On May 20, 2020, Elisa Perego, a researcher at University College London, posted a pair of words on Twitter that would enter the medical lexicon permanently: "long COVID." Perego had tested positive for SARS-CoV-2 in February and had never fully recovered. Weeks, then months after the acute infection, she was still experiencing fatigue, chest pain, shortness of breath, and what patients would later call "brain fog" — a cognitive difficulty that felt like thinking through wet concrete. She was not writing as a physician or virologist; she was writing as a patient who needed language for an experience she and thousands of others were having but that medicine had not yet recognized. The term spread within days across social media platforms. Survivor Corps, Patient-Led Research Collaborative, and dozens of other patient communities formed around it, aggregating symptom data, running surveys, publishing findings, and pushing for research that the mainstream medical establishment was only slowly beginning to pursue.

What followed was a rapid and unusual reversal. The World Health Organization published a formal clinical case definition in October 2021, defining long COVID as a condition occurring in individuals with a history of probable or confirmed SARS-CoV-2 infection, usually three months from onset, with symptoms lasting at least two months and not explained by an alternative diagnosis. The United States National Institutes of Health launched the RECOVER initiative — Researching COVID to Enhance Recovery — in 2021 with $1.15 billion in dedicated funding, the largest research investment in a post-acute infection syndrome in history. Long COVID had moved from a patient hashtag to a formal diagnostic category and a major scientific priority in under two years. The speed of that recognition was remarkable; the years of illness that millions experienced before it arrived were not.

That delay had structural roots. The dismissal of prolonged post-viral illness as psychological, as deconditioning, as anxiety — these were not individual physician failures but institutional habits, reproduced across medical training and reinforced by the absence of biomarkers that could make invisible symptoms legible to laboratory instruments. They were the same habits that had defined medicine's response to myalgic encephalomyelitis/chronic fatigue syndrome for four decades. Long COVID's emergence did not merely add a new condition to the catalog of post-viral illnesses. It cracked open a question medicine had long preferred to leave closed: whether the dismissal of post-infectious syndromes had ever been scientifically justified, or whether it had always been a failure of attention.

"Patients with long COVID have taught us that COVID can be a complex, multisystem disease, and that the acute phase is just the beginning of the story." — Eric Topol, Ground Truths (2022)

Key Definitions

Long COVID — The WHO clinical case definition (October 2021) describes post-COVID condition as occurring in individuals with a history of probable or confirmed SARS-CoV-2 infection, usually three months from the onset of COVID-19, with symptoms lasting at least two months and not explained by an alternative diagnosis. Also referred to as post-acute sequelae of SARS-CoV-2 (PASC) in research literature.

Post-exertional malaise (PEM) — A hallmark symptom in which physical or cognitive exertion produces a disproportionate, delayed worsening of symptoms — a "crash" that may last hours, days, or weeks. PEM is not ordinary fatigue after exercise; it is a pathological response that may worsen all symptoms and requires days of recovery. PEM is a cardinal feature of both long COVID and ME/CFS and is the primary clinical reason that graded exercise therapy is contraindicated in these conditions.

Viral persistence — The hypothesis that SARS-CoV-2 viral components — proteins, RNA, or potentially replication-competent virus — persist in tissue reservoirs beyond the resolution of acute infection, continuing to drive immune activation and inflammation.

Autoantibodies — Antibodies directed against the body's own proteins rather than foreign antigens. Several studies have identified elevated autoantibodies in long COVID patients, including against G-protein-coupled receptors that regulate autonomic nervous system function and other physiological processes.

Microclots — Fibrin-rich, amyloid-like microscopic clots identified in the blood of long COVID patients by Resia Pretorius and colleagues, resistant to normal enzymatic breakdown, hypothesized to impair microcirculation and contribute to tissue oxygen delivery deficits.

POTS (Postural Orthostatic Tachycardia Syndrome) — A form of dysautonomia in which heart rate increases abnormally upon standing, typically by 30 beats per minute or more within ten minutes. Associated with dizziness, fatigue, cognitive impairment, and fainting. Substantially elevated in prevalence among long COVID patients.

ME/CFS (Myalgic Encephalomyelitis/Chronic Fatigue Syndrome) — A disabling multi-system illness with striking symptomatic overlap with long COVID, characterized by profound fatigue not improved by rest, post-exertional malaise, cognitive impairment, and orthostatic intolerance. Estimated to affect 17-24 million people worldwide. Historically dismissed as psychosomatic; long COVID's emergence has substantially changed that scientific and social framing.

NIH RECOVER — Researching COVID to Enhance Recovery. The NIH initiative launched in 2021 with $1.15 billion in funding to study long COVID in adults, children, and through autopsy, representing the most substantial federal investment in post-acute infection research in US history.

