Review: 17 Studies Tested Interventions for Long-COVID Brain Fog, What Each One Found

What the review covered

The authors noted that the World Health Organization had reported more than 774,075,242 COVID-19 cases, and that approximately 10–35% of survivors experience persistent symptoms.^[1] Under the National Institute for Health and Care Excellence (NICE) definition cited in the paper, long-COVID describes symptoms that persist for more than 12 weeks (3 months) and cannot be attributed to an alternative diagnosis, and roughly one in three people with COVID-19 was diagnosed with neurological symptoms within 6 months of infection. "Brain fog", the paper's focus, was described as a cluster of cognitive impairments including confusion, short-term memory loss, and difficulty concentrating.

A team from the Azrieli Faculty of Medicine at Bar-Ilan University (Safed, Israel) and Maccabi Healthcare Services (Tel Aviv-Yafo, Israel) set out to map which interventions had been tested. Their systematic review followed the PRISMA checklist, and the protocol was pre-registered on PROSPERO (CRD42024502977).^[1] Searches ran across PubMed, MEDLINE, Central, Scopus, and Embase for articles published between January 1, 2020 and December 31, 2023.

Eligible studies enrolled adults (≥ 18 years) with brain fog confirmed at least four weeks after COVID-19 infection and required objective assessment of cognition; studies without objective brain-fog measurement, with unclear diagnoses, or involving pre-existing neurological conditions affecting cognition were excluded, as were non-English papers, reviews, and meta-analyses. From 5,770 records, 2,613 were screened by title and abstract, 287 met criteria, and 17 studies were included after quality assessment. Risk of bias was rated with the Newcastle–Ottawa Scale for cohort studies and the revised Cochrane ROB2 tool for randomized trials.

What it found across four intervention families

Across the 17 studies, sample sizes ranged from 1 to 208 patients, with a combined 806 patients drawn from 13 countries, the USA, Norway, the United Kingdom, Spain, Italy, Germany, Taiwan, Israel, Poland, the United Arab Emirates, Brazil, France, and Japan. The authors sorted the interventions into four families.^[1]

Noninvasive brain stimulation appeared in 6 studies, spanning transcranial magnetic stimulation, theta burst stimulation, transcranial direct current stimulation, transcranial alternating current stimulation, and photobiomodulation. The review reported that all six showed improvements in brain fog symptoms. Hyperbaric oxygen therapy appeared in 3 studies; in all three there was improvement in brain perfusion (assessed by perfusion MRI) and a reduction in brain fog symptoms. In one of those trials, 73 patients were randomized to 40 daily oxygen sessions or control, and the report described a significant group-by-time interaction in global cognitive function, attention, and executive function.^[3]

The PEA-LUT compound (palmitoylethanolamide and luteolin) was studied in 2 studies, both of which reported improved cognitive impairment. Rehabilitation appeared in 6 studies and produced the most mixed results: 4 reported gains and 2 reported no significant improvement. In one cohort of 208 patients assessed with the Barrow Neurological Institute Screen, the proportion scoring below the cut-off fell from 54% to 33% at follow-up.^[4]

Where computerized cognitive training fit in

Within the rehabilitation family, the review distinguished generic multimodal programs from personalized computerized cognitive training. The two rehabilitation studies that showed no improvement both used a one-size-fits-all multimodal protocol in which every patient received the same interventions; the authors attributed the null results in part to that lack of personalization and to low adherence, in one telerehabilitation trial, only 25% of participants made significant use of the mobile application.^[1]

By contrast, the review highlighted a feasibility study by Duñabeitia and colleagues in which 73 post-COVID patients with brain fog completed at least 10 sessions of personalized computerized cognitive training across 8 weeks, tailored to each individual's cognitive profile.^[2] The authors reported "uniform improvements across various cognitive areas in posttest evaluations compared to the initial assessments," extending to all five measured cognitive domains. That study is the corpus study this review surfaces; its full methods and results are summarized on the companion page, 8 weeks of personalized cognitive training in long-COVID brain fog.

Treatment dose and adverse effects

The review compared the median treatment burden of each intervention. Noninvasive brain stimulation was the most expedient, with a median treatment duration of 13.5 days and a median of 12 sessions. Rehabilitation had a median duration of 32.5 days; hyperbaric oxygen therapy ran a median of 56 days across a median of 40 sessions; and pharmacological treatment required the longest median duration, 75 days.^[1]

On safety, the review reported that noninvasive brain stimulation and hyperbaric oxygen therapy caused only mild adverse effects, dizziness, headache, and scalp pain at the stimulation site for stimulation, and barotrauma, ear pain, palpitations, headache, and fever for oxygen therapy. Rehabilitation and pharmacological (PEA-LUT) treatment were reported with no adverse effects. The authors framed rehabilitation as "the safest choice among the interventions," while noting that its efficacy remained uncertain because of the mixed results.

Limitations, what this review does not show

The authors were explicit about the boundaries of their conclusions.

It cannot rank the interventions. Because the included studies used different outcome measures, the authors wrote that they "could not determine which intervention showed the most success regarding the primary outcome, treating brain fog symptoms." No effect sizes were pooled.

The evidence base is small and heterogeneous. Only 17 studies met criteria, several were case reports or case series with very small samples (as few as 1 patient), and the review listed its own limitations as "the limited number of studies included, high heterogeneity due to the use of different scoring methods, the inclusion of case reports and case series with very small sample sizes, and the lack of meta-regression analysis."

Rehabilitation results were mixed. Two of the six rehabilitation studies showed no benefit over control, and the authors stressed that further research is necessary to establish the efficacy of rehabilitation. They also called for future studies to test combinations of interventions and longer follow-up to assess long-term effects.

Where this fits in the broader evidence

This review maps the early intervention landscape for a symptom that affects a large share of COVID-19 survivors, and it places personalized computerized cognitive training inside the rehabilitation family alongside brain stimulation, hyperbaric oxygen, and pharmacology. Its central message is that the field is still too small and too heterogeneous to crown a single best treatment, and that personalized, adherence-driven approaches fared better than generic multimodal ones in the included studies. For the underlying personalized-training study mapped here, see the long-COVID brain-fog training study. For related evidence, see the scoping review of brain training apps in older adults and condition pages such as cognitive training studied in multiple sclerosis and cognitive rehabilitation studied after stroke. The full catalog of published research is on our research studies index.