A Neighbor-Based Approach to Identify Tuberculosis Exposure, the Kopanyo Study

Contact investigation is one public health measure used to prevent tuberculosis by identifying and treating persons exposed to Mycobacterium tuberculosis. Contact investigations are a major tenet of global tuberculosis elimination efforts, but for many reasons remain ineffective. We describe a novel neighbor-based approach to reframe contact investigations.

Contact investigation is one public health measure used to prevent tuberculosis by identifying and treating persons exposed to Mycobacterium tuberculosis. Contact investigations are a major tenet of global tuberculosis elimination efforts, but for many reasons remain ineffective. We describe a novel neighbor-based approach to reframe contact investigations.
future-related patients as culture-positive patients with matching genotypes diagnosed after exposure to an index patient. Concurrent disease was TB diagnosed in a contact within 90 days of the index patient.
When we applied a neighbor-based approach, we noted that 257 (24%) index patients could have subsequent TB patients living next door (n = 749), 41 of which could have concurrent disease. Among nextnearest neighbors of index patients, 390 (36%) could have subsequent TB, 23 of which could have concurrent disease (Table). In addition, 29 (2.7%) index patients could have future-related patients among their nearest neighbors (n = 42), and 5 (0.5%) future-related patients among next-nearest neighbors (n = 10), 3 with concurrent TB disease (Table).
We found that a neighbor-based approach could identify 1,565 additional subsequent TB patients, including 175 future-related patients, and 102 patients with concurrent TB disease. The number of persons living with a bacteriologically positive patient varied by geography; however, ≈23,630 contacts potentially could benefit from TPT. Of note, 9% (97/1,072) of index patients interviewed stated they lived alone, but 91 (94%) had subsequent patients identified in the home, and 84 (87%) had subsequent future-related patients living in the home.

Conclusions
We explored the use of a nearest-neighbor approach to expand TB contact investigations. This approach does not rely on name-based contact identification, which has been shown to be ineffective (6,(16)(17)(18). In addition, the neighbor-based approach would not require mobile screening units or mass screening campaigns in the community. By simply expanding the number of homes visited to nearest and next-nearest neighbors, the Botswana National TB Program could increase the number of TB case diagnoses by 146% and potentially interrupt 175 secondary patient transmission events.
Preventing future TB disease through TPT could also hasten TB elimination in at-risk neighborhoods and reduce deaths in the community (11,12). Cegielski et al. effectively used TPT to eliminate TB from 2 atrisk neighborhoods in Texas, USA (11). The focus on nearest and next-nearest neighbors gives programs a tangible and practical approach to locating persons at risk for TB exposure and progression to TB disease. The neighbor-based approach differs from a neighborhood screening, which places an additional burden on TB programs by unnecessarily screening many persons at lower risk. For example, 59,100 persons reside in neighborhood C in Gaborone (data not shown). Under the neighbor-based approach, only 5,470 (9%) persons, including in-home and nearest neighbor residents, would be targeted for testing.
Previous reports suggest that contact investigations fail to identify key relationships, even within households (16,17). Potential stigma and lack of trust in government officials also play a role in contact investigations (16)(17)(18). In our cohort, many (n = 97) index patients said they lived alone, but 94% of them had subsequent cases identified in the home. In addition, 48% of future-related patients were linked to index patients who claimed no household contacts during name-based contact solicitation interviews conducted at the enrollment clinic. Household membership composition could have changed over time, and some connections might not have existed at the time of the interview. However, our study reinforces the necessity of home visits at times convenient to the index patient and when most household members are in the home, which might warrant home visits outside of business hours and flexibility in staff workplans.
Our analysis emphasizes the opportunity to prevent future TB patients and future-related TB patients by providing TPT. Household contacts, especially young children and persons living with HIV, are eligible for TPT by national policy, but TPT has not been practiced routinely in Botswana. As the Botswana Ministry of Health scales up access to TPT throughout the country, the neighbor-based approach could improve identification of most likely contacts and help target interventions where they are most needed.
Our study has limitations. Living in proximity to an index patient is not the only opportunity for transmission and might not always translate into time spent together. In addition, our analysis of futurerelated patients included only patients with culturepositive disease and genotyping results; excluding them did not affect the main analysis enumerating subsequent patients but might have underestimated the number of future-related patients. Also, our estimates for TPT represent the maximum number of persons who could benefit because we used the average number of persons per household and assumed all household members would be eligible for TPT without a reliable and available test for infection.
A neighbor-based approach should not supplant household investigations, and community-based interventions should not divert essential resources from those already devoted to finding and treating TB patients. Wide-scale implementation of this approach would require adequate resources to ensure that all patients complete the full cascade of treatment. To reach the ambitious global goal of TB elimination, we need simple, easy to implement, location-based approaches. Screening index patient households and nearest neighbors could help identify additional TB patients and persons who could benefit from TPT.