Tuberculosis remains the leading cause of death among hospitalised people with HIV, and the standard diagnostic pathway depends on the patient having symptoms and producing sputum. EXULTANT tested whether removing both conditions would help. This pragmatic randomised trial at 11 hospitals in Tanzania and Mozambique enrolled 1,172 adults with HIV within 24 hours of admission. The intervention group received Xpert MTB/RIF Ultra on sputum, stool and urine, plus lateral flow urine lipoarabinomannan, regardless of symptoms. Controls received WHO-recommended symptom-triggered sputum Xpert Ultra and LF-LAM. Median CD4 count was 232 cells per microlitre and 75.4% were on antiretroviral therapy.
Nothing moved. Microbiologically confirmed tuberculosis started on treatment within 72 hours occurred in 93 of 582 (16.0%) with expanded testing and 90 of 590 (15.3%) with standard care, a difference of 0.7% (95% CI -3.4 to 4.8, p=0.73). Eight-week all-cause mortality was 25.8% versus 28.8% (hazard ratio 0.86, 95% CI 0.69 to 1.07, p=0.18). Median time to treatment was about a day in both arms.
One number in the control arm explains a great deal. Of the 505 control patients eligible for sputum Xpert, only 306 — 60.6% — actually provided a sample. So the standard-of-care arm was already missing four in ten eligible patients at the point of sample collection, and expanding the test menu still did not close the gap.
The mortality figure is the one to sit with: more than a quarter of these patients were dead at eight weeks. That is not a diagnostic problem that better assays solve. It reflects late presentation with advanced immunosuppression, and the intervention that changes it happens months earlier, in the community, not at the hospital door.
- Do not expect broader specimen testing on admission to improve tuberculosis detection or survival in advanced HIV disease
- Check whether sputum is actually being collected — 4 in 10 eligible patients did not produce a sample even in a trial setting
- Keep using LF-LAM in the WHO-recommended group; the trial questions universal expansion, not the test itself
- Treat an admission with advanced HIV disease as a failure of earlier care and address retention and viral suppression at discharge
- Empirical tuberculosis treatment decisions in a sick patient still cannot wait for a confirmatory result
The statistics, in plain English
The primary difference of 0.7 percentage points with a confidence interval from -3.4 to 4.8 straddles zero and is narrow enough to make this a genuine null: the trial can exclude anything better than about a 5 percentage point gain. The mortality hazard ratio of 0.86, interval 0.69 to 1.07, is more interesting — it leans towards benefit but crosses 1.0, so it cannot be claimed. With 320 deaths across both arms this was reasonably powered, and a real mortality benefit of the size clinicians would hope for would probably have shown. Note that both arms received LF-LAM in the WHO-eligible group, so this compares expansion against a floor that already included urine testing.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for infectious diseases, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free