Six years after SARS-CoV-2 emerged, researchers are no longer asking whether patients are sick. They are trying to understand why the same infection can leave one person with debilitating fatigue, another with a racing heart, another with neurological problems — and another apparently healthy.
There is a moment in medicine when a disease stops being an argument and becomes a problem.
Long COVID has reached that moment.
For the first years of the pandemic, patients often had to establish the reality of their illness before anyone could investigate its cause. They described exhaustion that did not resemble ordinary tiredness; cognition that seemed to flicker on and off; hearts that raced when they stood; headaches, gastrointestinal disturbances, neuropathic sensations and sleep that no longer restored them. Physical or mental exertion could produce a delayed deterioration that sometimes lasted for days.
The symptoms were real.
What remained uncertain was the biology.
That uncertainty has not disappeared. But the nature of it has changed.
The most consequential Long-COVID studies now point in several directions at once: persistent SARS-CoV-2 material in some tissues; abnormal immune activation; autoantibodies in some patients; vascular and endothelial dysfunction; autonomic nervous-system abnormalities; small-fiber injury; altered metabolism and mitochondrial function; and reactivation of viruses that had been latent in the body before COVID-19.
The important word is some.
The evidence increasingly suggests that Long COVID is not one mechanism producing hundreds of symptoms. It may be several related biological disorders produced by a common initiating infection.
A 2023 Nature study of 275 people with and without Long COVID found differences in immune, hormonal and inflammatory measures and identified distinct immunologic signatures. The investigators themselves framed viral persistence, autoimmunity, dysbiosis, latent-virus reactivation and chronic inflammation as competing or complementary possibilities rather than a settled single cause.
A 2025 analysis of 3,659 participants in the NIH RECOVER cohort subsequently identified eight different longitudinal symptom trajectories, including patients whose symptoms remained persistently high, patients whose symptoms fluctuated, and a group whose symptoms became more prominent only later in the follow-up period.
The clinical heterogeneity is therefore not an inconvenience at the margins of the disease.
It may be one of its defining characteristics.
The patient whose heart could no longer tolerate standing
Consider a person who has recovered from the respiratory infection but discovers that standing has become an ordeal.
When a healthy person rises from a chair, gravity pulls blood into the legs and abdomen. The autonomic nervous system immediately responds. Blood vessels constrict, the heart beats somewhat faster and blood flow to the brain is preserved.
For some people after COVID, that compensation becomes abnormal.
Their heart rate may jump dramatically. They may feel lightheaded, weak, breathless or mentally impaired. Some develop postural orthostatic tachycardia syndrome, or POTS; others have orthostatic intolerance without satisfying conventional criteria for POTS.
The distinction is important.
POTS is a recognizable autonomic disorder. Long COVID is much broader.
Dr. Gemma Lladós, an infectious-disease physician in Spain who has studied autonomic and vagal abnormalities after COVID, has cautioned that nerve dysfunction may explain many cases but almost certainly cannot explain all of them.
That observation encapsulates the emerging problem.
The search for the cause of Long COVID may be asking the wrong question.
A disease of syndromes
The NIH’s enormous RECOVER cohort provides some of the clearest evidence.
In the foundational 2023 JAMA analysis, researchers examined nearly 9,800 adults, including more than 8,600 people with prior COVID-19. Twelve symptoms proved particularly useful for distinguishing Long COVID from prior infection without persistent illness:
- post-exertional malaise;
- fatigue;
- cognitive dysfunction or “brain fog”;
- dizziness;
- gastrointestinal symptoms;
- heart palpitations;
- sexual or reproductive dysfunction;
- altered smell or taste;
- unusual thirst;
- chronic cough;
- chest pain;
- and abnormal movements such as tremor or muscle twitching.
That list is more revealing than it first appears.
It contains symptoms from the brain, cardiovascular system, gastrointestinal tract, respiratory system, endocrine system, sensory system and autonomic nervous system.
The researchers identified four clusters of symptoms that tended to occur together.
Andrea Foulkes, the RECOVER scientist who helped lead the work, summarized the implication succinctly: Long COVID is not simply one syndrome but a “syndrome of syndromes.”
The NIH RECOVER (Researching COVID to Enhance Recovery) Initiative is the largest coordinated U.S. research effort devoted to understanding Long COVID. Launched in 2021, it is not a single study but a research network encompassing longitudinal observational cohorts, electronic-health-record analyses, tissue and autopsy studies, laboratory/pathobiology investigations, and randomized clinical trials.
