Time of stable liquid culture conversion was the visit corresponding to the 1st of 2 consecutive MTB-neg cultures obtained at least 7 days apart without an intervening MTB-pos culture. Inability to produce sputum with or without induction was considered an MTB-neg culture. A participant was MTB-pos at a visit if at least 1 of the liquid cultures was MTB-pos. A participant was MTB-neg at a visit if both liquid cultures were MTB-neg or if 1 liquid culture was MTB-neg and the other was missing or indeterminate. If the 1st MTB-neg liquid culture was at week 26, then conversion was met. If a participant did not convert by week 26, they were censored at their last culture result. If a participant died (any cause except trauma), they were censored at week 26 (worst possible outcome). Within-arm Kaplan-Meier estimates of proportions at week 26 and the Cox proportional hazards regression model hazard ratio were calculated; between-arm differences were tested with the log rank test.

Time of permanent discontinuation of at least one anti-TB drug due to side effects that did not lead to a protocol-required discontinuation, due to participant non-compliance with at least one anti-TB drug or study visits, or due to participant request was the corresponding date of discontinuation. If a participant did not permanently discontinue at least one anti-TB drug due to side effects that did not lead to a protocol-required discontinuation, due to participant non-compliance with at least one anti-TB drug or study visits, or due to participant request by week 26, they were censored at the date of permanent treatment discontinuation of all study drugs or, if still on study treatment, at week 26. Within-arm Kaplan-Meier estimates of proportions of participants with permanent discontinuation were calculated with 2-sided 95% confidence intervals using standard errors based on Greenwood's formula.