A carefully designed trial has delivered a disappointing but genuinely useful answer for people living with long Covid: an extended course of Paxlovid did not help.
Researchers at Mass General Brigham published the findings in The Lancet Infectious Diseases, with NIH funding.
The hypothesis being tested
Long Covid has several competing explanations, and one of the most prominent is viral persistence — the idea that fragments of virus, or actively replicating virus, remain in tissue reservoirs after the acute illness and continue driving symptoms through ongoing immune stimulation.
The hypothesis has support. Viral proteins and RNA have been detected in tissue months after infection in some patients, and it offers a straightforward explanation for why symptoms persist long after someone tests negative on a nasal swab.
It also generates an obvious, testable prediction: if virus is still there, an antiviral that eliminates it should produce improvement. Paxlovid is the most effective available oral antiviral against SARS-CoV-2, so the standard five-day acute course was extended substantially to give it a fair chance at a persistent reservoir.
Why the trial design is better than most
The study was randomised and placebo-controlled with three arms: the full Paxlovid combination, each of its two components alone (nirmatrelvir and ritonavir), and placebo — using extended 15-to-25-day courses.
Separating the components is the clever part, and it is not routine. Paxlovid is a combination in which only nirmatrelvir is the antiviral; ritonavir is included to slow nirmatrelvir’s breakdown by the liver, raising its concentration. But ritonavir is a drug in its own right with effects on drug metabolism and, potentially, on immune function.
Testing each separately means a positive result could be attributed correctly, and a negative one cannot be explained away by the components cancelling each other out. Placebo control matters especially here, because long Covid symptoms fluctuate and unblinded studies in this population have produced misleading improvements.
The result
“Unfortunately, treating viral persistence with this antiviral medication did not show evidence of clinical benefit,” said co-principal investigator Lindsey Baden. An extended course did not improve long Covid symptoms compared with placebo.
What it does and does not rule out
The distinction matters, and the investigators are careful about it. The trial tested one antiviral, at one set of doses, over one duration, in one patient population. It does not disprove viral persistence as a contributor to long Covid.
Several explanations survive. The drug may not reach the tissue reservoirs where virus persists at adequate concentration — distribution into tissue differs from blood levels. Persistence may drive symptoms in only a subset of patients, diluted within a general long Covid population. Or the damage may already be done: clearing residual virus would not reverse tissue injury or a dysregulated immune state that has become self-sustaining.
What the trial does establish is that this particular intervention is not the answer, which is a real and actionable finding.
Why negative results matter here specifically
Long Covid has produced an unusual dynamic. With millions affected, no approved treatment and slow-moving research, patients have turned to off-label medicines, unproven protocols and private clinics offering interventions with no controlled evidence behind them. Antivirals have featured prominently in that landscape, sometimes at considerable personal cost.
A rigorous negative result gives patients and clinicians something firm. It closes off a plausible avenue that was consuming attention, money and hope, and redirects effort toward hypotheses that remain open.
What else is being investigated
Viral persistence is one candidate among several. Others include immune dysregulation that fails to resolve after infection, autoimmunity triggered by the initial illness, reactivation of dormant viruses already carried by most adults, microvascular and clotting abnormalities, and autonomic nervous system dysfunction producing the intolerance to upright posture and exertion many patients describe.
These are not mutually exclusive, and the most likely explanation for a syndrome this varied is that long Covid is several conditions sharing a trigger. If so, trials enrolling broadly will keep producing null results even where a treatment works for a subgroup — which argues for better patient stratification as much as for better drugs.
How the field advances
The trial-design lesson for a contested field
The way this study was built is worth separating from what it found, because long Covid research has been badly served by weaker designs.
Symptoms fluctuate substantially on their own, and many patients improve slowly over months regardless of intervention. Any uncontrolled study, or any study in which participants know what they received, will therefore record improvement — and that improvement will be attributed to the treatment. A great deal of the enthusiasm around unproven long Covid interventions traces to exactly this pattern.
Randomisation with a placebo arm is what separates a drug effect from natural recovery and expectation. Adding arms for each drug component goes further still, closing off the retrospective explanations that so often follow a null result. The finding is disappointing, but it is trustworthy in a way that much of the surrounding literature is not — and building that standard into the field is worth something independent of this particular answer.
Long Covid remains urgent and unsolved. Rigorous, well-controlled trials — including ones that fail — are how a field narrows the search, and this study was designed carefully enough that its negative answer can be trusted. Research news, not medical advice.