The core observational studies have followed thousands of adults and children over years, including people who had COVID-19 and comparison groups without prior infection. Researchers repeatedly collect symptom information, clinical measurements and, in selected participants, blood and other biological specimens. This longitudinal design is particularly important because it allows investigators to study who develops Long COVID, which symptoms occur together, how symptoms evolve, and whether distinct biological phenotypes emerge over time.
RECOVER also connects these cohorts with more than 60 million electronic health records and more than 60 pathobiology studies, allowing investigators to examine Long COVID from the population level down to molecular and tissue-level mechanisms. Findings from the observational and laboratory studies are then used to design treatment trials, including trials directed at such problems as viral persistence, autonomic dysfunction, neurological symptoms, sleep disorders and exercise intolerance.
The later longitudinal RECOVER analysis made the point even more strongly. Among 3,659 prospectively followed participants, five percent had persistently high symptom burden, 12 percent had non-resolving but intermittently high symptoms, and another 14 percent developed increasing symptoms later despite not meeting the research threshold at three months.
A disease that changes its phenotype over time is unlikely to have a single static mechanism. (A phenotype is the observable characteristics produced by genes and environment.)
What patients mean by “brain fog”
The term sounds almost trivial.
It is not.
Patients use brain fog to describe difficulty retrieving words, sustaining attention, remembering recently learned information, organizing tasks, following conversations, reading and performing activities that previously required little conscious effort.
One patient may describe losing a word halfway through a sentence.
Another may read the same paragraph repeatedly without retaining it.
Another may know what he wants to say but be unable to retrieve the word.
Some describe losing the thread of a conversation. Others struggle with multitasking, planning or remembering appointments.
And for some, the problem fluctuates dramatically.
A person who appears cognitively normal at breakfast may find himself unable to perform a familiar task later in the day.
The RECOVER studies put this subjective experience into a much larger epidemiological framework: brain fog was among the most characteristic symptoms of Long COVID. In the 2024 RECOVER research index, 63.8 percent of participants classified as having likely Long COVID reported brain fog.
The biology is still being worked out.
Brain imaging studies have reported alterations in cerebral metabolism and structure; studies of cerebrospinal fluid have identified abnormalities in some patients; and investigations of immune and vascular pathways suggest several possible routes by which systemic disease could affect cognition. A major review has catalogued evidence involving neuroinflammation, microvascular dysfunction, altered neurotransmission, autonomic disturbance and viral or immune effects on the nervous system.
But an important recent study illustrates why caution matters. A 2026 Scientific Reports investigation found no significant elevation of neurofilament light chain, a marker of neuronal injury, in its Long-COVID cohort and argued that subjective cognitive impairment need not reflect widespread structural neuronal destruction.
In other words, a patient can have profound cognitive symptoms without having the sort of neuronal injury that conventional blood biomarkers detect.
That distinction is clinically important.
Headaches are not merely headaches
Long-COVID headaches can resemble migraine, tension-type headache or new daily persistent headache.
Some patients experience pressure or throbbing pain.
Others report sensitivity to light or sound, nausea, visual disturbance or worsening after exertion.
Some describe a headache that begins after prolonged concentration rather than physical activity.
The symptom is therefore not necessarily evidence of a single headache disorder.
It may represent one downstream expression of a larger neurological or vascular disturbance.
The same caution applies to vertigo and dizziness.
“Dizziness” can mean several different things: lightheadedness when standing, a spinning sensation, imbalance, visual-motion intolerance or a vague sense that the environment is unstable.
In Long COVID, these may arise from different mechanisms—including autonomic dysfunction, vestibular disturbance, migraine biology, medication effects or cardiovascular changes.
A single symptom label can therefore conceal several physiological problems.
Sleep: exhausted but unable to sleep
Sleep disturbance is another paradox.
Some patients become profoundly sleepy and may sleep 12 or 14 hours yet wake unrefreshed.
Others develop insomnia.
Some have fragmented sleep and frequent awakenings.
Some experience delayed sleep schedules.
Still others report abnormal dream activity or parasomnias.
Sleep can therefore be both excessive and inadequate.
The distinction matters because sleep is not merely a symptom downstream of fatigue. Sleep regulates immune function, autonomic tone, metabolism, memory consolidation and endocrine signaling.
Disordered sleep can consequently become part of a feedback loop:
illness → poor sleep → impaired cognition and autonomic regulation → greater fatigue → greater physiological stress → worse sleep.
NIH’s RECOVER program has treated sleep as a distinct research domain, including trials for insomnia and hypersomnia rather than assuming that all sleep problems are the same.
