Long COVID Treatments: What Medications Show Promise and What Safety Risks We Still Don’t Know

More than 15 million Americans are living with Long COVID. For many, the fatigue, brain fog, heart palpitations, and muscle pain didn’t go away after the initial infection. They’re still here-months, even years later. And yet, there’s still no FDA-approved treatment for it. That’s not for lack of trying. Doctors are prescribing off-label drugs, patients are self-experimenting, and clinical trials are racing to catch up. But behind the hope, there’s a growing list of safety signals and unanswered questions that no one is talking about enough.

What’s Actually Being Tried Right Now?

Right now, the most studied drugs for Long COVID aren’t new. They’re old ones repurposed. The big ones? Baricitinib, metformin, low-dose naltrexone (LDN), and Paxlovid. Each comes with a known safety profile-but from different patients in different contexts.

Baricitinib, a JAK inhibitor approved for rheumatoid arthritis and alopecia, showed promise in reducing death during acute COVID. Now, the NIH’s REVERSE-LC trial is testing it on Long COVID patients. Early data suggests it might help with fatigue and brain fog. But here’s the catch: in rheumatoid arthritis patients, serious infections happen in 10-20% of cases. There’s also a 3.3-4.1% annual risk of major heart events. These aren’t theoretical risks. They’re documented. And Long COVID patients? Many are young, previously healthy. Are they at the same risk? We don’t know yet.

Metformin, the cheap, widely used diabetes drug, delivered a surprise in a 2023 University of Minnesota trial. When taken within the first few days of acute infection, it cut the chance of developing Long COVID by 41%. That’s huge. But 35.7% of people had stomach problems-nausea, diarrhea, bloating. For someone already battling chronic fatigue, that’s not just inconvenient. It can be debilitating. And this was prevention, not treatment. Does it help people who’ve had symptoms for six months? No data yet.

Low-dose naltrexone (LDN) is another off-label favorite. Dosed at 1-5 mg daily-far below the 50 mg used for opioid addiction-it’s being used for pain, fatigue, and inflammation. A 2024 study from Nova Southeastern University found 62% of Long COVID patients reported less fatigue. But 28% had trouble sleeping. 19% got headaches. These aren’t minor side effects. They’re the kind that make people quit. And because LDN isn’t FDA-approved for this use, there’s no standardized dosing. Some take 1 mg. Others take 4.5. What’s safe? We’re guessing.

Paxlovid, the antiviral combo of nirmatrelvir and ritonavir, is the most controversial. One small study showed symptom improvement in 38% of Long COVID patients after a 15-day course. Another, much larger NIH trial found no difference compared to placebo. The bitter taste? Reported by 79% of users. The drug interactions? Ritonavir messes with over 100 common medications-statins, blood thinners, anxiety drugs. One patient on warfarin could end up in the ER. Another on a heart medication might get dangerous spikes in blood levels. And yet, doctors are still prescribing it.

The Trials That Failed-and What They Taught Us

Not every drug that looked promising made it. BC007, a therapy designed to neutralize harmful autoantibodies, was halted in March 2025 after a phase II trial showed no benefit over placebo. Worse, three patients in the treatment group had serious infusion reactions. One needed hospitalization. The placebo group had one. That’s not a coincidence. It suggests the treatment itself triggered something dangerous in a subset of patients.

Then there’s sipavibart, a monoclonal antibody approved for preventing COVID in Europe and Japan. Now being tested for Long COVID, it’s given as a monthly injection. So far, safety data from prevention use shows 15% get injection site reactions. Less than 1% have allergic responses. But we don’t know what happens when you give it to someone whose immune system is already misfiring. Could it make inflammation worse? Could it suppress needed immune activity? The trial is still enrolling. No answers yet.

Even newer candidates like AER002 (a human immunoglobulin) and polymerized collagen show early promise. AER002 caused mild infusion reactions in 18% of patients. Collagen? Only 12% had temporary pain at the injection site. No serious events. But these are tiny studies-40 people, 60 people. Can we trust them? Not yet.

