To evaluate the applicability and accuracy of the urine lipoarabinomannan (LAM) test in tuberculosis (TB)/HIV co-infected patients and HIV-negative patients with disseminated TB.

Frozen urine samples obtained at baseline from patients in the TB research cohort with proven culture-positive TB were selected for blinded urine LAM testing. One hundred and nine patients were categorized into four groups: (1) HIV-positive patients with TB; (2) HIV-negative patients with disseminated TB; (3) HIV-negative immunocompromised patients with TB; and (4) patients with diseases other than TB. The sensitivity of urine LAM testing for culture-positive TB, specificity of urine LAM testing for patients without TB, positive predictive value (PPV), and negative predictive value (NPV) were assessed.

The sensitivity of the urine LAM test in group 1 patients with a CD4 T-cell count of >100, ≤100, and ≤50 cells/mm³ was 38.5%, 40.6%, and 45%, respectively. The specificity and PPV of the urine LAM test were >80%. The sensitivity of the test was 20% in group 2 and 12.5% in group 3, and the specificity and PPV were 100% for both groups. A positive urine LAM test result was significantly associated with death.

This promising diagnostic tool could increase the yield of TB diagnosis and may predict the mortality rate of TB infection, particularly in TB/HIV co-infected patients.