There are few rapid point-of-care tests (POCT) for tuberculosis (TB) for use in resource-constrained settings with high levels of human immunodeficiency virus (HIV). This hinders early tuberculosis (TB) treatment. This cross-sectional study evaluates the recently developed urine Determine tuberculosis lipoarabinomannan (TB LAM) antigen test. A total of 122 participants with signs and symptoms of TB, including 21 (17.1%) participants positive for HIV, were enrolled from September 2011 to March 2012 at three selected health centers in Addis Ababa, Ethiopia. Blood, sputum and urine samples were collected. Löwenstein-Jensen (LJ) solid culture was used as a gold standard to evaluate the performance of the Determine TB LAM antigen test. Data were analyzed using STATA (Statacorp LP, USA).

Of the 122 participants with suspected TB, 35 (28.7%) had TB confirmed bacteriologically by LJ culture. The overall sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of Determine TB LAM (for both HIV-positive and HIV-negative participants) was 37.1% (95% CI 21.5–55.1), 97.7% (95% CI 91.9–99.7), 86.7% (95% CI 59.5–98.3) and 79.4% (95% CI 70.5–86.6), respectively. However, in participants who were co-infected with TB and HIV, sensitivity, specificity, PPV and NPV were 55.6% (95% CI 21.2–86.3), 100% (95% CI 73.5–100), 100% (95% CI 47.8–100) and 75.0% (95% CI 47.6–92.7). Moreover, the level of immunosuppression of the HIV-infected TB patients was found to have a significant association with the performance of Determine TB LAM (χ² = 7.89, p = 0.002).

The Determine TB LAM test is a potential alternative in peripheral health settings for TB diagnosis in patients who are co-infected with HIV, with advanced immunosuppression.