Tuberculosis is the leading cause of death among hospitalised people living with HIV, and the reasoning behind non-sputum diagnostics is straightforward: the sickest patients cannot produce sputum, and symptom screening misses disease in the immunosuppressed. EXULTANT tested the maximal version of that argument. Across 11 hospitals in Tanzania and Mozambique, 1,172 adults with HIV and no existing tuberculosis diagnosis were randomised within 24 hours of admission. The intervention arm received Xpert MTB/RIF Ultra on sputum, stool and urine plus lateral flow urine lipoarabinomannan, regardless of symptoms. The control arm got WHO-recommended symptom-triggered testing. Median CD4 count was 232 cells per microlitre.
Nothing moved. Microbiologically confirmed tuberculosis with treatment started within 72 hours occurred in 93 of 582 (16.0%) in the intervention arm and 90 of 590 (15.3%) in the control arm — a difference of 0.7% (95% CI -3.4 to 4.8, p=0.73). Eight-week mortality was 25.8% against 28.8% (hazard ratio 0.86, 95% CI 0.69 to 1.07, p=0.18). Median time to starting treatment was about a day in both arms.
The control arm is where the explanation sits. Of controls, 85.6% had tuberculosis-compatible symptoms and were eligible for sputum testing anyway, and 91.2% met WHO criteria for urine LAM. In a population this sick, symptom-based screening is not selective — almost everyone qualifies. There was very little diagnostic ground for the expanded strategy to gain, because standard care was already testing nearly everybody.
That is a useful negative result rather than a discouraging one. It says the current WHO algorithm is performing close to its ceiling in hospitalised patients with advanced HIV, and that adding specimen types will not move mortality. The 26% eight-week mortality in both arms is the number that should stay with you: these patients are dying at a rate that better tuberculosis detection alone cannot fix. The problem is what happens after diagnosis, and how late in their illness they arrive.
- Keep applying the WHO symptom-triggered algorithm; expanded routine testing added nothing here.
- Recognise that in advanced HIV nearly every admitted patient meets criteria for both sputum Xpert and urine LAM — so test them.
- Only 60.6% of eligible controls actually produced a sputum sample; chasing that gap is worth more than adding specimen types.
- Eight-week mortality was over 25% in both arms — focus on earlier presentation and post-diagnosis care.
- This does not argue against urine LAM, which was part of standard care in both arms.
The statistics, in plain English
A difference of 0.7% with an interval of -3.4 to 4.8 is a clean null: the data are equally compatible with a small benefit or a small harm. The mortality result is more interesting — a hazard ratio of 0.86 with an interval of 0.69 to 1.07 does not reach significance, but it leaves open a reduction of up to 31%, so the trial has not excluded a worthwhile mortality effect, only failed to demonstrate one. The near-universal eligibility of the control arm for the same tests explains why: when the comparator is already doing most of what you are proposing, even a good intervention has little room to show a difference.
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