Intended for healthcare professionals

Clinical Review State of the Art Review

Pathophysiology, diagnosis, and management of neuroinflammation in covid-19

BMJ 2023; 382 doi: (Published 18 August 2023) Cite this as: BMJ 2023;382:e073923

This article has a correction. Please see:

  1. Rachel L Brown123,
  2. Laura Benjamin24,
  3. Michael P Lunn12,
  4. Tehmina Bharucha25,
  5. Michael S Zandi12,
  6. Chandrashekar Hoskote6,
  7. Patricia McNamara2,
  8. Hadi Manji12
  1. 1University College London, Queen Square Institute of Neurology, London, UK
  2. 2National Hospital for Neurology and Neurosurgery, Queen Square, London, UK
  3. 3UCL Institute of Immunity and Transplantation, London, UK
  4. 4UCL Laboratory of Molecular and Cell Biology, London, UK
  5. 5Department of Biochemistry, University of Oxford, UK
  6. 6Lysholm Department of Neuroradiology, National Hospital for Neurology and Neurosurgery, University College London Hospitals NHS Foundation Trust, London, UK
  1. Correspondence to H Manji hadi.manji{at}


Although neurological complications of SARS-CoV-2 infection are relatively rare, their potential long term morbidity and mortality have a significant impact, given the large numbers of infected patients. Covid-19 is now in the differential diagnosis of a number of common neurological syndromes including encephalopathy, encephalitis, acute demyelinating encephalomyelitis, stroke, and Guillain-Barré syndrome. Physicians should be aware of the pathophysiology underlying these presentations to diagnose and treat patients rapidly and appropriately. Although good evidence has been found for neurovirulence, the neuroinvasive and neurotropic potential of SARS-CoV-2 is limited. The pathophysiology of most complications is immune mediated and vascular, or both. A significant proportion of patients have developed long covid, which can include neuropsychiatric presentations. The mechanisms of long covid remain unclear. The longer term consequences of infection with covid-19 on the brain, particularly in terms of neurodegeneration, will only become apparent with time and long term follow-up.


In December 2019, clusters of cases of severe pneumonia epidemiologically linked to a seafood and wet market in Wuhan, China were admitted to health facilities.1 In January 2020, a novel strain of coronavirus, SARS-CoV-2, belonging to the same family of viruses causing, severe acute respiratory syndrome (SARS) in 2002, and Middle East respiratory syndrome (MERS) in 2012, was isolated from respiratory tract samples of affected patients. On 11 March 2020, the World Health Organization declared the SARS-CoV-2 (covid-19) outbreak a global pandemic.2

Although the numbers of patients infected were relatively small (SARS 8096, MERS 2591) anecdotal case series documented neurological complications including stroke, seizures, and Guillain-Barré syndrome.3456 The death rate was highest for MERS (34.5%), followed by SARS 9.6% and SARS-CoV-2, 1.4%.47 The growth in the total number of covid-19 cases worldwide has been exponential: over 760 million cases and 6.8 million deaths have been recorded, with 13 billion doses of vaccine administered.8

The Opensafely study on behalf of NHS England, which used a health analytics platform linked to general practitioners, found that death related to covid-19 was associated with the following risk factors: male sex, greater age, deprivation, and comorbidities including diabetes and severe asthma. Compared with the white population, individuals of black and South Asian ethnicity were also at higher risk of death, even after controlling for other risk factors.9

The covid-19 pandemic took the world by surprise. The immediate medical response was confused and bewildering, owing to a lack of precedence and experience of any such previous pandemic. However, the rapid and international coordinated response by the medical community has been outstanding. Within 12 months of the onset of the pandemic, the first vaccines were administered. This review summarizes the evidence on the neurological consequences of covid-19. We provide a summary of the complications encountered by clinicians, propose a pathophysiological basis for them, and discuss potential treatments. This review is for all clinicians and scientists who were, and still are, caring for patients affected by covid-19, both in the acute and chronic clinical setting. The current evidence is that the virus will become endemic in the global population. However, the development of new variants is still a great unknown that only time will reveal.

This review will include a granular understanding of subtypes of syndromes (including acute disseminated encephalomyelitis), a critical review of stroke management in covid-19, and navigate neurological complications in an established vaccine era.


The first large retrospective case series of 214 patients from Wuhan in 2020 documented neurological manifestations in 36.4%.10 These manifestations were classified into three categories: central nervous system (headache, impaired consciousness, ataxia, stroke), peripheral nervous system (anosmia, ageusia, nerve pain), and skeletal muscle injury. Neurological complications were more frequent in those with severe covid-19 infection. However, the underlying pathophysiological mechanisms were unclear. Direct viral invasion, associated immune and inflammatory processes, and complications of severe illness including sepsis, hypoxia, ventilation, and extracorporeal membranous oxygenation (ECMO) were all postulated.

An early observational study, describing a highly selected group of 43 inpatients, was able to classify, with some pathophysiological insight, a collection of neurological syndromes associated with covid-19. These syndromes were encephalopathy (presenting with delirium and psychosis but with normal magnetic resonance imaging and cerebrospinal fluid); inflammatory central nervous system syndromes including ADEM (acute disseminated encephalomyelitis), para-infectious and post-infectious encephalitis; myelitis; ischemic stroke often associated with pulmonary thromboembolic disease; and Guillain-Barré syndrome.11 With increasing experience of the pandemic, these neurological case definitions have been revised.12

The global consortium study of neurological dysfunction in covid-19 (GCS-Neuro-COVID) and the European Academy of Neurology neuro-covid registry (ENERGY), representing 28 centers, 13 countries, and four continents between them, documented neurological manifestations (self-reported neurological symptoms or neurological signs or syndromes) in 82% of patients admitted to hospital.13 The most common symptoms were headache (37%) and anosmia or ageusia (26%). The most prevalent signs or syndromes were acute encephalopathy (49%), coma (17%), stroke (6%), and meningitis/encephalitis (0.5%). In addition, this study reported an increased risk of in-hospital death (adjusted odds ratio 5.99; 95% confidence interval 4.33 to 8.28) in those with neurological signs or syndromes. Pre-existing neurological disorders were associated with an increased risk of developing neurological signs and syndromes or both with covid-19 infection (2.23; 1.80 to 2.75).

An individual patient data meta-analysis of 1979 patients admitted to hospital with covid-19 and acute onset neurological presentations, which included patients from low and middle income countries (LMIC), identified important geographical variance in terms of outcomes in these patients. The death hazard was higher in LMIC WHO regions when compared with high income countries.14 Such inequities should be at the forefront of consideration in the inclusion, design, and aims of future research and policy globally.

Sources and selection criteria

PubMed and Medline were searched for articles published in English using terms including covid-19 OR “SARS-CoV-2” AND “neuro*” OR “neuropathology” OR “encephalopathy” OR “encephalitis” OR “acute demyelinating encephalomyelitis” OR “ADEM” OR “acute haemorrhagic leukoencephalopathy” OR “AHLE” OR “demyelination” OR “acute necrotising encephalopathy” OR “PIMS-TS” OR “MIS-C” OR “stroke” OR “Guillain-Barré syndrome” OR myopathy OR myasthenia gravis OR “long covid”. Articles were also identified through the reference lists of selected publications. We included systematic reviews, large randomized controlled trials, high quality population based observational studies (which we considered at low risk of bias), and clinical guidelines. Where no other literature was identified, we report case series and limit presentations described in case reports to those with multiple reports in more than one geographical location in the narrative review. However, for table 1, which describes all neurological syndromes in covid-19, we included case reports with fewer examples.

Table 1

Neurological complications of covid-19 described in the literature15161718

View this table:

Given the recency of the pandemic, that few prospective or longer term follow-up studies have been done is not surprising. Estimates of case frequency are largely linked to hospital admissions, relying upon physician reporting. The incidence of neurological presentations in patients not admitted to hospital is less clear. A lack of standardized definitions across studies complicates the interpretation of many results. For example, cases of ADEM were often included in encephalitis studies. The application of best fit diagnoses to novel presentations and the presence of mixed pathologies complicated interpretation further. While most studies state the evidence for covid-19 for each case, coincidental infection cannot be excluded, and are a consequence of real time learning and observation of a new disease. The neuropathology and pathophysiology of acute covid-19, as well as those of long covid, is an ongoing area of research.


The epidemiological data described above suggest that, despite the myriad neurological presentations that have been described in covid-19 infection, SARS-CoV-2 is not a classic neurotropic virus. It does, however, have the potential for neuroinvasion, and clearly has the ability for neurovirulence (box 1).

