Paxlovid for Nonhospitalized Patients with COVID-19

The effectiveness of nirmatrelvir/ritonavir (Paxlovid) in reducing mortality or hospitalization in mild to moderate COVID-19 is uncertain

Benefits in NNT

Uncertain, likely none
Uncertain, likely none

Harms in NNT

1 in 5 were harmed (virologic rebound)
1 in 25 were harmed (experienced adverse events including dysgeusia)
19% higher rate of virologic rebound
4% higher rate of treatment-related adverse events
View As:


Johari F, Verma R. Paxlovid for nonhospitalized patients with COVID ‐19. Academic Emergency Medicine. Published online March 22, 2024:acem.14896.

Study Population: Two randomized controlled trials with 3286 nonhospitalized symptomatic adults with acute mild to moderate COVID-19

Efficacy Endpoints

All-cause mortality and hospitalization

Harm Endpoints

Adverse events (e.g., rebound, dysgeusia, diarrhea)


Paxlovid is an oral medication that combines two drugs, nirmatrelvir and ritonavir. Nirmatrelvir is a protease inhibitor, and ritonavir is used to increase the levels of nirmatrelvir. In 2021 Paxlovid was granted Emergency Use Authorization by regulatory authorities in various countries including the United States for the treatment of mild-to-moderate COVID-19 in high-risk individuals. In 2023 the U.S. Food and Drug Administration (FDA) granted Paxlovid full approval for this purpose.1, 2 Here we summarize data from randomized trials relevant to the FDA-approved use of the drug.

As of February 2024, we are aware of two relevant randomized controlled trials (RCTs), both sponsored and conducted by Pfizer.3, 4 A third trial is under way by a nonindustry group in the United Kingdom.5 The first trial, called EPIC-HR,3 enrolled 2246 outpatients with <5 days of symptomatic COVID infection and at least one high-risk criterion for worsening. The most common criteria were obesity (80%), smoking, and hypertension. Any patients vaccinated against COVID or previously exposed to COVID were excluded from the trial. In this group, the drug reduced a composite endpoint of hospitalization for COVID or death: 6.3% with placebo versus 0.8% with Paxlovid (absolute risk difference [ARD] 5.5%, p < 0.001, number needed to treat [NNT] 18). This includes 12 deaths during the study period, all in the placebo group. Of note, a Cochrane systematic review in 2023 included the EPIC-HR trial and did not note or mention the results of the unpublished EPIC-SR trial.6

The risk of treatment-specific adverse events increases with Paxlovid (relative risk [RR] 2.1, 95% confidence interval [CI] 1.4–3.0; ARD 4%; number needed to harm [NNH] 25).6 The most common adverse events are dysgeusia and gastrointestinal symptoms such as diarrhea.

The second trial, EPIC-SR, remains unpublished but data have been uploaded to a trial registry site.4 The results report on 1288 symptomatic outpatients with acute COVID-19 given Paxlovid or placebo. No high-risk criteria were necessary for enrollment and COVID-vaccinated and previously exposed people were eligible. The trial found no difference between groups in hospitalization for COVID or death (0.8% vs. 1.6%, p = 0.2) and was stopped early for futility.

According to the only high-quality, prospective study reporting on virologic rebound that we are aware of, 127 nonhospitalized subjects with acute COVID were followed and tested. Virologic rebound (a substantial increase in viral load occurring after initial recovery) occurred in 21% of those who took Paxlovid versus 2% in the control group (ARD 19%, p = 0.001, NNH 5).7 Patients experienced a second clinical illness in 26% of those taking Paxlovid versus 15% in the control group.


The data on Paxlovid for acute COVID-19 suffer from several limitations warranting caution. First, EPIC-HR, performed early in the pandemic, excluded people with a history of exposure to COVID or COVID vaccination.3 This makes the trial inapplicable to the world's current population. The second trial,4 with a more generalizable population, did not have this exclusion but the findings remain piecemeal and unpublished. Most participants in both trials, however, were younger than 65, of White ethnicity, and predominantly from high-income and upper-middle-income countries, an additional limitation for generalizing.

On the subject of efficacy Pfizer's first trial EPIC-HR, which found a benefit, it was ostensibly performed in high-risk subjects (hence “HR”), while the second trial EPIC-SR (standard risk) found no benefit. However, the actual risk levels in the two trials, as indicated by the rates of hospitalization or death in both control groups, were similar at 6% and 2%, respectively. The 6% in EPIC-HR would, moreover, be lower today since unvaccinated, unexposed people have significantly decreased and vaccination has proven to lower hospitalization or death.8 While factors like obesity and chronic illness theoretically differentiated high versus standard risk in the two trials, vaccination status may have been the main driver of the small difference. Therefore, EPIC-SR seems much more likely to be relevant to current practice. In addition, EPIC-SR's finding of no benefit is more consistent with Paxlovid's failure in RCTs for patients hospitalized with COVID,9 for long COVID,10 and as a preventive after exposure to COVID.11 Similar to EPIC-SR, Pfizer has not published the results of the latter two studies. Other than EPIC-HR, performed in unvaccinated, unexposed patients, we are not aware of any trial of Paxlovid finding a benefit in COVID. Perhaps the ongoing UK trial will shed further light.

Two additional concerns with Paxlovid include, firstly, the strong cytochrome P450 effects of ritonavir, the drug's boosting agent. Concomitant use with some medications might significantly interfere with drug levels and metabolism,12 which raises the specter of harm and has led to a daunting list of common medications that render patients ineligible.

Secondly, virologic rebound following Paxlovid is a concern. There have been multiple reports and several studies of the subject, and an MMWR review suggests most studies do not show differences between persons receiving Paxlovid and those not.13 However, only one study we are aware of included careful prospective data collection and sequential viral load measurement. Published after the MMWR report, the study found a much higher rate of virologic rebound with Paxlovid and a higher rate of clinical worsening, though there were limitations. The Paxlovid group included more elderly and immunologically compromised patients, and no matching or randomization was attempted. However, rebound was more common with Paxlovid regardless of immunocompromise and other characteristics.6 The clinical significance of these rebound changes remain fuzzy, but harms of a new drug are often missed or underestimated initially; therefore, we are approaching the data cautiously. Hopefully further prospective data will be forthcoming.

Despite these limitations and concerns, and despite the very low risk of hospitalization and death, the Infectious Diseases Society of America (IDSA) recommends treatment with Paxlovid in ambulatory patients with mild to moderate COVID-19 at high risk for progression to severe disease. Their guideline recommends starting Paxlovid as soon as possible, preferably within 5 days of onset of symptoms.14 This may partly be due to the IDSA being unaware of the three unpublished Pfizer trials including EPIC-SR, in which the drug failed in cohorts more directly relevant to current practice.

Based on the failure of Paxlovid in EPIC-SR, likely harms of the drug, and a consistent trend of not publishing trial results gives rise to concerns about potential deliberate lack of transparency, we have assigned an NNT color recommendation of yellow (unclear if benefits/more data needed) to Paxlovid for COVID. While the drug might have a role in treating high-risk and particularly unvaccinated individuals infected with COVID-19, until further data from independent trials become available, the current evidence does not support its widespread use.

The original manuscript was published in Academic Emergency Medicine as part of the partnership between and AEM.


Fatima Johari, MD; Rajesh Verma, MD
Supervising Editors: Shahriar Zehtabchi, MD


April 23, 2024