In a field where the majority of treatment discussions in Long COVID are built on mechanistic reasoning, case series, or extrapolation from other conditions, controlled trial data are unusually valuable. Metformin has more of it than almost any other agent discussed in the Long COVID literature — and yet its clinical uptake for this indication remains limited, partly because the data emerged from contexts (early COVID treatment, prevention of Long COVID) that don't map neatly onto existing prescribing patterns.
This article reviews what the trials actually showed, what they did not show, and how the evidence informs the decision to consider metformin in a Long COVID patient.
On the evidence bar: This article discusses randomized controlled trial data specifically. Mechanistic rationale for metformin in Long COVID is compelling and is discussed briefly, but the clinical case rests on the trial data. Mechanism without trial data is a hypothesis; trial data without mechanistic understanding is still useful clinical evidence. The combination strengthens confidence, but the trials are what justify clinical application.
The Trial Evidence
COVID-OUT Trial (Bramante et al., 2023)
Design: Double-blind, placebo-controlled, factorial randomized trial. Participants were non-hospitalized adults with confirmed COVID-19 infection, high-risk medical conditions or obesity, symptom onset within 3 days. Interventions: metformin, ivermectin, fluvoxamine (and combinations), versus placebo. N = 1323.
Primary endpoint: Severe COVID-19 (hypoxia, emergency or hospital visit, or death) within 14 days. Metformin did not significantly reduce the primary endpoint compared to placebo.
The COVID-OUT finding was notable for several reasons beyond the headline number. The effect was present across subgroups, the study was well-controlled, and the placebo arm's Long COVID rate was in the range expected from observational studies, suggesting the control group was representative. The 41% relative reduction in Long COVID diagnosis is a substantial effect size by the standards of this literature.
It is important to contextualize what this trial tested: early metformin (within 3 days of symptom onset) in high-risk outpatients. It does not directly test metformin started months or years after acute infection in patients who already have Long COVID. The prevention finding is not the same as a treatment finding. This distinction matters — and is often elided in lay summaries of the data.
Evidence in Established Long COVID
Direct randomized trial data for metformin as treatment (as opposed to prevention) of established Long COVID is more limited. Several studies have examined outcomes in diabetic patients who were already on metformin at the time of COVID infection and found lower rates of Long COVID, which is consistent with the prevention data but confounded by the underlying diabetes and the duration of prior metformin use.
Observational data and mechanistic data together support a plausible treatment rationale, but as of this writing, a dedicated placebo-controlled trial of metformin in established Long COVID has not been published. This is a meaningful gap. The absence of trial data for established Long COVID does not mean the intervention is ineffective — it means we do not yet have the direct evidence to know, and clinical decisions must be made under that uncertainty.
Mechanistic Rationale
The mechanistic case for metformin in Long COVID runs through several pathways, none of which are fully established but several of which are supported by converging lines of evidence:
- mTOR inhibition: Metformin activates AMPK, which inhibits mTOR signaling. mTOR activation has been implicated in viral persistence, immune dysregulation, and cellular senescence — all of which have been proposed as contributors to Long COVID chronicity.
- Mitochondrial effects: Complex I inhibition by metformin has complex downstream effects on oxidative metabolism. Post-acute infection mitochondrial dysfunction is an active area of investigation in ME/CFS and Long COVID, and metformin may partially modulate some of these effects.
- Anti-inflammatory properties: Metformin has consistently demonstrated anti-inflammatory effects in non-diabetic populations independent of its glucose-lowering mechanism. Reduced cytokine signaling (particularly IL-6 and TNF-α) has been observed in multiple contexts.
- Viral persistence: Preliminary evidence suggests metformin may reduce viral reservoir persistence, which has been proposed as one mechanism maintaining Long COVID symptoms. This remains under active investigation.
Taken together, this mechanistic profile is consistent with the observed preventive effect. But it is worth noting that multiple mechanisms have been proposed for Long COVID and multiple agents have compelling mechanistic rationale without translating into demonstrated clinical benefit. Mechanism supports the hypothesis; the trial data support the clinical application.
Practical Considerations
Who to consider
In clinical practice, metformin is most readily considered in Long COVID patients who also have pre-diabetes, insulin resistance, or metabolic syndrome — because in those cases the indication overlaps and the benefit-risk calculation is straightforward. In patients without metabolic indications, the decision requires weighing the strength of the prevention data against the absence of direct treatment trial data, the patient's comorbidity profile, and their preferences after a genuine informed consent discussion.
Dosing and titration
The COVID-OUT protocol used 500 mg twice daily for the first two days, then 500 mg with each meal (three times daily) for the 13 remaining days of treatment. For longer-duration use in Long COVID management, standard titration from 500 mg daily, increasing by 500 mg every 1–2 weeks to a target of 1000–1500 mg/day as tolerated, is used. The extended-release formulation significantly reduces GI side effects — nausea and diarrhea are the most common reasons for early discontinuation.
Contraindications and monitoring
Metformin is contraindicated in significant renal impairment (eGFR < 30), active liver disease, and conditions predisposing to lactic acidosis. Basic metabolic panel and renal function at baseline and periodically thereafter is standard. Vitamin B12 depletion with long-term use is a well-documented effect — B12 monitoring annually and supplementation if needed should be part of the management plan.
Important context: Metformin for Long COVID is currently an off-label use (except in patients with diabetes or pre-diabetes). The decision to use it should involve a full discussion of the evidence, its limitations, and the patient's individual circumstances. The trial data support consideration — they do not mandate use.
Where the Evidence Stands
Metformin occupies an unusual position in the Long COVID pharmacotherapy landscape: it has the best prevention evidence of any agent studied, is inexpensive, has a decades-long safety record in general medicine, and has a mechanistic rationale that is consistent with current hypotheses about Long COVID pathophysiology. Its underuse for this indication likely reflects the fact that the primary evidence is in prevention, not treatment, and that clinicians are appropriately cautious about applying prevention data to an established chronic condition.
The next generation of trials — several of which are ongoing or in planning — should clarify the treatment question more directly. In the meantime, the current evidence is strong enough that metformin deserves serious consideration in the clinical management of Long COVID, particularly in patients with metabolic risk factors and those who are early in their post-acute course.