Mortality: results of fitting Cox proportional hazards regression models by HIV status.

aFor HIV-uninfected participants, the potential risk factors considered were resistance status, adherence to TB treatment, duration of TB symptoms, reason for being retreated, and BACTEC days to positivity at baseline, with age and sex as potential confounders.bFor HIV-infected participants, the potential risk factors and confounders considered were as for HIV-uninfected with in addition a composite factor measuring baseline CD4 count and ART status; this factor has five levels namely “CD4≥200,” “CD4 50–199+no ART,” “CD4 50–199+ART,” “CD4cCategory includes participants who were fully adherent with treatment and those that missed a few doses.CI, confidence interval; H, isoniazid; n/a, not available; R, rifampicin.