Better, Then Worse Again: COVID-19 Rebound After Nirmatrelvir/Ritonavir
A single patient, four days past finishing a full course of nirmatrelvir/ritonavir and feeling worse than she did on day two. The disagreement is whether rebound is a treatment failure asking for more drug, or a known phenomenon the drug never promised to prevent.
M.K., a 72-year-old retired court reporter with well-controlled hypertension and type 2 diabetes, tested positive for COVID-19 nine days ago and started nirmatrelvir/ritonavir on day one of symptoms, exactly the population EPIC-HR was built around. She improved steadily through the five-day course — fever gone by day three, energy mostly back by day five — and tested antigen-negative the day after finishing. Then, on what would have been day nine, the fatigue and low-grade fever returned, and a repeat antigen test she ran at home came back positive again, this time with a darker line than her original test ever showed. She called the clinic more frightened than she'd been the first time, certain something had gone wrong with the treatment itself — though what she describes is milder than her first week, low-grade rather than febrile, tired rather than laid flat, which is a different illness curve than the one she is afraid of.
Nothing about her second positive test means the drug failed in the way she fears. EPIC-HR measured hospitalization and death, not the absence of any detectable virus days after a course ends, and published case series — the pattern first drew attention through Charness and colleagues' 2022 report describing viral culture-positive rebound — describe symptomatic and virologic rebound occurring in a meaningful minority of treated patients, generally days after apparent resolution, generally mild, and generally self-limited without further antiviral intervention. The mechanism under discussion isn't drug failure in the pharmacokinetic sense; it's closer to a timing mismatch between how quickly nirmatrelvir suppresses replication and how long her own adaptive immune response needs to finish the job the drug started. Her second antigen line being darker than her first is the concrete detail the team keeps returning to: is that a meaningfully higher viral burden requiring more drug, or exactly the pattern rebound is already known to produce on its own? What nobody in the room can supply is the thing that would settle it: no trial has randomized rebound patients to retreatment or observation, so every position available today, the reassuring one included, is an inference from case series about a phenomenon that was only named after the drug was already in wide use.
Telehealth follow-up, day nine
I'd give her a second course. She's 72, diabetic, and hypertensive — every one of the features that made her high-risk the first time is still true, and now she's testing positive again with a darker line than her original result. I don't want to be the one who watched and waited on a patient with her risk profile if this turns into something worse.
If she were 40 and otherwise healthy, I'd be far more comfortable just watching — her age and comorbidities are doing most of the work in my own reasoning here, not the rebound phenomenon itself.
I wouldn't retreat. This is the pattern rebound has consistently shown since it was first described in Charness et al.'s 2022 report of viral culture-positive recurrence after a completed course — symptoms and antigen positivity return days after apparent resolution, generally mild, generally self-limited. Her trajectory the first time through was appropriately reassuring: fever gone by day three, near-full recovery by day five. That's not what a treatment failure looks like.
I understand the impulse to treat her risk factors as reason enough on their own, but retreatment hasn't been shown to shorten or prevent rebound at all — if the mechanism is immune-timing rather than incomplete suppression, more drug isn't actually addressing the thing that's happening.
I'd slow down on what the antigen-line darkness is actually telling either of you. A home antigen test is a qualitative readout, and "darker line" isn't a validated proxy for viral load the way a Ct value would be — it's suggestive, not diagnostic, and neither "she's sicker" nor "she's fine" should be concluded from it alone.
What should drive today's decision is her actual clinical trajectory: mild symptoms, no dyspnea, oxygen saturation normal, lungs clear. That picture supports observation with close follow-up over retreatment, but it's her exam and vitals doing that work, not the antigen test's visual appearance.
Agreed: observation with daily telehealth check-ins for three days, isolation precautions resumed until symptom resolution and a negative antigen test, and explicit return criteria — dyspnea, oxygen saturation below 94%, or fever persisting past five more days.
Not fully resolved: the primary care physician's underlying instinct — that her age and comorbidities alone justify a lower threshold for intervention — was acknowledged as reasonable in general, just not decisive against a specific, well-described phenomenon with no retreatment evidence behind it. The group agreed explicitly that a genuinely worsening trajectory at the 48-hour check-in, not the antigen line's appearance, would be what changes this plan.