A cluster-randomized trial of household vs incentive-based tuberculosis contact investigation in rural South Africa: implementation reach.

The World Health Organization recommends TB contact investigation in high-burden countries. We examined the implementation reach of two contact investigation strategies in South Africa. Kharitode TB, a cluster-randomized crossover trial, compared household- and incentive-based contact investigation in 28 clinics (July 2016-January 2020). Clinics used each strategy for 18 months (separated by a six-month "washout"). Adults recently diagnosed with TB (index participants) were enrolled. In the household-based arm, contact persons were screened and provided sputum samples at home. In the incentive-based arm, index participants distributed referral coupons to their contacts, who received a $3.50 incentive upon presenting for screening at clinics. We used mixed-effects logistic regression with random intercepts for clinics to examine factors associated with index participant enrollment and sputum collection from contact persons. In the household-based arm, 782/1,269 (61.6%) index participants consented, 1,882 contact persons were enrolled and sputum samples were collected from 988/1,882 (52.5%). In the incentive-based arm, 780/1,295 (60.2%) index participants consented, 1,940 contact persons were enrolled and sputum was collected from 1,431/1,940 (73.8%). Index participants living with HIV (adjusted odds ratio, aOR=0.56, 95% CI 0.38-0.83) or unknown HIV status (aOR=0.12, 95% CI 0.07-0.20) were less likely to participate in the study. Contact persons in the incentive-based arm were more likely to provide a sputum sample compared to those in the household-based arm (aOR=2.12, 95% CI 1.80-2.50). Regardless of the study arm, cough (aOR=2.27, 95% CI 1.87-2.77), current smoking (aOR=2.22, 95% CI 1.63-3.02), and living with HIV (aOR=1.89, 95% CI 1.36-3.62) were associated with higher likelihood of sputum collection. There were gaps in implementation reach at the stages of contacting and enrolling index participants, enrolling individuals with HIV, and obtaining sputum, especially among those under 18 years and household contact persons.