Tuberculosis and human immunodeficiency virus co-infections and their predictors at a hospital-based HIV/AIDS clinic in Uganda

SETTING: Mulago Hospital, Uganda. OBJECTIVE: To evaluate the burden of TB-HIV (tuberculosis-human immunodeficiency virus) co-infections and their predictors in an urban hospital-based HIV programme. DESIGN: Prospective observational study. METHODS: Clinicians screened all patients with HIV/AIDS (acquired immune-deficiency syndrome) for previous and current TB treatment at enrolment and throughout follow-up. RESULTS: Of 10 924 patients enrolled between August 2005 and February 2009, co-prevalent TB was 157/10 924 (1.4%), which included 88/157 (56%) with TB confirmed at enrolment and 65/157 (41%) with TB diagnoses established during follow-up in whom symptoms were present at enrolment. Male sex (adjusted odds ratio [aOR] 2.3, 95%CI 1.6-3.2) and body mass index (BMI) </=20 kg/m(2) (aOR 3.8, 95%CI 2.5-5.4) were associated with co-prevalent TB. Overall, 749/10 767 (7%) were diagnosed with incident TB at a higher rate among antiretroviral treatment (ART) patients (8/100 patient years of observation [PYO]) than non-ART patients (5/100 PYO, log rank P < 0.001). Female sex (adjusted hazard ratio [aHR] 1.4, 95%CI 1.2-1.7) and baseline BMI </= 20 (aHR 1.9, 95%CI 1.6-2.2) predicted incident TB. CONCLUSION: Routine TB screening in the HIV/AIDS care programme identified a significant number of TB-HIV co-infections among patients with and without ART, and is therefore a potential strategy to improve HIV treatment outcomes in resource-limited settings.