ARREST
ARrest RESpiraTory Failure From PNEUMONIA (ARREST Pneumonia)
Patient / Population Intervention / Exposure Comparison Outcome
In 465 patients hospitalised with pneumonia and hypoxaemia, aerosolised budesonide with formoterol did not reduce acute respiratory failure compared with aerosolised placebo.
Design
Phase 3, multicentre, double-blind, placebo-controlled trial (NCT04193878); patients hospitalised with pneumonia and hypoxaemia randomised to twice daily aerosolised budesonide with formoterol or to placebo for up to five days (ten doses). Patients with an indication for inhaled beta agonists or corticosteroids were excluded, and the protocol was amended to allow 6 mg dexamethasone in COVID-19 patients
Endpoint
Acute respiratory failure, defined as the need for high-flow nasal oxygen, non-invasive ventilation or invasive ventilation for more than 36 hours
ResultAcute respiratory failure in 47 of 235 patients (20.0%) on budesonide with formoterol vs 50 of 230 (21.7%) on placebo (adjusted odds ratio 0.91, 95% CI 0.56 to 1.46; P = 0.69). Intubation 6.0% vs 7.9%, ARDS 12.0% vs 12.8%, 60-day mortality 20.9% vs 23.5%, oxygen-failure-free days and length of stay all without a difference. Enrolment ran from May 2020 to July 2025 and the trial was stopped for futility at 75% of its target. Median time from emergency department arrival to the first dose was 20 hours, and a prespecified subgroup treated sooner trended better (odds ratio 0.6, 95% CI 0.3 to 1.2 under 20 hours vs 1.4, 95% CI 0.7 to 2.7 beyond; interaction P = 0.07).
Levitt J, Hedlin H, Rogers AJ, et al. Presented at ERS Congress 2026, ALERT 2, 8 September. Not yet published.
Discussion & critique
A pilot trial found better oxygenation with inhaled budesonide and formoterol in pneumonia, and the phase 3 trial built on it found nothing: one patient in five went on to respiratory failure either way, and enrolment stopped for futility at three quarters of target. The only signal is timing, a trend towards benefit when the first dose arrived within 20 hours, on an interaction P of 0.07 inside a negative trial. Five years of enrolment spanning the pandemic, with the protocol amended to allow dexamethasone in COVID-19 patients, leaves a population that changed underneath the question. Treat the timing finding as a hypothesis for a trial that starts in the emergency department, not as a subgroup that works.