Tuberculosis burden in China: a high prevalence of pulmonary tuberculosis in household contacts with and without symptoms

A few studies have indicated that close contacts have a higher risk of developing active tuberculosis [1‐5]. China national tuberculosis program has yet included case detection among this group due to inadequate resources in the past. By 2005, the coverage of DOTS has achieved 100% at county scale while the case detection rate and the cure rate of the new sputum smear positive patients reached 79%and 91% respectively in China, indicating that China met the WHO’s staggered global targets for detection and treatment of tuberculosis cases [6‐9]. Given the nature of tuberculosis surveillance being passive, most patients could not be detected until they present symptoms and seek health care [10]. In 2007, China’s Ministry of Health piloted a new screening strategy among suspects in household contacts of notified tuberculosis cases (Index Cases) in several regions to inform a possible screening strategy nationwide [11]. More positive pulmonary tuberculosis patients have been detected in this pilot than routine surveillance [2], which inspired a more extensive case detection among all household contacts considering the Mega health investment in health medical reform to eliminate tuberculosis from population [12]. Since 2008, a rigid screening strategy in household contacts with symptoms has been tentatively extended to the household contacts without symptoms.

The present study examines effectiveness of the screening strategies among the household contacts with or without symptoms, to inform whether the screening strategy should cover the household contacts without symptoms.

For the present study, the Index Case is defined as a smear positive pulmonary patient who was first to be diagnosed in the household, but not always the first real patient. To our knowledge, there is no consistent definition of household contacts [13]. In our study, the household contacts are either family members or not-family members who have been living in same house with the Index Case for more than two weeks in the past three months before the data when the Index Cases were diagnosed. Notified cases were the number of pulmonary tuberculosis cases registered in tuberculosis national surveillance system (TBNSS). We calculated notification rate by dividing the number of notified cases by the number of total population in a geographically defined location. The prevalence among household contacts was calculated as the percentage of pulmonary tuberculosis cases among 100,000 household contacts. Delayed reporting was defined as the interval being more than two weeks between the onset of any symptom and the presentation to a health care provider [14].

Out of 1,575 households of Index Cases, 3,381 household contacts were identified and 3,355 (99.2%) examined (Figure 2). Most Index Cases (1,412/1,575) were the new positive cases, 77% were men and 80% were reported more than two weeks (delayed reporting) (Table 1).

Of the 3,355 household contacts, 92 people had pulmonary tuberculosis (2,742 per 100,000 populations) with 21 being smear positive cases (626 per 100,000 populations). Of 92 pulmonary tuberculosis diagnosed, 46 cases (9 smear positive cases) had not yet presented any clinical symptoms (Table 1). For household contacts without symptom, the prevalence of pulmonary tuberculosis (1,483 per 100,000 populations) was 20 times higher than the notified rate (73 per 100,000 populations) and the prevalence of smear positive cases (290 per 100,000 populations) was 7 times higher than the notification rate (43 per 100,000 populations). Among household contacts with symptoms, the prevalence of pulmonary tuberculosis and smear positive cases were 248 (18,110 per 100,000 populations) and 107 (4,724 per 100,000 populations) times higher than the notification rates respectively (Table 1).

By adjusting household clustering, household contacts presenting symptoms had 14 times higher risk to be diagnosed as pulmonary tuberculosis in comparison with the those who have not yet presented any symptoms (OR13.94, P < 0.05) (Table 2). The pulmonary tuberculosis was more likely to be found among the spouses or sisters/brothers of the Index Cases and among those who were in contact with the female Index Cases (Table 2). However, the pulmonary tuberculosis was not found associated with age, cavitary lesions, treatment history of the Index Cases and living in the same bedroom with the Index Cases tuberculosis (Table 2). Among household contacts, the elder people (aged above 65 years old) (AOR4.96, P < 0.05) and men (AOR1.52, P < 0.05) were most likely to be diagnosed tuberculosis (Table 2).

