The Biden administration and other governments around the world have bet big on Pfizer’s Covid-19 pill, Paxlovid (Nirmatrelvir/Ritonavir). The administration touts widespread availability, and White House Covid czar Ashish Jha says that it’s “good news” that people are using it more.

But despite ringing endorsements for Paxlovid, what remains under-discussed is the lack of available data about how well it works in the populations using it. Since the drug costs the government more than $500 per treatment course, this lack of data means that we have no way to evaluate whether Americans are getting a fair deal.

Let’s review the evidence for Paxlovid. The drug label given by the FDA when it is authorized relies exclusively on a published trial called EPIC-HR (the HR stands for high risk of progression to severe Covid-19). EPIC-HR reported an 89 percent risk reduction for being hospitalized for Covid-19 or dying, and that much-promoted figure provides the basis for the claim that Paxlovid lowers bad outcomes by 90 percent. But that figure is the relative difference, just like 2 cents is 100% more than 1 cent, but the actual difference they found was ~6 percent. Seven percent of people without paxlovid had the bad outcome, and this was lowered to 1 percent. This too is impressive, but it applies only to this very high-risk population.

To participate in EPIC-HR, you had to be unvaccinated and could not have previously had Covid-19. You had to have symptoms and could not have been dealing with Covid-19 for more than five days. Beyond this, you had to have at least one high-risk condition.

This list excludes most Americans. Most Americans do not have high-risk criteria, and 78 percent of Americans have gotten at least one dose of the vaccine. Another 140 million Americans have had Covid-19 already, and this figure is likely an underestimate. Being vaccinated, having had Covid, or both, markedly lower your risk of hospitalization and death after getting Covid (again). Because of when the trial was conducted—between July and early December of 2021—few, if any, subjects were infected with the Omicron variant—also relevant, given that the individual risk of hospitalization with Omicron appears to be lower. The 7 percent figure seen in the trial is not representative of the risk faced by most Americans.

The moment we start giving Paxlovid to people who don’t fit EPIC-HR is the moment we can’t be sure of what we will get. Some scientists think that the 89 percent relative reduction would be preserved. Let’s assume they’re right for the moment. With a lower risk of hospitalization in vaccinated people (vaccines still work well for preventing severe disease), let’s assume the risk is 2 percent if you also have high-risk features. Well, instead of 7 percent to 1 percent risk, we might go from 2 percent to 0.2 percent risk—but this is just a guess. The truth is that both the absolute and relative benefits may fall.

Already some evidence exists for this supposition. A second trial, EPIC-SR (standard risk) was press released in December. EPIC-SR was said to include unvaccinated people without high-risk features and high-risk vaccinated patients. The study suggests a 70 percent reduction, but this did not meet statistical significance. The risk of a bad outcome without Paxlovid was 2.4 percent and 0.7 percent with the drug. But, despite the promise of data being ready in December 2021, these results have not been updated or published. Indeed, the current listing for EPIC-SR was changed in April 2022 to exclude patients who have had any vaccine within one year and those with prior Covid-19. The aforementioned vaccinated patients included in the December 2021 press release? Gone.

Moreover, we are now learning about rebound events when Paxlovid stops— people who stop using Paxlovid see their Covid symptoms return. What is the significance of this? How common will it be? Could this undo any short-term gain? We don’t know.

All this matters because as you use Paxlovid to treat healthier people, and those who have been vaccinated, boosted, and previously infected (or all three), the number of people you treat to prevent a single hospitalization will swell. It might be 100, or 1,000, or even 10,000 in the lowest-risk categories. This means that we will be spending substantial sums to avert a single hospitalization— perhaps even millions of dollars—and that’s assuming that Paxlovid works at all in these populations.

What about reducing Covid-19 symptoms? It’s an important goal to help people feel better faster and get back to work. Unfortunately, despite being a specified secondary outcome for EPIC-HR, the data were not included in the publication of the study. For EPIC-SR, there was no difference between groups in resolution of symptoms, according to the press release. What about preventing transmission? No data. Preventing long Covid? No data.

Worse, there could be rare safety signals that we don’t yet see. When Covid vaccines debuted, we had no idea about the blood-clot risk of the J&J vaccine, nor the cardiac risk of mRNA shots (Moderna and Pfizer). With Paxlovid, we don’t know what we don’t know: data from a randomized controlled trial of 2,100 patients are not nearly enough to appreciate rare, but important, side effects that might be acceptable for people with a one-in-20 risk of hospitalization but might not be in those with a one-in-1,000 risk.

The U.S. government is poised to buy more than $5 billion worth of Paxlovid. Other governments have followed suit. But before governments become major investors, shouldn’t they know what they’re buying?

Photo By EUROPA PRESS/C.Lujan.POOL via Getty Images


City Journal is a publication of the Manhattan Institute for Policy Research (MI), a leading free-market think tank. Are you interested in supporting the magazine? As a 501(c)(3) nonprofit, donations in support of MI and City Journal are fully tax-deductible as provided by law (EIN #13-2912529).

Further Reading

Up Next