fan Zhang, Jue-Wei Pan, Gui-Hua Yang, Yi-Wen Li, Sun-Huan Su, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8693705/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Purpose: To analyze the spatiotemporal evolution pattern of tuberculosis incidence in China from 2004 to 2023 and reveal various driving factors. Methods: Using national tuberculosis surveillance data, economic indicators, and environmental information, we employed spatiotemporal econometric models, geographic detectors, and random forests for a comprehensive analysis. Results: Overall, tuberculosis incidence rates in China have been declining; however, significant regional disparities persist. The western region consistently demonstrates higher incidence rates compared to the eastern region. Furthermore, there has been an annual increase in rifampicin-resistant cases from 2017 to 2020. Upon comparing five distinct spatiotemporal econometric models, the spatiotemporal geographically weighted regression (GTWR) model (R² = 0.950) emerged as the most effective, indicating that the impact of various factors exhibits both spatial and temporal variability. Population density (PD) and particulate matter 10 (PM 10 ) concentration were correlated with elevated incidence rates. In contrast, the proportion of the urban population, normalized difference vegetation index (NDVI), and ozone (O 3 ) concentration were correlated with reduced rates. Geographic detector analysis further identified NDVI, PD, and PM 10 as critical determinants, revealing statistically significant interactive effects among these variables. The random forest model demonstrated a complex, non-linear relationship between various factors and the incidence rate. Conclusions: This study emphasizes the importance of integrating socioeconomic, environmental, and population factors to understand tuberculosis transmission dynamics and provides a strong foundation for developing targeted prevention and control strategies. Funding: The research was supported by the 2024 Sanming City Health and Wellness Science and Technology Innovation Joint Project (No. 2024-S-011) and the Startup Fund for Scientific Research at Fujian Medical University (No.2023QH1286). Tuberculosis Spatiotemporal Analysis Spatiotemporal Geographically Weighted Regression Geographic Detector Machine Learning Public Health Policy Figures Figure 1 Figure 2 Figure 3 Figure 4 INTRODUCTION Tuberculosis (TB) remains a significant global public health threat as one of the most lethal infectious diseases. The World Health Organization's "Global Tuberculosis Report 2025" indicates that in 2024, there were approximately 10.7 million new tuberculosis cases globally, resulting in 1.2 million deaths [ 1 ]. Worldwide, tuberculosis remains one of the top ten causes of death and the leading cause of death from a single infectious source [ 1 , 2 ]. Specifically, as a country with a high tuberculosis burden, China has made significant progress in tuberculosis prevention and control. However, it still faces severe challenges, including considerable regional variation in incidence—meaning tuberculosis cases vary greatly across different areas [ 3 ]—and an increase in rifampicin-resistant tuberculosis cases [ 4 , 5 ]. Previous studies indicate that a complex interplay of socioeconomic, environmental, and demographic factors influences the transmission and distribution of tuberculosis [ 6 – 9 ]. Traditional global regression models, which assume spatial stationarity, often fail to capture the pronounced spatiotemporal heterogeneity underlying these associations [ 10 ]. With advances in geographic information systems and spatial statistical techniques, models such as geographically weighted regression (GWR) and geographically and temporally weighted regression (GTWR) have proven more effective in revealing the spatial non-stationarity of these relationships [ 11 ]. In addition, the driving factors of tuberculosis transmission exhibit intricate interactions and nonlinear effects, necessitating the integration of machine-learning approaches to achieve deeper mechanistic insights and more accurate predictive modeling. This study utilizes China’s pulmonary tuberculosis surveillance data from 2004 to 2023, along with socioeconomic and environmental indicators, to create a comprehensive spatiotemporal analysis framework. By integrating spatiotemporal econometrics, geographic detectors, and machine learning techniques, the research aims to systematically analyze the spatiotemporal dynamics and various factors influencing the incidence of pulmonary tuberculosis in China. Consequently, the study will provide a solid evidence base for precision-targeted prevention and control strategies for tuberculosis. METHODS Data sources The data on the incidence of pulmonary tuberculosis and its related influencing factors used in this study come mainly from the following sources. Pulmonary Tuberculosis Incidence Data These data come primarily from two sources. One is the "China Health and Wellness Statistics Yearbook" (https://www.zgtjnj.org), which provides detailed information on the incidence rate and mortality rate of pulmonary tuberculosis from 1997 to 2023. The other is the Public Health Science Data Center (https://www.phsciencedata.cn/Share/), which has compiled annual statistical data on the incidence rate, number of new cases, number of deaths, and mortality rate of rifampicin-resistant pulmonary tuberculosis across 31 provinces nationwide (excluding Hong Kong, Macau, and Taiwan) from 2017 to 2021. Population data Population data, including population density (PD), proportion of urban population (PUP), and the proportion of the population aged 65 and over, are sourced from the annual "China Statistical Yearbook" (http://www.stats.gov.cn/tjsj/ndsj/). Environmental data Environmental data are sourced from various public databases. The normalized difference vegetation index (NDVI) data are obtained from the official website of the National Aeronautics and Space Administration (https://ladsweb.modaps.eosdis.nasa.gov/search/), which quantifies vegetation coverage by measuring the difference between near-infrared and red-light bands. The annual average concentration data of air quality indices such as particulate matter 2.5 (PM 2.5 ), particulate matter 10 (PM 10 ), sulfur dioxide (SO 2 ), nitrogen dioxide (NO 2 ), carbon monoxide (CO), and ozone (O 3 ) are sourced from the National Qinghai-Xizang Plateau Scientific Data Center (https://data.tpdc.ac.cn/). Geographic information data Geographic information is sourced from the National Basic Geographic Information Center (http://www.ngcc.cn/). The base maps of China used in this study are derived from standard maps provided by the Ministry of Natural Resources' standard map service website (http://bzdt.ch.mnr.gov.cn/) with the review number GS (2019) 1823. Other data Economic factor data, such as regional per capita GDP and per capita consumption expenditure in urban and rural residents, are sourced from the "China Statistical Yearbook." Healthcare resource factors include the number of healthcare technicians per 10,000 population, the number of practicing (assistant) physicians per 10,000 population, and the number of registered nurses per 1,000 population. These statistics are sourced from the "China Health and Wellness Statistics Yearbook." Influencing factors information Based on an extensive literature review, this study collected various factors that may influence the incidence of tuberculosis, mainly including population factors [8], socioeconomic factors [7], natural environmental factors [12], and healthcare resource factors [13]. Specifically, population factors include PD, PUP, proportion of the population aged 65 and over, employment ratios in primary, secondary, and tertiary industries, and the crude divorce rate. Economic factors encompass regional per capita gross domestic product (GDP) and per capita consumption expenditure for both urban and rural residents. Natural environmental factors encompass concentrations of SO 2 , NO 2 , CO, O 3 , PM 2.5 , and PM 10 , alongside vegetation coverage and the NDVI. Healthcare resource factors include the number of healthcare technicians per 10,000 population, the number of practicing (assistant) physicians per 10,000 population, and the number of registered nurses per 1,000 population. Data preprocessing All collected data were preprocessed to ensure accuracy and consistency before analysis. This study created a tuberculosis database using Microsoft Excel 2019. The tuberculosis incidence and mortality data of 31 provinces and cities across the country were linked with China's vectorized electronic map (*. shp format) using administrative region codes as the linking field in ArcGIS 10.8 software, thereby constructing a spatial database. The preprocessing steps are as follows. Data Cleaning We check the completeness and consistency of the data and handle missing values and outliers. For missing data, interpolation methods are used to fill the gaps. For outliers, identification and processing are carried out using the box plot method or the z-score method. Data Standardization To eliminate the influence of varying variable dimensions, Z-score standardization is applied to all influencing factor data. Data Integration Data from different sources is integrated into a unified database to ensure consistency and completeness. Stepwise Regression Analysis In stepwise regression analysis, key variables that may affect the incidence of tuberculosis, such as population density, per capita GDP, vegetation coverage, air quality, and the number of health technicians, are first screened. The forward selection method is used to introduce variables one by one and conduct significance tests, while also removing variables that have been introduced but are no longer significant. Ultimately, the optimal regression model is constructed to reveal the degree and direction of the impact of various factors on the incidence rate, providing a basis for subsequent spatiotemporal econometric analysis. Analysis of Factors Influencing the Incidence Rate of Tuberculosis : Spatiotemporal Econometric Model To identify the most suitable analytical model, this study employs various spatiotemporal econometric models for the empirical analysis of tuberculosis incidence. The specific steps are as follows. Multicollinearity Test Before constructing the regression model, a multicollinearity test is conducted on the independent variables considered for the model to avoid interference from high correlations among them. This study uses the variance inflation factor (VIF) for testing. If the VIF values of all independent variables are between 0 and 10, it indicates no severe multicollinearity, and subsequent regression analysis can proceed. Spatial Autocorrelation Analysis [14] Before constructing the geographically weighted regression model and the spatiotemporal geographically weighted regression model, spatial autocorrelation analysis is performed on the independent variables to assess spatial dependence among variables. This study determines whether there is spatial autocorrelation in tuberculosis incidence by calculating the global Moran's I index. Regression Model Construction We developed several statistical models to analyze the incidence of pulmonary tuberculosis, including an ordinary least squares model (OLS), a spatial lag model (SLM), a spatial error model (SEM), a geographically weighted regression model (GWR), and a spatial-temporal geographically weighted regression model (GTWR). By comparing the performance metrics of these models, we aim to identify the one that best captures the spatiotemporal dynamics of TB incidence [15]. Geographic Detector [ 16 ] This study employed the