Townspeople repairing rope bridges between rocky islands as fog lifts from the harbour.
Back to the library

Energy & sleep

Long COVID and post-viral fatigue

Researchers have run controlled trials of hyperbaric oxygen with people whose fatigue, brain fog and breathlessness persisted after COVID. The largest randomised trial reported improvements in attention, energy, sleep and quality of life in the group who had 40 sessions compared with the group who had a sham treatment.

Evidence so far
Moderate
Reading timeAbout 12 minutes

Start here

Why this is being studied

Long COVID describes symptoms that persist for months after the initial infection has cleared, most commonly fatigue, brain fog, disturbed sleep, pain and breathlessness. Imaging studies suggest that in some people the brain and other tissues are receiving less blood flow and running with low-grade inflammation, leaving cells alive but underperforming. Hyperbaric oxygen is being studied because repeatedly raising and then lowering oxygen levels appears to trigger the body's own repair pathways in exactly this kind of tissue.

Where the evidence stands

Flag text

The mechanism

What is going on, and why this might help

What is going on in the body

Long Covid is not one illness but a cluster of overlapping problems that persist after the acute infection has cleared. The most consistent findings in studies are low-grade, persistent inflammation with raised cytokines; dysfunction of the endothelium (the lining of small blood vessels) with microclots and impaired capillary flow; mitochondrial changes that reduce the cell's ability to make energy efficiently; and, in many people, altered brain perfusion and metabolism on functional imaging. Autonomic dysregulation and immune activation, sometimes with reactivated latent viruses, sit alongside these. The common thread is tissue that is inflamed, poorly perfused and metabolically inefficient, which maps onto fatigue, brain fog and exercise intolerance.

Where oxygen comes in

Breathing oxygen at raised pressure dissolves it directly into plasma, so oxygen can reach tissue whose capillary supply is compromised without depending on red-cell delivery. In cell and animal work, hyperbaric oxygen down-regulates pro-inflammatory cytokines such as TNF-α, IL-1β and IL-6 and modulates HIF-1α signalling. The hyperoxic-hypoxic paradox described by Hadanny and Efrati (2020) proposes that the repeated rise and fall of oxygen across sessions mimics a hypoxic signal, driving expression of genes for mitochondrial biogenesis, angiogenesis via VEGF and stem-cell mobilisation (Thom 2006). Improved endothelial function and microcirculation, and possibly effects on the coagulation pathway, are further proposed routes. In long Covid specifically, a sham-controlled trial reported improvements in cognition, fatigue and sleep alongside changes on brain perfusion MRI, which is consistent with, but does not prove, these mechanisms.

How solid is this

The underlying mechanisms are well characterised in laboratory and animal models and in other conditions such as radiation injury. In long Covid, human evidence comes from one randomised sham-controlled trial (Zilberman-Itskovich et al. 2022) and several small open-label series, so the mechanistic story is plausible and partly supported by trials, but not yet established. Key mechanism references: Hadanny and Efrati 2020; Thom 2006.

Explainer illustration

The studies

Every study is laid out the same way: what the researchers wanted to find out, who took part, what they actually did, what they found, why it might work and where the limits are. No jargon, and a link to the original if you want to read it yourself.

Each study is laid out the same way: what the researchers asked, who took part, what they did, what they found, why it might work, and where the limits are. The clinician layer at the end of each one carries the numbers.

Paper 1

Hyperbaric oxygen therapy improves neurocognitive functions and symptoms of post-COVID condition: randomized controlled trial

Zilberman-Itskovich et al. · Scientific Reports · 2022 · randomised, sham-controlled trial, n=73

Read the paper

What they asked

Many people who recover from the acute phase of COVID-19 are left with brain fog, poor concentration, exhaustion, disturbed sleep and pain for months afterwards. This Israeli team asked a simple question: if you give people with these lingering symptoms a full course of hyperbaric oxygen, and compare them with people who go through an identical-looking sham procedure, do the oxygen group think more clearly and feel better?

Who took part

Seventy-nine adults were randomised and 73 finished the study (37 in the oxygen group, 36 in the sham group). All were over 18, had a laboratory-confirmed COVID infection, and had cognitive symptoms that were still interfering with daily life more than three months later. Anyone with a previous brain injury, dementia or other brain disease was excluded. The average participant was in their late forties.

What they did

The treatment group breathed 100 per cent oxygen through a mask at 2 ATA (twice normal atmospheric pressure) for 90 minutes, with a five-minute air break every 20 minutes, daily, five days a week, for 40 sessions over two months. The sham group sat in the same chamber for 90 minutes breathing ordinary air at close to normal pressure (1.03 ATA); the chamber was briefly pressurised to 1.2 ATA at the start with the usual noise so that it felt real. Neither participants nor the people scoring the tests knew who was in which group. Everyone had computerised cognitive testing, questionnaires and a brain MRI before treatment and one to three weeks after the last session.

What they found

The oxygen group improved more than the sham group on the overall cognitive score, particularly on attention and executive function (planning and mental flexibility). These differences were moderate, roughly half a standard deviation. They also reported more energy, better sleep, less psychological distress and less interference from pain, the pain result being the largest effect. MRI showed increased blood flow in several regions involved in attention, emotion and memory, with subtle changes in frontal white matter microstructure. Side effects were common in both groups (around a third of each) and nobody stopped because of them.

