The first data on Paxlovid, out last November, hinted that the COVID antiviral would cut the risk of hospitalization and death by 89 percent. Pundits called the drug “a monster breakthrough,” “miraculous,” and “the biggest advance in the pandemic since the vaccines.” “Today’s news is a real game-changer,” said Albert Bourla, the CEO of Pfizer, which makes the drug. The pills are “a game changer,” President Joe Biden repeated a few months later.
Now, finally, the game is being changed. The government has ordered 20 million courses of Paxlovid, committing half of the $10 billion in additional COVID funding that is being negotiated in the Senate; and Pfizer says that the number of patients taking the drug increased by a factor of 10 between mid-February and late April.
But as the treatment spreads, so too does confusion over its effectiveness and side effects. Patients have complained of a bitter, metallic taste, or one like grapefruit juice mixed with soap, the whole time they were on the drug. More concerning, some have reported experiencing a second round of symptoms, and going back to testing positive, when the pills were done, a phenomenon that’s become known as “Paxlovid rebound.” Meanwhile, Pfizer has never published any final data on the use of the drug by vaccinated patients, leaving medical professionals with little information about how the drug works for people who have received their shots—which is to say, most of the adult population in the U.S. “We’re all riding on hope at this point,” Reshma Ramachandran, a family-medicine doctor at Yale, told me.
An individual patient would never know if Paxlovid worked for them, because you could never say how sick you would have gotten if you hadn’t taken the pills. If the drug doesn’t really do that much for vaccinated people—if it fails to have meaningful effects on their risk of severe disease, and doesn’t help resolve their symptoms—then giving it out widely could be a waste of the dwindling resources the United States has committed to fight the pandemic, not to mention physicians’ time and patients’ sense of taste. And because people who have just recovered from COVID might reasonably believe they’re in the clear, and mingle with abandon, surprise cases of Paxlovid rebound could end up causing more transmission. “We continue to monitor data from our ongoing clinical studies and post-authorization safety surveillance,” a Pfizer spokesperson told me in an email, noting that cases of viral rebound “are being reported at a rate consistent with observations” from the company’s published clinical trial.
Taste disruption (a.k.a. dysgeusia) is the most straightforward of the Paxlovid mysteries, because any sudden onset of soapy-grapefruit-penny flavor can be attributed to the antiviral with a decent amount of confidence. In its only published trial of the drug, conducted in unvaccinated, high-risk patients, Pfizer found that 5.6 percent of Paxlovid-takers experienced dysgeusia, compared with 0.3 percent of those who got the placebo. If you apply that rate to the hundreds of thousands of people who have now received the drug, you might expect to see some tens of thousands of cases by now. Given how people like to kvetch on social media, that side effect could very well seem like it’s occurring in a lot more than one out of 18 patients. Perhaps all that’s going on here is that rare events seem common on a large scale. Perhaps! But I’ve heard from dozens of patients on the drug in the course of my reporting, and every single one told me that they’d suffered through at least mild dysgeusia. Paul Sax, the clinical director of the Division of Infectious Diseases at Brigham and Women’s Hospital, told me he suspects “way more than half” of the people who’ve taken Paxlovid have experienced the taste.
As for rebound, a Pfizer executive said during an earnings call this month that the company had taken a “preliminary look” at its trial data and concluded that viral loads bounced back up in about 2 percent of patients. He also said they saw “the same or close to the same percent in the placebo arm.” (These findings have not been published.) “Reports from the beginning of the pandemic suggested some participants exhibit fluctuations in nasal viral RNA, and these fluctuations could be a phenomena [sic] of the disease itself,” the Pfizer spokesperson told me. In any case, if you apply that measured rate of 2 percent to the population who have now taken the drug, you’d expect thousands of people to have experienced Paxlovid rebound by this point (and many, many more cases of rebound occurring among all the COVID patients who didn’t take it).
The real number is—well, we have pretty much no idea what the real number is. The federal government is not tracking Paxlovid rebound in any public-facing database, and the CDC released an advisory on Tuesday saying the agency doesn’t know whether a recurrence of symptoms can be connected to the drug. The agency also clarified that “Paxlovid continues to be recommended for early stage treatment of mild to moderate COVID-19 among persons at high risk for progression to severe disease” and that anyone who rebounds should isolate for another five days. No researchers have yet published studies measuring the prevalence of rebound, but a good number of clinicians and Paxlovid patients are convinced that it’s higher than 2 percent. “To trust that number would’ve been to not believe my eyes,” Bob Wachter, the chair of medicine at UC San Francisco, told me.
In an attempt to find some clarity, Wachter decided to poll his Twitter connections on whether they’d taken Paxlovid and rebounded. (I’m legally obligated to tell you that Twitter polls are neither scientific nor particularly reliable—which Wachter knows perfectly well.) Of the respondents who said they had taken Paxlovid, 45 percent rebounded; Wachter said he guesses the real proportion is closer to 10 or 20 percent. A few hours after we spoke, Wachter tweeted that his wife, who had recently finished a course of the antiviral and recovered from COVID, just tested positive again.