geographic detector method to explore the explanatory power of various influencing factors on the incidence of tuberculosis. The geographic detector is a statistical tool designed to identify spatial heterogeneity and its driving factors. It operates on the principle that a significant influencing factor will align the spatial distribution of disease incidence with that of the factor. This method comprises four modules: factor detector, interaction detector, risk zone detector, and ecological detector. The factor detector assesses how potential influencing factors explain the spatial heterogeneity of tuberculosis incidence using the q value. The q value, ranging from 0 to 1, signifies the factor's explanatory power, with higher values indicating greater strength. The interaction detector is used to assess whether the interaction between two influencing factors enhances, weakens, or acts independently on the risk of incidence. Random Forest Model [17] This study used the random forest model to draw partial dependence plots to visually display the nonlinear relationships between each independent variable and tuberculosis incidence. The random forest is an ensemble learning algorithm suitable for handling high-dimensional data and complex nonlinear relationships. Statistical Methods This study used R 4.3.3 and GraphPad Prism 9 software for statistical description and chart drawing, and ArcGIS 10.8 software for spatial analysis, spatiotemporal econometric model parameter estimation, and model construction. RESULTS Spatiotemporal Trends of Tuberculosis Incidence and Mortality From 2004 to 2023, the number of tuberculosis cases nationwide peaked at 1.17 million in 2008, and has been declining continuously, dropping to 610,000 by 2023 (Figure 1A). During the same period, the number of deaths fluctuated between 1,435 and 3,783, with a mortality rate maintained at 0.1–0.3 per 100,000, resulting in 2,167 deaths in 2023 (Figure 1B). The incidence rate peaked at 96.9 per 100,000 in 2005, and fell to 43.5 per 100,000 in 2023; the ARIMA model predicts it will drop to 23.5 per 100,000 by 2030 (Figure 1C). The incidence rate in the western region is significantly higher than in the eastern region (Supplemental Figure 1A-D), with Tibet (150.1 per 100,000), Xinjiang (109.9 per 100,000), and Guizhou (96.5 per 100,000) having the highest incidence rates in 2020, compared to the lowest rates in Tianjin (20.5 per 100,000), Shanghai (24.3 per 100,000), and Shandong (24.7 per 100,000), which are 4 to 7 times lower (Supplemental Figure 1D). By 2020, 18 of the 31 Chinese provinces, autonomous regions, and municipalities had incidence rates below 50 per 100,000 people (Supplemental Figure 1D). Since 2017, the reported cases of rifampicin-resistant tuberculosis have increased from 7,607 (0.6 per 100,000) to 15,489 (1.1 per 100,000) in 2020 (Figure 1E, 1F). The number of deaths fluctuated between 40 and 108 during this period, with a mortality rate of 0.004 per 100,000 in 2020 (Figure 1E, 1F). The incidence of rifampicin-resistant tuberculosis also shows a distribution pattern characterized by higher rates in the west and lower rates in the east, with the highest incidence rates in Tibet (5.4 per 100,000), Qinghai (2.9 per 100,000), and Jilin (2.6 per 100,000), compared to the lowest rates in Shandong (0.4 per 100,000), Shanxi (0.5 per 100,000), and Jiangsu (0.6 per 100,000), which are 4 to 13 times lower (Supplemental Figure 2A-2D). Construction and Comparison of Spatial Regression Models Through stepwise regression analysis, the two models with the highest adjusted R² values (Model 12 and Model 13) were selected. Model 13, which includes nine variables—PUP, PPI, PD, CO, NO 2 , O 3 , PM 10 , SO 2 , and NDVI—performed similarly to Model 12, which adds vegetation coverage. To avoid issues with multicollinearity, given that vegetation coverage is closely related to NDVI, Model 13 was ultimately selected (Supplemental Table 1). The multicollinearity test showed that the variance inflation factor (VIF) for all variables was less than 10, indicating good model stability (Table 1). The global Moran's I index was 0.496 (P < 0.05), confirming significant spatial autocorrelation in the data, necessitating the use of spatial econometric models. The OLS, SEM, SLM, GWR, and GTWR models were developed and compared, as shown in Table 2. Among these, the GTWR model demonstrated the highest goodness of fit, with an R 2 value of 0.950 and an Adjusted R 2 of 0.947. Additionally, it had the lowest AICc value of 162.5. This indicates that the GTWR model most effectively captures the spatiotemporal variations in tuberculosis incidence, leading to its selection as the final analysis model (Table 2). GTWR The results of the GTWR model indicate that the effects of various influencing factors on tuberculosis incidence exhibit significant spatial-temporal heterogeneity (P < 0.05) (Figure 2A-2I). From the perspective of average regression coefficients, among socio-economic factors, the PUP (coefficient ≈ -0.40) and the PPI (coefficient ≈ -0.20) both show significant negative impacts on the incidence rate (Figure 2B, 2C). Regarding environmental factors, O 3 (coefficient ≈ -0.35) (Figure 2A) and NO 2 (coefficient ≈ -0.30) (Figure 2G) in air pollutants exhibit significant negative impacts, while PM 10 (coefficient ≈ 0.35) (Figure 2D) shows a significant positive impact, but its influence intensity has a trend of weakening year by year (the coefficient was 0.59 in 2017, dropping to -0.08 in 2021); the impacts of CO (coefficient ≈ 0.10) (Figure 2F) and SO 2 (coefficient ≈ -0.06) (Figure 2E) are relatively weak. In addition, the NDVI (coefficient ≈ -0.20) (Figure 2I) also shows a negative impact. Among population factors, population density (coefficient ≈ 0.20) positively affects the incidence rate (Figure 2H). Geographical Detector Analysis Factor Detection Between 2017 and 2021, the q-values of all influencing factors passed the significance test (P < 0.05). The explanatory power of the NDVI (0.6-0.9), PD (0.6-0.7), and PM₁₀ (0.5-0.8) is relatively strong. The q-value for the PUP continues to increase (0.3-0.7). Following this, O₃ has a q-value between 0.3 and 0.5, while NO₂ ranges from 0.3 to 0.7. CO shows a weaker explanatory power, with a q-value of 0.2 to 0.4, and SO₂ has the weakest explanatory power, ranging from 0.1 to 0.4 (Figure 3A). Interaction Detection The interactions among all factors show a dual-factor enhancement. The interaction combinations of PD∩CO, PD∩PM₁₀, and NDVI∩O₃ have the highest q-values, reaching above 0.9, but overall, there is a downward trend year by year (Figure 3B). Random Forest and Partial Dependence Plot (PDP) Analysis PD, NDVI, and O 3 are the three most important variables affecting the incidence of tuberculosis (Figure 4). The partial dependence plot further reveals the complex nonlinear relationships between these factors and the incidence rate: the incidence rate decreases with the increase of NDVI, O 3 , and PUP, and tends to stabilize after reaching a certain threshold; conversely, the incidence rate increases with the increase of the proportion of PPI, PM 10 , and CO (Figure 4). In addition, NO 2 and SO 2 show inverted U-shaped and U-shaped relationships with the incidence rate, respectively, while the effect of PD exhibits nonlinear characteristics, i.e., the influence weakens at lower (<20 people per square kilometre) and higher (500 people per square kilometre) levels, while being most significant in the medium density range (150–200 people per square kilometre) (Figure 4). Discussion This study comprehensively utilizes descriptive epidemiology, spatial-temporal statistics, geographical detectors, and machine learning methods to systematically reveal the spatial-temporal dynamic patterns of tuberculosis incidence in China from 2004 to 2023, as well as the complex influencing mechanisms behind them. The main findings indicate that, although the overall incidence rate of tuberculosis in China shows a downward trend, the issue of rifampicin resistance is becoming increasingly prominent, and the disease burden exhibits significant spatial heterogeneity, with the western regions having a higher incidence than the eastern regions. Core analyses further confirm that socio-economic, environmental, and population factors dynamically, nonlinearly, and interactively drive the risk of tuberculosis transmission. This study finds that the reported incidence rate of tuberculosis in China continues to decline, marking a significant success of the massive investments made in tuberculosis prevention and control over the past two decades, which can be attributed to the comprehensive coverage of modern tuberculosis control strategies (DOTS), the improvement of the medical insurance system, and the enhancement of diagnostic levels [ 18 ]. However, the incidence rate in western regions (such as Tibet and Xinjiang) continues to be several times higher than that in the eastern coastal areas, revealing a close association between disease burden and the level of regional socio-economic development [ 14 ]. This spatial heterogeneity is closely related to factors such as uneven distribution of medical resources, higher poverty levels, and harsh natural conditions. More alarmingly, the significant rise in reported cases of rifampicin-resistant tuberculosis after 2017 poses a severe challenge to the goal of ending the tuberculosis epidemic by 2035, emphasizing the urgency of strengthening laboratory rapid diagnostic capabilities and the standardized management system for second-line drugs [ 19 ]. In terms of driving factor analysis, the comparative results of the models in this study strongly demonstrate that the GTWR model performs optimally in analyzing the influencing factors of tuberculosis incidence due to its ability to capture both spatial dependence and temporal non-stationarity. This finding aligns with the forefront direction of current research on the driving factors of complex disease environments, recognizing that exposure-response relationships have significant contextual dependencies [ 10 ]. The results of the GTWR model reveal that the intensity and direction of the effects of various factors exhibit significant spatial-temporal variations. For example, the urban population ratio acts as a protective factor in most areas, reflecting the improvement in accessibility to medical and health services accompanying urbanization; however, its effect weakens or even reverses in highly urbanized areas, possibly indicating increased risks due to intensified internal population flow and aggregation within cities [ 8 ]. This heterogeneity of effects suggests that a single global conclusion may obscure the real risks in local areas, making it crucial to adopt a precise prevention and control approach of "one place, one policy." An in-depth analysis of specific influencing factors reveals that the results of this study corroborate existing evidence while also highlighting new phenomena that require further exploration. The positive impact of population density, the risk effect of the PPI (as a proxy variable for rural and specific occupational exposure), and the positive correlation between PM 10 and incidence rate have been repeatedly confirmed by numerous studies [ 20 ]. The protective effect of the normalized difference vegetation index may be related to green spaces promoting physical activity, improving mental health, and purifying the air, supporting the necessity of increasing green spaces in the "healthy city" concept [ 21 ]. However, the significant negative associations of gas pollutants such as O 3 and NO 2 differ from some research conclusions. One possible explanation is the complexity of atmospheric chemical processes: O 3 and