Fatigue is not simply being tired
This may be the most misunderstood symptom.
Ordinary fatigue improves with rest.
Long-COVID fatigue can be qualitatively different.
Patients describe feeling as though their physical reserves have been radically reduced. Tasks that once required little effort—showering, cooking, shopping, climbing stairs—may become disproportionately demanding.
The most distinctive phenomenon is post-exertional malaise, or PEM. (PEMS Symptoms worsen after exertion, often delayed and lasting days or longer.)
A patient may perform an activity today and feel reasonably well while doing it, only to deteriorate hours later or the following day.
The deterioration may involve:
- exhaustion;
- cognitive dysfunction;
- muscle pain;
- headaches;
- sleep disturbance;
- dizziness;
- flu-like symptoms;
- increased autonomic symptoms;
- and heightened sensitivity to light or sound.
The delay is clinically significant.
It means that simply asking a patient, “How did you feel during exercise?” may miss the central phenomenon.
RECOVER found PEM to be one of the strongest distinguishing symptoms of Long COVID; in its 2024 research index, 87.4 percent of participants classified as likely having Long COVID reported it.
The NIH has also reported that 4.5 percent of post-COVID participants in the RECOVER cohort met criteria for ME/CFS, compared with 0.6 percent of participants without SARS-CoV-2 infection. (ME/CFS is a chronic, disabling illness characterized by profound fatigue, post-exertional malaise, unrefreshing sleep, and often cognitive or autonomic dysfunction.); (Autonomic refers to involuntary body functions such as heart rate and digestion.)
That finding does not mean Long COVID and ME/CFS are identical.
It does mean that their biological and clinical overlap deserves serious investigation.
The exercise paradox
For decades, exercise has been one of medicine’s most universal prescriptions.
That creates an unusual problem when treating patients with PEM.
In ordinary deconditioning, carefully structured physical activity can restore capacity.
In PEM, excessive exertion can trigger a delayed deterioration.
A published first-person account by Kristin Houlihan, a writer and mother with Long COVID and ME/CFS, describes the dilemma particularly well. She reported that even a medical appointment or a few hours out of bed for her child’s birthday could be followed by days of increased pain and cognitive dysfunction.
The distinction between deconditioning and post-exertional physiological intolerance has therefore become a central clinical question.
It is one reason that simplistic prescriptions to “exercise your way out of Long COVID” have become controversial.
A body that forgets how to regulate itself
Autonomic dysfunction offers one possible explanation for the strange combination of cardiovascular, gastrointestinal, neurological and temperature symptoms.
The autonomic nervous system regulates functions that people normally never have to think about:
- heart rate;
- blood pressure;
- sweating;
- pupil size;
- gastrointestinal motility;
- bladder function;
- vascular tone;
- body temperature.
When it malfunctions, the symptoms can seem unrelated.
A patient may have a racing heart after standing, abdominal discomfort after eating, sweating abnormalities, temperature intolerance and cognitive impairment.
To the patient, these can look like five diseases.
To an autonomic physiologist, they may be manifestations of one physiological system.
Dr. Peter Rowe of Johns Hopkins, who has treated POTS and ME/CFS patients for decades, was already seeing the same syndromes in post-COVID patients early in the pandemic. In published reporting, he warned that the health system was poorly prepared for the influx of patients.
The significance of POTS is therefore larger than the syndrome itself.
It provides a demonstration that a symptom that once appeared vague—”I feel terrible when I stand”—can correspond to measurable cardiovascular physiology.
The nerves beneath the nerves
The autonomic nervous system depends partly on small peripheral nerve fibers.
These fibers regulate pain, temperature and vascular function.
Small-fiber neuropathy can produce burning, tingling, altered temperature sensation, abnormal sweating and pain that seems disproportionate to conventional neurological examination.
Some Long-COVID studies have found evidence of small-fiber abnormalities, including altered corneal nerve fibers. Reviews have therefore proposed small-fiber injury as one possible bridge between neuropathic symptoms and autonomic dysfunction.
The hypothesis is attractive because it provides a possible biological link:
small-fiber injury
→ impaired vascular regulation
→ venous pooling
→ reduced effective cardiac filling
→ compensatory tachycardia
→ dizziness, fatigue and cognitive dysfunction.
But again, it is not universal.
Some patients with severe Long COVID do not have demonstrable small-fiber neuropathy.
Some patients with small-fiber abnormalities do not have POTS.
The field is gradually learning that overlap is not identity.