Leading Biological Hypotheses for Long COVID Mechanisms

Hypothesis Core mechanism Supporting evidence Open questions
Viral persistence Replication-competent or antigen-shedding virus persists in gut, brain, or other tissue reservoirs beyond apparent clearance SARS-CoV-2 RNA and protein found in intestinal biopsies months post-infection (Zollner et al.); viral spike protein in blood of long COVID patients (Patterson et al.) Whether persistent virus is replication-competent or just debris; why it persists in some but not others
Immune dysregulation Chronic, abnormal immune activation; elevated inflammatory cytokines; dysregulated T and B cell profiles persist beyond acute infection Elevated CCL-11, low serotonin (Pretorius, Patterson); low cortisol (Tronstad); abnormal T cell exhaustion patterns in multiple cohorts Cause vs. consequence unclear; whether dysregulation precedes illness or results from it
Autoantibodies Antibodies directed against the body's own proteins — especially G-protein-coupled receptors regulating the autonomic nervous system Elevated autoantibodies in multiple long COVID cohorts vs. controls; linked to POTS and dysautonomia presentations Causality not established; autoantibodies found in some healthy controls; which are pathogenic?
Microclots Fibrin-rich, amyloid-like microscopic clots in circulation that resist normal enzymatic degradation; impair microcirculation and tissue oxygen delivery Pretorius et al. demonstrated microclots in long COVID blood; platelet activation elevated No randomized treatment evidence; microclots also found in some other conditions; clinical significance debated
Reactivation of latent viruses (EBV, HHV-6) COVID-19 immune suppression triggers reactivation of herpesviruses already present in most adults Elevated EBV antibody titers in some long COVID cohorts; EBV reactivation rate higher than comparators (Bhatt et al.) Whether reactivation is causal or incidental; not found consistently across studies
Mitochondrial dysfunction Impaired cellular energy production leads to exercise intolerance and fatigue independent of oxygen delivery Skeletal muscle biopsies show mitochondrial abnormalities in some patients (Appelman et al., Nature Communications 2024) Sample sizes small; unclear if universal or subtype-specific
Dysautonomia / POTS Autonomic nervous system dysfunction causes heart rate dysregulation, circulation abnormalities, and the full symptom cluster POTS diagnosed at significantly elevated rates post-COVID vs. controls; tilt-table testing abnormal in many patients Mechanisms of dysautonomia unclear; may be downstream of other processes

The WHO Definition and the Challenge of Heterogeneity

The WHO clinical case definition, reached through a Delphi consensus process involving patients, clinicians, and researchers across multiple countries, was deliberately inclusive. The three-month threshold was chosen to capture persistent illness while distinguishing it from the normal recovery period following severe acute infection; the two-month symptom duration requirement was set to exclude transient symptoms. Crucially, the definition requires that symptoms not be explainable by an alternative diagnosis — a criterion that has proven difficult to operationalize uniformly in clinical practice.

More than 200 individual symptoms have been reported in association with long COVID. The most common and disabling include fatigue (often described as unlike any tiredness previously experienced, unrelieved by rest), post-exertional malaise, cognitive impairment or "brain fog," breathlessness, chest pain, palpitations, headache, joint and muscle pain, sleep disturbance, anxiety and depression, gastrointestinal symptoms, and persistent changes in smell or taste. No single biomarker or confirmatory test exists; the diagnosis remains clinical.

Hannah Davis, Gina Assaf, and colleagues in the Patient-Led Research Collaborative published an early systematic symptom analysis in EClinicalMedicine in 2021, drawing on 3,762 respondents. This patient-generated dataset documented the multi-system character of the illness, the persistence of symptoms beyond six months, and the frequency of relapse and remission patterns — findings that preceded much of the formal academic literature. The study identified 203 symptoms across 10 organ systems, with cognitive impairment, post-exertional malaise, and fatigue among the most commonly reported at six-month follow-up.

Michela Antonelli and colleagues at King's College London, analyzing data from the ZOE COVID Symptom Study, proposed in 2022 that at least two to five symptom phenotypes could be distinguished within long COVID — clusters with distinct dominant features that may reflect different underlying biological mechanisms. This heterogeneity argument has become central to research design: clinical trials that fail to stratify by subtype may aggregate populations with different pathologies, diluting any treatment signal.

Prevalence: The Scale of the Problem

Estimating the prevalence of long COVID with precision is complicated by definitional inconsistency, reliance on self-report in many studies, and the changing epidemiology of SARS-CoV-2 variants. The most methodologically rigorous longitudinal surveillance comes from the UK Office for National Statistics COVID-19 Infection Survey, which follows a randomly selected household sample with consistent methodology over time.

