Effect of smoking on drug-resistant tuberculosis treatment outcomes and exploring potential pathways: A multicountry cohort study.

People who smoke are at increased risk of unfavorable tuberculosis (TB) treatment outcomes compared with those who do not, but the pathways explaining this effect are unclear. We estimated the effect of smoking on a successful end-of-treatment outcome for multidrug-resistant and rifampicin-resistant (MDR/RR) TB and examined if intervening on loss to follow-up mitigates this effect. The endTB Observational Study was a prospective cohort of people with MDR/RR-TB who were treated with longer regimens containing bedaquiline and/or delamanid. We used marginal standardization to examine the effect of smoking (≥1 cigarette daily at enrollment) on treatment success (cured/completed). To simulate intervening on lost to follow-up, we censored participants and applied inverse probability of censoring weights. Among 1786 participants in 12 countries, 539 (30.2%) reported smoking. At the end of treatment, 73.5% of people who smoked and 80.3% of people who did not smoke had treatment success (risk difference in percentage points: -6.8, 95% CI: -11.1, -2.6). After adjusting for baseline confounders including demographics, social history, and comorbidities, the risk difference was similar (-5.2 percentage points) but 95% CIs were less precise (-14.1, 3.2). In a pseudopopulation without loss to follow-up, the risk difference was reduced (-1.9 percentage points; 95% CI: -10.2, 5.1). People who smoked had less frequent MDR/RR-TB treatment success compared with those who did not smoke. A simulated intervention on loss to follow-up reduced this difference, suggesting that pathways related to retention in care were a driver of this effect.