Box 1

Glossary of terms19

  • Neuroinvasion: The ability of a virus to enter the nervous system (central or peripheral)

  • Neurotropism: The ability of a virus to infect and replicate within cells (neural or glial) of the nervous system

  • Neurovirulence: The ability of a virus to cause neuropathology and clinical disease. Does not require a virus to enter the nervous system, and can include the effect of systemic processes including a dysregulated inflammatory or immune response, vasculopathy, or hypoxia


Angiotensin converting enzyme 2 (ACE2) has been identified as the key receptor for the entry of SARS-CoV-2 into cells.20 Publicly available brain transcriptomic databases show that although total expression of ACE2 is lower in the brain than in the lungs, some brain regions express relatively high levels. These regions include the paraventricular nuclei of the thalamus, the raphe nuclei (brainstem), and the tuberomamillary nuclei (posterior hypothalamus), as well as the choroid plexus.21 Bulk and single cell RNA sequencing in human and mouse models show ACE2 expression in the olfactory sustentacular and vascular pericytes, rather than in olfactory sensory and olfactory bulb neurons.21 Postmortem studies in patients with covid-19 confirmed that the major cellular targets of SARS-CoV-2 replication are the respiratory mucosa ciliated cells and the olfactory mucosa sustentacular cells. Some cases showed viral RNA in the leptomeningeal layers surrounding the olfactory bulb.22 In theory, therefore, direct viral invasion of the brain parenchyma by hematogenous dissemination via the choroid plexus, an impaired blood-brain barrier, or retrograde invasion from the leptomeningeal regions around the olfactory bulb remain possibilities.

An early neuropathological study detected SARS-CoV-2 in the brains of 21/40 (53%) patients with covid-19.23 Viral proteins were found in cranial nerves IX (glossopharyngeal) and X (vagus), as well as in individual cells in the medulla oblongata. However, the presence of viral antigen was not associated with severity of neuropathological changes, including astrogliosis, activation of microglia, and infiltration by cytotoxic T lymphocytes. These changes were mainly seen throughout the brainstem, cerebellum, and meninges. In most studies where virus has been identified in the brain, the viral load is reported to be very low, and the findings are subject to limitations, including the possibility of hematogenous spread or viral contamination during autopsy. Another difficulty is distinguishing shed protein from whole virus.24 Reports of SARS-CoV-2 RNA isolation in cerebrospinal fluid are rare, including in those patients with covid-19 neurological syndromes. Thus, the evidence points to SARS-CoV-2 invading the nervous system (neuroinvasion), but less compelling data suggest neurotropism. The varied neurological presentations are instead a reflection of disparate underlying pathophysiology including inflammatory, immune mediated, and vascular mechanisms (fig 1a and 1b). A recent study in which SARS-CoV-2 was sequenced in autopsy brain tissue of unvaccinated patients during the first wave of the pandemic, up to 200 days after the original diagnosis, raises new questions about a possible sanctuary site in the brain.25 Although no correlation with neuropathology was indicated, this result warrants further investigation, especially in relation to long covid.

Fig 1a
Fig 1a

Hypothetical mechanisms for covid-19 neuroinflammation

Fig 1b
Fig 1b

Hypothetical immuno-thrombotic mechanisms for stroke in covid-19

The pathophysiology of covid-19 neurological syndromes is discussed in the sections below. However, it should be noted that neuropathological changes are not limited to patients with neurological syndromes, and can be found in patients with covid-19 without clinical neurological presentations. These changes could reflect the systemic consequences of a complex and mixed pathophysiology of severe SARS-CoV-2 infection, including hypoxia, sepsis, and metabolic derangement, often in the intensive care setting. Many features overlap with other systemic inflammatory syndromes, viral infections of the central nervous system, or critical illness requiring admission to intensive care, invasive ventilation, or ECMO.242627 In postmortem studies of patients with severe covid-19, the most frequent neuropathological changes included widespread reactive astrocytosis, and microglial activation with neuronal injury or loss attributed to hypoxic ischemic brain injury.24 The same studies reported innate immune activation with microglial nodule formation and variable neuronophagia, with a particular predilection for the brainstem.2324 Some authors suggest that this could contribute to symptoms of brainstem dysfunction, such as tachypnea.24 Adaptive immunity appears to be less prominent; cytotoxic T lymphocyte infiltration, described in the perivascular regions and brain parenchyma in some patients, was often absent or sparse, while mild meningeal infiltration by cytotoxic T cells was thought to be reactive.2426 Perivascular demyelination with some ADEM-like features is described in some small case studies.28 Pre-existing neuropathology (vascular or neurodegenerative) could contribute to both histological findings and the clinicopathological response to infection.2429

Immune activation is prominent in the pathophysiology of systemic covid-19 infection, and likely contributes to both neuropathology and neurological symptomatology of covid-19. Cytokine storms are central to the pathophysiology of acute severe SARS-CoV-2 infection,30 Several studies have investigated the association of cytokines with clinical severity and prognosis.3132 An early 2020 retrospective study from China showed an association between severity/mortality from covid-19 and increased pro-inflammatory and anti-inflammatory levels of cytokines, including interleukin 2R, interleukin 6, interleukin 8, interleukin 10, and tumor necrosis factor (TNF).33 A subsequent observational study measured serum levels of interleukin 6, TNF-alfa, and interleukin 1 beta in 1484 patients admitted to hospital with covid-19.34 After adjusting for disease variables including laboratory inflammation markers, hypoxia, patient demographics, and comorbidities, interleukin 6 and TNF-alfa serum levels remained independent and significant predictors of disease severity and death. Immune directed treatments improve outcomes in severe covid-19 infection; these include tocilizumab (monoclonal antibody blocking interleukin 6 receptor) and dexamethasone.3536

Several studies have attempted to correlate the association between brain injury and markers of immune dysregulation.3738 In one observational study of 175 patients with covid-19, sera from patients with covid-19 showed elevations of brain injury biomarkers, including neurofilament light chain and glial fibrillary acidic protein (GFAP), in a severity dependent manner, with evidence of ongoing brain injury after four months of follow-up.37 These markers were associated with elevated levels of pro-inflammatory cytokines and autoantibodies against brain proteins, such as myelin associated glycoprotein. However, the positive correlation between pro-inflammatory cytokines and brain injury biomarkers was also seen in the comparator group of 45 patients with influenza, suggesting that rises could be related to severity, rather than infection with a specific virus.37 This finding could also have implications for patients with long covid.

Humoral autoimmunity could contribute to the pathophysiology of covid-19 neurological presentations. Molecular mimicry between viral proteins and neuronal antigens, or delayed stimulation of post-viral autoantibodies, is described in a number of viral infections, including herpes simplex virus, Epstein-Barr virus, and HIV.39 In a limited uncontrolled study of 11 critically ill patients with covid-19 with varied neurological symptoms including myoclonus, delirium, and seizures, one in four specimens of cerebrospinal fluid showed IgA and IgG SARS-CoV-2 antibodies.40 Intrathecal synthesis is rarely reported, and in most cases, SARS-CoV-2 antibodies derive from sera in the context of blood-brain barrier leakage.41 Using cell based assays and indirect immunofluorescence techniques on unfixed murine brain sections, all 11 patients showed anti-neuronal antibodies in cerebrospinal fluid or serum, or both.40 Antigens included intracellular and neuronal surface proteins such as Yo and N-methyl-D-aspartate receptor (NMDAR), but also a number of non-specific epitopes. These included vessel endothelium, astrocytic proteins and neuropil of basal ganglia, hippocampi, and olfactory bulb epitopes.

Cerebromicrovascular dysfunction in covid-19 likely contributes to several neurological manifestations, including strokes, headaches, and encephalopathy. Experimental animal models show that the protease of SARS-CoV-2 induces cellular pathways, leading to string vessel formation of the small cerebral vessels, while subunit 1 of SARS-CoV-2 binds to ACE2, triggering pericyte mediated angiotensin evoked constriction of the cerebral capillaries; downstream effects of which are hypoxia and clotting abnormalities.4243 Some of these findings are corroborated in human histopathological and radiological examination. For example, evidence of endotheliitis (including subluminal fibrinogen deposition inferring blood-brain barrier permeability, with resulting multifocal microvascular injury and microhemorrhages) has been described.2326 However, severe disease, systemic hypoxia, and hemodynamic changes confounded these findings. Some studies have shown biomarkers of endotheliopathy persisting beyond three months, and have hypothesized a role in long covid.44 However, rigorous profiling of tissue, and peripheral blood biomarkers linking pathophysiology to clearly phenotyped syndromes and outcomes, supporting a causal mechanism, are still lacking.

