OBJECTIVES

Tuberculosis (TB) is a highly contagious infectious disease that mainly affects low- or middle-income countries. Tests based on the detection of mycobacterial lipoarabinomannan antigen (TB-LAM) were developed as possible rapid tests for TB. This study aims to evaluate the accuracy of the lateral flow urine TB-LAM for the tuberculosis diagnostic in people living with HIV with tuberculosis symptoms.

METHODS

: We searched the PubMed, Embase and Cochrane CENTRAL for studies of the diagnostic performance where lateral flow urine TB-LAM was analyzed against cultures or composite reference standards for HIV-positive people with tuberculosis symptoms. Study quality was evaluated using QUADAS-2 criteria. The software R studio was used to perform the meta-analysis.

RESULTS

: From 266 studies founded, 16 were systematic selected. According to QUADAS-2 the overall evidence quality was moderated to low due to inconsistencies and inaccuracy in the effects of studies.

To evaluate the TB-LAM overall accuracy compared to reference standards for HIV-positive people, 11 studies were included with 5,644 sample resulting in a sensitivity of 43.6% (95% CI 0.33-0.55) and a specificity of 88.2% (95% CI 0.81-0.93). To assess the the TB-LAM overall accuracy compared to reference standards in HIV-positive patient that has CD4 count under 100 cels/µl, 6 studies were included with 1,099 samples resulting in a sensitivity of 55.4% (95% CI 0.33-0.76) and a specificity of 87% (95% CI 0.77-0.93). In this case, the correlation between the sensitivity and specificity measures was estimated in 58% (95% CI 0.44-0.95).

CONCLUSIONS

: TB-LAM displayed high specificity but modest sensitivity across various samples for diagnosing people living with HIV compared to the reference standard. Evidences shows increased sensitivity in patients with severe disease, especially with low CD4 counts. Positive results may be considered to indicate a presumptive case of TB, whereas negative test results indicate that the possibility of TB should not be excluded.