The blood-vessel theory
Another research group has been looking at the endothelium—the cellular lining of blood vessels.
Endothelial cells regulate vascular tone, coagulation, permeability and inflammatory interactions.
Acute COVID-19 clearly affects the vascular system.
The more difficult question is whether abnormalities persist.
Evidence suggests that vascular and endothelial abnormalities occur in subsets of patients. Theories involving microvascular dysfunction and abnormal clotting have received substantial attention, although claims that “microclots” are the universal explanation for Long COVID go beyond the evidence.
This is an important area of disagreement.
Some investigators see abnormal fibrin structures and impaired fibrinolysis as potentially important contributors. (Fibrin is a meshwork to help stop bleeding).
Others argue that the field has sometimes moved too quickly from laboratory observations to sweeping causal claims.
Michael Peluso and Steven Deeks, in a major 2024 Cell review, explicitly describe clotting abnormalities as one possible biological driver but emphasize that upstream biological abnormalities may interact and that no single mechanism has been established as the explanation for the entire syndrome.
That is a recurring theme in Long-COVID research.
The biology is becoming richer faster than it is becoming definitive.
The virus that refuses to leave
Perhaps the most intuitively compelling hypothesis is viral persistence.
If SARS-CoV-2 remains in tissues after the acute infection, perhaps it continues to stimulate the immune system.
There is evidence that SARS-CoV-2 RNA and proteins can persist beyond the period when nasal PCR has become negative.
A 2023 Nature Immunology review summarized tissue studies finding viral RNA or protein in multiple organs weeks or months after infection. One cited autopsy investigation detected viral material in numerous tissues, including parts of the nervous system, months after symptom onset.
A particularly important 2024 Lancet study from China examined tissue samples from 225 people and detected viral RNA in multiple solid tissues months after infection, including liver, kidney, stomach, intestine, brain, blood vessels, lung, skin and thyroid.
Those observations are difficult to dismiss.
But they do not answer the central causal question.
Is the virus still alive?
Is it replicating?
Is it producing biologically active proteins?
Is it causing symptoms?
Or are researchers detecting molecular debris left behind after an infection that has already ended?
Those are very different propositions.
Michael Peluso’s UCSF group has found persistent viral antigen in some people, including individuals with substantial symptom burdens. But preliminary observations also found persistent antigen in some people who had recovered, illustrating why persistence cannot automatically be equated with disease.
That is exactly the kind of finding that should make scientists more—not less—interested.
The immune system may be the second disease
A virus can initiate an illness without being responsible for every subsequent stage.
That possibility has become central to Long-COVID immunology.
In 2022, a landmark Cell study led by researchers including Akiko Iwasaki analyzed 309 people longitudinally and identified four factors present around the time of acute COVID that were associated with later PASC:
- type 2 diabetes;
- SARS-CoV-2 RNAemia;
- evidence of Epstein-Barr virus viremia;
- and particular autoantibodies.
The finding was important because it suggested that Long COVID might be partly determined during the acute infection, rather than arising entirely months later.
But association is not causation.
The same study did not prove that autoantibodies cause Long COVID or that EBV reactivation causes it.
Those questions remain open.
The autoantibody experiment
The autoimmune hypothesis became more compelling when researchers began asking not merely whether antibodies were present, but whether they could actually do something.
A study reported in Nature found that antibodies isolated from some people with Long COVID produced increased pain sensitivity and reduced movement when transferred into mice. The results suggested that antibodies could contribute to symptoms in at least some patients, although the investigators and outside experts emphasized the need for replication and larger studies. (An Antibody is a Biomolecule made by immune cells that recognizes and helps neutralize foreign substances.)
That is a more consequential experiment than simply measuring antibodies in blood.
It asks whether the antibodies are functional.
Akiko Iwasaki and colleagues have argued that autoimmune mechanisms deserve further investigation and that therapies aimed at pathogenic antibodies could eventually become testable. Yale’s report on this work emphasizes that the next step is determining which antibodies are actually pathogenic and whether removing or suppressing them improves symptoms.
But here again there is disagreement.
Not every patient with Long COVID has the same autoantibodies.
Not every autoantibody is pathogenic.
And immune abnormalities may be downstream consequences of persistent viral material or tissue injury.
The immune system could be the cause, the consequence—or both.
The brain is not necessarily injured in one way
Neurological Long COVID has become its own research field.
Symptoms include:
- cognitive slowing;
- difficulty concentrating;
- impaired word retrieval;
- headaches;
- dizziness;
- altered smell and taste;
- sleep disturbance;
- neuropathic pain;
- muscle twitching;
- sensory hypersensitivity;
- and, in some patients, changes in mood or emotional regulation.