In contrast, the mechanisms of macrovascular complications resulting in ischemic stroke are less specific, but likely to comprise a combination of microvascular injury, modulating cerebral blood flow, SARS-CoV-2 mediated systemic hypercoagulability, and antiphospholipid syndrome.4546 The delay of about one week from the onset of covid-19 symptoms to developing ischemic stroke suggests an immune mediated trigger.4748 Thromboembolic complications might also reflect cardiac dysfunction such as myocarditis.4950

The impact of SARS-CoV-2 infection on long term brain health is of great interest. Markers of neuronal and glial injury are reported to be elevated in blood in patients with severe covid-19, and in patients with encephalopathy.51 A study of 60 inpatients with covid-19 and neurological presentations, including 25 with encephalopathy, found a correlation between cerebrospinal fluid 14-3-3 and levels of neurofilament light chain (both markers of neuronal degeneration) in the acute illness and neurological disability at 18 months, suggesting that neuronal injury is related to outcome (ρ=0.719, p=0.001 and ρ=0.583, p=0.006 for CSF 14-3-3 and CSF neurofilament light, respectively).52 Longer term trends in cerebrospinal fluid, serum, and imaging biomarkers, and association with neurodegenerative disease, are not yet known. Before covid-19, severe infection (bacterial or viral) requiring hospital admission was reported to increase long term risk of dementia (including vascular dementia and Alzheimer’s disease).53 Whether covid-19 engenders similar or additional individual or population level risk will only be borne out over longer term follow-up studies. The main drivers leading to increased risk of dementia are unclear (possible factors include hypoxia, systemic inflammation, and loss of blood-brain barrier integrity). Alteration of amyloid processing has been suggested, but requires further investigation.54

Most outcome studies are in patients admitted to hospital, and how this relates to mild SARS-CoV-2 infection or younger individuals is unclear. A UK Biobank study of 785 participants, including 401 with predominantly mild (not admitted to hospital) SARS-CoV-2 infection, reported reduction in gray matter volume, particularly in the orbitofrontal cortex and para-hippocampal gyrus.55 Additionally, changes in structures connected to the olfactory cortex, as well as greater reduction in global brain size, were found in individuals following SARS-CoV-2 infection compared with non-infected controls. Since group results cannot be extrapolated to individuals, these data require caution in interpretation.56 These findings could simply reflect olfactory hypofunction prevalent with the earlier strains of the pandemic. Individuals studied were predominantly older adults, and 56-62% of infected individuals showed volume loss greater than controls. Plasticity with potential reversibility has not yet been explored.

Neurological syndromes


Encephalopathy, which includes delirium and a spectrum of disordered levels of consciousness, is the commonest neurological complication reported in patients admitted to hospital with covid-19. In the intensive care setting, encephalopathy can be severe, manifesting as delayed wakening from sedation, agitation, or confusion. An early retrospective observational study of 58 patients in intensive care units with acute respiratory distress syndrome (ARDS) caused by covid-19 found that 49 (84%) had neurological features, including agitation on discontinuing sedation and neuromuscular blockade (n=40, 69%), and confusion (n=26/40, 65%, assessed by confusion assessment method for the intensive care unit (CAM-ICU).57 Pyramidal tract signs with brisk reflexes, ankle clonus, and extensor plantars were found in 39 (67%), and 15/45 (33%) had ongoing executive dysfunction at the point of discharge from intensive care. Encephalopathy might have a prolonged course in this group. In a study of 795 patients with severe covid-19, 571 patients recovered command following (motor score 6 on the Glasgow coma scale) admission to intensive care, with 25% doing so ≥10 days and 10% doing so ≥23 days after cessation of mechanical ventilation.58 Encephalopathy, especially delirium, is not limited to the intensive care population, and can occur at or before the onset of respiratory symptoms or presentation to hospital and in patients with mild respiratory symptoms.1159 Younger adults can also be affected, and rarely, encephalopathy can overlap with acute psychiatric symptoms, including paranoid or delusional beliefs.1160 Focal neurological signs including ataxia or long tract signs can also occur in these patients, but imaging is usually normal.11 Both hyperactive and hypoactive delirium, including akinetic mutism, are reported.

The causes of encephalopathy could be multifactorial, including well established causes of cerebral dysfunction. In a retrospective study of 4491 inpatients with covid-19, 559 (12%) had a toxic metabolic encephalopathy, 435 (78%) of whom had mixed etiology including sepsis, hypoxia, uremia, hyper/hyponatremia, and fever.59 Patients with toxic metabolic encephalopathy had a more severe course of covid-19 (requiring admission to intensive care, intubation, and also developing acute renal failure). Risk factors for toxic metabolic encephalopathy included older age, diabetes, and premorbid neurological (including neurodegenerative), psychiatric, or cardiovascular disorders. Hypoxia might delay wakening from sedation, with both degree and duration of hypoxemia associated with delayed recovery of command following intubation.58 Additional factors associated with prolonged recovery include duration of sedating or neuromuscular blocking drugs, use of renal replacement treatment, older age, and male sex.58 Metabolic disorders such as thiamine deficiency and the precipitation of Wernicke’s encephalopathy have also been implicated in covid-19 encephalopathy.61 In older patients, pre-existing, perhaps subclinical, neurodegenerative pathology including microglial activation related to Alzheimer’s might predispose them to delirium.29

The role of systemic or central nervous system inflammation and microvascular dysfunction in encephalopathic states is unclear. Phenotypically similar encephalopathies, presenting with dysexecutive syndromes and aphasia, are described in other cytokine driven hyperinflammatory states, such as secondary hemophagocytic lymphocytosis (sHLH) or immune effector cell associated neurotoxicity syndrome (ICANS) following chimeric antigen receptor T cell (CAR-T) treatment.3062 Elevated levels of pro-inflammatory cytokines including interleukin 6, interleukin 8, and macrophage inflammatory protein-1 beta (MIP-1β) have been found in the cerebrospinal fluid of some patients with covid-19 encephalopathy, and could correlate with measures of blood-brain barrier integrity.63 However, both the presence of cytokines and blood-brain barrier disruption are not restricted to specific neurological phenotype or abnormalities on brain imaging, and comparison with control groups, including patients with covid-19 without neurological presentations, is lacking.526364 Evidence of SARS-CoV-2 in cerebrospinal fluid is sparse.65

Some evidence has been found to support a humoral autoimmune mechanistic model. In a short case series, two of three teenagers with SARS-CoV-2 infection and neuropsychiatric complications, not fulfilling the criteria for encephalitis and with normal imaging and standard laboratory investigations, had evidence of intrathecal antibody against SARS-CoV-2 and neuronal autoantibodies on anatomical immunostaining.66 Antibodies targeting transcription factor 4 were isolated in one patient who responded well to immunotherapy. A further case series reported three patients with severe myoclonus (the commonest movement disorder in covid-19 infection) and somnolence with normal imaging and cerebrospinal fluid.67 Again, improvement was observed following corticosteroids and plasma exchange, or both.

Brain imaging accompanying encephalopathy is by definition usually normal, or might show leptomeningeal enhancement. However, leukoencephalopathic changes or microhemorrhages might be found, especially in the intensive care population (fig 2 U-W).1164 Cytotoxic lesions in the splenium of the corpus callosum, thought to be caused by cytokine mediated glutamate release, are also reported.6869 Functional imaging studies might show abnormal patterns of metabolism or perfusion, particularly in frontoparietal or frontotemporal patterns in patients with neurological symptoms following covid-19 infection.7071 In a study of 26 patients, with two or more neurological symptoms, 10/15 patients undergoing18fluorodeoxyglucose positron emission tomography had predominantly frontoparietal hypometabolism with a high correlation with poor performance on the Montreal cognitive assessment (R2=0.62).71 Electroencephalography might show diffuse or focal slowing, with some patients reported to have seizures or non-convulsive status epilepticus. Cerebrospinal fluid is normocellular in encephalopathy; cerebrospinal fluid pleocytosis warrants investigation for SARS-CoV-2 related or post-infective immune mediated encephalitis, or other viral or bacterial co-infection.126364

