Design: pragmatic, individually randomised controlled superiority trial across 11 hospitals
Population: 1,172 adults living with HIV admitted to hospital in Tanzania and Mozambique, 61.0% female, median CD4 232 cells per µL
Primary outcome: proportion with microbiologically confirmed tuberculosis starting treatment within 72 hours
Effect: 16.0% vs 15.3% (difference 0.7%, 95% CI −3.4 to 4.8, p=0.73); 8-week mortality 25.8% vs 28.8% (HR 0.86, 0.69-1.07)

EXULTANT was a pragmatic individually randomised trial across 11 hospitals in Tanzania and Mozambique. It enrolled 1,172 adults living with HIV within 24 hours of admission, with no existing tuberculosis diagnosis. The intervention arm had Xpert MTB/RIF Ultra on sputum, stool and urine plus lateral flow urine lipoarabinomannan, regardless of symptoms. The control arm followed the WHO symptom-based strategy. Median CD4 count was 232 cells per µL.

Microbiologically confirmed tuberculosis with treatment started within 72 hours occurred in 16.0% of the intervention group against 15.3% of controls (difference 0.7%, 95% CI −3.4 to 4.8). Eight-week all-cause mortality was 25.8% against 28.8% (HR 0.86, 0.69-1.07). Median time to treatment was about a day in both arms.

The control arm explains the result. Symptom screening caught 85.6% of patients as eligible for sputum Xpert and 91.2% as eligible for urine LAM — so the WHO strategy was already testing almost everyone, and the expansion had little left to find. The reading for practice is not that extra specimens are useless but that in a severely immunosuppressed inpatient population, symptom-based criteria are not the bottleneck. Only 60.6% of eligible patients actually produced a sputum sample, and that is where the yield is being lost.

Apply the WHO symptom criteria properly before adding specimen types — in this population they include almost everyone
Chase the sputum sample: two in five eligible patients never produced one
Use urine LAM in any admitted patient with advanced HIV disease; it needs no sputum and no symptoms
An eight-week mortality of over a quarter means the diagnostic question is not the only one worth asking on admission
In Indian practice the CD4 distribution and background TB prevalence differ; the finding is about screening strategy, not about test performance

Why it matters

It moves the problem from which specimens to test to whether the specimens are being collected at all.

The statistics, in plain English

A difference of 0.7 percentage points with an interval from −3.4 to +4.8 rules out any large gain in either direction — this is an informative null, not an underpowered one. The mortality hazard ratio of 0.86 (0.69-1.07) crosses 1.0, so a real mortality benefit of up to about 30% remains possible and would need a larger trial to detect.