**2. Methods**

### *2.1. Design and Study Population*

This was a retrospective, cross-sectional study utilizing *M.tb* isolates from the Botswana National Tuberculosis Reference Laboratory (NTRL) bio-repository which were collected as part of routine clinical care between 2012 and 2013. The study was approved by the University of Botswana Ethics Institutional Review Board and Health Research and Development Committee (HRDC) at the Ministry of Health and Wellness (Reference No: HPDME: 13/18/1 Vol. XI (140)). Sixty-six *M.tb* isolates were selected. The selected isolates were isoniazid (H) or rifampicin (R) monoresistant or resistant to both (MDR) based on first-line culture-based drug susceptibility testing (DST). The isolates included in this study are part of a previously described larger study and culture-based drug susceptibility testing for first-line drugs was done as previously described [17]. The clinical treatment outcome data was obtained from the Botswana National Tuberculosis Program (BNTP) patient database. At the time

of the study, the standardized MDR-TB treatment regimen in Botswana consisted of pyrazinamide, amikacin, levofloxacin, ethionamide, cycloserine, P-aminosalicylic acid (PAS) [18].

### *2.2. Treatment Outcome Definitions*

Treatment outcomes were defined according to the Botswana national guidelines. Briefly, "cured" was defined as a patient whose smear or culture sample was positive at the start of treatment but either converted to smear negative or had two consecutive negative cultures, one during treatment and the other at the end of treatment; "failed" was defined as a patient whose smear or culture was positive five months or later during treatment; "loss-to-follow-up" was defined as a patient whose treatment was interrupted for more than 30 consecutive days; "not evaluated" referred to patients whose treatment outcome could not be assigned since treatment conclusion has not been reported to the national TB program; treatment "completed" referred to patients who completed treatment but did not have a negative smear or culture result in the last month of treatment [18]. In bivariate comparisons, treatment outcomes were combined: "failed treatment" (i.e. remaining smear or culture positive throughout treatment), "loss-to-follow-up", death, "not initiated on treatment" and "not evaluated" as "unsuccessful treatment outcome" and "completed treatment", "cured" as "successful treatment outcomes".
