Baloxavir cut influenza transmission in a trial, but not symptomatic cases

A single oral dose of the influenza antiviral baloxavir marboxil lowered the chance that a treated patient passed the virus to someone in their household, according to a large multi-country trial published in The New England Journal of Medicine [s1]. It is the clearest randomised evidence yet that treating one person's flu can protect the people around them — but the result is narrower than the headline suggests, and it comes with two important caveats the trial itself makes plain.

Influenza is usually treated to help the patient recover faster. Whether treatment also blunts onward spread has been an open question: trials of the older neuraminidase inhibitors, the class that includes oseltamivir, have not shown convincing evidence that they prevent transmission to contacts [s1]. Baloxavir works differently — it blocks viral replication at an early step and rapidly reduces the amount of virus a patient sheds, which is the biological reason researchers suspected it might cut transmission [s1].

What the trial did

CENTERSTONE was a phase 3b, randomised, placebo-controlled trial run across the 2019–2024 influenza seasons [s1]. Influenza-positive index patients aged 5 to 64 years were assigned in a 1:1 ratio to receive baloxavir or placebo within 48 hours of symptom onset, and their household contacts were then followed [s1]. In all, 1457 index patients and 2681 household contacts were enrolled; 726 index patients were assigned to baloxavir and 731 to placebo [s1]. The primary endpoint was transmission of influenza virus from an index patient to a household contact by day 5 [s1].

The result, and its limits

By day 5, transmission of laboratory-confirmed influenza was significantly lower with baloxavir than with placebo: an adjusted incidence of 9.5% versus 13.4%, an adjusted odds ratio of 0.68 (95.38% confidence interval, 0.50 to 0.93; P=0.01), which the authors describe as an adjusted relative risk reduction of 29% (95.38% CI, 12 to 45) [s1].

That is a real effect, but read the endpoint carefully. It counts laboratory-confirmed infection in a contact, whether or not that contact felt unwell. The first secondary endpoint — transmission that actually resulted in symptoms — did not reach statistical significance: an adjusted incidence of 5.8% with baloxavir versus 7.6% with placebo, odds ratio 0.75 (95.38% CI, 0.50 to 1.12; P=0.16) [s1]. In other words, the trial showed fewer contacts testing positive, but did not demonstrate that fewer contacts became ill — a distinction that matters for anyone weighing what the drug delivers in practice.

The second caveat is resistance. Drug-resistant viruses emerged during follow-up in 7.2% (95% CI, 4.1 to 11.6) of the index patients treated with baloxavir, though no resistant viruses were detected in household contacts, and no new safety signals were identified [s1]. Baloxavir's tendency to select for resistance mutations after a single dose is a known liability of the drug, and a 7.2% emergence rate is not trivial for a medicine being considered as a tool to interrupt community spread [s1].

A separate analysis added confidence to the core finding. Because CENTERSTONE defined transmission from the timing of test positivity and subtype match, some counted events could in principle have been coincidental community infections rather than true within-household spread [s2]. Investigators sequenced the whole genomes of viruses from households with two or more infected members to distinguish the two, and reported that the sequencing confirmed baloxavir's effect in reducing transmission [s2].

Who funded it

The trial was funded by F. Hoffmann-La Roche, which markets baloxavir, together with the US Biomedical Advanced Research and Development Authority [s1]. That dual funding does not invalidate a randomised, placebo-controlled design, but it is the context in which an enthusiastic reading of the primary endpoint should be treated with caution — particularly when the endpoint that would matter most to a family, symptomatic illness in a contact, was not significantly reduced [s1].

What it means

CENTERSTONE moves the question of transmission-blocking antivirals from theory to evidence, and the direction of effect is consistent across the primary endpoint and the sequencing analysis [s1][s2]. But a drug that reduces laboratory-confirmed infection without a demonstrated cut in symptomatic illness, and that selects resistant virus in roughly one treated patient in fourteen, is not a settled public-health instrument [s1]. Whether health agencies recommend baloxavir specifically to curb household or community spread will depend on how they weigh a significant effect on infection against a non-significant one on symptoms, and on how they manage the resistance risk at scale. Readers should treat claims that antiviral treatment now "stops the spread" of flu as ahead of what this trial actually shows.

Sources

• [s1] Efficacy of Baloxavir Treatment in Preventing Transmission of Influenza. The New England Journal of Medicine, 2025. https://doi.org/10.1056/NEJMoa2413156
• [s2] Whole genome sequencing of influenza virus transmission pairs confirms the efficacy of baloxavir in reducing transmission in the CENTERSTONE study. Clinical Infectious Diseases, 2026. https://doi.org/10.1093/cid/ciag489