Changes in Incidence and Epidemiological Characteristics of Pulmonary Tuberculosis in Mainland China, 2005-2016

This cross-sectional study assesses the recent epidemiological trend of pulmonary tuberculosis in mainland China based on national surveillance data.


Introduction
Tuberculosis (TB), an infectious disease caused by Mycobacterium tuberculosis, is not only among the top 10 causes of death in the world but also the leading fatal single infectious disease. 1 Three nationwide surveys regarding TB epidemiology in mainland China showed a decline in the prevalence of TB by more than 50% during the past 20 years owing to the broad adoption of a directly observed treatment, short-course strategy (DOTS) and the End TB Strategy. 2,3 The annual TB incidence and mortality in China decreased by a mean of 3.2% and 7.7%, 4 respectively, which is higher than the world mean (2.0% and 3.0%, respectively). 5 Nevertheless, the number of new cases in China in 2017 was the second highest in the world despite the ongoing interventions.
After the severe acute respiratory syndrome outbreak in 2003, China established its first web-based National Notifiable Infectious Disease Surveillance System for 39 reportable infectious diseases, 6 including TB. To collect more comprehensive information about patients with TB, starting January 1, 2005, China developed an additional web-based national TB surveillance system, the Tuberculosis Information Management System (TBIMS), to which all TB health facilities are required to report diagnosed cases of TB. 7 The TBIMS allows real-time monitoring of TB diagnosis, treatment, and outcomes in China, especially for pulmonary TB (PTB).
In this study, we collected PTB data reported to the TBIMS from January 1, 2005, to November 21, 2016, and characterized the epidemiology of PTB in mainland China, focusing on changes in annual incidence, demographic characteristics, geographic patterns, seasonal patterns, and delay in diagnosis. To our knowledge, this is the largest retrospective epidemiological study on PTB based on the TBIMS in China.

Ethical Approval
Data from patients with TB were reported to the TBIMS as part of routine public health surveillance, and no informed consent was required according to the National Health Commission of the People's Republic of China. The ethics committee at the Beijing Chest Hospital concluded that this crosssectional study was exempt from institutional review because only deidentified data were analyzed.
The methods and findings from this study are reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Case Definition
We defined PTB as bacteriologically confirmed or clinically diagnosed TB in the lung parenchyma or the tracheobronchial tree. Because of lesions in the lungs, miliary TB was considered PTB. A patient with both pulmonary and extrapulmonary TB was classified as having PTB. 8,9 Bacteriological diagnosis was based on test results of sputum smear or isolated culture as the reference standard. Clinical diagnosis was based on chest imaging (radiography or computed tomography), supplemented by epidemiological investigation, clinical manifestation (coughing, expectoration Ն2 weeks, or hemoptysis), or results of an immunology test (tuberculin skin test and/or interferon gamma release assay). 10 from 2005 to 2016 are the longest and most recently available we could obtain. Meanwhile, we obtained annual population statistics in different provinces from the National Bureau of Statistics of the People's Republic of China to calculate the PTB incidence. 12

Statistical Analysis
We divided the 31 provinces in mainland China into 3 regions: western, central, and eastern. We calculated the overall and provincial PTB annual incidence (per 100 000 population) by testing method and age group. Because the data for December 2016 were not available, we estimated the number of cases in December using the mean number of cases in the first 11 months to calculate the incidence of 2016. To quantify seasonal patterns of PTB by province, we used a heat map of proportions of weekly case numbers among the annual total case number, with the means calculated during the study years. A joinpoint regression of annual incidences over time was used to identify change points in the temporal trend of PTB incidence. We stratified subsequent analyses by the periods defined by the change points. A ring map was made using ArcGIS, version 10.4 (Environmental Systems Research Institute, Inc) to demonstrate the spatiotemporal pattern of PTB incidence at the provincial and annual levels.
We fitted parametric (Weibull, gamma, and log-normal) and nonparametric (kernel density) distributions to the time from illness onset to diagnosis. Model fitness was visually examined by comparing a fitted density curve to the observed frequencies, and, when necessary, parametric distributions were compared using the Akaike information criterion. 13 This analysis of diagnostic delay was performed by study period, age group, region, sex, and occupation.
We used SAS, version 9.4 (SAS Institute Inc) and R, version 3.6.0 (R Project for Statistical Computing) for data management and analysis. All statistical tests were 2-sided with a significance level of P < .05. years; 28.53% were 60 years or older, and 0.8% were younger than 15 years (