NO 2 , as secondary pollutants, have complex nonlinear relationships with primary pollutants (such as PM 2.5 ), and collinearity in the model may lead to seemingly contradictory results [ 22 ]. Another hypothesis that needs verification is that a certain concentration of O 3 has broad-spectrum bactericidal effects, which may inhibit the survival and transmission of Mycobacterium tuberculosis in the external environment to some extent [ 23 ]. The nonlinear relationships revealed by the random forest model (such as the U-shaped curve of SO 2 and the threshold effect of PD and NDVI) greatly enrich our understanding of the action patterns of these factors, indicating that their impacts are not simple monotonic linear relationships, providing scientific reference for accurately setting critical values for environmental governance and public health interventions [ 24 ]. Moreover, the strong dual-factor enhancement interactions revealed by the geographical detector are an important finding of this study. For example, the interaction q-values of "PD ∩ CO" and "PD ∩ PM 10 " are extremely high, strongly suggesting that in densely populated areas with heavy air pollution, the risk of tuberculosis transmission is synergistically amplified. This highlights that the epidemic of tuberculosis is the result of the interaction of the "social-economic-environmental" complex ecosystem, urgently requiring the breaking down of departmental barriers and promoting collaborative governance based on the concept of "integrated health [ 25 ]." Public health policies must be deeply integrated with environmental protection (emission reduction), urban-rural planning (optimizing population distribution and green space allocation), and industrial development (reducing agricultural population exposure) policies to fundamentally and effectively control tuberculosis. Based on the above findings, we propose the following policy implications: (1) Implement differentiated regional prevention and control strategies: for high-burden provinces in the west, strengthen investment in funding, technology, and personnel, focusing on enhancing the diagnostic capacity at the grassroots level and the prevention and control of drug-resistant tuberculosis; for low-incidence areas in the east, emphasize the management of tuberculosis among the floating population, active case finding, and epidemic warning [ 26 ]. (2) Promote cross-departmental collaborative interventions: clearly incorporate tuberculosis prevention and control goals into national and local air pollution management, green space system planning, and rural revitalization strategies to achieve health integration into all policies [ 27 , 28 ]. Limitations This study also has several limitations. First, ecological studies based on aggregated data cannot infer individual-level causal relationships and fail to control for the effects of individual behaviors and comorbid factors such as nutritional status, smoking, and diabetes. Second, the data rely on infectious disease reporting systems, and differences in diagnostic levels and reporting quality across regions may introduce some bias into the results. Finally, although we employed various statistical models to control for confounding factors, the possibility of residual confounding cannot be completely ruled out. Conclusion China has reduced the incidence of tuberculosis by half in the past 20 years; however, the burden remains significant in the West, with the presence of resistant strains. The risk of tuberculosis is influenced by factors such as population density, employment, and exposure to pollutants. To eliminate TB by 2035, it is essential to implement tailored regional strategies that include laboratory networks and targeted screening, alongside policies that address air quality and drug resistance. Future research should focus on the following directions: (1) Conduct research designs that integrate multi-level data (individual-community-environment) to assess the independent effects of various factors more accurately. (2) Combine whole-genome sequencing technology of Mycobacterium tuberculosis to clarify how environmental and social factors influence the transmission chain [ 29 ] of tuberculosis from the perspective of pathogen transmission dynamics. (3) Use high-precision models established in this study, such as GTWR, for prediction and design prospective intervention trials to assess the actual effects of specific policies (such as park green space construction) on reducing tuberculosis incidence, thereby providing strong evidence [ 30 ] for formulating more cost-effective prevention and control strategies. Declarations Acknowledgments The study was funded by the 2024 Sanming City Health and Wellness Science and Technology Innovation Joint Project. The funder had no involvement in the study's design or execution, the collection, management, analysis, or interpretation of the data, the preparation, review, or approval of the manuscript, or the decision to submit the manuscript for publication. Author contributions HWS designed the study, performed the statistical analyses, and drafted the initial manuscript. CFZ and JWP contributed to data collection and cleaning, and provided critical revisions to the manuscript. YWL and SHS assisted in the interpretation of the results and helped to refine the study's conclusions. All authors read and approved the final manuscript. Availability of Data Data is provided within the supplementary information files. Ethics approval The study utilized data from publicly available databases, and ethical approval was not needed. Consent to Publish declaration Not applicable. Conflicts of interest None. Funding The research was supported by the 2024 Sanming City Health and Wellness Science and Technology Innovation Joint Project (No. 2024-S-011) and the Startup Fund for Scientific Research at Fujian Medical University (No. 2023QH1286). References Global tuberculosis report 2025 [https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2025] Global tuberculosis report 2024 [https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2024] Li XX, Wang LX, Zhang J, Liu YX, Zhang H, Jiang SW, Chen JX, Zhou XN: Exploration of ecological factors related to the spatial heterogeneity of tuberculosis prevalence in P. R. China . Global health action 2014, 7 :23620. Wang L, Zhang H, Ruan Y, Chin DP, Xia Y, Cheng S, Chen M, Zhao Y, Jiang S, Du X et al : Tuberculosis prevalence in China, 1990-2010; a longitudinal analysis of national survey data . Lancet 2014, 383 (9934):2057-2064. Su W, Ruan YZ, Li T, Du X, Jiang JW, Li RZ: Characteristics of rifampicin-resistant tuberculosis detection in China, 2015-2019 . Infectious diseases of poverty 2021, 10 (1):99. Mohidem NA, Hashim Z, Osman M, Shaharudin R, Muharam FM, Makeswaran P: Demographic, socio-economic and behavior as risk factors of tuberculosis in Malaysia: a systematic review of the literature . Reviews on environmental health 2018, 33 (4):407-421. Liu JJ, Yao HY, Liu EY: Analysis of factors affecting the epidemiology of tuberculosis in China . Int J Tuberc Lung Dis 2005, 9 (4):450-454. Ren H, Lu W, Li X, Shen H: Specific urban units identified in tuberculosis epidemic using a geographical detector in Guangzhou, China . Infectious diseases of poverty 2022, 11 (1):44. Ren H, Lu W, Li X, Shen H: Specific urban units identified in tuberculosis epidemic using a geographical detector in Guangzhou, China . Infectious Diseases of Poverty 2022, 11 (1). Yu H, Yang J, Yan Y, Zhang H, Chen Q, Sun L: Factors affecting the incidence of pulmonary tuberculosis based on the GTWR model in China, 2004-2021 . Epidemiol Infect 2024, 152 :e65. Amaral AVR, González JA, Moraga P: Spatio-temporal modeling of infectious diseases by integrating compartment and point process models . Stoch Environ Res Risk Assess 2023, 37 (4):1519-1533. Yang J, Zhang M, Chen Y, Ma L, Yadikaer R, Lu Y, Lou P, Pu Y, Xiang R, Rui B: A study on the relationship between air pollution and pulmonary tuberculosis based on the general additive model in Wulumuqi, China . Int J Infect Dis 2020, 96 :42-47. Mahara G, Yang K, Chen S, Wang W, Guo X: Socio-Economic Predictors and Distribution of Tuberculosis Incidence in Beijing, China: A Study Using a Combination of Spatial Statistics and GIS Technology . Med Sci (Basel) 2018, 6 (2). Wang Q, Guo L, Wang J, Zhang L, Zhu W, Yuan Y, Li J: Spatial distribution of tuberculosis and its socioeconomic influencing factors in mainland China 2013–2016 . Tropical Medicine & International Health 2019, 24 (9):1104-1113. Wang Q, Guo L, Wang J, Zhang L, Zhu W, Yuan Y, Li J: Spatial distribution of tuberculosis and its socioeconomic influencing factors in mainland China 2013-2016 . Trop Med Int Health 2019, 24 (9):1104-1113. Hu Y, Xia C, Li S, Ward MP, Luo C, Gao F, Wang Q, Zhang S, Zhang Z: Assessing environmental factors associated with regional schistosomiasis prevalence in Anhui Province, Peoples' Republic of China using a geographical detector method . Infectious diseases of poverty 2017, 6 (1):87. Zhao D, Zhang H, Wu X, Zhang L, Li S, He S: Spatial and temporal analysis and forecasting of TB reported incidence in western China . BMC Public Health 2024, 24 (1):2504. Zuo Z, Wang M, Cui H, Wang Y, Wu J, Qi J, Pan K, Sui D, Liu P, Xu A: Spatiotemporal characteristics and the epidemiology of tuberculosis in China from 2004 to 2017 by the nationwide surveillance system . BMC Public Health 2020, 20 (1):1284. Su W, Ruan Y, Li T, Du X, Jiang J, He Y, Li R: Epidemiological Characteristics of Rifampicin-Resistant Tuberculosis in Students - China, 2015-2019 . China CDC Wkly 2021, 3 (26):549-552. Zhao M, Wu J, Figueiredo DM, Zhang Y, Zou Z, Cao Y, Li J, Chen X, Shi S, Wei Z et al : Spatial-temporal distribution and potential risk of pesticides in ambient air in the North China Plain . Environ Int 2023, 182 :108342. Song WM, Liu Y, Men D, Li SJ, Tao NN, Zhang QY, Liu SQ, An QQ, Zhu XH, Han QL et al : Associations of residential greenness exposure and ambient air pollutants with newly-diagnosed drug-resistant tuberculosis cases . Environ Sci Pollut Res Int 2024, 31 (18):27240-27258. Smith GS, Van Den Eeden SK, Garcia C, Shan J, Baxter R, Herring AH, Richardson DB, Van Rie A, Emch M, Gammon MD: Air Pollution and Pulmonary Tuberculosis: A Nested Case-Control Study among Members of a Northern California Health Plan . Environ Health Perspect 2016, 124 (6):761-768. Priĭmak AA, Kaliuk AN, Kirgintsev AG: [The effect of an ozone-oxygen mixture on Mycobacterium tuberculosis and conditionally pathogenic microorganisms] . Probl Tuberk 1991(4):7-10. Labib SM: Greenness, air pollution, and temperature exposure effects in predicting premature mortality and morbidity: A small-area study using spatial random forest model . Sci Total Environ 2024, 928 :172387. Huang L, He J, Zhang C, Liu J, Guo Z, Lv S, Zhang X, Li S: China's One Health governance system: the framework and its application . Sci One Health 2023, 2 :100039. Liu J, Yin H, Wang N, Wang Y, Guan L, Feng Y, Wu L, Liu W, Zhang H, Liu Z et al : Factors associated with exposure to tuberculosis education among internal migrants with diabetes in China: a multilevel regression analysis of cross-sectional data from the 2017 China Migrants Dynamic Survey . BMJ open 2025, 15 (4):e086915. Yu A: Socio-environmental determinants of tuberculosis in South Africa . African journal of reproductive health 2025, 29 (11):204-211. Nguyen MK, Nguyen TT, Sepehri A: Air pollution and tuberculosis incidence in Vietnam: short- and long-term effects from a provincial-level study . International archives of occupational and environmental health 2025, 98 (9-10):1023-1035. Gardy JL, Johnston JC, Ho Sui SJ, Cook VJ, Shah L, Brodkin E, Rempel S, Moore R, Zhao Y, Holt R et al : Whole-genome