Long COVID patients are often more informed about treatment options than their doctors. This page is an honest account of the treatments generating the most discussion — what's known, what's not, and how each fits into a framework-guided approach to care.
Patients who have spent months in Long COVID forums, following specialist podcasts, or researching clinics arrive at an appointment already familiar with names like low-dose naltrexone, ivabradine, Paxlovid, and plasmapheresis. That research is often accurate at the level of mechanism — these are real compounds with real pharmacological effects and some basis for interest. The problem is the inferential step from interesting mechanism or small positive study to this is what you should try.
No single treatment works for all Long COVID patients — not because Long COVID is a single disease that simply hasn't met its cure yet, but because the cluster of conditions we label Long COVID involves multiple distinct sub-syndromes with different underlying drivers. A treatment that addresses dysautonomia may have no effect on central sensitization. A treatment that reduces mast cell reactivity doesn't resolve sleep architecture disruption. Getting the treatment right requires getting the sub-syndrome identification right first.
The following is an honest account of where each prominent treatment stands — what the evidence actually supports, how it fits into the broader sub-syndrome framework, and how I use or don't use it in clinical practice. I've tried to be accurate in both directions: neither dismissing treatments with genuine potential nor overstating evidence that doesn't yet warrant confidence.
These are among the most-discussed treatments in POTS and dysautonomia-driven Long COVID. Each has legitimate rationale, but the sub-syndrome details matter significantly for whether and how they should be used.
LDN is one of the most frequently requested treatments among Long COVID patients, and the interest is not baseless. At standard doses, naltrexone blocks opioid receptors to treat addiction. At very low doses — typically 1–4.5 mg — the brief blockade appears to prompt the brain to upregulate its own endorphin production, with downstream effects on immune modulation and central sensitization. In clinical practice, I find it can be genuinely helpful in some patients, but the response is heterogeneous: some patients notice meaningful improvement in brain fog, others in muscle fatigue and recovery, and others notice little effect at all. That variability is not a failure of the drug — it reflects the underlying diversity of what's driving symptoms across patients.
My approach is to consider LDN when the dominant sub-syndrome target isn't clear, or when a patient needs a behavioral or lifestyle intervention to take hold before a more targeted pharmacological approach makes sense. It's also a reasonable option when other treatments haven't provided traction and the symptom profile suggests a central sensitization component without an obvious primary driver. I start low and titrate slowly, watching for sleep disruption — vivid dreams are a common side effect that can usually be managed by taking the dose in the morning rather than at bedtime.
On the evidence: There are no randomized controlled trials of LDN in Long COVID as of the most recent systematic reviews. The evidence base consists primarily of observational studies and pre-post case series across ME/CFS and fibromyalgia populations, with modest but generally positive findings in patients who tolerate it. The mechanistic hypothesis — particularly involving TRPM3 channel effects on natural killer cell function — is genuinely interesting, but a compelling mechanism doesn't establish clinical efficacy or tell us which patients are most likely to respond. LDN is in the category of treatments I use because the risk-to-benefit profile is favorable and I have seen clinical benefit, not because the evidence base is robust.
Ivabradine reduces heart rate by a different mechanism than beta-blockers — it works on the "funny current" (If) in the sinoatrial node rather than on adrenergic receptors. This distinction matters clinically: because it doesn't broadly block the adrenaline system, it can reduce the tachycardia of POTS without worsening the fatigue or exercise intolerance that beta-blockade sometimes produces. In patients who cannot tolerate beta-blockers, ivabradine is a meaningful alternative. Some find it better tolerated; others find the opposite.
I generally prefer to optimize a beta-blocker first, then add ivabradine if higher doses aren't tolerated, and use ivabradine alone when beta-blockers can't be used at all. One important clinical observation: the standard literature description of ivabradine's side effects mentions "luminous phenomena" — a visual flickering at the edges of vision. In my experience with Long COVID patients, this is part of a broader sensory sensitization response that can include worsening light sensitivity, increased sound sensitivity, and motion intolerance. This makes it a less straightforward option for patients whose symptom profile already includes significant sensory sensitization, migrainous features, or visual auras. Beta-blockers tend to be more globally useful across the various dimensions of dysautonomia; ivabradine is more targeted.
On the evidence: Supporting data comes from case series and small open-label trials in POTS populations. This is a more substantial evidence base than for many treatments discussed in this space, though it remains off-label. The current evidence is sufficient to guide its selective use — it's not guesswork — but the confidence level is different from medications with large randomized trials behind them.
Guanfacine is an alpha-2 agonist that reduces norepinephrine signaling in the prefrontal cortex. It's used in ADHD, where it "settles down" inefficient, high-noise cognitive processing — helping with attention and impulse regulation without the stimulant mechanism. The interest in Long COVID relates to two distinct targets that often coexist: cognitive dysfunction (brain fog) and the non-tachycardia features of dysautonomia, particularly the adrenaline surge pattern — hot flashes, night sweats, the sensation of adrenaline dumps that patients describe as a sudden rush of activation.
