Independent predictors of tuberculosis mortality in a high HIV prevalence setting: A retrospective cohort study

Abstract

Background: Identifying those at increased risk of death during TB treatment is a priority in resource-constrained settings. We performed this study to determine predictors of mortality during TB treatment. Methods: We performed a retrospective analysis of a TB surveillance population in a high HIV prevalence area that was recorded in ETR.net (Electronic Tuberculosis Register). Adult TB cases initiated TB treatment from 2007 through 2009 in Khayelitsha, South Africa. Cox proportional hazards models were used to identify risk factors for death (after multiple imputations for missing data). Model selection was performed using Akaike's Information Criterion to obtain the most relevant predictors of death. Results: Of 16,209 adult TB cases, 851 (5.3 %) died during TB treatment. In all TB cases, advancing age, co-infection with HIV, a prior history of TB and the presence of both pulmonary and extra-pulmonary TB were independently associated with an increasing hazard of death. In HIV-infected TB cases, advancing age and female gender were independently associated with an increasing hazard of death. Increasing CD4 counts and antiretroviral treatment during TB treatment were protective against death. In HIV-uninfected TB cases, advancing age was independently associated with death, whereas smear-positive disease was protective. Conclusion: We identified several independent predictors of death during TB treatment in resource-constrained settings. Our findings inform resource-constrained settings about certain subgroups of TB patients that should be targeted to improve mortality during TB treatment.