sequencing and social-network analysis of a tuberculosis outbreak . N Engl J Med 2011, 364 (8):730-739. Guo J, Liu C, Liu F, Zhou E, Ma R, Zhang L, Luo B: Tuberculosis disease burden in China: a spatio-temporal clustering and prediction study . Front Public Health 2024, 12 :1436515. Tables Table 1 and 2 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Table1.tif Table1 Results of multicollinearity test Table2.tif Table 2 Comparision of different spatial regression models SupplementalFigure1.tif Figure S1 Provincial-Level Tuberculosis Incidence Rates in China, 2005–2020 (A)​ 2005 (B)​ 2010 (C)​ 2015 (D)​ 2020 SupplementalFigure2.tif Figure S2 Incidence Rates of Rifampicin-Resistant Tuberculosis at the Provincial Level in China (2017–2020) (A)​ 2017 (B)​ 2018 (C)​ 2019 (D)​ 2020 SupplementalFigure3.tif Figure S3 Feature Importance Ranking of the Random Forest Model. The X-axis represents feature variables, and the Y-axis represents relative importance (%). The length of the bars in the figure indicates the relative contribution of each feature to the model's prediction. The values have been normalized and presented as percentages, which intuitively reflect the degree of influence of different features on the prediction results. SupplementalTable1.tif Table S1 Results of stepwise regression model Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8693705","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":584600723,"identity":"6accf80c-fb18-447f-9aab-601ebc770f81","order_by":0,"name":"Chen-Fan Zhang","email":"","orcid":"","institution":"Sanming First Hospital Affiliated with Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Chen-Fan","middleName":"","lastName":"Zhang","suffix":""},{"id":584600724,"identity":"e91833e0-2639-4eb9-be4c-d72c622c929c","order_by":1,"name":"Jue-Wei Pan","email":"","orcid":"","institution":"Sanming First Hospital Affiliated with Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jue-Wei","middleName":"","lastName":"Pan","suffix":""},{"id":584600725,"identity":"5e5b8b12-0705-47cb-b7cf-b0025e081a1d","order_by":2,"name":"Gui-Hua Yang","email":"","orcid":"","institution":"Sanming First Hospital Affiliated with Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Gui-Hua","middleName":"","lastName":"Yang","suffix":""},{"id":584600726,"identity":"5a7d5bbb-efb4-427a-8a97-b6b5be80892e","order_by":3,"name":"Yi-Wen Li","email":"","orcid":"","institution":"Sanming First Hospital Affiliated with Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yi-Wen","middleName":"","lastName":"Li","suffix":""},{"id":584600727,"identity":"b87df1d3-74e1-4574-910d-04aacf12c51c","order_by":4,"name":"Sun-Huan Su","email":"","orcid":"","institution":"Sanming First Hospital Affiliated with Fujian Medical University","correspondingAuthor":false,"prefix":"","firstName":"Sun-Huan","middleName":"","lastName":"Su","suffix":""},{"id":584600728,"identity":"48351476-f9e6-43b8-8070-6b47c80f7396","order_by":5,"name":"Hui-Wen Song","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA1UlEQVRIiWNgGAWjYDACCQaGAx8MbPgZG0A8NuK0MB6cUZEm2UiKFubDPGcOS4J1EKVFfnaPwQHetvMSzNPOGDB8KDvMwD+7Ab8WgztnDA5Itt2WYJydY8A449xhBok7BwhokcjdcMCw7XYdSAszb9thoEgCAYfNAGpJbDsHtoX5LzFaGG4AtRw4cwCihZEYLQY38j8cbKhIBmpJKzjYcy6dR+IGQYelJX/+Y2AnYTg7eeODH2XWcvwzCDkMBgwbgMkASPMQqR5kHfFKR8EoGAWjYKQBAJXVSKP2y8URAAAAAElFTkSuQmCC","orcid":"","institution":"Sanming First Hospital Affiliated with Fujian Medical University","correspondingAuthor":true,"prefix":"","firstName":"Hui-Wen","middleName":"","lastName":"Song","suffix":""}],"badges":[],"createdAt":"2026-01-25 16:23:14","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8693705/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8693705/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":101881143,"identity":"d1b4cedf-2e7b-4e21-ae43-5ef3c2f1b1da","added_by":"auto","created_at":"2026-02-04 15:10:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":270431,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTrends in the Tuberculosis Burden in China, 2004–2023\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Annual number of tuberculosis-related deaths in China, 2004–2023.\u003c/p\u003e\n\u003cp\u003e(B) Annual number of new tuberculosis cases (incidence) in China, 2004–2023.\u003c/p\u003e\n\u003cp\u003e(C) Trend of tuberculosis incidence in China, 2004–2023, and forecast to 2030.\u003c/p\u003e\n\u003cp\u003e(D) Trend in tuberculosis mortality rates in China, 2004–2023.\u003c/p\u003e\n\u003cp\u003e(E) Annual number of incidence cases and deaths attributed to rifampicin-resistant tuberculosis in China, 2017–2020.\u003c/p\u003e\n\u003cp\u003e(F) Incidence and mortality rates for rifampicin-resistant tuberculosis in China, 2017–2020.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/d4a1fa27eb34d5f76ce3bac4.png"},{"id":101790758,"identity":"24ac03fe-5981-4f17-923e-bfe196a6dde7","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":5102918,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSpatiotemporal Heterogeneity of Tuberculosis Incidence Influencing Factors: Spatio-temporal geographic weighted regression (GTWR) Model Analysis (2017–2021)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) O₃ (Ozone)\u003c/p\u003e\n\u003cp\u003e(B) PUP (Proportion of Urban Population)\u003c/p\u003e\n\u003cp\u003e(C) PPI (Proportion of Primary Industry)\u003c/p\u003e\n\u003cp\u003e(D) PM\u003csub\u003e10\u003c/sub\u003e (Particulate Matter 10 µm)\u003c/p\u003e\n\u003cp\u003e(E) SO\u003csub\u003e2\u003c/sub\u003e (Sulfur Dioxide)\u003c/p\u003e\n\u003cp\u003e(F) CO (Carbon Monoxide)\u003c/p\u003e\n\u003cp\u003e(G) NO\u003csub\u003e2\u003c/sub\u003e (Nitrogen Dioxide)\u003c/p\u003e\n\u003cp\u003e(H) PD (Population Density)\u003c/p\u003e\n\u003cp\u003e(I) NDVI (Normalized Difference Vegetation Index)\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/9add98447f451e63a53c5ed3.png"},{"id":101790760,"identity":"73d949dc-207c-4c63-86a9-1f578da224cb","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":207288,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGeodetector Analysis of Tuberculosis Incidence Influencing Factors (2017–2021)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Factor Detection: q-values of the influencing factors.\u003c/p\u003e\n\u003cp\u003e(B) Interaction Detection of the influencing factors.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/d3f579d33beb193bc517adc1.png"},{"id":101790757,"identity":"9789218b-e6d0-4fa2-a172-89f77f402306","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":280587,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003ePartial Dependence Plot (PDP) Analysis of Key Features. \u003c/strong\u003eThe X-axis represents the values of features, and the Y-axis represents the incidence rate. Each curve reflects the marginal dependence relationship between the corresponding feature and the model's prediction target: an upward trend indicates a positive correlation between the feature value and the incidence rate, a downward trend indicates a negative correlation, and a flat trend indicates a weak influence. Among them, the black curve represents the actual observed values, and the blue curve represents the model-fitted values.\u003c/p\u003e\n\u003cp\u003e(A) PD (Population Density)\u003c/p\u003e\n\u003cp\u003e(B) NDVI (Normalized Difference Vegetation Index)\u003c/p\u003e\n\u003cp\u003e(C) O₃ (Ozone)\u003c/p\u003e\n\u003cp\u003e(D) PM\u003csub\u003e10\u003c/sub\u003e (Particulate Matter 10)\u003c/p\u003e\n\u003cp\u003e(E) PUP (Proportion of Urban Population)\u003c/p\u003e\n\u003cp\u003e(F) PPI (Proportion of Primary Industry)\u003c/p\u003e\n\u003cp\u003e(G) NO\u003csub\u003e2\u003c/sub\u003e (Nitrogen Dioxide)\u003c/p\u003e\n\u003cp\u003e(H) CO (Carbon Monoxide)\u003c/p\u003e\n\u003cp\u003e(I) SO₂ (Sulfur Dioxide)\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/e620fa91f119e03d1366ba18.png"},{"id":103507097,"identity":"c3357848-8056-4fde-9e1f-6233f42a8212","added_by":"auto","created_at":"2026-02-26 13:40:25","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":7540545,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/f2f4a466-d3b4-437f-b8d2-a426970565c4.pdf"},{"id":101790766,"identity":"34068fad-de10-408c-93bd-d104353cb3ce","added_by":"auto","created_at":"2026-02-03 16:07:01","extension":"tif","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":12041536,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable1 Results of multicollinearity test\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Table1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/f19c10cadaefd6535ccc7dd5.tif"},{"id":101790761,"identity":"b6dd6b82-91b2-419c-ae3b-4b451d67bdb4","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"tif","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":4027164,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable 2 Comparision of different spatial regression models\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"Table2.tif","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/ee4be4524afec0b34759a068.tif"},{"id":101790762,"identity":"eb102588-8a9f-4eaa-83e3-90a73788c011","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"tif","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":2162276,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure S1 Provincial-Level Tuberculosis Incidence Rates in China, 2005–2020\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(A)​ 2005\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(B)​ 2010\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(C)​ 2015\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(D)​ 2020\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"SupplementalFigure1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/21cf429ff48a9239e103a860.tif"},{"id":101790764,"identity":"4bf43e6c-aa4a-4e6c-b0b4-e0f38ca25059","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"tif","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":730582,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure S2 Incidence Rates of Rifampicin-Resistant Tuberculosis at the Provincial Level in China (2017–2020)\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(A)​ 2017\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(B)​ 2018\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(C)​ 2019\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003e(D)​ 2020\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"SupplementalFigure2.tif","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/3015288455352dec3e505e36.tif"},{"id":101880579,"identity":"52130497-2eeb-45c1-89cd-471539e9670a","added_by":"auto","created_at":"2026-02-04 15:03:50","extension":"tif","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":163404,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure S3 Feature Importance Ranking of the Random Forest Model. The X-axis represents feature variables, and the Y-axis represents relative importance (%). The length of the bars in the figure indicates the relative contribution of each feature to the model's prediction. The values have been normalized and presented as percentages, which intuitively reflect the degree of influence of different features on the prediction results.