Long COVID Treatments: What 24 RCTs Actually Show Works

The best-supported long COVID treatment is not a pill

If you expected the drug pipeline to win this round, the randomized evidence says otherwise. In the most comprehensive synthesis of long COVID treatment trials available through early 2024, structured rehabilitation outperformed pharmacology. Zeraatkar et al.'s 2024 BMJ living systematic review pooled 24 randomized controlled trials with 3,695 adults who met WHO long COVID criteria: symptoms starting after SARS-CoV-2 infection, still present at least three months later, and lasting at least two months. Across that evidence base, online cognitive behavioral therapy and combined physical-mental rehabilitation showed the most consistent gains in fatigue, functional capacity, and quality of life. No drug class produced a compelling cross-trial efficacy signal.

That does not mean every rehab program works, or that drugs are doomed forever. It means that if you want the best-supported option from randomized trials searched through December 2023, rehabilitation is where the evidence sits.

Why this review matters more than one flashy trial

The BMJ paper matters because it did the unglamorous work properly. The authors used Cochrane-standard methods, graded certainty with GRADE, and applied the TRACT trustworthiness checklist. They excluded trials with fewer than 25 participants per arm, searched the literature through December 2023 with a supplementary update in February 2024, and grouped studies by predominant symptom profile: general fatigue and functional symptoms, respiratory symptoms, and neurological or cognitive symptoms. That structure matters in long COVID, where "one condition" can mean very different problems in different patients.

The review also handled crossover trials cautiously by using first-phase data only. That choice makes sense in a relapsing-remitting illness, where later periods can get messy fast. The tradeoff is lower statistical power, especially for smaller pharmacotherapy studies.

The bottom line from this synthesis is simple enough to be annoying: the boring interventions won. Online CBT and combined rehabilitation showed the most consistent benefit. Antivirals, anti-inflammatories, anticoagulants, and immunomodulators did not show convincing benefit across trials. After years of speculation about silver bullets, the strongest randomized signal favored supervised, structured work over pharmacologic hope.

Through the end of 2023, long COVID RCTs support rehabilitation more than drugs.

What "rehabilitation" actually means here

This is not a vague instruction to "exercise more" and power through symptoms. The helpful interventions in the BMJ review were structured programs, often combining physical and psychological support, with outcomes measured against controls. That distinction matters because long COVID patients vary widely in fatigue burden, exercise tolerance, autonomic symptoms, breathing limitation, and cognitive complaints. A good rehab program is a treatment protocol, not a motivational poster.

For patients with respiratory symptoms, the evidence gets more specific. Li et al.'s 2025 systematic review and meta-analysis focused on pulmonary rehabilitation RCTs in long COVID and found improvements in lung function, fatigue, and quality of life. But the authors also reported substantial heterogeneity. Effects varied across pulmonary rehab modalities and program durations, so the right summary is not "all pulmonary rehab works equally well." It is "pulmonary rehab has evidence of benefit, but program design matters."

Oliveira et al.'s 2024 meta-analysis reached a similar conclusion in patients with post-COVID-19 syndrome. The review found improvements in exercise capacity measured by the six-minute walk test, dyspnea, fatigue, and peripheral muscle strength. It also compared face-to-face pulmonary rehab with telerehabilitation and found comparable efficacy. That is unusually useful. A therapy is much easier to recommend when it does not require a hospital parking garage and half a day off work.

Why rehab may help more than one symptom cluster at once

One reason rehabilitation may beat drugs in mixed long COVID populations is that it can hit several problems at the same time. Deconditioning, reduced exercise tolerance, breathlessness, anxiety, depression, and poor quality of life often travel together. A targeted drug may help one mechanism if you picked the right mechanism. A structured rehab program can improve function even when the biology is mixed.

The psychological side is not hand-waving. Depression and anxiety are common in long COVID, and the CBT signal in the BMJ review fits with broader randomized exercise evidence. Zeng et al.'s 2025 meta-analysis examined nine RCTs with 514 participants with depression and found high-intensity exercise improved Hamilton Rating Scale for Depression scores versus control, with a standardized mean difference of -0.44 and a 95% confidence interval from -0.69 to -0.18. Overall depression scores also improved, with an SMD of -0.23 and a 95% confidence interval from -0.39 to -0.07. Those trials were not in long COVID patients, so they do not prove the same effect size applies here. They do make the rehab result more biologically plausible, especially for neurological and psychiatric symptom clusters.

Why the drug story is still unfinished

The strongest case against overreading the BMJ verdict is timing. Its search cutoff was December 2023. That means major pharmacotherapy trials published in 2024 through 2026 were not part of the synthesis. The review therefore supports a narrower claim than many headlines would like: no drug had shown compelling cross-trial efficacy in the randomized evidence available through the end of 2023.

That is not the same as saying no drug works, full stop. It is a timestamped conclusion. The authors called the review "living" for a reason. Long COVID pharmacology remains a moving target, and the field still needs updated synthesis that includes later trials.

There is also a mechanistic reason to keep an open file on drugs. Mo et al.'s 2024 review describes endothelial injury and dysfunction in post-acute sequelae of SARS-CoV-2. That vascular biology gives researchers a rational target for drug development. It also offers a clean explanation for why pooled drug results may look weak in heterogeneous long COVID cohorts: if only a subgroup has an endothelial-driven phenotype, a mixed-population trial can wash out a real subgroup effect. That is still a hypothesis, not proof. But it is a better hypothesis than "nothing pharmacologic could ever work."

The evidence has limits, and they are not trivial

The rehabilitation-first conclusion is evidence-based, but it is not airtight in every direction.

• The BMJ review excluded studies with fewer than 25 participants per arm. That improves reliability, but it may also remove legitimate phase II drug signals more often than rehab signals, because rehabilitation trials can be easier to enroll.
• Long COVID is heterogeneous by symptom pattern, vaccination status, variant exposure, time since infection, and likely mechanism. Pooling such patients can hide subgroup responses.
• Pulmonary rehabilitation results were inconsistent across modalities and durations in Li et al., so there is no single effect size you can apply to every patient and every program.
• The long-term durability of long COVID rehab remains incompletely established in high-certainty RCTs. Tan et al.'s 2024 meta-analysis found that cancer survivors maintained gains in physical function, activity, and quality of life six to twelve months after exercise-based rehabilitation, which supports biological plausibility for durable benefits after illness recovery. But cancer survivorship is not long COVID, and analogy is not evidence.

Those caveats weaken overstatement, not the main result. The current randomized record still favors structured rehabilitation over any drug strategy that had been adequately tested by the end of 2023.

What this means in practice

If you are choosing based on evidence rather than wishful branding, structured rehabilitation deserves first consideration. That includes pulmonary rehabilitation for respiratory limitation and exercise intolerance, and programs that incorporate CBT when fatigue, function, mood, or cognitive symptoms are prominent. Telerehabilitation is not a consolation prize; Oliveira et al. found it performed comparably to in-person delivery.

That does not justify reckless "push through it" advice. Long COVID patients can have post-exertional symptom worsening, autonomic dysfunction, or severe fatigue that make generic exercise advice counterproductive. The interventions with evidence were structured and supervised, not improvised from a fitness influencer's comment section.

As of the best synthesized randomized evidence through December 2023, the verdict is plain. Rehabilitation has the strongest support. Drugs had not earned it yet. That may change as newer trials accumulate, but it had not changed in the published evidence base captured by the BMJ living review. Until an updated synthesis shows otherwise, the most evidence-grounded move is not to wait for a miracle molecule. It is to start the kind of rehabilitation program that trials have already shown can help.