These should not be collapsed into a single neurological disease.
A patient with migraine-like headaches may have a different mechanism from a patient with small-fiber neuropathy.
A patient with orthostatic dizziness may have autonomic dysfunction rather than a primary vestibular disorder. (Orthostatic refers to symptoms or changes in the body that occur when a person stands upright.)
A patient with cognitive dysfunction may have altered cerebral blood flow, immune signaling, sleep disruption or another mechanism rather than widespread neuronal death.
This heterogeneity is one reason why the field has increasingly moved toward phenotyping rather than merely counting symptoms.
The gut is part of the story
The gastrointestinal tract is another important clue.
Long COVID can include:
- nausea;
- abdominal pain;
- diarrhea;
- constipation;
- reflux;
- altered appetite;
- bloating;
- impaired motility.
A massive Veterans Affairs analysis involving 154,068 people who survived acute COVID-19 found increased risks of multiple gastrointestinal disorders during the following year, including motility disorders, dyspepsia, reflux, peptic ulcer disease, functional intestinal disorders, pancreatitis and hepatobiliary (Liver) disease.
The finding is epidemiological, not mechanistic.
But the gastrointestinal tract also provides one of the strongest settings in which viral persistence can be investigated.
Researchers have detected SARS-CoV-2 RNA and protein in intestinal tissue months after acute infection, and some studies have reported associations between tissue persistence and persistent symptoms.
The gut may therefore represent both a target organ and a possible reservoir.
The metabolic hypothesis
Then comes one of the most tantalizing questions:
Why does the body sometimes seem unable to produce or utilize energy normally?
Mitochondria convert nutrients into usable cellular energy while simultaneously participating in signaling, oxidative balance and immune regulation.
Long-COVID research has reported changes in metabolic pathways, mitochondrial markers and immune-cell metabolism.
But mitochondrial dysfunction is especially vulnerable to overinterpretation.
A metabolic abnormality may be:
- a primary defect;
- a consequence of inflammation;
- a consequence of inactivity;
- an adaptation to chronic illness;
- or a combination.
The NIH RECOVER program has therefore explicitly made metabolomics and mitochondrial biology a research priority, particularly in relation to post-exertional malaise.
The field is not yet ready to say that mitochondria are “the cause” of Long-COVID fatigue.
But the question has become experimentally approachable.
The strange viruses already inside us
And then there are the viruses that were there before COVID.
Epstein-Barr virus is the most familiar example.
Most adults who have been infected carry it for life.
Usually, the immune system keeps it under control.
A severe infection can perturb that equilibrium.
The 2022 Cell study found EBV viremia among the factors associated with subsequent PASC.
A 2023 NIH/RECOVER review similarly identified latent-virus reactivation as an important research priority while emphasizing that causality remained unresolved.
Then came the new multi-omic evidence.
The new clue: when the other viruses wake up
The August 2026 Nature study of 1,154 hospitalized COVID-19 patients adds another dimension to the story.
Researchers followed patients longitudinally and examined viral RNA alongside immune-cell profiles, cytokines, transcriptomic signatures, proteins and metabolites.
They found evidence of reactivation involving viruses in the Herpesviridae and Anelloviridae families, with associations between reactivation, systemic immune changes and later clinical outcomes.
The significance is not that COVID has somehow turned into another infection.
It is that a major systemic infection may disrupt the immune environment sufficiently to alter the behavior of viruses that have been quietly coexisting with the host.
The crucial scientific qualification is that the study demonstrates association, not proof that viral reactivation causes Long COVID.
That distinction is essential.
Reactivation might contribute to disease.
It might be a marker of severe immune disturbance.
It might occur because the host is already physiologically compromised.
Or all three could occur simultaneously.
The clinicians treating the mystery
While laboratories search for mechanisms, clinicians have to treat patients now.
That often means something less glamorous than a breakthrough drug.
Dr. Jeffrey Parsonnet, an infectious-disease physician working in a Long-COVID clinic, has described treatment as a form of sophisticated trial and error. In published reporting, he emphasized the importance of occupational therapy, psychological support and, above all, having a knowledgeable clinician listen carefully to patients.
Dr. Brad Nieset, who leads a post-COVID recovery program in Montana, has emphasized individualized goals—sometimes as modest as being well enough to sing in church or resume an outdoor activity—rather than chasing every newly published biological hypothesis.
That philosophy can sound surprisingly conservative beside the molecular biology.
