Contact Investigation: A Priority for Tuberculosis Control Programs

Tuberculosis (TB) is one of the greatest threats to global health, productivity, and socioeconomic development (1). In 2014, TB killed 1.5 million people, 0.4 million of whom were HIV infected, and it now ranks as the leading cause of death from infectious disease worldwide (2). To address the persistent human suffering associated with TB, the World Health Organization has proposed a highly ambitious, multisectoral strategy to end the TB epidemic by 2035 (3). The enormity of this task is highlighted by the fact that at least one-third of all people suffering from TB are never diagnosed and treated, leading to persistent TB transmission by infectious cases within households and communities (4). Furthermore, an estimated 2 billion people have latent TB infection and therefore serve as a reservoir from which new cases of TB arise, propagating the global epidemic (5).

Contact investigation is defined as the systematic evaluation for TB disease and/or latent TB infection in people who have close contact with TB “index” cases and is recommended by the World Health Organization (6). Such “contacts” have a high risk of concurrently having or subsequently developing TB disease themselves and therefore represent an accessible population from which new cases may be promptly diagnosed and treated, and to which TB preventive therapy may be targeted (7). Despite these potential opportunities, contact investigation has generally been neglected as an intervention in low- and middle-income countries because of inadequate human and material resources, insufficient programmatic emphasis, ineffective tools for predicting which contacts are at highest risk of developing TB disease, and a shortage of evidence with which to optimize guidelines (7).

In this issue of the Journal, the study by Martinez and colleagues (pp. 1152–1163) contributes important evidence to this field (8). In their large study of household contacts of TB index cases in a high TB- and HIV-burden setting, the authors aimed to characterize the relationship between index case HIV status and the rate of latent TB infection (defined through tuberculin skin testing) among contacts in order to make inferences about infectiousness. In this study, contacts of an HIV-infected index case were less likely to have latent TB infection, regardless of the threshold used for a positive tuberculin skin test or the age of the contact. In addition, the contacts of HIV-infected index cases with sputum smear–positive or apparent cavitary TB disease were as likely to have latent TB infection as the contacts of non–HIV-infected index cases. These findings highlight the interaction between contagion and the presentation of TB disease in HIV-infected persons, who are less likely to have a prolonged illness, lung cavities, or be smear positive and therefore are considered to be less infectious (9). Of note, all index cases recruited to the study were culture positive, partially explaining the unusually high smear-positivity rates and classic radiological findings observed within the cohort, which are uncommon in HIV/TB coinfection (10).

Although limited by their inability to confirm transmission directly between index cases and contacts, the current findings suggest that the subgroup of HIV-infected index cases with sputum smear–negative or noncavitary disease may be less infectious than are non–HIV-infected index cases with the same disease characteristics (8). While a smear-negative culture-positive status indicates paucibacillary TB disease, HIV-infected index cases may have had more pronounced immunosuppression, thus were more unwell and had less frequent or weaker cough to produce infectious aerosols (11). Indeed, a limitation of the current study is the absence of data on severity of immunosuppression, antiretroviral use, and indication of general health status or frailty of the index case, including duration of hospitalization. Furthermore, the study baseline characteristics imply that households with HIV-infected index cases had a different demographic and socioeconomic structure. The authors’ findings may have been strengthened by more detailed characterization of the household environment, which plays an important role in the dynamics of TB transmission (12). Another possibility is that the host–pathogen interaction in immune-competent individuals may change the Mycobacterium to a more infectious phenotype (13), a phenomenon that may not occur in immunosuppressed individuals.

So what do these findings really mean for TB control programs in high TB- and HIV-burden settings where contacts are typically defined by sharing a household with the index case? Overall, the rates of latent TB infection and TB disease in this cohort were high, with approximately 70% latently infected and 6% having or subsequently developing disease over 2 years. Even among contacts of the less infectious HIV-infected index cases, more than half had latent TB infection and, importantly, the rates of coprevalent and incident TB disease did not differ by index case HIV status. These findings emphasize the limited value of tuberculin skin testing for predicting which individuals actually become ill with TB disease (14). It is well established that latent TB infection and subsequent development of TB disease in contacts is the result of a complex interplay between index case, environmental, and contact characteristics, which are frequently clustered within households (12). In high-burden settings, this complexity is exacerbated by the fact that community transmission accounts for a higher proportion of contact TB cases than does household transmission, meaning that it is frequently difficult to identify who the index case actually is (15).

These observations beg the question: in high-burden settings, why use index case characteristics to prioritize contact investigation at all? Once a household, through the diagnosis of one household member, has revealed itself to a TB control program as a TB hotspot, contact investigation to identify coprevalent cases should be prioritized immediately and contacts should be evaluated systematically for their risk of developing TB disease in order to provide TB preventive therapy to those who are at highest risk. This process may include HIV testing and could incorporate locally developed risk factor assessments, enabling TB control programs to move away from the current “one size fits all” approach that is often dependent on unreliable and impractical tuberculin skin testing (16).

If the highly ambitious targets set out in the End TB Strategy are to be made a reality, the missing third of people suffering from TB are to be diagnosed and treated, and TB is to be prevented in vulnerable people, then contact investigation should rapidly become a priority for TB control programs. This should be accompanied by further research better characterizing the intricate relationship between index case, environmental, and contact characteristics with TB infection and disease.