Longitudinal analysis of HIV risk behaviour patterns and their predictors among public primary care patients with tuberculosis in South Africa

The goal of this study was to identify various HIV risk behaviours among tuberculosis (TB) patients in a longitudinal study design in South Africa. In 42 public primary healthcare facilities in three districts in three provinces, adult new TB and TB retreatment patients with hazardous or harmful alcohol use were interviewed within 1 month of initiation of anti-TB treatment and were followed up at 6 months. The total sample with a complete 6-month follow-up assessment was 853. At the follow-up assessment, several HIV risk behaviours significantly reduced from baseline to follow-up. In multivariate Generalized Estimating Equations logistic regression analyses, high poverty (odds ratio (OR): 2.68, 95% confidence interval (CI): 1.56-4.62), Posttraumatic Stress Disorder (PTSD) symptoms (OR=1.55, 95% CI=1.03-2.36), and sexual partner on antiretroviral therapy (ART) (OR=1.84, 95% CI=1.09-3.10) were associated with a higher odds, and excellent/very good perceived health status (OR: 0.61, 95% CI: 0.37-0.98), severe psychological stress (OR=0.51, 95% CI=0.34-0.77), and HIV non-disclosure to most recent sexual partner (OR=0.40, 95% CI=0.25-0.65) were associated with a lower odds of inconsistent condom use. Being HIV positive (OR=4.18, 95% CI=2.68-6.53) and excellent/very subjective health status (OR=2.98, 95% CI=1.73-5.13) were associated with a higher odds, and having PTSD symptoms (OR=0.60, 95% CI=0.36-0.99), being on ART (OR=0.48, 95% CI=0.25-0.95), having a sexual partner on ART (OR=0.41, 95% CI=0.18-0.96), and HIV status non-disclosure (OR=0.25, 95% CI=0.15-0.41) were associated with a lower odds of having sex with an HIV-positive or HIV status unknown person. High poverty index (OR=1.97, 95% CI=1.19-3.25) and having a sexual partner on ART (OR=4.37, 95% CI=1.82-10.48) were associated with a higher odds, and having a partner with HIV-negative status (OR=0.29, 95% CI=0.16-0.51) and inconsistent condom use (OR=0.39, 95% CI=0.24-0.64) were associated with a lower odds of HIV status non-disclosure at last sex. The study found that among TB patients with problem drinking over a 6-month TB treatment period, the frequency of some HIV risk behaviours (inconsistent condom use) declined (OR=0.64, 95% CI=0.41-0.98), but also persisted at a high-level calling for a strengthening and integration of HIV prevention into TB management.