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"SupplementalFigure3.tif","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/48c74d07a94595f7e0fec37c.tif"},{"id":101790765,"identity":"f8342e50-60d9-420e-867c-41915c9e2537","added_by":"auto","created_at":"2026-02-03 16:07:00","extension":"tif","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":7683908,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTable S1 Results of stepwise regression model\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"SupplementalTable1.tif","url":"https://assets-eu.researchsquare.com/files/rs-8693705/v1/67434a5049e9ad0e90d45cfb.tif"}],"financialInterests":"No competing interests reported.","formattedTitle":"Spatiotemporal trends and drivers of pulmonary tuberculosis incidence in China in the past two decades","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eTuberculosis (TB) remains a significant global public health threat as one of the most lethal infectious diseases. The World Health Organization's \"Global Tuberculosis Report 2025\" indicates that in 2024, there were approximately 10.7\u0026nbsp;million new tuberculosis cases globally, resulting in 1.2\u0026nbsp;million deaths [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Worldwide, tuberculosis remains one of the top ten causes of death and the leading cause of death from a single infectious source [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e, \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. Specifically, as a country with a high tuberculosis burden, China has made significant progress in tuberculosis prevention and control. However, it still faces severe challenges, including considerable regional variation in incidence\u0026mdash;meaning tuberculosis cases vary greatly across different areas [\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]\u0026mdash;and an increase in rifampicin-resistant tuberculosis cases [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e, \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePrevious studies indicate that a complex interplay of socioeconomic, environmental, and demographic factors influences the transmission and distribution of tuberculosis [\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. Traditional global regression models, which assume spatial stationarity, often fail to capture the pronounced spatiotemporal heterogeneity underlying these associations [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. With advances in geographic information systems and spatial statistical techniques, models such as geographically weighted regression (GWR) and geographically and temporally weighted regression (GTWR) have proven more effective in revealing the spatial non-stationarity of these relationships [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e]. In addition, the driving factors of tuberculosis transmission exhibit intricate interactions and nonlinear effects, necessitating the integration of machine-learning approaches to achieve deeper mechanistic insights and more accurate predictive modeling.\u003c/p\u003e \u003cp\u003eThis study utilizes China\u0026rsquo;s pulmonary tuberculosis surveillance data from 2004 to 2023, along with socioeconomic and environmental indicators, to create a comprehensive spatiotemporal analysis framework. By integrating spatiotemporal econometrics, geographic detectors, and machine learning techniques, the research aims to systematically analyze the spatiotemporal dynamics and various factors influencing the incidence of pulmonary tuberculosis in China. Consequently, the study will provide a solid evidence base for precision-targeted prevention and control strategies for tuberculosis.\u003c/p\u003e"},{"header":"METHODS","content":"\u003cp\u003e\u003cstrong\u003eData sources\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data on the incidence of pulmonary tuberculosis and its related influencing factors used in this study come mainly from the following sources.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePulmonary Tuberculosis Incidence Data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThese data come primarily from two sources. One is the \u0026quot;China Health and Wellness Statistics Yearbook\u0026quot; (https://www.zgtjnj.org), which provides detailed information on the incidence rate and mortality rate of pulmonary tuberculosis from 1997 to 2023. The other is the Public Health Science Data Center (https://www.phsciencedata.cn/Share/), which has compiled annual statistical data on the incidence rate, number of new cases, number of deaths, and mortality rate of rifampicin-resistant pulmonary tuberculosis across 31 provinces nationwide (excluding Hong Kong, Macau, and Taiwan) from 2017 to 2021.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePopulation\u003c/strong\u003e\u003cstrong\u003edata\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003ePopulation data, including population density (PD), proportion of urban population (PUP), and the proportion of the population aged 65 and over, are sourced from the annual \u0026quot;China Statistical Yearbook\u0026quot; (http://www.stats.gov.cn/tjsj/ndsj/).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eEnvironmental data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEnvironmental data are sourced from various public databases. The normalized difference vegetation index (NDVI) data are obtained from the official website of the National Aeronautics and Space Administration (https://ladsweb.modaps.eosdis.nasa.gov/search/), which quantifies vegetation coverage by measuring the difference between near-infrared and red-light bands. The annual average concentration data of air quality indices such as particulate matter 2.5 (PM\u003csub\u003e2.5\u003c/sub\u003e), particulate matter 10 (PM\u003csub\u003e10\u003c/sub\u003e), sulfur dioxide (SO\u003csub\u003e2\u003c/sub\u003e), nitrogen dioxide (NO\u003csub\u003e2\u003c/sub\u003e), carbon monoxide (CO), and ozone (O\u003csub\u003e3\u003c/sub\u003e) are sourced from the National Qinghai-Xizang Plateau Scientific Data Center (https://data.tpdc.ac.cn/).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGeographic information data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGeographic information is sourced from the National Basic Geographic Information Center (http://www.ngcc.cn/). The base maps of China used in this study are derived from standard maps provided by the Ministry of Natural Resources\u0026apos; standard map service website (http://bzdt.ch.mnr.gov.cn/) with the review number GS (2019) 1823.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eOther data\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eEconomic factor data, such as regional per capita GDP and per capita consumption expenditure in urban and rural residents, are sourced from the \u0026quot;China Statistical Yearbook.\u0026quot; Healthcare resource factors include the number of healthcare technicians per 10,000 population, the number of practicing (assistant) physicians per 10,000 population, and the number of registered nurses per 1,000 population. These statistics are sourced from the \u0026quot;China Health and Wellness Statistics Yearbook.\u0026quot;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInfluencing factors information\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBased on an extensive literature review, this study collected various factors that may influence the incidence of tuberculosis, mainly including population factors [8], socioeconomic factors [7], natural environmental factors [12], and healthcare resource factors [13]. Specifically, population factors include PD, PUP, proportion of the population aged 65 and over, employment ratios in primary, secondary, and tertiary industries, and the crude divorce rate. Economic factors encompass regional per capita gross domestic product (GDP) and per capita consumption expenditure for both urban and rural residents. Natural environmental factors encompass concentrations of SO\u003csub\u003e2\u003c/sub\u003e, NO\u003csub\u003e2\u003c/sub\u003e, CO, O\u003csub\u003e3\u003c/sub\u003e, PM\u003csub\u003e2.5\u003c/sub\u003e, and PM\u003csub\u003e10\u003c/sub\u003e, alongside vegetation coverage and the NDVI. Healthcare resource factors include the number of healthcare technicians per 10,000 population, the number of practicing (assistant) physicians per 10,000 population, and the number of registered nurses per 1,000 population.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData preprocessing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll collected data were preprocessed to ensure accuracy and consistency before analysis. This study created a tuberculosis database using Microsoft Excel 2019. The tuberculosis incidence and mortality data of 31 provinces and cities across the country were linked with China\u0026apos;s vectorized electronic map (*. shp format) using administrative region codes as the linking field in ArcGIS 10.8 software, thereby constructing a spatial database. The preprocessing steps are as follows.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Cleaning\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe check the completeness and consistency of the data and handle missing values and outliers. For missing data, interpolation methods are used to fill the gaps. For outliers, identification and processing are carried out using the box plot method or the z-score method.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Standardization\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo eliminate the influence of varying variable dimensions, Z-score standardization is applied to all influencing factor data.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Integration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eData from different sources is integrated into a unified database to ensure consistency and completeness.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStepwise Regression Analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn stepwise regression analysis, key variables that may affect the incidence of tuberculosis, such as population density, per capita GDP, vegetation coverage, air quality, and the number of health technicians, are first screened. The forward selection method is used to introduce variables one by one and conduct significance tests, while also removing variables that have been introduced but are no longer significant. Ultimately, the optimal regression model is constructed to reveal the degree and direction of the impact of various factors on the incidence rate, providing a basis for subsequent spatiotemporal econometric analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAnalysis of Factors Influencing the Incidence Rate of\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003eTuberculosis\u003c/strong\u003e\u003cstrong\u003e: Spatiotemporal Econometric Model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo identify the most suitable analytical model, this study employs various spatiotemporal econometric models for the empirical analysis of tuberculosis incidence. The specific steps are as follows.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMulticollinearity Test\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBefore constructing the regression model, a multicollinearity test is conducted on the independent variables considered for the model to avoid interference from high correlations among them. This study uses the variance inflation factor (VIF) for testing. If the VIF values of all independent variables are between 0 and 10, it indicates no severe multicollinearity, and subsequent regression analysis can proceed.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSpatial Autocorrelation Analysis\u003c/strong\u003e [14]\u003c/p\u003e\n\u003cp\u003eBefore constructing the geographically weighted regression model and the spatiotemporal geographically weighted regression model, spatial autocorrelation analysis is performed on the independent variables to assess spatial dependence among variables. This study determines whether there is spatial autocorrelation in tuberculosis incidence by calculating the global Moran\u0026apos;s \u003cem\u003eI\u003c/em\u003e index.