Fig 2
Fig 2

Neuroimaging in covid-19.11A-D show imaging from a patient with limbic encephalitis: axial fluid attenuated inversion recovery (FLAIR) images show bilateral hyperintensity in the mesial temporal lobes (A), hypothalamus (B) and thalamus (C,D); E-H show axial magnetic resonance images (E-H), and histopathology (I-K) from a patient diagnosed with acute disseminated encephalomyelitis (ADEM): axial T2 weighted (E), susceptibility weighted images (F), post gadolinium (G, H) images show extensive confluent tumefactive lesions involving the white matter of the right cerebral hemisphere, corpus callosum, and corona radiata with mass effect, subfalcine herniation (E), clusters of prominent medullary veins (F, short arrows), and peripheral rim enhancement (H, arrows). The white matter shows scattered small vessels with surrounding infiltrates of neutrophils and occasional foamy macrophages extending into the parenchyma (I, arrow). The endothelium is focally vacuolated but no evidence indicates vasculitis or fibrinoid vessel wall necrosis in any region. The white matter contains a few perivascular T cell but the cortex appears normal (not shown). CD68 stain confirms foci of foamy macrophages in the white matter, mainly surrounding small vessels (J). Myelin basic protein stain (SMI94) shows areas with focal myelin debris in macrophages around vessels in the white matter (arrows) in keeping with early myelin breakdown (K). Bar is equivalent to 45 microns in E, 70 microns in J and K. L-O show imaging from a patient with ADEM-like changes following severe covid-19 and admission to intensive care: axial T2 weighted (L), diffusion weighted imaging (M), susceptibility weighted images (N), and post contrast T1 weighted (O) images show multifocal clusters of lesions involving the deep white matter of both cerebral hemispheres, intralesional cyst-like areas of varied sizes and some peripheral rims of restricted diffusion (M), some hemorrhagic changes (N), and T1 hypointense “black holes” without contrast enhancement (O). P-S show imaging from a patient with cerebral infarction and pulmonary thromboembolism: CT pulmonary angiogram (P) showed large emboli in the right and left pulmonary arteries (arrows). Diffusion weighted imaging (Q), T2 weighted fast spin echo (R), and susceptibility weighted images (S) show restricted diffusion (Q) and T2 hyperintensity (R) in the left basal ganglia and cortical territory of left middle cerebral artery. The susceptibility weighted images (S) show hemorrhagic transformation in the basal ganglia (short arrow) and a long intravascular thrombus in a Sylvian branch of the left middle cerebral artery (long arrow). T-W shows images from a patient with severe covid-19 infection and cerebral microhemorrhages: chest CT (T) shows severe covid-19 pneumonitis. Susceptibility weighted images (U-W) show numerous cerebral microbleeds in the temporal, frontal, and parietal lobes, predominantly located at the grey/white matter junction. Adapted from Paterson, Brown et al,11 with permission from journal Brain

In our experience, encephalopathy tends towards improvement, especially if imaging and laboratory investigations including cerebrospinal fluid parameters are normal. Published longer term outcome data are lacking. Management of encephalopathy includes dealing with contributing factors such as hypoxia, electrolyte imbalance, and sepsis. Anecdotal reports suggest that intravenous immunoglobulin (IVIg) might have a role in management, thus supporting a neuroinflammatory pathophysiological mechanism.72 The effects of IVIg are pleiotropic, and include binding to Fc receptors on macrophages and microglia, as well suppression of inflammation mediated by cytokines and chemokines. Encephalopathy in covid-19 is associated with a greater risk of death within 30 days, poorer functional outcome, and risk of a dementia at six months.7374

Inflammatory central nervous system syndromes

Acute necrotizing encephalopathy

Acute necrotizing encephalopathy presents with acute, rapid onset neurological deterioration, typically progressing to coma, often with seizures, in the context of a febrile illness.7576 Although more common in children, adult cases have been described, including case reports in the context of covid-19.11777879 Magnetic resonance imaging shows bilateral, symmetrical signal abnormalities in the deep gray matter, usually involving the thalami, progressing to necrosis, and often hemorrhage. Cerebrospinal fluid is typically non-cellular but with elevated protein, and histology reveals edema with lack of inflammatory infiltrate.75 Acute necrotizing encephalopathy is associated with a range of viruses, including influenza A and H1N1. Cytokine storming has been implicated in the pathogenesis, which could be relevant to covid-19, though systemic inflammatory markers might not be significantly raised in some cases.79 Metabolic or mitochondrial mechanisms are also postulated. A genetic predisposition linked to mutations in the RANBP2 nucleoporin encoding gene is described, and might provide insights into the cellular and metabolic response to infection and inflammation.76 Treatment targeting the systemic immune response including corticosteroids, intravenous immunoglobulins, and tocilizumab have been used with benefit in some cases, though outcomes are often poor with high case fatality.7678


Encephalitis is reported rarely in the context of covid-19. Most cases are presumed to be immune mediated, without clear evidence for direct brain infection. A study of 10 384 patients treated for covid-19 at the Mayo clinic identified five (0.05%) patients with definite, probable, or possible autoimmune encephalitis, using the Graus criteria.8081 However, three had only possible autoimmune encephalitis, reported as difficult to distinguish from toxic metabolic encephalopathy. Two large multicenter retrospective studies which screened consecutive covid-19 admissions determined incidences of 40 and 50 in 100 000 admissions meeting International Encephalitis Consortium criteria.828384 In a retrospective study of 71 904 covid-19 emergency department admissions in Spain, 29 patients met diagnostic criteria for meningoencephalitis, though 32.2% had presumed bacterial etiology, and detailed case descriptions were lacking82 The ENCOVID study identified 22 patients with encephalitis among 43 139 admissions84 The range of pathologies and radiological abnormalities described include potential encephalitis mimics, highlighting the need for careful case assessment and use of diagnostic criteria. These mimics could include stroke, reversible posterior leukoencephalopathy (RPLS) syndrome, and reversible cerebral vasoconstriction syndrome (RCVS).

No dominant encephalitis phenotype associated with covid-19 has emerged. Our own experience, reflected by case reports in the literature, included occasional cases of limbic or rhomboencephalitis, responsive to immune suppression. Other presentations reported are highly heterogeneous, including altered mental state with aphasia, seizures, or non-convulsive status epilepticus, movement disorders including opsoclonus myoclonus or parkinsonism, and ataxia.84 Focal motor signs corresponding with lesions within white matter tracts are also described.80 Reported imaging features are equally varied, including cases with temporal lobe hyperintensity on magnetic resonance imaging typical of limbic encephalitis, T2 weighted/fluid attenuated inversion recovery (T2/FLAIR) hyperintensities within the cerebral white matter, brainstem or cerebellum, or normal brain imaging (fig 2 A-D).11

Timing of onset of encephalitis can vary. In the ENCOVID study, 11 (44%) patients had onset of neurological and respiratory symptoms concomitantly, while 12 (48%) had onset after (median eight days, range 5-12) and two (8%) preceding respiratory symptoms.84 The variable relation between infection and symptom onset, as well as the phenotypical heterogeneity, suggests a possible range of contributing para-infective or post-infective immune processes and host susceptibilities, including cellular, cytokine, humoral, or other immune factors.8586 Associations with known cell surface antibodies including to the NMDAR and voltage gated potassium channel complex have been described, albeit rarely, such that a direct causal link is difficult to establish.8788 In some cases, the presence of SARS-CoV-2 could be coincidental or non-specific; however, some of these observations could contribute to the general understanding of neurological autoimmunity and its association with infection and inflammation, rather than of covid-19 as such. More specific biomarkers are needed to identify cases of autoimmune encephalitis among those patients with significant cognitive and neuropsychiatric symptoms.


ADEM is an immune mediated inflammatory demyelinating disorder. It usually affects the white matter regions of the brain and spinal cord. ADEM is considered a para-infectious or post-infectious syndrome, usually triggered by viruses. Acute hemorrhagic leukoencephalopathy (AHLE, Weston-Hurst disease) is a rare fulminant form of ADEM. Both have been reported rarely in the context of covid-19, with reporting bias and under-reporting of cases both likely. In a UK nationwide surveillance study, 13 of 267 reported covid-19 neurological presentations, in patients admitted to hospital from March to October 2020, were classified as leukoencephalopathy or demyelination.60 In a prospective study of 47 572 patients with covid-19 in Singapore, also in the first phase of the pandemic from March to July 2020, one case each of ADEM and AHLE were identified, both in critically unwell male adults.89

Outside of the covid-19 setting, ADEM and AHLE are most often described in children. Clinical presentation of ADEM includes encephalopathy, focal neurological symptoms or signs, and sometimes seizures. Magnetic resonance imaging is characterized by diffuse, large, poorly demarcated lesions in the white matter and in some cases in deep gray matter, while histology reveals perivenular inflammatory infiltration and demyelination.90 Some children might have antibodies to myelin oligodendrocyte glycoprotein.91 Most respond to immunosuppression with good prognosis and monophasic course. AHLE can cause raised intracranial pressure requiring neurosurgical intervention, with cerebral edema, petechial hemorrhage on magnetic resonance imaging, and high mortality.92 Histology shows small vessel fibrinoid necrosis, perivascular neutrophil infiltration, hemorrhage, and demyelination.

