Background: Tuberculosis (TB) is the most common opportunistic infection among HIV-infected persons in Africa and Asia and a major cause of mortality. The ideal duration of preventive therapy for TB for HIV-infected persons in TB-endemic countries is not known. The objective of this study was to compare the efficacy of a short (6-month) regimen with a long (36-month) regimen in preventing tuberculosis in HIV-infected persons.

Methods: HIV-infected persons attending the Tuberculosis Research Centre, Chennai and Madurai, India between 2000 and 2005 without active TB were randomized to receive either 6 EH (Ethambutol 800 mg and Isoniazid 300 mg daily for 6 months) or 36 H (Isoniazid 300 mg daily for 3 years) and followed for 3 years. Drugs were supplied every 15 days and adherence checked by surprise pill counts. Patients were evaluated clinically every 3 months and investigations for TB performed every 6 months or when clinically indicated. Primary outcome measures were incidence of TB and all-cause mortality. Intent to treat analysis was performed and proportion of TB and death as well as incidence rates compared. Approximately 25% of patients were initiated on anti-retroviral therapy during follow-up, equally distributed in both arms. 

Results: There were 712 patients with the following characteristics: 63% female, mean age 30 (SD 7) years, median CD4 327 cells/mm³ were enrolled. There were 683 patients included in the analysis (29 excluded for active TB diagnosed at baseline). Mean adherence was 95% for EH arm and 93% for INH arm respectively and severe AE necessitating termination of drugs was observed in 1 patient. Twenty-two of 344 patients in the EH arm (2.4 per 100 patient-years, 95%CI 1.4 to 3.5) and 13 of 339 patients in the INH arm (1.6 per 100 patient-years, 95% CI0.7 to 2.4) developed TB, p = ns. Death rates were similar in the 2 arms (2.8 and 2.2 per 100 patient-years, respectively). TB incidence was 4.4 times higher in patients with CD4<200 cells/mm³ and 1.4 times higher in those with a TST >5 mm at baseline. Among the patients who developed TB, 16 had positive cultures - drug susceptibility test (DST) was available for 14 of whom 8 had resistance to isoniazid and 2 had multidrug-resistant TB.

Conclusions: A 6-month regimen of Ethambutol and Isoniazid was as effective as 3 years of Isoniazid in prevention of TB in HIV-infected persons. Culture and DST should be available to rule out drug resistant incident TB.