Risk-benefit analysis of tuberculosis infection testing for household contact management in high-burden countries : a mathematical modelling study

Background Preventive therapy for tuberculosis reduces the risk of disease in people who have been infected but who are not sick. Countries with a high burden of tuberculosis that are expanding preventive therapy use must decide how tuberculosis infection testing should be used for risk stratification among household contacts of patients with tuberculosis. Methods We modelled the risks of tuberculosis disease and severe adverse events, comparing the following two preventive therapy strategies: preventive therapy for all household contacts, or preventive therapy for only household contacts with a positive tuberculin skin test (TST) result. We used data from clinical trials and literature on tuberculosis natural history to model outcomes, assuming different preventive therapy regimens, ages, and TST positivity prevalence. Findings Assuming 25% prevalence of TST positivity among 1000 household contacts aged 0–17 years, a treat-all approach with isoniazid and rifapentine compared with a treat-TST-only approach led to 13 fewer incident tuberculosis cases (IQR –5 to –18) and four additional severe adverse events (2 to 6). With rifampicin, the difference was 11 fewer incident tuberculosis cases (–3 to –17) and two additional severe adverse events (1 to 3). For adults, a treat-all approach led to fewer incident tuberculosis cases, and additional adverse events increased with age. Assuming 25% prevalence of TST positivity among adult contacts, a treat-all approach would lead to around two fewer tuberculosis cases per 1000 contacts for all regimens; the number of additional severe adverse events ranged from seven (IQR 5 to 8) for 18 to 34-year-olds treated with rifampicin to 63 (50 to 74) for people older than 64 years treated with isoniazid and rifapentine. A rifampicin-only regimen was associated with the fewest additional severe adverse events (seven [IQR 5 to 8] per 1000 adults aged 18–34 years and 35–64 years, and 17 [9 to 23] per 1000 adults older than 64 years). Interpretation Based on the available data, giving preventive therapy to all household contacts would probably reduce the incidence of tuberculosis cases in high-burden settings. Adverse events could be minimised by using nonisoniazid regimens and, in adults older than 18 years, focusing treatment on individuals with a positive infection test.