In both syndromes, the (presumed) precipitating infective agent is frequently not identified; thus, the pandemic, with a prevalent and known agent, provides an unusual scenario. Despite this, the series of events leading from infection to inflammation in both presentations remains poorly delineated and can still only be presumed. Evidence for a preceding SARS-CoV-2 infection in reported cases varies from positive nasopharyngeal polymerase chain reaction (PCR) test, to SARS-CoV-2 antibody seropositivity or clinical symptoms temporally associated with peaks in infection. SARS-CoV-2 PCR has been found to be positive in cerebrospinal fluid in a small number of cases; however, the relevance of this is unclear.93

A systematic review comparing 46 covid-19 related ADEM and AHLE cases to typical pre-covid-19 cases, found increased numbers of adult compared with pediatric cases. The authors postulated that this could be related to immunological naivety to SARS-CoV-2 in the general population, and the age related increase in severity of SARS-CoV-2 respiratory infection. Historically, infections relating to ADEM are usually mild.94 Other differences between pre-covid-19 and post-covid-19 cases identified were more frequent preceding severe respiratory illness (29/46, 63% ventilated), less frequent myelin associated glycoprotein antibody positivity (1/15, 7% tested), more common hemorrhagic lesions (42%), less frequent response to standard immune suppression, and higher mortality.94 These differences were suggested to be due in part to disproportionate numbers of AHLE in the cohort (15/46), with delays to diagnosis and treatment owing to ventilation contributing to worse treatment response and outcomes. However co-reporting of ADEM and AHLE, alongside more novel presentations to which these labels are applied as a best fit, or in which there might be mixed (for example vascular/microvascular) pathology, could contribute to some differences, making interpretation difficult.

Most diagnoses of ADEM and AHLE reported in relation to covid-19 were based on pre-existing pediatric diagnostic criteria. We reported the only case to date with fulminant presentation and extensive edema requiring hemicraniectomy (fig 2 E-K).11 Histological descriptions are limited to the few patients who undergo biopsy. In our case, brain histology showed perivenular inflammation and myelin laden macrophages typical for ADEM.11 Brain tissue was negative for SARS-CoV-2. Similar histological features were present in another case without preceding respiratory signs but with positive SARS-CoV-2 serology in 2020.94 Other demyelinating presentations reported as post-infective complications of SARS-CoV-2 infection include myelitis and optic neuritis.95

The treatment of ADEM in the context of covid-19 has been extrapolated from the experience of pre-covid-19 cases. This extrapolation has been based on case reports and small case series in adults and children. Treatment consists of immunomodulation, usually with corticosteroids (1 g methylprednisolone for 3-5 days), followed by an oral prednisolone taper. In poor responders, treatment escalation often includes intravenous immunoglobulin and plasma exchange or both.96

ADEM-like presentations in the intensive care cohort

Atypical deep white matter lesions, especially in patients “slow to wake” after intubation and ventilation for critical covid-19, have been difficult to classify, and could be included in different studies as ADEM, ADEM-like, or hypoxia-ischemic brain injury.1197 Lesions are multifocal and appear hyperintense on FLAIR sequences, with variable degrees of hemorrhagic change, intralesional cyst-like areas, peripheral rim enhancement, and restricted diffusion (fig 2 L-O). Clinical features associated with these changes include abnormal wakefulness after sedation was stopped, confusion, or agitation, with variable pyramidal tract signs and seizures and headache reported in some cases. Given the requirement of magnetic resonance imaging, and the identification of clinical signs only on sedation withdrawal, diagnosis of these lesions is often delayed, and the timing and mechanism of brain injury is unclear. Histological correlation is lacking. While both ADEM and vascular pathologies have been described together, other contributors could include hypoxia, sepsis, coagulopathy, and cytokine storms.28 In one study, such lesions were correlated with severity of systemic and respiratory illness.97

Given the rarity of reporting of these cases, few treatment data are available. Whether inflammation is persistent is unclear. Some reported cases were treated with steroids, and others with supportive management only.11 In our own experience, these cases tended towards slow recovery with variable persistent cognitive and physical outcomes. On magnetic resonance imaging, we have found these lesions to be largely static over time, in contrast to ADEM, in which lesions typically resolve.

Pediatric inflammatory multisystem disorder temporally associated with SARS-CoV-2 (PIMS-TS) or multisystem inflammatory syndrome in children (MIS-C)

While children were less likely to present with severe respiratory symptoms, concerns of a cytokine mediated, pediatric, multisystem inflammatory syndrome were raised early in the covid-19 pandemic.9899 This syndrome, termed PIMS-TS or MIS-C, disproportionately affected black, Asian, and Hispanic children, and included older adolescents or young adults under adult neurology services.100101102 Characteristic systemic features include fever, rash, and frequent gastrointestinal symptoms including abdominal pain and diarrhea, with cardiac involvement and hypotension requiring inotropic support in some cases.99101 Neurological involvement is reported in 14.8-20.5% cases.102103 In a retrospective UK multicenter study including 25 children with PIMS-TS and neurological involvement, encephalopathy was the commonest symptom (88%), followed by headache or meningism (40%), behavioral change (36%), visual hallucinations (24%), focal neurological signs (24%), and seizures (16%).100 Peripheral neurological symptoms (40%) were mostly attributed to critical illness neuromyopathy, although focal nerve or muscle involvement is also reported.100103 Other covid-19 neurological complications, including acute stroke or ADEM, might co-occur.95100102

Elevated cytokine levels (particularly interleukin 6 and interleukin 8) and infrequent cerebrospinal fluid pleocytosis supported a cytokine driven etiology.100101 Additionally, markers of systemic inflammation are significantly elevated or abnormal, including elevated C reactive protein, erythrocyte sedimentation rate, D-dimer, ferritin, lymphopenia, and thrombocytopenia.99101 Magnetic resonance imaging abnormalities are common. In a retrospective imaging study including 11 PIMS-TS cases, abnormalities included splenial lesions of the corpus callosum (64%), ADEM-like changes (64%), cranial nerve (18%) or cauda equina (9%) enhancement, myelitis (9%), features suggestive of microthrombi (9%), and enhancing myositis of the facial and neck musculature (36%).95 Cerebral edema has also been rarely reported.102104 Splenial lesions are reported to represent cytotoxic edema secondary to cytokine mediated glutamate release, to which the corpus callosum is especially vulnerable. These splenial lesions have previously been described in children and adults with febrile or viral illnesses including influenza.6995100102105106107 Most reported cases were treated with various forms of immunomodulation including corticosteroids, IVIg, and in a small number of cases, tocilizumab or anakinra.100101 A more recent randomized controlled trial that recruited 75 patients with PIMS-TS to receive either intravenous methylprednisolone or IVIg concluded that intravenous methylprednisolone could be an acceptable first line treatment for children with PIMS-TS.108 Outcome data are limited to short term only; encephalopathy appears to respond well to treatment in most patients; however, ongoing disability related to neuromyopathy is reported.100103


The risk of stroke is more than twofold greater in patients with covid-19 compared with matched controls.4748109110 Further, the risk appears to peak between 1-2 weeks following primary respiratory covid-19 infection and persists for up to one month; the stroke sometimes occurred 1-3 days before manifestation of the respiratory illness.4748110 Acute stroke has been commonly reported as a neurological complication, but the overall frequency in the covid-19 population is low (1.4%; 95% confidence interval 1.0 to 1.9).111 A meta-analysis suggested that the absolute incidence of stroke was 0.168%, lower than previously estimated.112 The most common manifestation was acute ischemic stroke (87.4%); intracerebral hemorrhage was less common (11.6%).111 Emerging data indicate an association between covid-19 and intracerebral hemorrhage, occurring in patients admitted to hospital and intensive care.113 Notably, the characteristics of intracerebral hemorrhage in covid-19 differed from historic populations; in lobar location +/- fluid levels, multifocality (often in the presence of coagulopathies or anticoagulation use), and in its association with high rates of mortality.113

Similarly, in ischemic stroke, multiple focal large vessel arterial disease, usually in the absence of vasculopathy and often accompanied by systemic venous thrombosis (eg, pulmonary embolism), was characteristic of covid-19 (fig 2 P-S)60114115 Systemic venous thrombosis was likely due to widespread thromboinflammation, with inflammatory markers such as D-dimer and C reactive protein invariably linked to the covid-19 ischemic stroke phenotype.114116 Although the exact mechanism is not fully understood, a stroke secondary to a cardiac shunt is less likely to fully explain these observations.