The ONS estimated that 1.9 million people in the United Kingdom were experiencing self-reported long COVID as of late 2022 — approximately 2.8% of the UK population. Among those, 71% reported that long COVID adversely affected day-to-day activities, and 20% reported that activities were limited substantially.

In the United States, the CDC Household Pulse Survey reported in 2022 that 19% of US adults who had ever had COVID-19 reported current long COVID symptoms. Because an estimated 60% or more of American adults had experienced COVID-19 by that time, this represented tens of millions of current cases. Separate analysis by Katie Bach at the Brookings Institution in 2022 estimated that approximately 4 million Americans were out of the workforce at any given time due to long COVID — a labor market disruption comparable in scale to a moderate recession, largely absent from economic forecasting models.

Rates differ substantially by exposure context. Vaccination reduces the risk of developing long COVID by approximately 50%, according to meta-analyses including a 2022 Lancet Infectious Diseases study by Agrawal and colleagues analyzing UK healthcare worker cohorts. Omicron-wave infections are associated with lower long COVID rates than Delta-wave infections in matched cohorts, though because Omicron infected far more people, the absolute number of long COVID cases did not decline commensurately. Children have substantially lower long COVID rates than adults, though a non-trivial proportion — estimated at 4-12% depending on definition — experience persistent symptoms following infection.

A consistently surprising epidemiological finding is that long COVID disproportionately affects working-age middle-aged adults rather than the elderly who bore the highest acute mortality. The ONS data showed peak self-reported long COVID rates in the 35-69 age range. Long COVID is not simply a complication of severe acute disease: many patients had only mild initial illness, were never hospitalized, and in some cases were never even tested during their acute phase.

Biological Mechanisms: A Heterogeneous Disease

The most important thing to understand about long COVID biology as of 2025 is that it is almost certainly not one condition with one mechanism. Several distinct pathological processes have been identified with supporting evidence, and different patients likely experience different dominant mechanisms. This heterogeneity is not a defect in the research; it is the finding.

Viral Persistence

The hypothesis that SARS-CoV-2 or its components persist in tissue reservoirs after apparent clearance has accumulated substantial supporting evidence. Steven Peluso and colleagues at the University of California, San Francisco, detected SARS-CoV-2 nucleocapsid antigen and RNA in gut biopsies from long COVID patients months after initial infection, published in Nature in 2023. Viral material has been found in lymph nodes, appendix tissue, and other gut-associated lymphoid tissue even in vaccinated individuals who appeared clinically recovered. Amy Proal and Michael VanElzakker have proposed that reservoir tissue — sites with high ACE2 receptor density and reduced immune surveillance — may harbor replication-competent virus that continuously seeds systemic inflammation.

If viral persistence is confirmed as the primary driver in a substantial long COVID subgroup, it has direct therapeutic implications: antiviral drugs targeting active viral replication would be a rational treatment. The NIH RECOVER-VITAL trial tested this logic by administering nirmatrelvir-ritonavir (Paxlovid) for 15 days to long COVID patients with the hypothesis that clearing persistent virus would resolve symptoms. Results reported in 2024 showed mixed findings — no statistically significant benefit on the primary endpoint across the full study population, though some secondary analyses suggested possible benefit in specific symptom domains. The absence of a clear positive result does not definitively refute the persistence hypothesis: Paxlovid inhibits active viral replication but may not clear established viral protein in tissue reservoirs.

Autoimmunity and Autoantibodies

Multiple research groups have identified elevated autoantibodies in long COVID patients — antibodies targeting the body's own proteins rather than viral antigens. Harald Heidecke and colleagues, as well as Aristo Vojdani's laboratory, have documented elevated autoantibodies against G-protein-coupled receptors in long COVID patients, particularly receptors involved in autonomic nervous system regulation. These autoantibodies could functionally block or dysregulate receptor signaling, producing the autonomic dysfunction — POTS, orthostatic intolerance, abnormal sweating — that characterizes a significant long COVID subgroup.

Akiko Iwasaki's group at Yale University has documented a broader autoantibody landscape, with long COVID patients showing antibodies against interferon pathway components and other immune regulatory proteins. The autoimmunity hypothesis is mechanistically coherent: SARS-CoV-2 infection, like other viral infections, could trigger molecular mimicry (viral peptides resembling host proteins provoking cross-reactive antibodies) or bystander activation (nonspecific immune activation producing autoreactive antibodies). BC007, a compound designed to neutralize GPCR autoantibodies, is in European clinical trials testing this mechanism specifically in long COVID patients with documented autoantibody elevations.