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRegression Model Construction\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe developed several statistical models to analyze the incidence of pulmonary tuberculosis, including an ordinary least squares model (OLS), a spatial lag model (SLM), a spatial error model (SEM), a geographically weighted regression model (GWR), and a spatial-temporal geographically weighted regression model (GTWR). By comparing the performance metrics of these models, we aim to identify the one that best captures the spatiotemporal dynamics of TB incidence [15].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGeographic Detector\u003c/strong\u003e \u003cstrong\u003e[\u003c/strong\u003e16\u003cstrong\u003e]\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study employed the geographic detector method to explore the explanatory power of various influencing factors on the incidence of tuberculosis. The geographic detector is a statistical tool designed to identify spatial heterogeneity and its driving factors. It operates on the principle that a significant influencing factor will align the spatial distribution of disease incidence with that of the factor. This method comprises four modules: factor detector, interaction detector, risk zone detector, and ecological detector. The factor detector assesses how potential influencing factors explain the spatial heterogeneity of tuberculosis incidence using the q value. The q value, ranging from 0 to 1, signifies the factor\u0026apos;s explanatory power, with higher values indicating greater strength. The interaction detector is used to assess whether the interaction between two influencing factors enhances, weakens, or acts independently on the risk of incidence.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRandom Forest Model\u003c/strong\u003e [17]\u003c/p\u003e\n\u003cp\u003eThis study used the random forest model to draw partial dependence plots to visually display the nonlinear relationships between each independent variable and tuberculosis incidence. The random forest is an ensemble learning algorithm suitable for handling high-dimensional data and complex nonlinear relationships.\u003c/p\u003e\n\u003ch2\u003eStatistical Methods\u003c/h2\u003e\n\u003cp\u003eThis study used R 4.3.3 and GraphPad Prism 9 software for statistical description and chart drawing, and ArcGIS 10.8 software for spatial analysis, spatiotemporal econometric model parameter estimation, and model construction.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003ch2\u003eSpatiotemporal Trends of Tuberculosis Incidence and Mortality\u003c/h2\u003e\n\u003cp\u003eFrom 2004 to 2023, the number of tuberculosis cases nationwide peaked at 1.17 million in 2008, and has been declining continuously, dropping to 610,000 by 2023 (Figure 1A). During the same period, the number of deaths fluctuated between 1,435 and 3,783, with a mortality rate maintained at 0.1\u0026ndash;0.3 per 100,000, resulting in 2,167 deaths in 2023 (Figure 1B). The incidence rate peaked at 96.9 per 100,000 in 2005, and fell to 43.5 per 100,000 in 2023; the ARIMA model predicts it will drop to 23.5 per 100,000 by 2030 (Figure 1C). The incidence rate in the western region is significantly higher than in the eastern region (Supplemental Figure 1A-D), with Tibet (150.1 per 100,000), Xinjiang (109.9 per 100,000), and Guizhou (96.5 per 100,000) having the highest incidence rates in 2020, compared to the lowest rates in Tianjin (20.5 per 100,000), Shanghai (24.3 per 100,000), and Shandong (24.7 per 100,000), which are 4 to 7 times lower (Supplemental Figure 1D). By 2020, 18 of the 31 Chinese provinces, autonomous regions, and municipalities had incidence rates below 50 per 100,000 people (Supplemental Figure 1D).\u003c/p\u003e\n\u003cp\u003eSince 2017, the reported cases of rifampicin-resistant tuberculosis have increased from 7,607 (0.6 per 100,000) to 15,489 (1.1 per 100,000) in 2020 (Figure 1E, 1F). The number of deaths fluctuated between 40 and 108 during this period, with a mortality rate of 0.004 per 100,000 in 2020 (Figure 1E, 1F). The incidence of rifampicin-resistant tuberculosis also shows a distribution pattern characterized by higher rates in the west and lower rates in the east, with the highest incidence rates in Tibet (5.4 per 100,000), Qinghai (2.9 per 100,000), and Jilin (2.6 per 100,000), compared to the lowest rates in Shandong (0.4 per 100,000), Shanxi (0.5 per 100,000), and Jiangsu (0.6 per 100,000), which are 4 to 13 times lower (Supplemental Figure 2A-2D).\u003c/p\u003e\n\u003ch2\u003eConstruction and Comparison of Spatial Regression Models\u003c/h2\u003e\n\u003cp\u003eThrough stepwise regression analysis, the two models with the highest adjusted R\u0026sup2;\u0026nbsp;values (Model 12 and Model 13) were selected. Model 13, which includes nine variables\u0026mdash;PUP, PPI, PD, CO, NO\u003csub\u003e2\u003c/sub\u003e, O\u003csub\u003e3\u003c/sub\u003e, PM\u003csub\u003e10\u003c/sub\u003e, SO\u003csub\u003e2\u003c/sub\u003e, and NDVI\u0026mdash;performed similarly to Model 12, which adds vegetation coverage. To avoid issues with multicollinearity, given that vegetation coverage is closely related to NDVI, Model 13 was ultimately selected (Supplemental Table 1). The multicollinearity test showed that the variance inflation factor (VIF) for all variables was less than 10, indicating good model stability (Table 1). The global Moran\u0026apos;s I index was 0.496 (P \u0026lt; 0.05), confirming significant spatial autocorrelation in the data, necessitating the use of spatial econometric models.\u003c/p\u003e\n\u003cp\u003eThe OLS, SEM, SLM, GWR, and GTWR models were developed and compared, as shown in Table 2. Among these, the GTWR model demonstrated the highest goodness of fit, with an R\u003csup\u003e2\u003c/sup\u003e value of 0.950 and an Adjusted R\u003csup\u003e2\u003c/sup\u003e of 0.947. Additionally, it had the lowest AICc value of 162.5. This indicates that the GTWR model most effectively captures the spatiotemporal variations in tuberculosis incidence, leading to its selection as the final analysis model (Table 2).\u003c/p\u003e\n\u003ch2\u003eGTWR\u003c/h2\u003e\n\u003cp\u003eThe results of the GTWR model indicate that the effects of various influencing factors on tuberculosis incidence exhibit significant spatial-temporal heterogeneity (P \u0026lt; 0.05) (Figure 2A-2I). From the perspective of average regression coefficients, among socio-economic factors, the PUP (coefficient \u0026asymp; -0.40) and the PPI (coefficient \u0026asymp; -0.20) both show significant negative impacts on the incidence rate (Figure 2B, 2C). Regarding environmental factors, O\u003csub\u003e3\u003c/sub\u003e(coefficient \u0026asymp; -0.35) (Figure 2A) and NO\u003csub\u003e2\u003c/sub\u003e(coefficient \u0026asymp; -0.30) (Figure 2G) in air pollutants exhibit significant negative impacts, while PM\u003csub\u003e10\u0026nbsp;\u003c/sub\u003e(coefficient \u0026asymp; 0.35) (Figure 2D) shows a significant positive impact, but its influence intensity has a trend of weakening year by year (the coefficient was 0.59 in 2017, dropping to -0.08 in 2021); the impacts of CO (coefficient \u0026asymp; 0.10) (Figure 2F) and SO\u003csub\u003e2\u003c/sub\u003e(coefficient \u0026asymp; -0.06) (Figure 2E) are relatively weak. In addition, the NDVI (coefficient \u0026asymp; -0.20) (Figure 2I) also shows a negative impact. Among population factors, population density (coefficient \u0026asymp; 0.20) positively affects the incidence rate (Figure 2H).\u003c/p\u003e\n\u003ch2\u003eGeographical Detector Analysis\u003c/h2\u003e\n\u003cp\u003eFactor Detection\u003c/p\u003e\n\u003cp\u003eBetween 2017 and 2021, the q-values of all influencing factors passed the significance test (P \u0026lt; 0.05). The explanatory power of the NDVI (0.6-0.9), PD (0.6-0.7), and PM₁₀\u0026nbsp;(0.5-0.8) is relatively strong. The q-value for the PUP continues to increase (0.3-0.7). Following this, O₃\u0026nbsp;has a q-value between 0.3 and 0.5, while NO₂\u0026nbsp;ranges from 0.3 to 0.7. CO shows a weaker explanatory power, with a q-value of 0.2 to 0.4, and SO₂\u0026nbsp;has the weakest explanatory power, ranging from 0.1 to 0.4 (Figure 3A).\u003c/p\u003e\n\u003cp\u003eInteraction Detection\u003c/p\u003e\n\u003cp\u003eThe interactions among all factors show a dual-factor enhancement. The interaction combinations of PD\u0026cap;CO, PD\u0026cap;PM₁₀, and NDVI\u0026cap;O₃ have the highest q-values, reaching above 0.9, but overall, there is a downward trend year by year (Figure 3B).\u003c/p\u003e\n\u003ch2\u003eRandom Forest and Partial Dependence Plot (PDP) Analysis\u003c/h2\u003e\n\u003cp\u003ePD, NDVI, and O\u003csub\u003e3\u003c/sub\u003e are the three most important variables affecting the incidence of tuberculosis (Figure 4). The partial dependence plot further reveals the complex nonlinear relationships between these factors and the incidence rate: the incidence rate decreases with the increase of NDVI, O\u003csub\u003e3\u003c/sub\u003e, and PUP, and tends to stabilize after reaching a certain threshold; conversely, the incidence rate increases with the increase of the proportion of PPI, PM \u003csub\u003e10\u003c/sub\u003e, and CO (Figure 4). In addition, NO\u003csub\u003e2\u003c/sub\u003e and SO\u003csub\u003e2\u0026nbsp;\u003c/sub\u003eshow inverted U-shaped and U-shaped relationships with the incidence rate, respectively, while the effect of PD exhibits nonlinear characteristics, i.e., the influence weakens at lower (\u0026lt;20 people per square kilometre) and higher (500 people per square kilometre) levels, while being most significant in the medium density range (150\u0026ndash;200 people per square kilometre) (Figure 4).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study comprehensively utilizes descriptive epidemiology, spatial-temporal statistics, geographical detectors, and machine learning methods to systematically reveal the spatial-temporal dynamic patterns of tuberculosis incidence in China from 2004 to 2023, as well as the complex influencing mechanisms behind them. The main findings indicate that, although the overall incidence rate of tuberculosis in China shows a downward trend, the issue of rifampicin resistance is becoming increasingly prominent, and the disease burden exhibits significant spatial heterogeneity, with the western regions having a higher incidence than the eastern regions. Core analyses further confirm that socio-economic, environmental, and population factors dynamically, nonlinearly, and interactively drive the risk of tuberculosis transmission.