Molecular Urine and Stool Tests Do Not Improve Early TB Treatment in Hospitalized HIV Patients
Posted on 13 May 2026
Tuberculosis is the leading cause of death among people living with HIV, and diagnosis in hospital settings remains difficult. Symptoms are often non-specific, disease can be extrapulmonary, and many patients cannot produce sputum for microbiological confirmation. Non-invasive testing of urine and stool has been explored to close these gaps. A new randomized study shows that expanded molecular testing of these samples in HIV wards did not speed treatment initiation or lower short-term mortality.
Barcelona Institute for Global Health (ISGlobal), together with FIND (Foundation of Innovative New Diagnostics; Geneva, Switzerland), evaluated an expanded tuberculosis (TB) screening strategy for hospitalized adults with HIV. The approach added molecular testing of urine and stool to standard sputum-based workflows and incorporated a urine lateral flow lipoarabinomannan assay. Investigators sought to determine whether integrating these assays into World Health Organization (WHO)-recommended protocols would accelerate treatment start in routine inpatient care.
The strategy combined Xpert MTB/RIF Ultra molecular assays on sputum, urine, and stool with a urine lateral flow lipoarabinomannan (LF-LAM) test. The comparator was the WHO-recommended standard pathway based on symptom assessment with selective use of diagnostic tests. Outcomes included the proportion of microbiologically confirmed TB cases initiating therapy within 72 hours, as well as time to diagnosis and treatment, and eight-week mortality.
In the multicenter EXULTANT randomized study across 11 hospitals in Tanzania and Mozambique, 1,172 adults living with HIV were enrolled irrespective of TB symptoms. The primary endpoint was treatment initiation within the first 72 hours after admission. Secondary measures captured diagnostic timelines and short-term survival.
Expanded screening did not improve early treatment starts: 16.0% of patients in the intervention group initiated therapy within 72 hours compared with 15.3% under standard care, with no statistically significant difference. Eight-week mortality was also similar (25.8% versus 28.8%). Although additional TB cases were identified when urine and stool were molecularly tested—and the urine LF-LAM yielded the highest number of detections—the overall approach neither increased confirmed diagnoses nor shortened time to treatment.
About half of participants could provide sputum, underscoring urine and stool as useful alternative specimens. The authors note that these findings raise questions about clinical impact and point to a need for further evaluation of test accuracy and optimization of diagnostic algorithms. Results are published in The Lancet Infectious Diseases.