Latent Virus Reactivation

Epstein-Barr virus infects approximately 95% of adults worldwide and persists latently in B lymphocytes indefinitely after primary infection. EBV reactivation — the shift from latency to active viral replication — can be triggered by immune perturbation including other infections. Multiple research groups have found elevated EBV reactivation markers (elevated viral capsid antigen IgG, early antigen IgG) in long COVID patients compared to recovered COVID patients without persistent symptoms and healthy never-infected controls. Research by Jim Bhatt and colleagues found that early EBV reactivation during acute COVID-19 infection was predictive of subsequent long COVID development. EBV reactivation itself can produce fatigue, cognitive symptoms, and immune dysregulation independently of ongoing SARS-CoV-2 activity — suggesting a mechanism by which acute COVID could trigger persistent illness through reactivation of a dormant co-infection.

Microclots and Microvascular Pathology

Resia Pretorius at Stellenbosch University and colleagues have documented a distinctive pathology in long COVID: fibrinogen-rich, amyloid-like microclots detectable by fluorescence microscopy in blood samples from long COVID patients but not healthy controls. These microclots incorporate inflammatory proteins including fibrinogen, alpha-2-antiplasmin, von Willebrand factor, and SARS-CoV-2 spike protein, and resist normal fibrinolytic digestion. The hypothesis is that these microclots impair microcirculation — blood flow through capillaries — producing oxygen delivery deficits in muscles, brain, and other tissues, explaining fatigue, cognitive symptoms, and exercise intolerance without requiring large-vessel thrombosis detectable on standard imaging.

The microclot hypothesis has been partially replicated in independent laboratories and has generated therapeutic investigation: several research groups are studying anti-platelet and anticoagulant approaches. The evidence base remains contested, with some researchers questioning whether the staining technique artifacts or co-morbidities could explain the findings and whether the clots are causal or epiphenomenal.

Neuroinflammation and CNS Involvement

Brain imaging and autopsy studies have provided evidence of neuroinflammation in long COVID patients. Michelle Monje and colleagues at Stanford, in a 2023 Nature paper, found microglial activation — a marker of neuroinflammation — in brain regions associated with cognitive function in long COVID patients with prominent cognitive symptoms, using PET imaging. White matter changes consistent with demyelination or axonal injury have been reported on MRI. Avindra Nath's group at the NIH found microglial nodules and T-cell infiltration in brainstem tissue from long COVID autopsies, providing neuropathological evidence of ongoing central nervous system inflammation.

The neuroinflammation hypothesis explains the "brain fog" symptom cluster — difficulty concentrating, memory failures, word-finding problems, slowed processing — that is among the most disabling and poorly understood aspects of long COVID for many patients.

Autonomic Dysfunction

A substantial proportion of long COVID patients, estimated at 30-50% in some cohorts, meet diagnostic criteria for POTS or other dysautonomias. The autonomic nervous system regulates heart rate, blood pressure, digestion, sweating, and other involuntary functions; dysregulation produces the tachycardia, dizziness, fatigue, and cognitive difficulties characteristic of a significant long COVID subgroup. Proposed mechanisms include autoantibodies against autonomic receptors, direct viral damage to autonomic ganglia, neuroinflammation in autonomic regulatory brain regions, and functional hypovolemia from altered vascular tone.

Risk Factors

Epidemiological analyses have consistently identified female sex as one of the strongest risk factors for long COVID, with women representing 60-70% of cases in most cohorts. This pattern parallels other post-viral syndromes and autoimmune conditions, and may reflect biological differences in immune response, hormonal influences on autoimmune susceptibility, or reporting patterns.

Additional risk factors include higher BMI, pre-existing asthma, type 2 diabetes, and a higher number of acute symptoms during the first week of COVID-19 illness — this last factor independent of illness severity as measured by hospitalization or clinical deterioration. Two immunological markers have shown prospective predictive value: elevated EBV reactivation early in acute COVID infection, identified by multiple research groups as a predictor of subsequent long COVID; and low cortisol levels during acute illness, a finding that recapitulates the HPA axis dysregulation seen in ME/CFS research and suggests a shared neuroendocrine vulnerability.

Vaccination halves risk. Delta variant infection carries higher long COVID risk than Omicron infection in matched cohorts. Middle-aged adults 35-69 are disproportionately affected relative to both the very young and the elderly.

