Influenza at Hour Sixty: Treating Outside the 48-Hour Window
A single patient, confirmed influenza A two and a half days into her illness. The disagreement isn't about whether an antiviral works in general — it's about whether either drug still means anything once the trials that proved it stop covering the patient in front of you.
Renata S., a 34-year-old software quality-assurance analyst who has worked from her apartment for six years, spent the first two days of this illness doing what she always does with a cold: pushing through client calls with a mug of tea at her elbow, telling herself it would pass. It didn't. By the evening of day two, the ordinary head-cold ache had become a measured fever of 39.4°C, a headache she describes as sitting "behind my eyes," and myalgias severe enough that she skipped a scheduled call for the first time in years. She has mild, well-controlled asthma — an albuterol inhaler she reaches for maybe twice a month, no history of an exacerbation requiring steroids or an emergency visit — and is otherwise healthy, vaccinated against influenza six weeks ago. She arrived at urgent care sixty hours after her first symptom, not sixty hours after diagnosis: today's rapid antigen test, confirmed by RT-PCR as influenza A, is simply catching up to an illness that has already been running two and a half days on its own.
Both of the trials that established the two drugs now under discussion enrolled patients within 48 hours of symptom onset — CAPSTONE-1's single-dose baloxavir arm against its oseltamivir comparator in otherwise-healthy adults, and the decades of oseltamivir evidence behind it. Renata no longer belongs to either population, not by a little. That gap is mechanical, not just administrative: baloxavir's cap-dependent endonuclease inhibition blocks the step that initiates new viral mRNA synthesis, a target with no obvious biological reason to stop mattering at hour 48 if virus is still replicating — but "no obvious reason" is a mechanistic argument, not trial evidence, and CAPSTONE-1 itself reported treatment-emergent PA/I38 substitutions in 9.7% of baloxavir recipients — one in ten, in a population dosed inside the window she has already left. What that trial cannot tell anyone is whether the figure changes when the single dose lands at hour sixty instead of hour thirty, because nobody was dosed at hour sixty. The nearest handle on who selects resistance comes from Hirotsu and colleagues, who found variant emergence associated with low baseline hemagglutination-inhibition antibody titers — and that reading points the other way for her. She was vaccinated six weeks ago, which is precisely the interval at which a titer should be at or near its peak, placing her in the better-defended half of the population the resistance signal concentrates outside of. Her fever curve is what the team keeps returning to: still climbing at hour sixty, not past a peak the way an illness already resolving on its own would be showing by now.
Urgent care, hour sixty
Give her baloxavir marboxil today, as a single oral dose. The polymerase acidic endonuclease she'd be blocking initiates new viral mRNA synthesis — if her fever is still climbing at hour sixty, that machinery is still running, and there's no pharmacologic reason the drug stops mattering the moment we cross the calendar line CAPSTONE-1 happened to draw at 48 hours. One dose, done, no adherence problem across a workweek she's already fighting to salvage.
If she'd presented at hour twelve I wouldn't be making an argument at all — this is entirely about whether the mechanism still applies once the trial's own enrollment window has closed, not about whether baloxavir works in general.
I'd rather give oseltamivir, five days, twice daily. The mechanism argument for baloxavir isn't wrong, but it cuts both ways. CAPSTONE-1 found PA/I38 substitutions in 9.7% of its own baloxavir recipients — and miniSTONE-2, which is a separate pediatric trial rather than an extension of CAPSTONE-1, found roughly double that in children. I want to be careful here: I can't tell you that rate rises when the dose lands late, because no trial dosed anyone late. What I can tell you is that one dose is one chance, and the consequence of that chance failing is a selected variant rather than a missed day of therapy. Five days of oseltamivir spends the same uncertainty differently.
You're right that the endonuclease target itself doesn't expire at 48 hours — but "the target is still there" isn't the same claim as "hitting it late still works the way it did in the trial," and the resistance data are a real, documented cost specifically for late or incomplete suppression, not a hypothetical one.
Both of you are arguing about which drug to give outside the population either drug was shown to help — I'd rather name that plainly before picking one. She's not high-risk, she's not hospitalized, and neither CAPSTONE-1 nor the oseltamivir literature behind it describes a healthy adult treated this late. That's a real gap, not a technicality.
I take the resistance point seriously, and it's part of why I'd lean away from baloxavir specifically if we treat at all — but the more honest fork isn't "which drug," it's "treat or don't," and her fever curve is the one fact in the room that argues for treating rather than watching: a still-climbing temperature at hour sixty, in someone otherwise not past the point where oseltamivir's broader safety margin makes it the lower-cost bet if we're extrapolating either way.
Agreed: oseltamivir 75mg twice daily for five days, started today despite the sixty-hour delay, with explicit return precautions — worsening dyspnea, fever persisting past day five, or any change in mental status.
Not agreed: whether treating at all was the right call outside the trial evidence base, versus supportive care and closer outpatient follow-up. The primary care physician's framing — that both drug choices were arguing past the more basic "treat or don't" question — was heard but not adopted as the final answer; the fever curve carried the room toward treating. Nobody claimed the oseltamivir choice was proven to work at this hour, only that it carried the lower downside if it didn't.