Paxlovid does not reduce symptoms, definitive Pfizer trial finds.

Pfizer finally published its study of Paxlovid’s effects on symptoms for standard-risk and high-risk vaccinated patients with Covid-19.

No difference in symptoms.

The upshot of the trial, known as EPIC-SR and published in the New England Journal of Medicine, is that Paxlovid did not reduce the amount of time until patients got symptom relief. For anyone who thinks Paxlovid helps reduce symptoms, we now have high-quality, randomized, blinded, placebo-controlled clinical trial data—from the makers of the drug and published in the most prestigious medical journal in the world—saying otherwise. Of study participants randomized to receive Paxlovid, the median number of days until sustained symptom relief was 12 days, compared to 13 days among placebo recipients That difference was not statistically significant (and even if it were, it would hardly align with the anecdotes people tell about taking it and getting immediately better; Inside Medicine readers know better, of course).

No difference* in hospitalizations and deaths.

Pfizer’s study also found that Paxlovid had no statistical effect* on severe Covid outcomes (hospitalizations and deaths). This would be headline news, but for that asterisk. The explanation for the caveat comes from the trial’s design. EPIC-SR was designed to measure symptom duration, primarily. While the researchers decided ahead of time to look at severe outcomes as well, the trial did not enroll enough patients for those findings to provide definitive answers. (In research terms, the study may have lacked statistical power to detect any reasonable differences.)

Here’s what played out: In the Paxlovid group, 0.8% (5 of 654 participants) were either hospitalized for Covid or died (any cause), compared to 1.6% (10 of the 634 participants) in the placebo group. Though some might see that as a 50% relative reduction, the 0.8% absolute difference (1.6%-0.8%=0.8%) was not statistically significant.

The researchers also looked at high-risk participants who were vaccinated. This is massively important data to have. Keep in mind, this is the first randomized, blinded, placebo-controlled trial data we have studying Paxlovid among vaccinated people. And it found that those randomized to get Paxlovid had no statistical improvement in hospitalizations or deaths (2.2% in the placebo group, versus 0.9% in the Paxlovid group). This is a far cry from the eye-popping results from pre-vaccine data that made Paxlovid an overnight sensation when it came out in 2022.

Does this mean that Paxlovid should not be given to patients with a high risk of severe Covid-19 if they’ve been vaccinated?

Not necessarily, but it certainly is a blow. The main argument against the conclusion that Paxlovid should no longer be given to high-risk vaccinated patients is, as above, that this study was not really designed to have the statistical power to definitively answer whether bad outcomes were prevented in this population. And, while I tend to think that Paxlovid’s benefits have been overstated lately (that is, the benefits are not what they were in the pre-vaccine era), I spent a fair amount of time squinting at these new results this evening. When I did that, I was not convinced these new findings render Paxlovid useless for high-risk vaccinated people, and for two reasons. First, the confidence intervals on the severe outcome results were mostly in the negative terrain, hinting that a larger study designed to study this question might have found Paxlovid remains helpful. Second, a bunch of other outcomes bent towards Paxlovid retaining some benefit—like how many ICU visits occurred, and how long ER visits and hospitalizations lasted.

So, while not a death knell for Paxlvoid, this study should profoundly change our “priors” on what to expect from the forthcoming PANORAMIC trial out of the United Kingdom (a massive study of thousands of vaccinated patients randomized to get Paxlovid or not).

Why does this study go against what you’ve heard?

But isn’t Paxlovid still a great drug for high-risk patients? What about studies that have shown a signal of benefit? Well, the blockbuster clinical trial establishing Paxlovid’s benefit was conducted in the pre-vaccine era. Since then, all of the studies showing Paxlovid helps vaccinated people have been observational ones, without randomization or placebo. These studies are interesting, but they have great potential for being confounded. In observational studies, even though we try to match drug recipients to apparently very similar people who did not get the drug of interest, there are often differences we just can’t see in the medical records. Those differences may (or may not) negate the findings. Look, I say this as a person who published a study that found Paxlovid helped some vaccinated adults under age 50, but not others. When faced with contradictory data from high-quality clinical trials and observational studies, usually the trial data should win out.

The best we got ain’t great anymore. What’s next?

We are in a weird moment here. Paxlovid remains the best anti-viral we have, but the data supporting it are not nearly as impressive as they once were. I imagine that the drug still helps very high-risk populations, such as immunocompromised people. But the data from EPIC-SR leaves me wondering: will the trove of data from the PANORAMIC trial be very disappointing or merely somewhat disappointing? (I’d love to be wrong on this; I doubt I will be.)

It’s time to think ahead. The search for better anti-viral options for high-risk Covid-19 patients should start to feel more urgent. I hope the drug companies see an opportunity here.

*Note: in an earlier version, I accidentally switched the Paxlovid and placebo numbers. But it doesn’t matter really, because in either case, the difference between the two rates is not statistically significant.

Questions? Comments? Chime in!