Treatments and the Current State of Evidence

No pharmacological treatment had demonstrated efficacy in a completed Phase 3 randomized controlled trial specifically designed for long COVID as of 2025. This reflects both the time required to design and conduct such trials and the challenge of enrolling a heterogeneous population across a condition with no confirmed biomarker. The NIH RECOVER Clinical Trials network had active or completed trials targeting antiviral therapy, autonomic dysfunction, cognitive rehabilitation, and sleep disturbance.

The absence of proven pharmacological treatment makes current clinical management symptom-directed. For autonomic dysfunction and POTS, salt and fluid loading, compression garments, beta-blockers, and other dysautonomia treatments drawn from the pre-COVID literature provide partial relief. For cognitive symptoms, occupational therapy, cognitive rehabilitation, and accommodating pacing strategies are the primary tools. For sleep disturbance, sleep hygiene and short-term pharmacological support may help. For post-exertional malaise — the most disabling symptom for many patients — pacing is the cornerstone intervention.

Pacing means staying within one's energy envelope, avoiding the boom-and-bust cycle in which feeling relatively well triggers activity that then causes crashes. Heart rate monitoring to stay below the anaerobic threshold is a common practical tool. The principle is that PEM is not a motivational challenge but a physiological signal: exceeding the energy envelope produces real biological harm. This understanding is the direct inverse of graded exercise therapy, which assumed that increasing activity despite symptoms would produce adaptation and recovery. For patients with PEM, GET is not merely ineffective — it is contraindicated and can produce lasting deterioration. This evidence, established in ME/CFS research and now extended to long COVID, was among the most important clinical practice findings to emerge from the long COVID experience.

Several experimental treatments are in investigation or early trials. Low-dose naltrexone, which modulates microglial activity and endorphin signaling at doses far below those used for opioid use disorder, is being studied for potential neuroinflammatory benefit. Fluvoxamine, an SSRI with anti-inflammatory properties beyond serotonin reuptake inhibition — including sigma-1 receptor agonism and anti-cytokine effects — showed early promise and is under investigation. BC007 for GPCR autoantibodies is in European clinical trials. Anticoagulant and antiplatelet approaches targeting the microclot hypothesis are in early investigation. Whether any of these will demonstrate efficacy at Phase 3 scale remains to be determined.

ME/CFS: The Historical Context

To understand the scientific and social significance of long COVID, it is necessary to understand myalgic encephalomyelitis/chronic fatigue syndrome. ME/CFS has been a recognized clinical syndrome since at least the 1980s, with documented case clusters dating to the 1930s. It is characterized by profound fatigue not improved by rest, post-exertional malaise, cognitive impairment, orthostatic intolerance, and unrefreshing sleep — symptoms that are subjective, cannot be confirmed by standard laboratory tests, and respond poorly to treatments premised on a psychosomatic model.

For most of its recognized history, ME/CFS was systematically marginalized by the medical establishment. The dominant UK and US psychiatric tradition classified it as a condition perpetuated by maladaptive illness beliefs, advocated CBT to correct those beliefs and graded exercise therapy to address deconditioning, and repeatedly dismissed patient reports of biological abnormality and GET-induced worsening. NIH funding for ME/CFS research in the late 2010s was approximately $7 per affected patient annually — compared to over $2,000 per patient for HIV/AIDS. The National Academy of Medicine's 2015 report Beyond Myalgic Encephalomyelitis/Chronic Fatigue Syndrome found that ME/CFS patients faced average diagnostic delays of five years and that most physicians had little or no training in the condition.

Long COVID changed this landscape. With COVID-19 as an unambiguous viral precipitant tracked from its moment of onset, the argument that post-infectious illness was psychosomatic became impossible to sustain for the long COVID population. Researchers who had spent decades studying ME/CFS watched the long COVID scientific community independently rediscover the same biological abnormalities — immune dysregulation, autonomic dysfunction, disrupted energy metabolism, neuroinflammation — and observed them receive research funding and mainstream attention within months that ME/CFS research had failed to generate in decades. The overlap between the two conditions is not incidental: many long COVID patients meet formal diagnostic criteria for ME/CFS, and the conditions appear to share biological mechanisms and risk factors.

Anthony Komaroff at Harvard, Lucinda Bateman at the Bateman Horne Center, and others have argued explicitly that long COVID is providing the visibility and resources that ME/CFS research has always needed, and that the two conditions should be studied and treated as closely related post-viral syndromes. The NIH RECOVER initiative has incorporated ME/CFS cohort data and expertise into its research framework. Whether the scientific attention and funding that long COVID has generated will persist long enough to produce treatments that benefit both populations remains the central question.

For related science, see how pandemics spread, how the human immune system works, and how stress damages the body.

References

Tags

long covid COVID-19 post-viral illness immune system chronic illness ME/CFS public health virology

Share this article