Six-month regimen matches the nine-month standard for rifampicin-resistant tuberculosis in South African trial

A six-month all-oral drug regimen was non-inferior to South Africa's nine-month standard of care for rifampicin-resistant tuberculosis, the BEAT-Tuberculosis trial reported in the New England Journal of Medicine on 25 June 2026. Treatment succeeded in 86.1 per cent of participants on the six-month strategy and 86.0 per cent of those on the nine-month regimen.

BEAT-Tuberculosis, registered on ClinicalTrials.gov as NCT04062201, was a phase 3, open-label, pragmatic, randomised, controlled non-inferiority trial conducted entirely in South Africa. Of 432 people screened, 403 were randomised: 203 to the six-month trial strategy and 200 to the nine-month regimen that was then the country's standard of care.

The trial regimen was bedaquiline, linezolid, delamanid and levofloxacin or clofazimine or both, given for six months. Treatment in both arms was adjusted on the basis of second-line drug-susceptibility testing.

Enrolment was deliberately inclusive. Participants were aged 6 years or older, and people who were pregnant or breastfeeding and those with fluoroquinolone-resistant tuberculosis were eligible. All three groups are routinely excluded from tuberculosis drug trials, so the result speaks to patients who are usually treated without direct trial evidence.

The primary efficacy endpoint was cure or treatment completion at the end of treatment and at 76 weeks after randomisation. It was reached by 174 of 202 participants, or 86.1 per cent, in the trial-strategy group, and by 172 of 200, or 86.0 per cent, in the control group. The adjusted risk difference was minus 0.2 percentage points (95% CI minus 6.9 to 6.5), P equal to 0.001 for non-inferiority against a pre-specified margin of 10 percentage points.

On safety, grade 3 or higher adverse events occurred in 63 of 202 participants, or 31.2 per cent, in the six-month arm, against 74 of 200, or 37.0 per cent, in the nine-month arm. Ten participants died in each group.

Two features bound the finding. The trial was open-label, so neither participants nor clinicians were blinded to the regimen being given, and it was conducted entirely in South Africa.

The first author, Francesca Conradie, leads the Isango Lethemba TB Research Unit at the University of the Witwatersrand in Gqeberha, South Africa. Co-authors include Gary Maartens of the Division of Clinical Pharmacology, Department of Medicine, University of Cape Town; Graeme Meintjes of the University of Cape Town and the Queen Mary and Barts Health Tuberculosis Centre at the Blizard Institute, Queen Mary University of London; H. Simon Schaaf and Jennifer Hughes of the Desmond Tutu TB Centre at Stellenbosch University; Norbert Ndjeka of TB Control and Management at South Africa's National Department of Health; and Patrick P. J. Phillips of the UCSF Center for Tuberculosis at the University of California, San Francisco.

The trial was funded by the U.S. Agency for International Development under award 72067418CA00006, and others. It appears in the New England Journal of Medicine, volume 394, issue 24, pages 2429 to 2439, with the DOI 10.1056/NEJMoa2503687 and the PubMed identifier 42341301.

The burden the result addresses is large. The Global Fund's 2026 results report states that an estimated 10.7 million people fell ill with tuberculosis in 2024 and more than 1.2 million died, and that 115,000 people were treated for drug-resistant tuberculosis through its grants in 2025.

Shortening drug-resistant tuberculosis treatment from nine months to six cuts cost, pill burden and loss to follow-up in precisely the high-incidence settings that carry the disease. The finding adds to the evidence base the World Health Organization draws on when revising its drug-resistant tuberculosis treatment guidelines.