But it reflects the current state of the evidence.
Researchers may know considerably more about what might be happening than clinicians know about what will reliably make an individual patient better.
The patient who used to run 15 miles
Lucy Keighley, a former gym owner and personal trainer in England, provides an unusually vivid example of what the syndrome can mean in ordinary life.
Before COVID, she trained intensely and ran long distances. She had completed a 15-mile race and described herself as exceptionally fit.
After infection, her capacity progressively collapsed.
Walking upstairs became exhausting. Showering, cooking and tying her shoelaces became difficult. At one point she slept for more than 15 hours a day. An attempt at graded exercise produced a major deterioration.
Her cognitive symptoms were equally striking.
She described her mind as a “Magna Doodle”: information could be present one moment and disappear the next. She struggled to follow storylines and remember characters in books and television programs.
Her symptoms extended far beyond fatigue:
- breathlessness;
- cough;
- headaches;
- muscle and joint pain;
- swelling;
- sensory hypersensitivity;
- temperature dysregulation;
- itching and rashes;
- gastrointestinal symptoms;
- and cognitive dysfunction.
The collection is almost a parody of a multisystem disease.
Except that it was her life.
The physician who became her own patient
Elizabeth Dawson’s experience illustrates a different phenotype.
She was a dermatologist who had been accustomed to seeing dozens of patients a day, caring for her young daughter and running long distances.
After COVID, her heart raced when she stood.
She developed severe headaches, nausea, profound fatigue and cognitive dysfunction. She could not remain standing for long without becoming dizzy.
She recognized the pattern of POTS herself and eventually received the diagnosis.
Her story is important because the diagnosis was not a vague psychological label.
POTS can be tested.
The patient’s heart rate and blood pressure can be measured supine and then after standing or during tilt-table testing.
That does not prove that POTS explains Long COVID.
It demonstrates something more basic:
at least some apparently mysterious Long-COVID symptoms correspond to measurable physiological abnormalities.
My Own Story
The people who don’t fit the theory
A serious scientific story also needs the patients who don’t fit.
Some people have persistent symptoms without detectable evidence of viral persistence.
Some have no clear autonomic abnormality.
Some have no obvious autoimmune marker.
Some improve spontaneously.
Others remain severely ill for years.
Some patients with persistent viral antigen have recovered.
Some people without detectable antigen remain profoundly symptomatic.
These inconvenient observations are scientifically valuable.
They prevent a plausible hypothesis from becoming a doctrine.
Michael Peluso and Steven Deeks make essentially this point in their Cell review: the mechanisms proposed for Long COVID are not isolated boxes but interacting biological drivers, and evidence for viral persistence is substantial but incomplete and sometimes conflicting.
That is probably closer to the truth than any single grand theory.
The skeptical voice
There is also a broader scientific disagreement about how quickly researchers should move from biological associations to causal claims.
Epidemiologists Tracy Beth Høeg, Shamez Ladhani and Vinay Prasad have argued for particular attention to study quality and methodological rigor in Long-COVID research. Their criticism is not that persistent illness does not exist; it is that conclusions should be anchored in the strongest available comparative evidence rather than in preliminary or highly selected studies.
That skepticism is valuable.
Long COVID is an unusually difficult condition to study because there is no single diagnostic biomarker, symptoms fluctuate, the population is heterogeneous and many symptoms are common in the general population.
A biological abnormality can therefore be seductive.
But seductive is not synonymous with causal.
The same caution applies to viral persistence.
The evidence that SARS-CoV-2 material can persist is real.
The evidence that it persists in every Long-COVID patient is not.
The evidence that persistent material can cause biological effects is compelling enough to justify trials.
The evidence that it explains all Long COVID is absent.
That distinction should remain in every serious account of the disease.
A disease with several clocks
The longitudinal RECOVER data add another surprising dimension.
Long COVID does not simply begin at three months and then remain constant.
Some people improve.
Some remain severely symptomatic.
Some fluctuate.
Some become worse later.
The 2025 trajectory study identified eight distinct symptom courses over 15 months.
That suggests that Long COVID may have several biological clocks.
A patient may begin with acute viral injury.
Then immune activation.
Then autonomic dysfunction.
Then metabolic consequences.
Then partial recovery.
Another patient may have persistent antigen that continues to stimulate immunity.
Another may experience viral reactivation months later.
A third may enter a cycle dominated by PEM and altered energy metabolism.
The clinical label stays the same.
The underlying biology may change.
Why genomics matters
Eventually, genetics may help explain why.