Incidence and Epidemic Characteristics
The annual incidence of PTB in mainland China initially increased from 72.95 per 100 000 population in 2005 to 77.53 per 100 000 population in 2007, followed by a gradual but steady decline to 52.18 per 100 000 population in 2016 (Figure 1A  At the national level, the reporting of PTB exhibits a clear seasonal pattern, jumping from the valley near January or February to the peak in March or April and then declining gradually over the rest of the year ( Figure 1C). Spring Festival holidays in February (occasionally in January) could have partially contributed to the lower reporting in the month. The holiday effect is also seen from the slightly higher case numbers in November than in October for most of the years due to the holiday week associated with National Day on October 1.

Delay in Diagnosis
There was a median delay of 32 (IQR, 15-67) days from disease onset to diagnosis among all patients with reported PTB during the study period ( Figure 3A).     Figure 3D). Pediatric and adult patients had similar delay times ( Figure 3E), but the delay tended to be longer among the group 60 years or older (median, 34 [IQR,    Figure 3F).

Discussion
We summarized epidemiological characteristics of more than 10 million patients with PTB reported from 2005 to 2016 in mainland China, which is, to our knowledge, the largest epidemiological study of TB in the country. We provided a thorough descriptive analysis of the temporal trend of PTB incidence at the national level as well as by demographic and geographic subpopulations. TB to local public health authorities via an internet-based reporting system within 24 hours. 15 In addition, the Ministry of Health issued a policy to strengthen collaboration between the hospitals and TB dispensaries on diagnosis and treatment, which doubled the TB discovery rate. 15 Last, free diagnostic testing and therapies for patients with TB were made available nationwide starting in 2004, which greatly reduced the economic burden of patients, shortened the diagnosis delay, and improved treatment adherence and outcomes. 16,17 Although the incidence of TB has been declining in China since 2007, the targets of a 90% reduction in incidence and a 95% reduction in mortality by 2035 remain difficult to achieve, even if all existing interventions are scaled up. 5 It is therefore crucial to target effective interventions at highrisk populations and areas that have been underserved. Our study found that male farmers and herdsmen, especially those in the west, constituted the underserved high-risk subpopulation in China, likely due to tobacco use, corticosteroid use, immunity levels, migration, and living environment. 18 Tuberculosis-related education programs and community-based TB screening strategies should be tailored to this subpopulation. 5,19 Compared with the eastern and central regions, the western region has a much lower population density (53.74/km 2 vs 310.09/ km 2 to 413.05/km 2 ), 12,20 yet its incidence was higher and declining at a slower rate. A major contributing factor is the less developed socioeconomic infrastructure in the west. 21 Tuberculosis is known to be a disease associated with poverty. 4 The gross domestic product per capita during 2005 to 2016 in the vast western region was ¥21.95 thousand compared with ¥47.10 thousand in the eastern and ¥26.22 thousand in the central regions. 22 Although a low population density helps deter transmission of infectious diseases, it also increases the difficulty of health services reaching people in need. More cost-effective prevention and control strategies, possibly aided by mobile phone technologies, such as TB-dedicated self-screening and compliance-monitoring apps, should be designed for the western region. 5,23 In addition to economic and logistic factors, social and cultural barriers could explain the higher incidences and slower decline in western and southwestern provinces such as Xinjiang (135.03 per  Delay in diagnosis of TB is common in low-and middle-income countries 25 despite the key role of early diagnosis for effective TB treatment and desired outcomes. 26,27 Our study showed

Conclusions
This cross-sectional study found that, from 2005 to 2016, the incidence of PTB in mainland China showed a downward trend. Comprehensive, scalable, and cost-effective interventions should target high-risk populations, particularly those in rural areas and among ethnic minority groups, to sustain or accelerate the decline toward achieving the World Health Organization's goal of eliminating TB by 2035.