Spatiotemporal epidemiology of, and factors associated with, the tuberculosis prevalence in northern China, 2010-2014.

BACKGROUND: Tuberculosis (TB) is an important public health issue worldwide. However, evidence concerning the impact of environmental factors on TB is sparse. We performed a retrospective analysis to determine the spatiotemporal trends and geographic variations of, and the factors associated with, the TB prevalence in Inner Mongolia. METHODS: We performed a retrospective analysis of the epidemiology of TB. A Bayesian spatiotemporal model was used to investigate the spatiotemporal distribution and trends of the TB prevalence. A spatial panel data model was used to identify factors associated with the TB prevalence in the 101 counties of Inner Mongolia, using county-level aggregated data collected by the Inner Mongolia Center for Disease Control and Prevention. RESULTS: From January 2010 to December 2014, 79,466 (6.36‱) incident TB cases were recorded. The TB prevalence ranged from 4.97‱ (12,515/25,167,547) in 2014 to 7.49‱ (18,406/ 24,578,678) in 2010; the majority of TB cases were in males, and in those aged 46-60 years; by occupation, farmers and herdsmen were the most frequently affected. The Bayesian spatiotemporal model showed that the overall TB prevalence decreased linearly from 2010 to 2014 and occupation-stratified analyses yielded similar results, corroborating the reliability of the findings. The decrease of TB prevalence in the central-western and eastern regions was more rapid than that in the overall TB prevalence. A spatial correlation analysis showed spatial clustering of the TB prevalence from 2011 to 2014 (Moran's index > 0, P < 0.05); in the spatial panel data model, rural residence, birth rate, number of beds, population density, precipitation, air pressure, and sunshine duration were associated with the TB prevalence. CONCLUSIONS: The overall TB prevalence in Inner Mongolia decreased from 2010 to 2014; however, the incidence of TB was high throughout this period. The TB prevalence was influenced by a spatiotemporal interaction effect and was associated with epidemiological, healthcare, and environmental factors.