Design: Phase 2, double-blind, placebo-controlled RCT, 69 US sites
Population: 959 adults with long COVID ≥12 weeks in cognitive, autonomic or exercise phenotypes
Primary outcome: Clinically significant change in phenotype-specific patient-reported outcome at day 90
Effect: No benefit in any phenotype; e.g. cognitive 25-day +3.2% (95% CI −10.4 to 16.8)

RECOVER-VITAL was a double-blind, placebo-controlled phase 2 trial at 69 US sites. It randomised 959 adults (median age 49, 67% women) with long COVID of at least 12 weeks, enrolled into cognitive, autonomic or exercise-intolerance phenotypes, to nirmatrelvir-ritonavir for 25 days, for 15 days followed by placebo, or to ritonavir with placebo. It was published in The Lancet Infectious Diseases in August.

No phenotype improved on its primary patient-reported outcome at 90 days. For the cognitive phenotype the 25-day difference was 3.2% (95% CI −10.4 to 16.8); autonomic −6.4% (−18.5 to 5.7); exercise −7.8% (−19.5 to 3.8). Secondary performance measures did not differ. There were no deaths and no safety signal.

Viral persistence is one of the main hypotheses for long COVID, and this was a direct test of it with the most widely available oral antiviral. Off-label courses for long COVID are not supported. The drug remains appropriate for acute COVID-19 in high-risk patients, which is a separate question.

Do not prescribe nirmatrelvir-ritonavir for long COVID.
Continue to use it for acute COVID-19 in eligible high-risk patients.
Tell patients who ask that a large trial found no benefit, even with longer courses.
Refer to structured post-COVID services where they exist.

Why it matters

It removes the most accessible drug from the list of plausible long COVID treatments.

The statistics, in plain English

All differences are close to zero with intervals crossing zero, and some point estimates favour placebo. Each phenotype had about 330 people across three arms, so small benefits cannot be excluded, but a meaningful effect is unlikely.