Abstract
The sensitivity of microscopy for the diagnosis of tuberculosis (TB) is around 50% but decreases by about 15% in patients with suspected TB who are coinfected with HIV. Here, we compared the accuracies of three microscopy methods for processing sputum smears (concentration by centrifugation with or without N-acetyl-L-cysteine [NALC] and concentration by filtration on a polycarbonate membrane) to that of culture on Ogawa-Kudoh medium as the gold standard method. Sputum samples were obtained from 432 patients with suspected pulmonary TB, of whom 60% were infected with HIV. Analysis was performed using the first specimen. Compared to the gold standard culture, the small-membrane-filter (SMF) method was the most sensitive microscopic method. In HIV-infected TB patients, the sensitivity of the SMF method was significantly higher than those for centrifugation of sputum samples with or without NALC treatment (61.9%, 47.6%, and 45.2%, respectively; P = 0.001). Similarly, in TB patients without HIV infection, the sensitivity of the SMF method was significantly higher than those for centrifugation of sputum samples with or without NALC treatment (81.8%, 63.6%, and 57.5%, respectively; P = 0.001). In the two study groups, TB patients with or without HIV, no significant differences between the specificities of the three methods were observed. Handling of the second sputum sample similarly by centrifugation with or without NALC and by the SMF method increased positivities by 13%, 11%, and 4%, respectively. The overall agreement between microscopy and culture was above 90% for all groups. Microscopic evaluation of the sputum samples treated with NALC compared to those not treated with NALC did not show any increase in sensitivity. Altogether, the sensitivity of the SMF method is higher than those of the other two microscopic methods studied without a loss of specificity.