Background
Tuberculosis (TB) is still one of the most infectious diseases worldwide with an estimated 10.0 million new cases and 1.3 million deaths according to the Global Tuberculosis Report from 2018 [
1]. China is ranked second among the 30 high burden countries for TB and accounted for 9% of the world’s cases. Previous studies indicated that TB epidemics were influenced by three principal aspects: environment, host (human) and pathogen (
M. tuberculosis) [
2,
3]. The environmental aspect mainly refers to the social environment such as socioeconomic level, medical research and demographic factors [
4‐
6]. The prevalence of TB in China has halved since 1991 due to the strategy of short-course chemotherapy following WHO guidelines to control TB [
7]. However, if we rely only on the original strategy, the global TB epidemic, including China’s epidemic, will not end by 2035, which is the aim of the World Health Organization (WHO)'s “End TB Strategy”. To complete this aim as early as possible, and to further improve the Chinese TB control strategy, we should pay attention to new or previously neglected risk factors, such as weather [
8]. At present, most studies have focused on the relationship between TB incidence and meteorological factors at the provincial level [
9‐
11]. To further control TB incidence, we need to study the association between TB and meteorological factors from a narrower geographical area level. Additionally, biological research suggests that a temperature of 37 °C and sufficient oxygen and water are favoured for the propagation of
M. tuberculosis, but the pathogen is very sensitive to ultraviolet (UV) light, which is found in sunlight [
12]. Therefore, it was speculated that meteorological factors may affect the tuberculosis incidence by affecting the growth and reproduction of
M. tuberculosis to a certain extent. Therefore, the aim of this study was to explore whether meteorological factors from the prefecture level are potential determinants of TB incidence in China in order to provide a theoretical basis for the prevention and control measures of TB.
Discussion
During 2005–2015 in mainland China, the spatial distribution presented a high incidence in the west and a low incidence in the east. The western region with high incidence included the Xinjiang Uygur Autonomous Region, the Tibet Autonomous Region, Heilongjiang Province and Gansu Province. In contrast, the eastern region including Beijing, Tianjin and Shandong Province enjoyed a low TB incidence. The possible reasons for the geographic differences are as follows: first, the western region has a relatively underdeveloped economy, poor medical resources, a low educational level and a high proportion of minorities [
21]; in contrast, eastern China has a relatively developed economy, adequate medical resources, a high educational level and a good living and work environment contribute to the low TB incidence. A recent survey of tuberculosis hospitals in China found that 34.6% of TB patients lived in western China, but only 12.8% of hospitals and 14.8% of beds were located in that region in 2010 [
4]. In addition to the environment (socioeconomic) and host factors, we can also analyse these factors to observe how they relate to the pathogenic bacteria.
M. tuberculosis lineage 2 was dominant in the eastern region, while
M. tuberculosis lineages 2, 3 and 4 were present in the western region [
22,
23]. Many studies have suggested that
M. tuberculosis lineage 2 is more pathogenic than other lineages [
22] but the western China presented a more serious condition than the east, which implies that the environment may play a more important role in TB onset than the pathogen itself, so it is meaningful to conduct research on the meteorological factors. Additionally, from the cluster and hotspot analyses, we found that six areas (Fig.
3a) (four areas were in the west (①, ②, ④ and ⑤), one was east of Heilongjiang (③), one was in the Zhujiang Delta, Guangdong (⑥)) had high TB incidence, which is the key areas for TB control and provention.
A correlation analysis with the GWR model showed that AR had a positive correlation with TB incidence. The increased rainfall may create a suitable environment for the growth and reproduction of
M. tuberculosis [
10,
24]. However, a negative correlation was observed between ASD and TB incidence. For ASD, on the one hand, a long sunshine duration with large amounts of UV light would restrict the development of
M. tuberculosis [
25]. On the other hand, the UV light could help the synthesis of vitamin D, which could protect people from TB to a certain extent [
26,
27]. Additionally, biological research [
12] support our study of meteorological factors across the country; that is, TB incidence is positively correlated with the average precipitation and is negatively corrected with the average sunshine duration.
M. tuberculosis is the pathogen of human TB, and humans are hosts. The time from inhalation of M. tuberculosis to onset is long for humans, and meteorological factors are not relevant at this stage. However, meteorological factors may play an important role when TB patients expel M. tuberculosis into the surrounding environment by spitting. Our study also suggested that ARH is negatively correlated with the incidence of TB in the whole country (p < 0.05). In aerosol form only, M. tuberculosis can be inhaled into the lung to cause TB. Humanity has reduced dust emissions and the formulation of aerosols, and has thus reduced the spread of TB. In addition, TB incidence was associated with AWS and AAP in many prefectures. The higher AWS could accelerate ventilation, dilute the concentration of bacteria and help reduce the risk of becoming infected. For the AAP, increased atmosphere flow usually forms from high air pressure regions to low air pressure regions, so the mechanism of negative correlation between air pressure and TB incidence may be similar to wind speed, but further explorations are needed.
There are some limitations of the study. First, the potential for under-reporting of cases is inevitable in surveillance data, which is dependent on healthcare-seeking behaviours. In addition, if there is less healthcare available in certain areas, one might expect less reporting, so TB incidence would be underestimated. Second, this is a cross-sectional study with all data collapsed from 2005 to 2015, so the time effects, such as a lag effect, were ignored. Third, we took only meteorological factors into consideration, and some other risk factors associated with TB incidence, such as healthcare access, socioeconomic status and individual-level factors that correlate with geography were not considered due to the unavailability of data. Therefore, further studies should control for other factors affecting TB at a more detailed time scale. Moreover, an ecological study cannot provide conclusive results but can only generate and develop hypotheses.