Compared with a historic pre-pandemic stroke cohort, patients with stroke associated with covid-19 were younger (about one third were <60 years old) and had a greater number of conventional, modifiable cerebrovascular risk factors (eg, atrial fibrillation, diabetes, hypertension, dyslipidemia, and congestive cardiac failure).60115 Inpatient mortality and morbidity was considerably higher compared with non-covid-19 or historical controls.111114115116117118

In terms of management, a large individual patient data meta-analysis suggested that anticoagulation could be beneficial in this population.14 However, no trials have provided definitive guidance. The value of treatment with mechanical thrombectomy and thrombolysis remains unclear. Although small studies suggest a negative impact with mechanical thrombectomy, intravenous thrombolysis was not associated with an increased risk of disability, mortality, or hemorrhagic transformation.119120

Overall, vaccinations appear to reduce covid-19 associated stroke, despite the rare post-vaccine complication of arterial and venous cerebral infarction among those taking vector based vaccines.121

Guillain-Barré syndrome

Guillain-Barré syndrome, a disabling post-infectious polyradiculoneuropathy, was an early target of interest for potential neurological complications of SARS-CoV-2. Guillain-Barré syndrome has been described with other viral infections including MERS, and more convincingly with Zika virus, but also established viral infections such as cytomegalovirus and HIV; Campylobacter jejuni remains the best characterized and most frequent cause worldwide.122 Alongside the historical concern from influenza vaccinations, reports of a causative association of Guillain-Barré syndrome with SARS-CoV-2 were therefore inevitable.

The first cautiously worded case report from Wuhan in January 2020123 was swiftly followed by many reports of cases, clusters, and series of patients with Guillain-Barré syndrome associated with covid-19. These descriptions were retrospective, many with insecure reporting methods, and a paucity of reliable diagnostic confirmation. In addition, the small event numbers even in the larger series resulted in unfeasibly high Guillain-Barré syndrome cases rates. For example, 1.8% of covid-19 cases, a figure reported in one study, would equate to more than 250 000 excess Guillain-Barré syndrome cases in the UK in 2020 and 2021.124 Questions were therefore appropriately raised to challenge the covid-19-Guillain-Barré syndrome association. Counting cases of Guillain-Barré syndrome independently of the pandemic, through a UK-wide mandatory intravenous immunoglobulin treatment register for Guillain-Barré syndrome without reporting bias, showed fewer than 2300 cases in the UK in 2020 and 2021 combined. In the first six months of the pandemic, no geographical clustering of cases was observed with the recorded geographical spread of covid-19.125 Similar data were reported from Singapore, and a parallel pandemic of Guillain-Barré syndrome has not materialized.126 One population based study from the US veterans affairs population, investigating long term sequalae, suggested a link with covid-19 and Guillain-Barré syndrome beyond six months, but this requires corroboration in other settings.15

Guillain-Barré syndrome is a post-infectious disease triggered by molecular mimicry of neural epitopes on invading pathogens, driving a break in immunological tolerance to these epitopes, and targeted attack on axons and myelin mediated by macrophages. To date, and despite a number of different approaches, no linear, non-linear, protein, glycoprotein or carbohydrate epitopes, or immunological responses that might result from such drivers have been found in covid-19.127

A greater likelihood of causation exists if an excess of plausible cases can be shown, as well as specific or unique disease features, or a phenotype that makes the cases identifiable from background disease. All case reporting is biased unless one is able to obtain a truly random sample or complete acquisition. However, our own cohort, and the review of all the published case series, found no distinctive features of Guillain-Barré syndrome reported as being related to covid-19, compared with Guillain-Barré syndrome not associated with covid-19.125128 The excess morbidity and ventilation of patients in the covid-19 cohort was most likely related to covid-19 lung diseases, rather than neuromuscular complications of Guillain-Barré syndrome.128

If any extremely rare cases of Guillain-Barré syndrome are caused by covid-19, since they are not phenotypically different from other spontaneous cases, the treatment remains the same: intravenous immunoglobulin or plasma exchange. Early in the pandemic, an initial shift towards more plasma exchange was performed in some cohorts, as (subsequently unfounded) concerns arose about an adverse effect of IVIg. In the long term, patients have been treated in the same way and had very similar outcomes.

Other neuromuscular syndromes

Anecdotal case reports have described new onset myasthenia gravis in patients infected with covid-19. One explanation for these cases is that, as with any other viral infection, those with subtle symptoms or subclinical disease will become manifest.129130

In the early series from Wuhan, 11% of patients had evidence of muscle injury, defined as having an elevated creatine kinase level and myalgia, or both.10 More severely ill patients had higher levels of injury. However, many confounding variables were present, including hypoxia, sepsis, drug toxicity, and mechanical ventilation, making it difficult to disentangle the possible underlying mechanisms. A large autopsy study comparing non-covid-19 infected patients and patients with covid-19 who had died found significant signs of muscle inflammation in the covid-19 group.131 These included upregulation of major histocompatibility complex, endomysial infiltration with leukocytes, and occasionally, abnormal expression of myxovirus resistance protein A on capillaries suggestive of type 1 interferonopathy; which can be seen in dermatomyositis, but also as a tissue response to viral infection.131132 Although SARS-CoV-2 RNA was detected by reverse transcriptase PCR, evidence for direct invasion of myofibers was not confirmed using immunohistochemistry and electron microscopy. Occasional cases of rhabdomyolysis in patients with covid-19 have also been described.133 However, owing to the restrictions imposed during the pandemic, muscle biopsies were not routinely performed.

A number of focal neuropathies, including facial palsy and neuralgic amyotrophy, have been described in the context of SARS-CoV-2 infection.134135 Both presentations, however, can occur in the setting of other viral infections such as herpes simplex, herpes zoster, and HIV. Causation is difficult to prove.136137

In a follow-up series of patients who had been severely ill with covid-19 and required mechanical ventilation, a portion (11/69, 16%) presented with multiple mononeuropathies.138 These were not at the usual sites attributable to compression or traction. The authors postulated a possible vasculitic or immune thrombotic mechanism affecting the peripheral nerves or plexuses, rather than attributing the findings to critical care neuromyopathy. However, nerve biopsies were not performed.

Neurological presentations related to long covid

Long covid refers to prolonged symptoms following infection with SARS-CoV-2 that are not explained by an alternative diagnosis. It includes the National Institute for Health and Care Excellence terms “ongoing symptomatic covid-19” (symptoms lasting 4-12 weeks) and “post-covid syndrome” (symptoms beyond 12 weeks).139

In a review of the first year of the post-covid-19 clinic at University College London Hospitals (n=1325; 566 not admitted to hospital), the most common neurological symptoms were headache (17.6%), brain fog (15.1%), disturbed sleep (10.7%), and anosmia (9.2%).140 Anecdotally, at the Queen Square covid-19 neurology clinic, the most common neurological symptoms are attentional cognitive symptoms (brain fog), chronic daily headache, sensory symptoms, and autonomic symptoms. This list of symptoms is similar to a review of the first 100 patients seen at a neurology covid-19 clinic in Chicago (brain fog 81%, headache 68%, numbness/tingling 60%, anosmia 55%).141 In an international cohort, an online survey was completed by 3762 patients who had symptoms for more than six months, with onset before June 2020. The most frequent symptoms were fatigue, post-exertional malaise, and cognitive dysfunction. Of those, 85% experienced a relapse in symptoms with physical or mental activity.142 In a meta-analysis of 51 studies assessing 16 neuropsychiatric symptoms, the most common symptoms were sleep disturbance, fatigue, cognitive impairment, anxiety and post-traumatic stress. Little evidence indicated a relation between symptoms and the severity of the acute illness.143 In the UK post-hospitalization study (1077 adults, multicenter observational study), factors associated with failure to recover were female gender, middle age, ≥2 comorbidities, and a more severe acute illness.144

The chronic daily headache phenotype is most consistent with migraine. In the Queen Square cohort, imaging has been normal in the majority; a small number had some non-specific white matter T2 hyperintensities or incidental findings. In a Spanish case-control study, the presence of headache at onset was associated with a higher prevalence of headache and fatigue in the post-infectious phase.145 The headache had features of migraine and tension headache. Headache at onset was associated with a history of migraine (odds ratio 2.9; 95% confidence interval 1.41 to 5.98) and with development of persistent tension-type headache post-covid-19 (2.65; 1.66 to 4.24). Among those who had a history of migraine, worsening of migraine post infection was seen in 58.6% of those with headache at onset of the acute illness and 34.3% of those who did not have headache acutely (p<0.001). Fatigue was more common in those who had headache acutely (1.55; 1.07 to 2.24).

Cognitive symptoms are common (box 2) and on neuropsychometry assessment, both anecdotally from the Queen Square clinic and other published studies, mainly affect anterior cognitive pathways including executive function, impaired attention, and reduced verbal fluency.146147 In one University of California, San Francisco study, patients with cognitive symptoms had a greater median number of pre‐existing cognitive risk factors than control patients without cognitive symptoms. A small number of patients also had mildly raised cerebrospinal fluid protein. Matched oligoclonal bands were present, and therefore intrathecal synthesis/inflammation was not indicated.148

Box 2

Long covid-19; a patient experience

Long covid for me was characterized by severe daily headache, dizziness, nausea, fatigue, and difficulty concentrating. At one stage I experienced word finding difficulties during conversation. I needed six months off work. Two years later, I am progressively much improved. I am still unable to work as hard as I used to due to ongoing problems with mental endurance and frequent headaches following periods of focus. With careful activity pacing, I am on the whole able to function normally at work, and also, socially.