The Biggest Unknowns

The biggest problem isn’t just that drugs might be risky. It’s that we don’t even know who they’re for.

Long COVID isn’t one disease. The NIH now recognizes at least four different types-some driven by lingering virus, others by autoimmunity, some by nerve damage, others by mitochondrial failure. If you give a drug that targets autoimmunity to someone whose problem is viral persistence, it won’t work. And it might hurt them.

There are no blood tests to tell you which type you have. No biomarkers. No scans. No clear way to predict who will respond to baricitinib versus metformin versus LDN. So doctors are flying blind. Patients are too. That’s why 68% of people in the Body Politic support group have tried at least one off-label drug. And why 57% say it didn’t help enough-and 41% say the side effects were worse than the symptoms.

Even the drugs that work might not work long-term. We don’t know how long to take them. Should you take metformin for six months? A year? Forever? What happens if you stop? Does the fatigue come back? Does the risk of infection rise? We have no data.

What’s Coming Next-and Why It Matters

The NIH’s RECOVER initiative is the largest Long COVID research project in history. By 2025, it had poured $1.15 billion into the field. Now, it’s testing even more drugs: GLP-1 agonists like tirzepatide (Mounjaro), originally for diabetes and weight loss. These drugs reduce inflammation and might help with brain fog. But they cause nausea in up to 25% of users. For someone already nauseous from Long COVID, is that helpful or harmful?

Stellate ganglion blocks-tiny nerve injections used for chronic pain-are being tested for Long COVID-related anxiety and heart rhythm issues. Early data from pain clinics shows 15% get hoarse voices after the procedure. 5% get a bruise or bleed. But no one has studied this in Long COVID patients. What if it triggers a flare-up? What if it makes brain fog worse?

And then there’s the new mouse study from WEHI in Australia. A compound that completely prevented Long COVID symptoms in animals. No side effects seen. But it’s never been tested in humans. Toxicity studies start in early 2026. If it works, it could be a breakthrough. If it fails? We’ll be back to square one.

What Should You Do Right Now?

If you’re considering a medication for Long COVID, here’s what you need to know:

• Baricitinib is still experimental. Don’t ask your doctor for it unless you’re in a trial. The infection and heart risks are real.
• Metformin is the most evidence-backed option-but only if you start early. If you’ve had symptoms for months, it’s unclear if it helps. GI side effects are common. Start low (250 mg once daily) and go slow.
• LDN might help fatigue, but it can mess with sleep. Try 1.5 mg at night. If you can’t sleep, stop. No one knows the long-term effects of daily LDN use.
• Paxlovid should only be used under close supervision. Tell your doctor every medication you take. Check for interactions. The bitter taste? It’s normal. The drug interactions? Not worth the risk unless you’re in a trial.
• Never self-prescribe. Even “safe” drugs can be dangerous without proper monitoring. LDN isn’t regulated for Long COVID. Metformin can cause lactic acidosis in rare cases. Baricitinib requires blood tests.

There’s no magic pill. No quick fix. But there is progress. The first FDA-approved treatment for Long COVID could arrive by late 2027-if the trials pan out. Until then, the safest approach isn’t chasing the newest drug. It’s tracking your symptoms, working with a doctor who understands Long COVID, and avoiding treatments with more unknowns than answers.

Why This Matters Beyond the Individual

This isn’t just about one person feeling better. It’s about how medicine responds to a crisis it didn’t anticipate. We’ve spent billions on vaccines and acute treatments. Now we’re scrambling to fix what’s left behind. The drugs we’re testing now will shape how we treat future post-viral syndromes-like ME/CFS, Lyme disease, or even long-term effects of other viruses.

If we get this right, we’ll have a blueprint for treating chronic illness after infection. If we get it wrong, we’ll add more patients to the pile-those harmed by drugs that were supposed to help.

The science is moving fast. But the human cost is moving faster. Every patient who tries a drug without knowing the risks is a data point. Every side effect is a warning. Every failed trial is a lesson. We’re not just waiting for answers. We’re building them-one patient, one trial, one safety signal at a time.