Background:  The role of a direct amplification test for the diagnosis of HIV-associated pulmonary tuberculosis (TB) is currently not fully established. The use of this technique is recommended for smear-positive respiratory samples (to distinguish TB from infection with a non-tuberculous mycobacteria) but not for smear-negative sputums. We evaluated this technique in HIV⁺ patients with proven pulmonary TB who were smear-negative.

Methods:  We included 73 HIV-infected patients with a first proven episode of pulmonary TB (clinical and radiological suspicion of infection, positive culture for Mycobacterium tuberculosis) were evaluated between 1995 and 2004. The mean CD4 level was 92/mm³. Direct amplification techniques were performed in smear-negative patients (i.e. 3 negative samples).

Results:  Of the 73 cases, 30 were smear-negative TB (41%) (mean CD4 88/mm³). Direct amplification test could be realized in 20 of 30 specimens (mean CD4 95/mm³) and was positive in 15 of 20 (75%) (9 on sputum and 6 on bronchoalveolar lavage) (mean CD4 66/mm³).Treatment against M. tuberculosis was initiated in all patients as soon as amplification test results were made available.

Conclusions:  Among HIV⁺ patients with culture proven pulmonary TB who are smear-negative, direct amplification techniques contributed to establish early diagnosis in 75% of the cases. This is higher than published studies, where HIV and non-HIV patients where mixed. Amplification technique costs US$20, and can be performed in 1 day, allowing early initiation of appropriate therapy and control of risk of transmission of M. tuberculosis. The clinical usefulness and the cost benefit ratio of this approach should be further evaluated in larger studies.