The fatigue associated with long covid for me has been a very different experience from normal tiredness. The physical fatigue is a whole body experience that is like a brick wall exhaustion. That has now settled. The mental fatigue is characterized by an inability to focus, think through complex problems and, sometimes, even have a conversation. It is triggered by long sustained periods of mental activity. I still experience this in a much milder way two years later, but have learnt that I can avoid it with pacing and by taking proper rest breaks at work.

As a doctor myself, I am all too aware of poorly understood clinical presentations. I now carefully frame how I tell a patient that their test results are normal. Instead I tell them that the results don’t explain their symptoms. This avoids a misunderstanding that I might think their symptoms are imagined or due to anxiety.

A pandemic with large numbers of patients with the same condition provides a perfect pool for research. I hope this will help patients with myalgic encephalomyelitis/chronic fatigue syndrome too. It is noteworthy that many of these patients report an infective trigger prior to symptom onset.


The syndrome is heterogeneous and the mechanisms behind the neurological manifestations are still poorly understood. Some studies have shown an association between inflammatory markers (elevated D-dimer, C reactive protein, and lymphopenia) and post-covid-19 symptoms,149 but this is not borne out in other studies.150151152 Similarities exist with other post-viral syndromes and chronic fatigue syndrome.153154155 Some patients might also experience somatoform symptoms.156 Ongoing research could yield further insights into the pathogenesis of post-covid-19 syndrome, and perhaps help unravel the mechanisms behind chronic fatigue syndrome or other post-viral syndromes.

Effect on pre-existing inflammatory neurological disease (central and peripheral)

Initially, concerns were raised that covid-19 or vaccination would worsen disease or precipitate relapse in patients with pre-existing inflammatory central nervous system and neuromuscular disease.157 Some evidence suggested that more substantial harms resulted from difficult and delayed diagnosis of general neurological disorders, anxiety of the unknown for patients, and reduced disease control resulting from fewer healthcare interactions and alterations in treatment strategies in the first year of covid-19, than resulted from relapses precipitated by infection or vaccination.157158

Early concerns were also raised that immunosuppressants used in neurological and rheumatological disease would affect humoral seroconversion to virus infection or covid-19 vaccination, with subsequent risks and adverse outcomes. A very large prospective national study showed that only mycophenolate mofetil, anti-CD20 monoclonal treatments, and sphingosine 1-phosphate (SIP-1) inhibitors (eg, fingolimod or siponimod) reduced humoral seroconversion.159

That the chronic relapsing and remitting peripheral nerve (and some central) diseases, especially chronic inflammatory demyelinating polyneuropathy (CIDP) or multifocal motor neuropathy with conduction block (MMNCB), worsen briefly with any infection through inflammatory decompensation, is well recognized. Early in the pandemic, case reports of recurrent Guillain-Barré syndrome and CIDP were inevitably published, but none has been substantiated by large series. In contrast, the main risk of an adverse outcome to patients with pre-existing multiple sclerosis who contracted covid-19 was not the infection, but previous recent treatment with anti-CD20 agents (rituximab and ocrelizumab), whereas interferons and glatiramer acetate were protective against severe covid-19.160161162 Patients with multiple sclerosis had normal T cell responses to SARS-CoV-2 infection and covid-19 vaccine in all treated groups compared with controls, and indeed, patients on anti-CD20 treatments with multiple sclerosis had enhanced IFN-γ T cell responses compared with untreated patients.163 No good studies have been performed on the effects of SIP-1 inhibitors on SARS-CoV-2 infection or vaccination T cell responses, which are much more likely to be associated with deficient T cell function; these studies are urgently needed.

In patients with myasthenia gravis, a global electronic health record database found that compared with the general covid-19 population, patients with myasthenia gravis and covid-19 were at a higher risk of admission to hospital (odds ratio 3; 95% confidence interval 2.4 to 3.8), admission to intensive care (5.2; 3.7 to 7.3), intubation (odds ratio 4.6), and death (4.3; 2.9 to 6.4).164

Concerns about antiviral vaccines precipitating relapses of inflammatory diseases are extrapolated from the rare association of influenza vaccine with Guillain-Barré syndrome.165 However, a study of more than 700 patients with previous Guillain-Barré syndrome found they were at no additional risk of another episode of Guillain-Barré syndrome following vaccination.166 Recent reassurance that influenza vaccination does not precipitate further episodes of Guillain-Barré syndrome or CIDP influenced two Dutch studies showing that although 5% of patients with CIDP and MMNCB reported a worsening of symptoms, these were mild, and only half required a modification of treatment.167168169 This finding was similar for patients with central nervous system diseases (multiple sclerosis, neuromyelitis optic spectrum disorders (NMOSD)).

Management of covid-19 associated neurological disorders

The key to the management of neurological disorders associated with covid-19 is a detailed history, including the time between infection and symptom onset, and examination together with the appropriate interpretation of radiological and laboratory data. This protocol is crucial, as the diagnosis will imply specific disease mechanism(s) which will guide best management (fig 3). At Queen Square, a multidisciplinary input in challenging cases, including complex stroke, has been invaluable.

Fig 3
Fig 3

Management strategies for key neurological complications in covid-19
ADEM=acute-demyelinating encephalomyelitis; AMPAR=alfa-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor; ANA=antinuclear antibodies; ANCA=antineutrophil cytoplasmic antibodies; ANE=acute necrotizing encephalitis; AQP4=aquaporin 4; MOG=myelin oligodendrocyte glycoprotein; CK=creatinine kinase; CNS=central nervous system; CRP=C reactive protein; CSF=cerebrospinal fluid; CSVT=cerebral sinus venous thrombosis; CXR=chest x ray; dsDNA=double stranded DNA; ECG=electrocardiogram; EEG=electroencephalogram; ENA=extractable nuclear antigen; ESR=electrolyte sedimentation rate; FBC=full blood count; FVC=forced vital capacity; GABA-R= γ-aminobutyric acid receptor; GBS=Guillain-Barré syndrome; GCS=Glasgow coma scale; HBA1C=glycated hemoglobin A1C; HIV=human immunodeficiency virus; ICP=intracranial pressure; MDT=multidisciplinary team; IgG-immunoglobulin G, ILR=implantable loop recorder; IPD=individual patient data; IVIg=intravenous immunoglobulin; LFT=liver function test; LGI1=leucine rich glioma inactivated 1; LP=lumbar puncture; LVO=large vessel arterial occlusion; mGluR5=metabotropic glutamate receptor 5; NCS/EMG=nerve conduction study/electromyography; NCSE=non-convulsive status epilepticus; NMDA-R=N-methyl-D-aspartate receptor; OP=opening pressure; PCR=polymerase chain reaction; PET=positron emission tomography; PF4=platelet factor 4; PIMS-TS =pediatric multisystem inflammatory syndrome-temporally associated with SARS-CoV-2; PLEX=plasma exchange; SARS-CoV-2=severe acute respiratory coronavirus-2; TOE=transesophageal echocardiogram; U&Es=urea and electrolyte; WCC=white cell count.
*Reports of increased incidence of secondary syphilis and HIV are risk factors for hospital admission/severe covid-19 and increased mortality.179180
†Evidence of reactivation during covid-19 and post vaccination.181
‡Indication includes an unexplained intracranial arterial stenosis.
§To investigate for evidence of a hypercoagulable state

In most cases, including encephalitis, ADEM, and Guillain-Barré syndrome, the recommended management corresponds with pre-covid-19 guidance. For syndromes with multifactorial causes such as encephalopathy, identification and optimization of contributing factors including hypoxia, sepsis, and metabolic disturbances form an essential part of management. Specific treatment guidelines have been required for the syndromes more particular to covid-19 such as hyperinflammation in PIMS-TS.

Emerging treatments

Emerging treatment data largely highlight the secondary benefit of systemic treatment on neurological complications of covid-19. A recent retrospective UK based observational study of 184 986 adult hospital inpatients (January 2020 to June 2021) showed reduction of neurological events (stroke, seizure, meningitis/encephalitis, or other neurological complication) in patients treated with dexamethasone and remdesivir, compared with standard care.182 In patients with severe covid-19, neurological events were reduced in patients treated with dexamethasone (odds ratio 0.76, 95% confidence interval 0.69 to 0.90), remdesivir (0.69, 0.51 to 0.90), and dexamethasone and remdesivir combined (0.54, 0.47 to 0.61). In patients with non-hypoxic covid-19, neurological events were reduced in patients treated with dexamethasone (0.78, 0.62 to 0.97), with a trend towards reduced events with dexamethasone and remdesivir combined (0.63, 0.31 to 1.15). The impact on neurological complications of other drugs recommended by NICE with shown efficacy in covid-19, including tocilizumab and baricitinib, nirmatrelvir, and ritonavir, are not yet reported.183 However, these similarly target either systemic inflammation or viral replication or entry, and thus would be predicted to have a similar effect.