The question is not necessarily whether there is a single Long-COVID gene.
More likely, researchers will identify genetic differences affecting:
- antiviral immune responses;
- inflammatory regulation;
- tissue repair;
- autonomic signaling;
- vascular biology;
- metabolism;
- and susceptibility to autoimmune reactions.
This could explain why two people exposed to the same virus can have radically different outcomes.
One clears the infection and returns to normal.
Another develops an autonomic syndrome.
Another develops ME/CFS-like illness.
Another develops neurological symptoms.
Another develops gastrointestinal disease.
The pathogen is the same.
The host response is not.
The treatment problem
All of this leads to an uncomfortable conclusion.
There may never be one “Long-COVID treatment.”
If the disease consists of multiple biological endotypes, the rational treatment strategy will eventually resemble precision medicine.
A patient with evidence of persistent viral replication might be evaluated for antiviral therapy.
A patient with immune-mediated disease might require an immunologic strategy.
A patient with POTS might need autonomic treatment.
A patient with small-fiber neuropathy might need neurological management.
A patient dominated by PEM may require carefully calibrated energy management and treatment of associated sleep, autonomic and metabolic abnormalities.
A patient with several mechanisms may need several interventions.
The problem is that medicine does not yet have sufficiently validated biomarkers to make this approach routine.
The next generation of Long-COVID research therefore needs to accomplish something more difficult than finding another abnormality.
It needs to link abnormality to patient, symptom, time course and treatment response.
The most important experiment has not yet been completed
Imagine a future clinical trial.
A thousand people with Long COVID are not simply enrolled under one label.
Instead, every participant undergoes:
- immune profiling;
- viral-persistence testing;
- autonomic testing;
- neurological assessment;
- metabolic profiling;
- genetic analysis;
- evaluation for small-fiber dysfunction;
- and longitudinal symptom measurement.
The participants are then divided according to biological phenotype.
A drug is tested only in the subgroup whose biology suggests that it should work.
That trial could succeed where a conventional Long-COVID trial fails.
And if it does, the field will finally have crossed the line from description to mechanism.
The question has changed
The early Long-COVID debate asked whether the symptoms were real.
The current science makes that question increasingly irrelevant.
The more difficult question is what kind of disease Long COVID actually is.
The evidence now supports several serious possibilities:
Persistent SARS-CoV-2 may continue to stimulate tissues or the immune system in some patients.
Immune dysregulation may persist long after the acute infection.
Autoantibodies may be pathogenic in a subset rather than merely incidental.
Endothelial and vascular dysfunction may impair the delivery and regulation of blood flow.
Autonomic dysfunction may explain orthostatic intolerance, palpitations, temperature abnormalities and some gastrointestinal symptoms.
Small-fiber injury may link neurological symptoms to autonomic disturbances.
Metabolic and mitochondrial abnormalities may contribute to impaired exercise tolerance and post-exertional malaise.
Latent-virus reactivation may add another inflammatory stimulus in some patients.
And these mechanisms may not be competing explanations.
They may be interacting pieces of the same biological network.
The network
The simplest old model was:
SARS-CoV-2 → illness → recovery.
The emerging model is more complicated:
SARS-CoV-2 infection
↓
viral replication and tissue injury
↓
immune activation
↓
persistent antigen and/or altered immune regulation
↓
vascular, autonomic, neurological and metabolic dysfunction
↓
altered physiology
↓
possible viral reactivation
↓
persistent disease
But the arrows do not necessarily go in only one direction.
Immune dysfunction can permit viral reactivation.
Viral persistence can stimulate immunity.
Inflammation can alter metabolism.
Metabolic stress can alter immune-cell behavior.
Autonomic dysfunction can influence vascular regulation.
Vascular dysfunction can affect the nervous system.
What looks like six separate diseases may therefore be a network of interacting systems.
The mystery is no longer whether Long COVID is biological
It is how many biological diseases we have been calling Long COVID.
That may ultimately prove to be the most important discovery of all.
The patient with POTS may not have the same biological illness as the patient whose dominant symptom is post-exertional malaise.
The patient with persistent viral antigen may not have the same disease as the patient with an autoimmune neurological phenotype.
The patient with profound cognitive dysfunction may not have the same mechanism as the patient with predominantly gastrointestinal symptoms.
Yet all may have started with SARS-CoV-2.
This is why the phrase “Long COVID” may eventually function less like a diagnosis than like a family name.
The next task is to identify the members of that family.
And once researchers can do that, the question changes one final time.
Not:
“What is Long COVID?”