In short, Pfizer’s clinical-trial results may not be matching up with physicians’ and patients’ real-world experience. When I asked the company why, the spokesperson replied, “We cannot speculate on why some people may or may not experience dysgeusia, but we can reiterate that 5.6% of participants in a well-controlled clinical trial experienced that event compared to 0.3% in the placebo arm.” As for rebound, he said, the company continues to monitor the data but hasn’t yet seen any unexpected numbers. “We are actively reviewing but, thus far, have not seen an association with subsequent severe disease (i.e., hospitalization or death),” he added.
Discrepancies between the trial data and real-world experience might arise from the timing of the original research. Pfizer announced its results in early November, which means that participants received Paxlovid to help fight off infections caused by the Delta variant, which is naught but an unpleasant memory today. Three Omicron subvariants are currently floating through Americans’ airways. Perhaps one of them simply causes more rebound cases than Delta did, by keeping viral levels high enough that five days of antiviral therapy are not enough to wipe it out. Anthony Fauci announced last week that the National Institutes of Health is in talks with Pfizer to test out a longer course of Paxlovid to see if it reduces rates of rebound. (“We will share updates when we have them,” the Pfizer spokesperson said.) In the meantime, Bourla, Pfizer’s CEO, has suggested that those who experience a rebound should simply take another round of Paxlovid. But the FDA was less than enthusiastic about the idea.
Patients’ immunization status remains the most obvious difference between Pfizer’s published clinical trial and present-day reality. That study was conducted exclusively in unvaccinated participants who were at high risk of complications from COVID. The drug is now authorized for use in vaxxed and unvaxxed patients alike. Could this explain the apparent gulf in the prevalence of bad tastes and rebound? Ali Ellebedy, an immunologist at Washington University in St. Louis, told me he couldn’t imagine any direct link between vaccination and Paxlovid rebound or dysgeusia. And if anything, he said, immunological principles suggest that, compared with unvaccinated people, the vaccinated should have fewer cases of rebound, not more. Then again, “with COVID over the last two and a half years, we have been wrong—I have been wrong—so many times,” he said.
Providers are certainly anxious to know how many of their vaccinated patients experience Paxlovid rebound. But on a more basic level, they’re anxious to know how well the drug works in vaccinated people at all. “We really know nothing about the magnitude of its benefit or its risk in people who are vaccinated, let alone triple or quadruple vaccinated,” Walid Gellad, who directs the University of Pittsburgh’s Center for Pharmaceutical Policy and Prescribing, told me. Without that information, if a doctor has to decide whether to prescribe Paxlovid to a patient who’s eligible, “you make your best guess,” Gellad said.
Pfizer has hinted at some sunny results in vaccinated people, but no data have been made publicly available. Also, way back in December, the company said it had finished enrolling participants for a follow-up study of Paxlovid in people who are not at high risk of severe COVID. But then, a few months later, it changed the eligibility criteria to leave out anyone who had received a vaccine dose in the preceding 12 months.
Today, that follow-up trial is still listed as in the “recruiting” phase on ClinicalTrials.gov. Reshma Ramachandran said the changes are a “red flag” that Pfizer might have uncovered some preliminary results among vaccinated participants that weren’t so flattering. David Boulware, a clinical-trial expert at the University of Minnesota, told me that he thinks the rationale for Pfizer’s actions is “pretty obvious”: The company will have an easier time proving the drug works in people who are more likely to be hospitalized—that is, the unvaccinated (and those whose vaccinations are more than a year old). “From a pure researcher perspective, I can understand exactly why they did this. But from a public-health and just, like, being-a-physician perspective, it’s a terrible idea.” The Pfizer spokesperson told me that the company had limited enrollment to patients who had not received a vaccine dose for at least a year in order to “enrich the study population for individuals whose immunity may be waning and who may be at elevated risk of severe COVID-19, hospitalization or death.”
To make the best decisions possible regarding prescribing Paxlovid, doctors and patients would need to understand how common rebounds are, whether the drug causes them, and whether people are infectious during the rebound period. They’d also need to know whether the drug has any meaningful benefits for people who have gotten a primary vaccine dose or booster shot since May 2021. Boulware said he expects researchers in the United Kingdom to have data on Paxlovid’s efficacy in vaccinated people within the next couple of months. Gellad is also eager to know whether Paxlovid helps stave off long COVID, a hypothesis that would by definition take months or years to test. Ali Ellebedy is curious about whether taking Paxlovid dulls a patient’s immune response to the virus and therefore could leave them more vulnerable the next time they encounter it.
Until those questions are answered, the providers I spoke with are all erring on the side of prescribing Paxlovid. “You’re kind of stuck as a prescriber,” Gellad said. Doctors could wait and advise their vaccinated patients not to take the drug until they’re certain it’ll help, but vaccinated people are still getting sick, developing long COVID, going to the hospital, and dying. To draw a balance between caution and action, Ramachandran said that when she prescribes Paxlovid to her vaccinated patients, she also explains that the clinical trials weren’t conducted in people like them, and so exactly what they’ll get out of the drug is uncertain. “When we’re trying to look for options for COVID-19, especially for treatment, we just have so few options,” she said. For now, Paxlovid is the best bet.