Systemic treatment could either directly or indirectly affect the pathophysiology of neurological complications of covid-19. In subgroup analysis of the above study, incidence of stroke was reduced by use of dexamethasone in both severe and non-hypoxic disease (odds ratio 0.57, 95% confidence interval 0.48 to 0.68 and 0.47, 0.29 to 0.72, respectively), supporting the role of inflammation in stroke related to covid-19.182184 The impact on other presentations including seizures (reduced in patients with severe covid-19 treated with dexamethasone (0.41, 0.33 to 0.52), remdesivir (0.28, 0.1 to 0.61), or both (0.37, 0.27 to 0.51)), could be multifactorial. Meningitis/encephalitis events were also reduced in the severe covid-19 groups with dexamethasone or combination treatment, suggesting a potential benefit of reduction in central nervous system inflammatory events. However, further interrogation of this group with detailed phenotyping data is needed. ADEM and other inflammatory events were studied separately, alongside syndromes not expected to respond to immune treatment.

The success of the vaccine programs in reducing the incidence and severity of covid-19 is well documented. Although adenovirus vector based vaccines have been linked to a small number of neurological complications (cerebral venous sinus thrombosis and Guillain-Barré syndrome), self-controlled studies conclude that the risk of neurological events from SARS-CoV-2 infection far outweighs that following vaccination.121134 Incidentally, following the introduction of vaccination programs and increasing population immunity, the number of cases of acute neurological syndromes, including stroke and ADEM, has fallen at our institution. However, this finding is yet to be corroborated at a population level. A recent systematic review and meta-analysis showed that vaccination reduces risk of developing long covid; patients who had received two doses of vaccination had significantly lower risk compared with those not vaccinated (odds ratio 0.57, 95% confidence interval 0.43 to 0.76).185

Optimization of comorbidities associated with severe covid-19 in reducing risk of neurological complications needs further investigation. In a retrospective cohort study of 10 541 patients admitted to hospital with covid-19 between January and September 2020, outpatient use of statins and hypertensives, or both, reduced the risk of death (adjusted odds ratio 0.68, 95% confidence interval 0.58 to 0.81) or severe disease (0.80, 0.69 to 0.93) in patients with known cardiovascular disease and hypertension, or both.186 Although incidence of neurological events was not explored in this study, covid-19 severity increases risk of neurological complications. The public health need to catch up on cardiovascular risk modifying treatments missed because of pandemic disruption has also been highlighted, to reduce risk of not just severe covid-19, but also excess cerebrovascular events resulting from missed treatment opportunities.187


The Association of British Neurologists and other national and international neurological organizations published regularly updated guidelines throughout the pandemic, for clinicians and patients living with neurological diseases, and their carers.188189 These guidelines included guidance on relative risk of severe covid-19 in neurology patients, shielding and reducing risk of infection, information on potential drug interactions in the context of covid-19 and common neurological conditions, and guidance on immune suppression in patients with neuroinflammatory disorders not related to covid-19. Neurological diseases such as multiple sclerosis, motor neuron disease, myasthenia gravis, and Huntington’s disease have been highlighted as high risk patient groups warranting early covid-19 treatment in the community, following confirmation of a positive test.177

Consensus guidelines have been published for the investigation and management of encephalopathy and PIMS-TS.170178 Encephalopathy guidelines recommended investigating and optimizing secondary causes of encephalopathy or central nervous system pathology including encephalitis, ADEM, and posterior reversible encephalopathy syndrome (PRES), when suggestive clinical features are present.170 Guidelines for PIMS-TS were published in 2020 relating to prescription of steroids, IVIg, and biological treatments, as well as highlighting the need for multidisciplinary input.178 In the UK, NICE has published guidelines on the management of long covid, and also vaccine induced immune thrombocytopenia and thrombosis.139171 More integrated guidelines could emerge as understanding of covid-19 neuropathology evolves. These guidelines are likely to overlap with general preventative and treatment guidelines reducing severity of systemic covid-19, in the expectation that these measures will also reduce neurological burden. The need for increased surveillance and early recognition of neurological injury, especially in the intensive care population, should be highlighted.


The covid-19 pandemic has had seismic global consequences in terms of morbidity and mortality, as well as in overwhelming health systems, even in wealthy settings. Nevertheless, the medical and scientific community has risen to the challenge with impressive results. Although neurological complications in the acute setting are relatively rare in comparison with the total number of infections, the consequences of neurological injury for individual patients are significant. The mechanism of neurovirulence of SARS-CoV-2 appears to be related primarily to immune dysregulation and an immunothrombotic milieu. Treatments therefore need to be focused on this aspect of the pathophysiology. The large numbers of patients now with long covid will be a legacy of the pandemic. Elucidating the mechanism of long covid to manage and treat these patients is vital. Additionally, the unknown long term consequences in relation to neurodegeneration will require longer term study.

Questions for further research

  • Can a granular pathophysiological mechanistic model for both acute and long covid-19 be established?

  • What are the differing risk factors globally for short term and long term neurological complications of covid-19?

  • What are the benefits versus complications of anticoagulation in ischemic stroke associated with covid-19 with consideration of subtypes (ie, large v small arterial strokes) on outcome?

  • What are the optimal treatment strategies for long covid?

  • What are the long term neurological consequences of covid-19 infection?

How patients were involved in the creation of this article

The authors are grateful to a patient who reviewed the manuscript and provided a patient account to show the individual patient burden of neurological symptoms following covid-19 (box 2).

Glossary of abbreviations

  • SARS: severe acute respiratory syndrome

  • MERS: Middle East respiratory syndrome

  • ADEM: acute disseminated encephalomyelitis

  • ECMO: extracorporeal membranous oxygenation

  • LMIC: low and middle income countries

  • ACE2: angiotensin converting enzyme 2

  • TNF-alfa: tumor necrosis factor

  • GFAP: glial fibrillary acidic protein

  • NMDAR: N-methyl-D-aspartate receptor

  • ARDS: acute respiratory distress syndrome

  • sHLH: secondary hemophagocytic lymphocytosis

  • ICANS: immune effector cell associated neurotoxicity syndrome

  • CAR-T: chimeric antigen receptor T cell

  • MIP-1β: macrophage inflammatory protein-1 beta

  • IVIg: intravenous immunoglobulin

  • RPLS: reversible posterior leukoencephalopathy

  • RCVS: reversible cerebral vasoconstriction syndrome

  • T2/FLAIR: T2 weighted/fluid attenuated inversion recovery

  • AHLE: acute hemorrhagic leukoencephalopathy

  • PIMS-TS: pediatric inflammatory multisystem disorder temporally associated with SARS-CoV-2

  • MIS-C: multisystem inflammatory syndrome in children

  • SIP-1: sphingosine 1-phosphate:

  • CIDP: chronic inflammatory demyelinating polyneuropathy

  • MMNCB: multifocal motor neuropathy with conduction block

  • NMOSD: neuromyelitis optic spectrum disorders

  • PRES: posterior reversible encephalopathy syndrome

Ethics statements

Ethical approval

Ethical approval was not required for the review. Figures 1 and 2 are adapted with permission from Paterson, Brown et al, Brain, 2020 (Open Access).


We thank our patient reviewer, who reviewed the manuscript, and contributed a patient account to show the individual patient burden of neurological symptoms following covid-19 (box 2).


  • Series explanation: State of the Art Reviews are commissioned on the basis of their relevance to academics and specialists in the US and internationally. For this reason they are written predominantly by US authors

  • Funding: RB is supported by an MRC clinical research fellowship (MR/T001313/1); LB is supported by a Wellcome Trust fellowship (222102/Z/20/Z); MPL is supported by the UCL/UCLH NHS Foundation Trust Biomedical Research Centre; TB is supported by the University of Oxford and the Medical Research Council (MR/N013468/1). No funding was provided for this project. This review was completed by the authors independently of funding sources.

  • Competing interests: All authors have read and understood the BMJ policy on declaration of interests. No competing interests were declared.

  • Contributors: RB and HM developed the structure for the review and led the manuscript preparation and editing; all authors defined intellectual content, conducted literature review, and participated in manuscript preparation, editing, and critical review. HM is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.

  • Provenance and peer review: commissioned; externally peer reviewed.