The study presents very high prevalence of pulmonary tuberculosis in the household contacts with or without any symptoms, which are about six times or nine times higher in comparison with report of 2010 national tuberculosis prevalence survey (2742 per 100,000 populations vs. 459 per 100,000 populations and 626 per 100,000 populations vs. 66 per 100,000 populations) respectively [16]. Assuming the difference in prevalence between the present study and 2010 national tuberculosis survey can be extrapolated to nationwide, there might have been 30,748 undiagnosed pulmonary tuberculosis cases nationwide if all the household contacts of all notified cases in 2008 were screened. This presents a great missed opportunity to detect a large pool of pulmonary tuberculosis in the population to further advance the remarkable achievement in past two decades with fighting against tuberculosis in China [9].

Not surprising, our study find that the household contacts presenting symptoms have 14 times the likelihood (OR95% CI, 13.94 (8.61, 2.59)) to be diagnosed as pulmonary tuberculosis comparing with those have not yet presented any symptoms (Table 2). This finding is consistent with previous studies but we could not determine whether the household contacts have been infected by TB cases inside or outside the family [1, 2]. However, we find the prevalence of pulmonary tuberculosis and positive tuberculosis are respectively 20 times and 7 times higher than the notification rates which is publicly reported (Tables 1 & 2). This implies that early active case finding and treatment of tuberculosis patients without presenting symptoms among close contacts might bring benefit for the general public health. Although whether the screening strategy among the household contacts is cost-effective remains controversial, it is obvious that eradicating tuberculosis from the population totally needs to identify all potential transmission sources and eliminate them [17, 18]. This concept of cost-effectiveness should be observed in a longer period of term and from a global perspective rather than measuring the cost and in local areas only. Furthermore, most Index Cases had a history of TB before being found and might contribute to drug resistance which will worsen TB burden. The on-going medical reform in China may increase the budget and expenditure on tuberculosis control [12]. Screening all household contacts of Index Cases is feasible and also a key step to bring tuberculosis control to sustainability in China. In fact, Index Cases are not always detected by passive surveillance now and increased numbers of tuberculosis patients have been detected by improved education and routine physical examinations [19].

In contrast to men, women are about 1.7 times more likely to transmit pulmonary tuberculosis to the household contacts while the male household contacts are more easily infected (AOR1.5) (Table 2). This may be due to the fact that women have more chance and longer time to be in contact with other family members. A local study and the national survey in rural areas in 2008 all indicate that about 74% of rural women’s work is responsible for taking care children and elderly people at home [20, 21]. In the present study, more than 80% of Index Cases are in the rural areas, consistent with the distribution of tuberculosis in China [7, 16].

The data from the present study do not support our hypothesis that cavitary tuberculosis, new smear positive cases and delayed reported patients may be associated with a higher probability of transmission within the home (Table 2). These findings may imply that Index Cases may not be the first patient in the family although they are the cases notified. Another unexpected outcome is the higher rate of non-tuberculosis cases in household contacts with symptoms (34/46 with symptoms vs. 37/46 without symptoms), which might be expected to be lower (<50%) in view of the low HIV prevalence in China. This unusual result might be due to wide mobilization during the study which may result in overstated symptoms when the household contacts were interviewed. Clearly further study is needed.

No bacterial genotyping was performed the present study. This disables us to identify whether the household contacts are infected by Index Cases or in the community. However, convincing evidence in the present study indicates that household contacts have more exposure to tuberculosis. There are a large number of potential pulmonary tuberculosis cases yet to be detected in China, irrespective of having symptoms or not. We also did not assessed HIV status for both Index Cases and the household contacts, which might influence our results, but China is still low HIV prevalence, which limits HIV/tuberculosis co-infection as a major confounder [22]. The type of building and ventilation of house might be associated with TB infection, but our study focused on rural areas where TB cases account for more than 80% in China. In these areas, there is little difference in the type of building and ventilation of house, so this factor will is not main contribution to difference in TB infection.

Despite the limitation and controversies, the present study highlights that screening only the symptomatic contacts is clearly not enough to sustain the achievement of tuberculosis control in China, as well as in the world.