Menu
Menu
Menu
Menu
BACKGROUND: Household contact tracing for tuberculosis (TB) may facilitate diagnosis and access to TB preventive treatment (TPT). We investigated whether household contact tracing and intensive TB/human immunodeficiency virus (HIV) screening would improve TB-free survival. METHODS: Household contacts of index TB patients in 2 South African provinces were randomized to home tracing and intensive HIV/TB screening or standard of care (SOC; clinic referral letters). The primary outcome was incident TB or death at 15 months. Secondary outcomes included tuberculin skin test (TST) positivity in children ≤14 years and undiagnosed HIV. RESULTS: From December 2016 through March 2019, 1032 index patients (4459 contacts) and 1030 (4129 contacts) were randomized to the intervention and SOC arms. Of intervention arm contacts, 3.2% (69 of 2166) had prevalent microbiologically confirmed TB. At 15 months, the cumulative incidence of TB or death did not differ between the intensive screening (93 of 3230, 2.9%) and SOC (80 of 2600, 3.1%) arms (hazard ratio, 0.90; 95% confidence interval [CI], .66-1.24). TST positivity was higher in the intensive screening arm (38 of 845, 4.5%) compared with the SOC arm (15 of 800, 1.9%; odds ratio, 2.25; 95% CI, 1.07-4.72). Undiagnosed HIV was similar between arms (41 of 3185, 1.3% vs 32 of 2543, 1.3%; odds ratio, 1.02; 95% CI, .64-1.64). CONCLUSIONS: Household contact tracing with intensive screening and referral did not reduce incident TB or death. Providing referral letters to household contacts of index patients is an alternative strategy to home visits. CLINICAL TRIALS REGISTRATION: ISRCTN16006202.
We conducted an open, 2-arm, cluster-randomized trial of household contact tracing and intensive TB/HIV screening in South Africa (ISRCTN16006202). Methods have been described previously (Supplementary Materials, Protocol) [19]. The Mangaung Municipality in Free State Province is predominantly urban with an estimated population of 780 755, an antenatal HIV prevalence of 31.7%, and estimated annual TB incidence in 2019 of 476/100 000. In 2018, the more rural Capricorn Health District in Limpopo Province had an estimated population of 1 338 763 [20], an antenatal HIV seroprevalence in 2015 of 21.6%, and estimated annual TB incidence in 2019 of 201/100 000 [21, 22]. During the study period, there were few programmatic attempts made to identify and screen household contacts for TB. Study teams identified consecutive eligible index TB patients at government clinics and hospitals within study site boundaries. We included TB patients of any age but required those aged ≥7 years to have laboratory-confirmed pulmonary TB, whereas those aged <7 years could have physician-diagnosed TB of any organ, with or without laboratory confirmation. We additionally included TB patients who died within 8 weeks of TB diagnosis. We excluded institutionalized TB patients and withdrew participants whose households we could not locate or from where no household member could be recruited. A list of household contacts was obtained at enrollment. Households of index patients were defined as people living together within a set of rooms under a contiguous roof linked by doorways or windows through which air moved and where household members had shared airspace by either sleeping overnight at least once or had shared at least 2 meals in the same household as the index case in the 14 days prior to the index case’s diagnosis of TB. Index cases and their households were block-randomized to either intervention or standard of care (SOC) in a 1:1 ratio, stratified by district. Investigator blinding was maintained until after the final participant household follow-up was completed. In the intervention group, research fieldworkers visited households within 14 days of index TB patient enrollment (maximum 3 attempts), obtaining written individual or parental consent for adults and children aged <18 years, respectively, with assent from older children. A questionnaire was administered to each household member (Supplementary Materials, Questionnaires), and sputum specimens were obtained where possible (but not required from children aged <5 years) and were tested using Xpert and mycobacterial growth indicator tube (MGIT) culture. Household contacts received TST (from a variety of sources due to global shortages), administered and read within 72 hours [23]. Study nurses dispensed the first month of TPT (6 months of daily isoniazid) to participants living with HIV who tested negative for TB, participants not living with HIV with positive TST (≥10 mm), and children aged <5 years. Subsequent TPT was obtained from local clinics. For household members without a confirmed HIV diagnosis, rapid point-of-care HIV testing was offered to participants aged ≥18 months, and polymerase chain reaction on dried blood spot was provided for children aged <18 months whose maternal HIV status was unknown or positive. Participants living with HIV had a CD4 count measured and were referred to their nearest clinic for assessment and initiation of ART. Intervention households were visited approximately 3 months after enrollment to support treatment linkage. In the SOC arm, index TB patients (or their representative, if deceased or a child) were given referral letters for every household member by the recruiting team at the health facility, recommending that each household contact take the letter to their local clinic and be screened for TB and HIV. At 15 months after randomization, study teams visited all households, updated the household membership list, and recorded episodes of incident TB and death. We investigated household members for HIV (if untested) and TB (if symptomatic). All children aged ≤14 years had TST placed, read at 48–72 hours. The primary outcome was time to TB or death, measured among all household members included in the household census at baseline, from 1 month after randomization through the final 15-month ascertainment visit. Primary analysis included all incident TB diagnoses, irrespective of diagnostic method; sensitivity analyses included only bacteriologically confirmed incident cases of TB. Secondary outcomes were prevalence of TB infection (TST induration ≥10 mm) at month 15 among household children aged ≤14 years, time to initiation of TB treatment, and prevalence of undiagnosed or untreated HIV at month 15. Primary analyses for all outcomes were restricted to household contacts resident at baseline enumeration; supplementary analyses included all household contacts regardless of baseline residency. In protocol-specified subgroup analysis, we compared outcomes by trial site and TST positivity by household contact age (<5 years, ≥5 years). The University of Witwatersrand Human Research Ethics Committee (Medical) and the London School of Hygiene and Tropical Medicine granted ethical approval. Assuming a mean household size of 5.5 and a primary outcome incidence of 2000/100 000 person-years, 1200 index cases per site (total 2400) provided 80% power to detect a 30% overall difference in the primary outcome between groups with alpha 0.05 and intracluster correlation coefficient 0.3. All statistical analyses were performed using Stata v16 (StataCorp, College Station, TX). Analyses were done on an intention-to-treat basis. This study is reported following CONSORT guidelines for cluster-randomized trials (Supplementary Materials, Checklist). We summarized baseline index and household characteristics by trial arm. For the primary outcome, follow-up time began 1 month after randomization (to avoid counting prevalent TB cases) and ended at the month-15 visit or the date of TB or death. Cox proportional hazards regression with robust standard errors was used to assess the impact of the intervention on the primary outcome, with a time-by-treatment interaction term fitted to assess the proportionality assumption. Logistic regression with generalized estimating equations was used to assess the impact of the intervention on binary outcomes. Interaction terms were fitted to assess effect modification in planned subgroup analyses.
