CONTEXT: Tuberculosis preventive therapy reduces tuberculosis incidence among human immunodeficiency virus (HIV)-infected individuals in clinical trials, but implementation has been limited and there are no data on effectiveness under routine conditions. OBJECTIVE: To determine the effect on tuberculosis incidence of a clinic providing isoniazid preventive therapy to HIV-infected adults under routine conditions. DESIGN, SETTING, AND PARTICIPANTS: Randomized intervention study with a novel incremental recruitment design. Between 1999 and 2001 (before antiretroviral therapy was available), 1655 HIV-infected male employees of a South African gold-mining company (median age, 37 years) were enrolled in the study. Median follow-up was 22.1 months. INTERVENTION: Employees were invited in random sequence to attend a workplace HIV clinic. Isoniazid, 300 mg/d, was self-administered for 6 months among attendees with no evidence of active tuberculosis. MAIN OUTCOME MEASURE: Incidence of tuberculosis (including both first and recurrent episodes) during the periods before and after clinic enrollment. RESULTS: A total of 1016 of 1655 men included in the analysis attended the clinic at least once. Six hundred seventy-nine (97%) of 702 men eligible to start primary isoniazid preventive therapy did so. The tuberculosis incidence rate before vs after clinic enrollment was 11.9 vs 9.0 per 100 person-years, respectively (incidence rate ratio [IRR] after adjustment for calendar period, 0.68; 95% confidence interval [CI], 0.48-0.96). In a multivariable analysis adjusting for calendar period, age, and silicosis grade, the tuberculosis IRR for clinic enrollment was 0.62 (95% CI, 0.43-0.89). In a further analysis excluding individuals with a history of tuberculosis (and, hence, ineligible for isoniazid preventive therapy), the adjusted IRR for clinic enrollment was 0.54 (95% CI, 0.35-0.83). CONCLUSIONS: Enrollment in a clinic offering primary isoniazid preventive therapy to HIV-infected adults reduced tuberculosis incidence by 38% overall and by 46% among individuals with no history of tuberculosis prior to the study. Tuberculosis incidence remained high despite isoniazid preventive therapy, and further work is needed to determine how to use additional interventions most effectively to reduce morbidity and mortality due to tuberculosis in HIV-infected persons.
JAMA (2005) 293 (22) pp. 2719-2725.