\u003c/p\u003e \u003cp\u003eThis study finds that the reported incidence rate of tuberculosis in China continues to decline, marking a significant success of the massive investments made in tuberculosis prevention and control over the past two decades, which can be attributed to the comprehensive coverage of modern tuberculosis control strategies (DOTS), the improvement of the medical insurance system, and the enhancement of diagnostic levels [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. However, the incidence rate in western regions (such as Tibet and Xinjiang) continues to be several times higher than that in the eastern coastal areas, revealing a close association between disease burden and the level of regional socio-economic development [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. This spatial heterogeneity is closely related to factors such as uneven distribution of medical resources, higher poverty levels, and harsh natural conditions. More alarmingly, the significant rise in reported cases of rifampicin-resistant tuberculosis after 2017 poses a severe challenge to the goal of ending the tuberculosis epidemic by 2035, emphasizing the urgency of strengthening laboratory rapid diagnostic capabilities and the standardized management system for second-line drugs [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn terms of driving factor analysis, the comparative results of the models in this study strongly demonstrate that the GTWR model performs optimally in analyzing the influencing factors of tuberculosis incidence due to its ability to capture both spatial dependence and temporal non-stationarity. This finding aligns with the forefront direction of current research on the driving factors of complex disease environments, recognizing that exposure-response relationships have significant contextual dependencies [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. The results of the GTWR model reveal that the intensity and direction of the effects of various factors exhibit significant spatial-temporal variations. For example, the urban population ratio acts as a protective factor in most areas, reflecting the improvement in accessibility to medical and health services accompanying urbanization; however, its effect weakens or even reverses in highly urbanized areas, possibly indicating increased risks due to intensified internal population flow and aggregation within cities [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e]. This heterogeneity of effects suggests that a single global conclusion may obscure the real risks in local areas, making it crucial to adopt a precise prevention and control approach of \"one place, one policy.\"\u003c/p\u003e \u003cp\u003eAn in-depth analysis of specific influencing factors reveals that the results of this study corroborate existing evidence while also highlighting new phenomena that require further exploration. The positive impact of population density, the risk effect of the PPI (as a proxy variable for rural and specific occupational exposure), and the positive correlation between PM\u003csub\u003e10\u003c/sub\u003e and incidence rate have been repeatedly confirmed by numerous studies [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e]. The protective effect of the normalized difference vegetation index may be related to green spaces promoting physical activity, improving mental health, and purifying the air, supporting the necessity of increasing green spaces in the \"healthy city\" concept [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eHowever, the significant negative associations of gas pollutants such as O\u003csub\u003e3\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e differ from some research conclusions. One possible explanation is the complexity of atmospheric chemical processes: O\u003csub\u003e3\u003c/sub\u003e and NO\u003csub\u003e2\u003c/sub\u003e, as secondary pollutants, have complex nonlinear relationships with primary pollutants (such as PM\u003csub\u003e2.5\u003c/sub\u003e), and collinearity in the model may lead to seemingly contradictory results [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e]. Another hypothesis that needs verification is that a certain concentration of O\u003csub\u003e3\u003c/sub\u003e has broad-spectrum bactericidal effects, which may inhibit the survival and transmission of Mycobacterium tuberculosis in the external environment to some extent [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. The nonlinear relationships revealed by the random forest model (such as the U-shaped curve of SO\u003csub\u003e2\u003c/sub\u003e and the threshold effect of PD and NDVI) greatly enrich our understanding of the action patterns of these factors, indicating that their impacts are not simple monotonic linear relationships, providing scientific reference for accurately setting critical values for environmental governance and public health interventions [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eMoreover, the strong dual-factor enhancement interactions revealed by the geographical detector are an important finding of this study. For example, the interaction q-values of \"PD \u0026cap; CO\" and \"PD \u0026cap; PM\u003csub\u003e10\u003c/sub\u003e\" are extremely high, strongly suggesting that in densely populated areas with heavy air pollution, the risk of tuberculosis transmission is synergistically amplified. This highlights that the epidemic of tuberculosis is the result of the interaction of the \"social-economic-environmental\" complex ecosystem, urgently requiring the breaking down of departmental barriers and promoting collaborative governance based on the concept of \"integrated health [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e].\" Public health policies must be deeply integrated with environmental protection (emission reduction), urban-rural planning (optimizing population distribution and green space allocation), and industrial development (reducing agricultural population exposure) policies to fundamentally and effectively control tuberculosis.\u003c/p\u003e \u003cp\u003eBased on the above findings, we propose the following policy implications: (1) Implement differentiated regional prevention and control strategies: for high-burden provinces in the west, strengthen investment in funding, technology, and personnel, focusing on enhancing the diagnostic capacity at the grassroots level and the prevention and control of drug-resistant tuberculosis; for low-incidence areas in the east, emphasize the management of tuberculosis among the floating population, active case finding, and epidemic warning [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. (2) Promote cross-departmental collaborative interventions: clearly incorporate tuberculosis prevention and control goals into national and local air pollution management, green space system planning, and rural revitalization strategies to achieve health integration into all policies [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e].\u003c/p\u003e \u003cdiv id=\"Sec26\" class=\"Section2\"\u003e \u003ch2\u003eLimitations\u003c/h2\u003e \u003cp\u003eThis study also has several limitations. First, ecological studies based on aggregated data cannot infer individual-level causal relationships and fail to control for the effects of individual behaviors and comorbid factors such as nutritional status, smoking, and diabetes. Second, the data rely on infectious disease reporting systems, and differences in diagnostic levels and reporting quality across regions may introduce some bias into the results. Finally, although we employed various statistical models to control for confounding factors, the possibility of residual confounding cannot be completely ruled out.\u003c/p\u003e \u003c/div\u003e"},{"header":"Conclusion","content":"\u003cp\u003eChina has reduced the incidence of tuberculosis by half in the past 20 years; however, the burden remains significant in the West, with the presence of resistant strains. The risk of tuberculosis is influenced by factors such as population density, employment, and exposure to pollutants. To eliminate TB by 2035, it is essential to implement tailored regional strategies that include laboratory networks and targeted screening, alongside policies that address air quality and drug resistance.\u003c/p\u003e \u003cp\u003eFuture research should focus on the following directions: (1) Conduct research designs that integrate multi-level data (individual-community-environment) to assess the independent effects of various factors more accurately. (2) Combine whole-genome sequencing technology of Mycobacterium tuberculosis to clarify how environmental and social factors influence the transmission chain [\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e] of tuberculosis from the perspective of pathogen transmission dynamics. (3) Use high-precision models established in this study, such as GTWR, for prediction and design prospective intervention trials to assess the actual effects of specific policies (such as park green space construction) on reducing tuberculosis incidence, thereby providing strong evidence [\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e] for formulating more cost-effective prevention and control strategies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eAcknowledgments\u003c/h2\u003e\n\u003cp\u003eThe study was funded by the 2024 Sanming City Health and Wellness Science and Technology Innovation Joint Project. The funder had no involvement in the study\u0026apos;s design or execution, the collection, management, analysis, or interpretation of the data, the preparation, review, or approval of the manuscript, or the decision to submit the manuscript for publication.\u003c/p\u003e\n\u003ch2\u003eAuthor contributions\u003c/h2\u003e\n\u003cp\u003eHWS designed the study, performed the statistical analyses, and drafted the initial manuscript. CFZ and JWP contributed to data collection and cleaning, and provided critical revisions to the manuscript. YWL and SHS assisted in the interpretation of the results and helped to refine the study\u0026apos;s conclusions. All authors read and approved the final manuscript.\u003c/p\u003e\n\u003ch2\u003eAvailability of Data\u003c/h2\u003e\n\u003cp\u003eData is provided within the supplementary information files.\u003c/p\u003e\n\u003ch2\u003eEthics approval\u003c/h2\u003e\n\u003cp\u003eThe study utilized data from publicly available databases, and ethical approval was not needed.\u003c/p\u003e\n\u003ch2\u003eConsent to Publish declaration\u003c/h2\u003e\n\u003cp\u003eNot applicable.\u003c/p\u003e\n\u003ch2\u003eConflicts of interest\u003c/h2\u003e\n\u003cp\u003eNone.\u003c/p\u003e\n\u003ch2\u003eFunding\u003c/h2\u003e\n\u003cp\u003eThe research was supported by the 2024 Sanming City Health and Wellness Science and Technology Innovation Joint Project (No. 2024-S-011) and the Startup Fund for Scientific Research at Fujian Medical University (No. 2023QH1286).\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003e\u003cstrong\u003eGlobal tuberculosis report 2025 \u003c/strong\u003e[https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2025]\u003c/li\u003e\n\u003cli\u003e\u003cstrong\u003eGlobal tuberculosis report 2024 \u003c/strong\u003e[https://www.who.int/teams/global-programme-on-tuberculosis-and-lung-health/tb-reports/global-tuberculosis-report-2024]\u003c/li\u003e\n\u003cli\u003eLi XX, Wang LX, Zhang J, Liu YX, Zhang H, Jiang SW, Chen JX, Zhou XN: \u003cstrong\u003eExploration of ecological factors related to the spatial heterogeneity of tuberculosis prevalence in P. R. China\u003c/strong\u003e. \u003cem\u003eGlobal health action \u003c/em\u003e2014, \u003cstrong\u003e7\u003c/strong\u003e:23620.\u003c/li\u003e\n\u003cli\u003eWang L, Zhang H, Ruan Y, Chin DP, Xia Y, Cheng S, Chen M, Zhao Y, Jiang S, Du X\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eTuberculosis prevalence in China, 1990-2010; a longitudinal analysis of national survey data\u003c/strong\u003e. \u003cem\u003eLancet \u003c/em\u003e2014, \u003cstrong\u003e383\u003c/strong\u003e(9934):2057-2064.\u003c/li\u003e\n\u003cli\u003eSu W, Ruan YZ, Li T, Du X, Jiang JW, Li RZ: \u003cstrong\u003eCharacteristics of rifampicin-resistant tuberculosis detection in China, 2015-2019\u003c/strong\u003e. \u003cem\u003eInfectious diseases of poverty \u003c/em\u003e2021, \u003cstrong\u003e10\u003c/strong\u003e(1):99.\u003c/li\u003e\n\u003cli\u003eMohidem NA, Hashim Z, Osman M, Shaharudin R, Muharam FM, Makeswaran P: \u003cstrong\u003eDemographic, socio-economic and behavior as risk factors of tuberculosis in Malaysia: a systematic review of the literature\u003c/strong\u003e. \u003cem\u003eReviews on environmental health \u003c/em\u003e2018, \u003cstrong\u003e33\u003c/strong\u003e(4):407-421.