The World Health Organization recommends tuberculosis (TB) contact investigation in countries with a high TB burden. This study evaluated the implementation reach of two distinct contact investigation approaches in South Africa. The Kharitode TB trial, a cluster-randomized crossover study, compared household-based versus incentive-based contact investigation across 28 clinics from July 2016 to January 2020. Each clinic implemented both strategies sequentially for 18 months, with a six-month washout period in between. Adults with a recent TB diagnosis (index participants) were eligible for enrollment. In the household-based strategy, community health workers visited homes to screen contacts and collect sputum samples on-site. In the incentive-based strategy, index participants received referral coupons to distribute to their contacts, who were offered a $3.50 incentive upon attending the clinic for TB screening. Mixed-effects logistic regression models, incorporating random intercepts for clinics, were used to identify factors associated with index participant enrollment and successful sputum collection from contacts. In the household-based arm, 782 of 1,269 eligible index participants (61.6%) provided consent, leading to the enrollment of 1,882 contacts; sputum samples were obtained from 988 of these contacts (52.5%). In the incentive-based arm, 780 of 1,295 eligible index participants (60.2%) consented, resulting in 1,940 enrolled contacts and sputum collection from 1,431 (73.8%). Index participants 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 significantly less likely to enroll in the study. Contacts in the incentive-based arm had substantially higher odds of providing a sputum sample than those in the household-based arm (aOR 2.12, 95% CI 1.80–2.50). Across both arms, factors independently associated with greater likelihood of sputum submission included presence of cough (aOR 2.27, 95% CI 1.87–2.77), current smoking (aOR 2.22, 95% CI 1.63–3.02), and known HIV-positive status (aOR 1.89, 95% CI 1.36–3.62). Notable implementation gaps were observed at multiple stages, including outreach and enrollment of index participants, engagement of people living with HIV, and sputum collection—particularly among contacts under 18 years of age and those identified through household visits.
