HIV Remains a Risk Factor for Unfavorable Tuberculosis Treatment Outcomes in the Era of Universal Access to Antiretroviral Therapy in Botswana

Abstract

Botswana implemented its universal "Treat All" antiretroviral therapy (ART) policy in 2016, expanding treatment eligibility to all people living with HIV (PLHIV). HIV has been known to be a leading risk factor for tuberculosis (TB) and poor TB treatment outcomes. The primary goal of this study is to assess whether HIV infection and HIV-associated immunosuppression remain risk factors for unfavorable TB treatment outcomes in the Post-Treat All era. We analyzed 636 TB patients treated in Gaborone (2017-2023), of whom 54.4% were HIV-positive. Unfavorable outcomes (death, failure, or loss to follow-up) occurred in 19.7% of HIV-positive and 8.5% of HIV-negative patients. We used logistic regression to estimate unadjusted and covariate-adjusted associations between TB treatment outcome and HIV status and between TB treatment outcome and CD4+ T-cell count. HIV-positive patients had 2.5-fold higher odds of unfavorable outcomes compared with HIV-negative patients [adjusted OR: 2.51, 95% CI: (1.48, 4.38)], controlling for age, sex, TB history, distance to clinic, substance use, and occupational status. PLHIV with CD4+ T-cell < $200\text{cells}/\mu \text{L}$ was associated with approximately three-fold higher odds of unfavorable outcomes compared with HIV-negative participants [OR: 3.12, 95% CI: (1.65, 5.97)]. The secondary goal was to test whether the HIV effect changed following Treat All implementation. We combined the data from 2017-2023 with a Pre-Treat All cohort (2012-2016, n= 233, HIV prevalence 60.8%) and fit a frequentist logistic regression and Bayesian mixed-effects models with an interaction term that allows treatment era (Pre- vs. Post-Treat All) to modify the effect of HIV on TB treatment outcome. The estimated change in the HIV effect was uncertain [relative OR: 0.41; 95% CI: (0.11, 1.55)]. Combining the two Botswana data sets with 12 Pre- and Post-Treat All studies from neighboring Ethiopia showed that the pooled effect of HIV infection on unfavorable TB outcome has increased in the Post-Treat All period [relative OR: 2.39; 95% BCI: (1.36, 3.34)].

Figure 1:

Flow chart showing data cleaning steps and sample sizes for analysis datasets.