\u003c/li\u003e\n\u003cli\u003eLiu JJ, Yao HY, Liu EY: \u003cstrong\u003eAnalysis of factors affecting the epidemiology of tuberculosis in China\u003c/strong\u003e. \u003cem\u003eInt J Tuberc Lung Dis \u003c/em\u003e2005, \u003cstrong\u003e9\u003c/strong\u003e(4):450-454.\u003c/li\u003e\n\u003cli\u003eRen H, Lu W, Li X, Shen H: \u003cstrong\u003eSpecific urban units identified in tuberculosis epidemic using a geographical detector in Guangzhou, China\u003c/strong\u003e. \u003cem\u003eInfectious diseases of poverty \u003c/em\u003e2022, \u003cstrong\u003e11\u003c/strong\u003e(1):44.\u003c/li\u003e\n\u003cli\u003eRen H, Lu W, Li X, Shen H: \u003cstrong\u003eSpecific urban units identified in tuberculosis epidemic using a geographical detector in Guangzhou, China\u003c/strong\u003e. \u003cem\u003eInfectious Diseases of Poverty \u003c/em\u003e2022, \u003cstrong\u003e11\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eYu H, Yang J, Yan Y, Zhang H, Chen Q, Sun L: \u003cstrong\u003eFactors affecting the incidence of pulmonary tuberculosis based on the GTWR model in China, 2004-2021\u003c/strong\u003e. \u003cem\u003eEpidemiol Infect \u003c/em\u003e2024, \u003cstrong\u003e152\u003c/strong\u003e:e65.\u003c/li\u003e\n\u003cli\u003eAmaral AVR, Gonz\u0026aacute;lez JA, Moraga P: \u003cstrong\u003eSpatio-temporal modeling of infectious diseases by integrating compartment and point process models\u003c/strong\u003e. \u003cem\u003eStoch Environ Res Risk Assess \u003c/em\u003e2023, \u003cstrong\u003e37\u003c/strong\u003e(4):1519-1533.\u003c/li\u003e\n\u003cli\u003eYang J, Zhang M, Chen Y, Ma L, Yadikaer R, Lu Y, Lou P, Pu Y, Xiang R, Rui B: \u003cstrong\u003eA study on the relationship between air pollution and pulmonary tuberculosis based on the general additive model in Wulumuqi, China\u003c/strong\u003e. \u003cem\u003eInt J Infect Dis \u003c/em\u003e2020, \u003cstrong\u003e96\u003c/strong\u003e:42-47.\u003c/li\u003e\n\u003cli\u003eMahara G, Yang K, Chen S, Wang W, Guo X: \u003cstrong\u003eSocio-Economic Predictors and Distribution of Tuberculosis Incidence in Beijing, China: A Study Using a Combination of Spatial Statistics and GIS Technology\u003c/strong\u003e. \u003cem\u003eMed Sci (Basel) \u003c/em\u003e2018, \u003cstrong\u003e6\u003c/strong\u003e(2).\u003c/li\u003e\n\u003cli\u003eWang Q, Guo L, Wang J, Zhang L, Zhu W, Yuan Y, Li J: \u003cstrong\u003eSpatial distribution of tuberculosis and its socioeconomic influencing factors in mainland China 2013\u0026ndash;2016\u003c/strong\u003e. \u003cem\u003eTropical Medicine \u0026amp;amp; International Health \u003c/em\u003e2019, \u003cstrong\u003e24\u003c/strong\u003e(9):1104-1113.\u003c/li\u003e\n\u003cli\u003eWang Q, Guo L, Wang J, Zhang L, Zhu W, Yuan Y, Li J: \u003cstrong\u003eSpatial distribution of tuberculosis and its socioeconomic influencing factors in mainland China 2013-2016\u003c/strong\u003e. \u003cem\u003eTrop Med Int Health \u003c/em\u003e2019, \u003cstrong\u003e24\u003c/strong\u003e(9):1104-1113.\u003c/li\u003e\n\u003cli\u003eHu Y, Xia C, Li S, Ward MP, Luo C, Gao F, Wang Q, Zhang S, Zhang Z: \u003cstrong\u003eAssessing environmental factors associated with regional schistosomiasis prevalence in Anhui Province, Peoples\u0026apos; Republic of China using a geographical detector method\u003c/strong\u003e. \u003cem\u003eInfectious diseases of poverty \u003c/em\u003e2017, \u003cstrong\u003e6\u003c/strong\u003e(1):87.\u003c/li\u003e\n\u003cli\u003eZhao D, Zhang H, Wu X, Zhang L, Li S, He S: \u003cstrong\u003eSpatial and temporal analysis and forecasting of TB reported incidence in western China\u003c/strong\u003e. \u003cem\u003eBMC Public Health \u003c/em\u003e2024, \u003cstrong\u003e24\u003c/strong\u003e(1):2504.\u003c/li\u003e\n\u003cli\u003eZuo Z, Wang M, Cui H, Wang Y, Wu J, Qi J, Pan K, Sui D, Liu P, Xu A: \u003cstrong\u003eSpatiotemporal characteristics and the epidemiology of tuberculosis in China from 2004 to 2017 by the nationwide surveillance system\u003c/strong\u003e. \u003cem\u003eBMC Public Health \u003c/em\u003e2020, \u003cstrong\u003e20\u003c/strong\u003e(1):1284.\u003c/li\u003e\n\u003cli\u003eSu W, Ruan Y, Li T, Du X, Jiang J, He Y, Li R: \u003cstrong\u003eEpidemiological Characteristics of Rifampicin-Resistant Tuberculosis in Students - China, 2015-2019\u003c/strong\u003e. \u003cem\u003eChina CDC Wkly \u003c/em\u003e2021, \u003cstrong\u003e3\u003c/strong\u003e(26):549-552.\u003c/li\u003e\n\u003cli\u003eZhao M, Wu J, Figueiredo DM, Zhang Y, Zou Z, Cao Y, Li J, Chen X, Shi S, Wei Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eSpatial-temporal distribution and potential risk of pesticides in ambient air in the North China Plain\u003c/strong\u003e. \u003cem\u003eEnviron Int \u003c/em\u003e2023, \u003cstrong\u003e182\u003c/strong\u003e:108342.\u003c/li\u003e\n\u003cli\u003eSong WM, Liu Y, Men D, Li SJ, Tao NN, Zhang QY, Liu SQ, An QQ, Zhu XH, Han QL\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eAssociations of residential greenness exposure and ambient air pollutants with newly-diagnosed drug-resistant tuberculosis cases\u003c/strong\u003e. \u003cem\u003eEnviron Sci Pollut Res Int \u003c/em\u003e2024, \u003cstrong\u003e31\u003c/strong\u003e(18):27240-27258.\u003c/li\u003e\n\u003cli\u003eSmith GS, Van Den Eeden SK, Garcia C, Shan J, Baxter R, Herring AH, Richardson DB, Van Rie A, Emch M, Gammon MD: \u003cstrong\u003eAir Pollution and Pulmonary Tuberculosis: A Nested Case-Control Study among Members of a Northern California Health Plan\u003c/strong\u003e. \u003cem\u003eEnviron Health Perspect \u003c/em\u003e2016, \u003cstrong\u003e124\u003c/strong\u003e(6):761-768.\u003c/li\u003e\n\u003cli\u003ePriĭmak AA, Kaliuk AN, Kirgintsev AG: \u003cstrong\u003e[The effect of an ozone-oxygen mixture on Mycobacterium tuberculosis and conditionally pathogenic microorganisms]\u003c/strong\u003e. \u003cem\u003eProbl Tuberk \u003c/em\u003e1991(4):7-10.\u003c/li\u003e\n\u003cli\u003eLabib SM: \u003cstrong\u003eGreenness, air pollution, and temperature exposure effects in predicting premature mortality and morbidity: A small-area study using spatial random forest model\u003c/strong\u003e. \u003cem\u003eSci Total Environ \u003c/em\u003e2024, \u003cstrong\u003e928\u003c/strong\u003e:172387.\u003c/li\u003e\n\u003cli\u003eHuang L, He J, Zhang C, Liu J, Guo Z, Lv S, Zhang X, Li S: \u003cstrong\u003eChina\u0026apos;s One Health governance system: the framework and its application\u003c/strong\u003e. \u003cem\u003eSci One Health \u003c/em\u003e2023, \u003cstrong\u003e2\u003c/strong\u003e:100039.\u003c/li\u003e\n\u003cli\u003eLiu J, Yin H, Wang N, Wang Y, Guan L, Feng Y, Wu L, Liu W, Zhang H, Liu Z\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eFactors associated with exposure to tuberculosis education among internal migrants with diabetes in China: a multilevel regression analysis of cross-sectional data from the 2017 China Migrants Dynamic Survey\u003c/strong\u003e. \u003cem\u003eBMJ open \u003c/em\u003e2025, \u003cstrong\u003e15\u003c/strong\u003e(4):e086915.\u003c/li\u003e\n\u003cli\u003eYu A: \u003cstrong\u003eSocio-environmental determinants of tuberculosis in South Africa\u003c/strong\u003e. \u003cem\u003eAfrican journal of reproductive health \u003c/em\u003e2025, \u003cstrong\u003e29\u003c/strong\u003e(11):204-211.\u003c/li\u003e\n\u003cli\u003eNguyen MK, Nguyen TT, Sepehri A: \u003cstrong\u003eAir pollution and tuberculosis incidence in Vietnam: short- and long-term effects from a provincial-level study\u003c/strong\u003e. \u003cem\u003eInternational archives of occupational and environmental health \u003c/em\u003e2025, \u003cstrong\u003e98\u003c/strong\u003e(9-10):1023-1035.\u003c/li\u003e\n\u003cli\u003eGardy JL, Johnston JC, Ho Sui SJ, Cook VJ, Shah L, Brodkin E, Rempel S, Moore R, Zhao Y, Holt R\u003cem\u003e et al\u003c/em\u003e: \u003cstrong\u003eWhole-genome sequencing and social-network analysis of a tuberculosis outbreak\u003c/strong\u003e. \u003cem\u003eN Engl J Med \u003c/em\u003e2011, \u003cstrong\u003e364\u003c/strong\u003e(8):730-739.\u003c/li\u003e\n\u003cli\u003eGuo J, Liu C, Liu F, Zhou E, Ma R, Zhang L, Luo B: \u003cstrong\u003eTuberculosis disease burden in China: a spatio-temporal clustering and prediction study\u003c/strong\u003e. \u003cem\u003eFront Public Health \u003c/em\u003e2024, \u003cstrong\u003e12\u003c/strong\u003e:1436515.\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 and 2 are available in the Supplementary Files section.\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Tuberculosis, Spatiotemporal Analysis, Spatiotemporal Geographically Weighted Regression, Geographic Detector, Machine Learning, Public Health Policy","lastPublishedDoi":"10.21203/rs.3.rs-8693705/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8693705/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003ePurpose: \u003c/strong\u003eTo analyze the spatiotemporal evolution pattern of tuberculosis incidence in China from 2004 to 2023 and reveal various driving factors.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e Using national tuberculosis surveillance data, economic indicators, and environmental information, we employed spatiotemporal econometric models, geographic detectors, and random forests for a comprehensive analysis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults: \u003c/strong\u003eOverall, tuberculosis incidence rates in China have been declining; however, significant regional disparities persist. The western region consistently demonstrates higher incidence rates compared to the eastern region. Furthermore, there has been an annual increase in rifampicin-resistant cases from 2017 to 2020. Upon comparing five distinct spatiotemporal econometric models, the spatiotemporal geographically weighted regression (GTWR) model (R² = 0.950) emerged as the most effective, indicating that the impact of various factors exhibits both spatial and temporal variability. Population density (PD) and particulate matter 10 (PM\u003csub\u003e10\u003c/sub\u003e) concentration were correlated with elevated incidence rates. In contrast, the proportion of the urban population, normalized difference vegetation index (NDVI), and ozone (O\u003csub\u003e3\u003c/sub\u003e) concentration were correlated with reduced rates. Geographic detector analysis further identified NDVI, PD, and PM\u003csub\u003e10\u003c/sub\u003e as critical determinants, revealing statistically significant interactive effects among these variables. The random forest model demonstrated a complex, non-linear relationship between various factors and the incidence rate.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e This study emphasizes the importance of integrating socioeconomic, environmental, and population factors to understand tuberculosis transmission dynamics and provides a strong foundation for developing targeted prevention and control strategies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding: \u003c/strong\u003eThe research was supported by the 2024 Sanming City Health and Wellness Science and Technology Innovation Joint Project (No. 2024-S-011) and the Startup Fund for Scientific Research at Fujian Medical University (No.2023QH1286).\u003c/p\u003e","manuscriptTitle":"Spatiotemporal trends and drivers of pulmonary tuberculosis incidence in China in the past two decades","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-03 16:06:55","doi":"10.21203/rs.3.rs-8693705/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"c2318449-7e3f-4c95-91bc-dfe8c76805fc","owner":[],"postedDate":"February 3rd, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-02-25T07:57:34+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-03 16:06:55","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8693705","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8693705","identity":"rs-8693705","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}