For the tachycardia component of POTS, guanfacine has some downstream effect, but it's not my first-line choice. Where it tends to be most useful is when the predominant dysautonomia features are those adrenergic surges rather than primarily orthostatic tachycardia, or when beta-blockers are not tolerated and there's a concurrent cognitive symptom burden worth addressing. I discuss it with patients as a treatment that may help both of those targets simultaneously, which makes it worth trialing when both are present.
On the evidence: A study of guanfacine combined with N-acetylcysteine (NAC) showed global improvement in Long COVID patients with cognitive and other symptoms. I use that study to support a conversation about guanfacine, but I'm careful to hedge appropriately: the combination design doesn't tell us which component — guanfacine, NAC, or the combination — was responsible for benefit, and the study didn't identify which sub-syndromes improved most. What I can say is that it showed global benefit in a patient population that overlaps significantly with the patients I see, and the side effect profile at low doses is generally manageable.
Probably the treatment with the clearest evidence base in Long COVID — but the evidence requires careful reading about what it actually shows.
Metformin has the strongest trial evidence of any treatment discussed on this page — but the specifics of what that evidence shows matter considerably. The COVID-OUT trial, a well-designed randomized controlled trial, found that metformin reduced the risk of developing Long COVID by approximately 41% (odds ratio 0.59 at 300 days) when started early during acute COVID infection. That is a meaningful, placebo-controlled result that supports metformin as a prevention strategy — potentially reducing the probability that someone who gets COVID goes on to develop persistent symptoms.
What this trial does not establish is whether metformin meaningfully treats established Long COVID in patients who already have it. Several plausible mechanisms have been proposed — including anti-inflammatory effects and mitochondrial effects — but the direct treatment evidence is substantially weaker than the prevention evidence, and most clinics promoting metformin for Long COVID do not clearly distinguish between these two populations or two claims. The distinction isn't academic; it changes what a patient can reasonably expect from the treatment.
I do discuss metformin with appropriate patients, particularly those with metabolic risk factors who might have independent benefit from it, or where the clinical picture suggests it might help with fatigue dimensions that respond to metabolic support. I am also straightforward about what the COVID-OUT trial showed versus what remains uncertain about its use in established Long COVID. Metformin is a safe, well-tolerated, inexpensive medication — the bar for trialing it is relatively low — but that doesn't mean the evidence justifies presenting it as a Long COVID treatment with the same confidence as a prevention strategy.
Mast cell activation syndrome is one of the most frequently discussed sub-syndromes in Long COVID — and one of the most clinically complex to treat, with a wide range of options at very different evidence levels.
When the MCAS sub-syndrome is clearly present — flushing, urticaria, GI reactivity, heightened allergic-type responses after COVID infection — a layered treatment approach makes sense, but the evidence varies considerably across the layers. My general framework is that the medications readily available now serve as a bridge until a biologic option can be pursued. The analogy I use with patients is the same one used in rheumatology: the way biologics in autoimmune disease are "steroid-sparing agents," I think of Xolair (omalizumab) and Dupixent (dupilumab) as "antihistamine-sparing agents" in MCAS — the goal being to reduce the mast cell burden at a more fundamental level rather than managing the downstream histamine release indefinitely with symptom-controlling agents.
H1 antihistamines (like fexofenadine, cetirizine, loratadine) are robustly effective at managing histamine-mediated symptoms but carry a meaningful cognitive trade-off. Histamine is not simply a problem to be suppressed — it plays important roles in wakefulness, attention, and cognitive processing. H1 agents that enter the central nervous system are linked to cognitive dulling, and this is not a trivial consideration in patients who are already experiencing significant brain fog. I prefer fexofenadine (Allegra) over other H1 agents specifically because it has lower CNS penetration than cetirizine or loratadine, though I don't consider it perfectly neutral on this dimension either. For acute MCAS events — when a patient feels an episode coming on — stronger antihistamines like diphenhydramine (Benadryl) or hydroxyzine can abort the event, but these are tools for individual episodes, not daily management.
H2 antihistamines (famotidine, ranitidine) target a different receptor subtype concentrated in the gastrointestinal tract, and tend to be better tolerated cognitively. However, suppressing gastric acidity long-term carries its own concerns, including microbiome disruption — an issue that overlaps with the significant patient interest in SIBO as a co-occurring problem in this population. I use H2 blockers but am attentive to the microbiome implications.
Cromolyn sodium (oral) is one of the safest options available — it works by stabilizing mast cells before they degranulate rather than blocking the downstream histamine. The difficulty is that the clinical literature supporting oral cromolyn in MCAS is thin; most available data is for inhaled cromolyn in allergic conditions. I find that it helps some patients to some degree, and its safety profile makes trialing it reasonable, but I can't point to robust data to guide dosing or confidently predict who will respond.
