Most of the existing studies on the influencing factors of pulmonary tuberculosis incidence are based on temporal or spatial regression models, and the results are limited.

To explore the temporal and spatial heterogeneity of pulmonary tuberculosis in China, and to analyze the temporal and spatial correlations between the incidence of pulmonary tuberculosis and meteorological and air quality factors, offering a scientific reference for the development of measures containing tuberculosis.

Monthly statistical data of pulmonary tuberculosis in China from 2016 to 2018 were collected. After being tested with multicollinearity and spatial-autocorrelation between incidence of pulmonary tuberculosis and meteorological and air quality factors, the incidence of pulmonary tuberculosis was used as the dependent variable, and meteorological and air quality factors as independent variables to construct OLS, GWR and GTWR models, respectively. Then the goodness of the three models was evaluated, and the optimal model was selected to describe the incidence of pulmonary tuberculosis. Kernel density plot and spatio-temporal graph were used to describe the spatio-temporal specificity of the fitting coefficients of each variable.

The overall incidence of pulmonary tuberculosis in China during 2016-2018 decreased annually, with clustered spatial distribution. The GTWR model had higher R² value and lower AICc value compared to other two models, indicating that it had better performance in explaining the influence of meteorological and air quality factors on the incidence of pulmonary tuberculosis. The kernel density plot of each variable showed that the increase of wind speed was associated with decreased pulmonary tuberculosis incidence in most cities. But the increase of humidity and air pollutant concentration was associated with increased incidence of pulmonary tuberculosis, and the strength of association varied across cities.

Meteorological and air quality factors may significantly influence the incidence of pulmonary tuberculosis, and the influence had spatio-temporal specificity. So prevention methods for pulmonary tuberculosis should be developed according to region-specific factors influencing the disease.

现有针对肺结核发病影响因素的研究大多是独立的时间或空间回归分析，研究结果存在局限性。

探索中国肺结核分布的时间和空间异质性，并分析肺结核发病情况与气象和空气质量因素在时间和空间上的相关关系，为制订相应结核病防控措施提供科学参考。

使用2016—2018年全国分地区肺结核分月统计数据，将肺结核发病率作为因变量，将气象和空气质量因素作为自变量，在预先进行多重共线性和空间自相关检验后，分别构建普通最小二乘（OLS）模型、地理加权回归（GWR）模型、时空地理加权回归（GTWR）模型，评估并比较模型优度，选取最优模型以描述肺结核发病情况。分别绘制各变量拟合系数的核密度分布图和时空分布图，以描述拟合系数的时空特异性。

我国肺结核总发病率在逐年下降，且空间分布较为集中。GTWR模型的R²值均比OLS和GWR模型要高，同时GTWR模型的修正后的赤池信息量（AICc）值均比OLS和GWR模型要小，表明GTWR模型能更好地解释自变量对肺结核发病情况的影响。各变量核密度图结果显示，风速的增加对大多数城市的肺结核发病呈现显著的保护作用；湿度及空气污染物浓度的增加将显著增加肺结核发病率，且在不同城市的影响程度不同。

气象和空气质量因素对肺结核发病情况具有显著影响，且该影响存在时空特异性，对于不同地区的不同影响因素，应制订针对性的疾病预防措施。