But:
“Which Long COVID does this patient have—and what can we do about it?”
That is a much harder question.
It is also the first question that offers a realistic path toward precision treatment.
Selected notes and major studies
1. RECOVER symptom definition. Thaweethai T, et al. Development of a Definition of Postacute Sequelae of SARS-CoV-2 Infection. JAMA. 2023. The study analyzed 9,764 adults and identified 12 symptoms that best differentiated PASC from prior infection without persistent disease.
2. RECOVER longitudinal phenotypes. Long COVID trajectories in the prospectively followed RECOVER-Adult US cohort. Nature Communications. 2025. The study followed 3,659 adults and identified eight distinct symptom trajectories over 3–15 months.
3. Su S, et al. Multiple early factors anticipate post-acute COVID-19 sequelae. Cell. 2022;185:881–895.e20. The longitudinal multi-omic study of 309 patients identified type 2 diabetes, SARS-CoV-2 RNAemia, EBV viremia and specific autoantibodies as early factors associated with PASC.
4. Klein J, et al. Distinguishing features of long COVID identified through immune profiling. Nature. 2023. The study used multidimensional immunophenotyping and machine learning in 275 participants to identify immune, inflammatory and hormonal differences associated with Long COVID.
5. Proal AD, VanElzakker MB, et al. SARS-CoV-2 reservoir in post-acute sequelae of COVID-19. Nature Immunology. 2023. The review synthesized evidence for persistent SARS-CoV-2 RNA/protein in tissues and discussed possible links to immune, coagulation, microbiome and neuroimmune abnormalities.
6. Xu E, Xie Y, Al-Aly Z. Long-term gastrointestinal outcomes of COVID-19. Nature Communications. 2023;14:983. The VA cohort included 154,068 COVID-19 survivors and found increased risks of multiple gastrointestinal disorders during the post-acute period.
7. The Chinese tissue-persistence study. A 2024 Lancet study analyzed 317 tissue samples from 225 individuals and detected SARS-CoV-2 RNA in multiple tissues up to four months after infection.
8. Peluso MJ, Deeks SG. Mechanisms of long COVID and the path toward therapeutics. Cell. 2024;187:5500–5529. This review is particularly useful because it treats viral persistence, immune dysregulation, clotting, metabolic changes and other mechanisms as potentially interacting rather than mutually exclusive.
9. Nature Immunology immune study. A longitudinal immunologic study reported persistent inflammatory and immune-exhaustion signatures beyond six months in people with Long COVID. The findings support persistent immune dysregulation but do not establish that inflammation is the universal cause.
10. Autoantibody-transfer study. Research reported in Nature found that antibodies from some people with Long COVID induced pain sensitivity and reduced movement in mice. This is important evidence for possible antibody-mediated pathology but requires replication and does not establish that autoimmunity explains all Long COVID.
11. ME/CFS and Long COVID. NIH’s RECOVER analysis found that 4.5 percent of post-COVID participants met ME/CFS criteria compared with 0.6 percent of uninfected participants, strengthening the case for studying Long COVID alongside other postinfectious illnesses.
12. Viral reactivation. The August 2026 Nature study of 1,154 hospitalized COVID-19 patients used longitudinal multi-omic measurements to investigate reactivation of Herpesviridae and Anelloviridae and associations with subsequent disease. It is an important new piece of evidence, but association should not be confused with proof of causation.
Published patients and clinicians used as narrative sources
The patient material above is based on published accounts:
Elizabeth Dawson, a dermatologist who developed post-COVID POTS, was profiled by TIME and described severe headaches, nausea, orthostatic tachycardia, dizziness and cognitive impairment.
Lucy Keighley, a former gym owner and personal trainer, was extensively profiled by The Guardian after developing severe Long COVID, PEM, cognitive dysfunction, breathlessness and multisystem symptoms.
Kristin Houlihan, a writer and mother, described living with Long COVID, ME/CFS, POTS and sleep disorders in a first-person STAT essay, including the delayed worsening that followed relatively modest activity.
Peter Rowe, Johns Hopkins physician and longtime POTS/ME-CFS researcher, has described the influx of post-COVID patients with these syndromes and the longstanding shortage of clinicians equipped to diagnose and treat them.
Jeffrey Parsonnet, an infectious-disease physician treating Long-COVID patients, has emphasized the present limits of mechanistic certainty and the practical importance of individualized symptom management.
Brad Nieset, who directs a post-COVID recovery program in Montana, has emphasized patient-specific functional goals rather than pursuing every speculative biological mechanism.