Abstract
Sputum induction, bronchoalveolar lavage, or gastric aspiration are often needed to produce adequate diagnostic respiratory samples from people with HIV in whom tuberculosis is suspected. Since these procedures are rarely appropriate in less-developed countries, we compared the performances of a simple string test and the gold-standard sputum induction. 160 HIV-positive adults under investigation for tuberculosis, and 52 asymptomatic HIV-positive control patients underwent the string test followed by sputum induction. The string test detected tuberculosis in 14 patients in whom this disease was suspected; sputum induction detected only eight of them (McNemar's test, p=0.03). These preliminary data suggest that the string test is safe and effective for retrieval of useful clinical specimens for diagnosis of pulmonary tuberculosis, and is at least as sensitive as sputum induction.