Why it might work

The authors' working theory is that long COVID involves patches of brain tissue that are alive but underperforming, with reduced blood flow and low-grade inflammation. Repeated exposure to high oxygen followed by a return to normal air seems to switch on repair signals that normally respond to oxygen shortage: new small blood vessels, mitochondrial renewal and changes in immune cell behaviour. The MRI changes fit with that idea without proving it.

Limits

This was a phase II exploratory trial: 73 completers is enough to detect moderate effects but not to settle the question. Outcomes were measured only one to three weeks after treatment, so this paper alone says nothing about durability (a later follow-up paper does). Blinding relied on participants not being able to tell sham from real. Several authors work for or hold shares in a commercial hyperbaric company, which the paper declares. The study was funded by the hospital's research fund.

Paper 2

Long term outcomes of hyperbaric oxygen therapy in post covid condition: longitudinal follow-up of a randomized controlled trial

Hadanny et al. · Scientific Reports · 2024 · one-year follow-up of the trial above

Read the paper

What they asked

The 2022 trial above showed benefits measured only a few weeks after the last session. The obvious next question was whether those benefits fade, hold, or continue to build once people stop treatment. This follow-up study went back to the same participants more than a year later to find out.

Who took part

Of the 40 people originally allocated to hyperbaric oxygen, 37 completed treatment and 31 agreed to be reassessed for this long-term analysis. They were the same adults with persistent post-COVID cognitive symptoms described in the original trial. There was no untreated comparison group at this stage, because the sham group had since been offered the real treatment.

What they did

No new treatment was given. Participants had completed the original protocol of 40 daily sessions at 2 ATA, 90 minutes each with air breaks, five days a week over two months. They were re-evaluated on average 486 days (about 16 months, with a spread of roughly two and a half months either way) after their final session. The same questionnaires were repeated: quality of life (SF-36), sleep quality (PSQI), psychological distress (BSI-18) and pain. Cognitive testing and MRI were not repeated.

What they found

The improvements seen shortly after treatment were still present more than a year later, and in most domains were of similar size. Sleep quality, psychological distress, pain severity and pain interference all remained significantly better than before treatment, with no meaningful drift back towards baseline between the short-term and long-term visits. Effect sizes were in the large range (around 0.6 to 0.95) at both time points. Several quality-of-life domains followed the same pattern.

Why it might work

If hyperbaric oxygen works by prompting structural repair (new small vessels, restored energy production in cells, resolution of inflammation) rather than by a temporary boost while oxygen is high, then you would expect gains to persist once the course ends. Persistence over 16 months is consistent with that, although natural recovery over time cannot be ruled out without a control group.

Limits

Only 31 people, no control group at follow-up, and the measures were self-reported questionnaires. Cognitive scores and brain imaging, which were the primary outcomes of the original trial, were not repeated, so we do not know whether the objective changes persisted. Some natural improvement in long COVID over 16 months is expected, which this design cannot separate from a treatment effect. The same commercial conflicts of interest apply as in the original paper.

Paper 3

Effects of hyperbaric oxygen therapy on functional and structural connectivity in post-COVID-19 condition patients: a randomized, sham-controlled trial

Catalogna et al. · NeuroImage: Clinical · 2022 · randomised, sham-controlled trial (brain imaging sub-study)

Read the paper

What they asked

This paper comes from the same randomised trial as the first study, but looks at a different question: does hyperbaric oxygen change the way brain regions communicate with one another, and do those changes line up with the cognitive and emotional improvements people reported? Rather than blood flow, it examines functional connectivity (which regions fire together at rest) and structural connectivity (the integrity of the white-matter cables linking them).

Who took part

Of the 79 people randomised, 56 had usable imaging for this analysis (28 oxygen, 28 sham) after excluding scans spoiled by head movement. Average age was 48. The oxygen group happened to contain more men (54 per cent) than the sham group (25 per cent), which is a chance imbalance worth noting.

What they did

The treatment was the same as in the parent trial: 40 daily sessions of 100 per cent oxygen at 2 ATA for 90 minutes with air breaks, versus a matched sham at 1.03 ATA breathing air. Everyone had a 3-tesla MRI before and one to three weeks after the course, including a seven-and-a-half-minute resting-state functional scan and a diffusion scan with 64 directions. The team analysed connections between major brain networks, connections from specific seed regions such as the hippocampus and amygdala, and white-matter tracts mapped with tractography. Cognition was tested with a computerised battery and distress with the BSI-18 questionnaire.

What they found

After treatment, the oxygen group showed reduced coupling between the motor network and the salience network (the system that flags what deserves attention), and increased connectivity between the right hippocampus and the medial prefrontal cortex, a circuit important for memory and emotional regulation. Improvements in attention and executive function correlated with these network changes. On the structural side, white matter around the left amygdala showed improved integrity in the oxygen group, and the degree of that change tracked with reductions in psychological distress. Cognitive gains mirrored the parent paper: moderate improvements in global cognition, attention and executive function.

Why it might work

Long COVID may leave networks in a state of inefficient over-coupling, with regions working harder to compensate. If hyperbaric oxygen improves the energy supply and reduces inflammation in affected tissue, networks may settle back towards more typical patterns. The hippocampus and amygdala are metabolically demanding and sensitive to oxygen supply, which would explain why changes were concentrated there. This is a plausible narrative, not an established mechanism.

Limits

Fifty-six people, with 17 excluded for head motion, which is a well-known confounder in resting-state imaging and may not be random. Imaging was done only weeks after treatment. Group differences in sex at baseline were not planned. Multiple comparisons were corrected for, but the number of analyses is large. Funding was from the hospital research fund; two authors work for and one holds shares in a commercial hyperbaric company.