Ketotifen has both some antihistamine properties and purported mast cell stabilizing activity, and patients frequently ask about it. My honest answer is that the dosing literature and clinical trial evidence in MCAS is insufficient to have well-grounded confidence about risk-benefit for any given patient. I don't discourage its pursuit when patients want to try it under close monitoring, but I can't guide it with the precision I'd prefer.
The idea that SARS-CoV-2 triggers reactivation of latent viruses — EBV, HHV-6, and others — as a perpetuating mechanism for Long COVID is biologically plausible. How confidently it translates to a treatment target is a different question.
The viral reactivation hypothesis holds that COVID infection disrupts immune surveillance, allowing latent herpesviruses (EBV, HHV-6) or other pathogens to reactivate and drive persistent symptoms. The hypothesis has biological plausibility and some observational support — elevated titers of reactivated viruses have been found in subsets of Long COVID patients in several studies. But a plausible mechanism generating a correlational biomarker finding is still a substantial distance from evidence that treating the virus changes the patient's clinical course.
My concern — and the consistent finding across analogous treatment trials in other post-infectious syndromes — is that even when antiviral treatment suppresses the viral load, patients often return to baseline after the treatment ends. The long-documented parallel is in Lyme disease: antimicrobials given for adequate durations reliably eliminate the infection, yet a subset of patients have persistent symptoms anyway. The most likely explanation is that what's perpetuating symptoms isn't the ongoing infection but rather the immune and neural changes the infection triggered. Treating the trigger after it has already done its work may not resolve those downstream effects.
RECOVER-VITAL, the RECOVER Initiative's randomized trial of Paxlovid for Long COVID, did not meet its primary endpoint, and the weight of negative evidence in this area is significant. Among the antivirals patients are being treated with elsewhere, Maraviroc has comparatively favorable long-term safety data — better than prolonged courses of acyclovir or valacyclovir. My clinical experience is that patients who complete antiviral courses often return to approximately where they started, with some having side effects that created new problems. I do not prescribe antivirals for Long COVID, but I don't discourage patients who want to pursue this through other providers, and I remain genuinely open to updating that position if adequately controlled trials produce different results.
These treatments generate the highest patient interest and the most emphatic practitioner promotion. They also have the largest gap between hypothesis and controlled clinical evidence.
These treatments are being promoted by some prominent clinics and specialists with genuine enthusiasm. I understand the motivation — these patients have exhausted standard options, the biology of Long COVID implicates immune dysregulation as a plausible target for plasma-based or cellular interventions, and specialists who believe in these approaches are trying to help people who are suffering. That good faith does not resolve the evidentiary problem.
Plasmapheresis has at least two negative controlled trials in the closest analogous conditions. The WASH trial and others were supposed to clarify its role — current data does not support its use as a standard treatment. Ongoing trials may produce different results, and I genuinely hope they do, but as of now the evidence does not justify it as a recommendation.
Stem cells, exosomes, and peptides occupy a category I would honestly describe as a clinical wild west. Each clinic using them has developed its own protocol — different sources, different doses, different preparation methods, different targets — and those protocols are not standardized, not primarily guided by controlled trial evidence, and in many cases not adequately regulated. Some may eventually produce meaningful results. The current situation, where confidence is high and data is thin, resembles many areas of medicine that later required significant course-correction.
I have hope that some of these approaches will, in time, be part of a useful toolkit for Long COVID. What I cannot do is recommend them to patients in a practice built on the principle that conclusions should not outrun evidence — that gap between hypothesis and data is exactly the gap the Trivium exists to expose, not paper over. If well-controlled trial data emerges, I will update this position accordingly, and I mean that non-defensively.
The treatments discussed on this page that I don't yet recommend are in that category because the evidence isn't there — not because the research questions aren't worth asking. I don't like to experiment on my patients, but I fully support the ongoing research efforts that will eventually settle some of these questions with the rigor they deserve. Those trials need participants, and patients who want to contribute to that work are doing something genuinely valuable.
I've tried to compile a searchable clinical trials page drawn from ClinicalTrials.gov, covering active Long COVID and ME/CFS studies across condition and intervention type. The categorical organization is pulled algorithmically and occasionally places trials imperfectly — I apologize in advance for any of that — but the underlying trial information is accurate and the contact details for each study are included where available.
Browse Clinical Trials →Every treatment discussed on this page can be the right choice for the right patient at the right point in their care — and irrelevant or counterproductive for a different patient with a superficially similar symptom profile. The sub-syndrome framework is what makes that distinction possible. Understanding whether fatigue is primarily orthostatic, migrainous, or driven by poor sleep architecture changes which of these treatments is even worth discussing.