Long COVID and Chronic Fatigue

What long COVID actually means, and why the numbers disagree

Long COVID is a name that arrived before an explanation. It was coined by patients, adopted by clinicians, and only later given a formal definition. That order of events explains almost every argument that has followed.

The World Health Organization, the National Academies, and RECOVER define long COVID differently

The World Health Organization built the first widely used definition through a method called a Delphi consensus. A panel of experts and patients answers the same questions in rounds, sees the group's anonymous answers, and revises, until the group converges. A team led by the respiratory physician Joan Soriano published the result. It defines post COVID-19 condition as symptoms usually beginning three months after infection. They must last at least two months and have no alternative diagnosis that explains them.

In 2024 a committee of the National Academies of Sciences, Engineering, and Medicine wrote a broader one. Their working definition of long COVID calls it a chronic condition that follows SARS-CoV-2 infection, is present for at least three months, and can affect any organ system. It can be continuous or it can come and go. It can arrive after an infection so mild the person barely noticed it.

A third approach came from RECOVER, the large American research program. A team led by the biostatistician Tanayott Thaweethai asked a different question. Rather than deciding in advance which symptoms count, they compared nearly ten thousand infected and uninfected people and let the data pick.

Twelve symptoms separated the groups most sharply, and the researchers scored each one and summed them. The list matters, because it is long COVID described from the outside. Post-exertional malaise. Fatigue. Brain fog. Dizziness. Palpitations. Chronic cough. Chest pain. Loss or change of smell and taste. Thirst. Gut symptoms. Change in sexual desire or capacity. Abnormal movements.

How common long COVID actually is

Estimates of how common long COVID is run from 2 percent to more than 30. The spread comes mostly from one methodological choice. Ask people who had COVID whether they still feel tired and a great many say yes. So do a great many people who never had COVID.

The Dutch Lifelines study handled this properly. A team led by the epidemiologist Aranka Ballering tracked symptoms in more than 76,000 people through the pandemic. They had readings from before infection as well as after, and they matched every infected participant to uninfected controls. Once both corrections were applied, the excess attributable to COVID was 12.7 percent. A global modeling effort pooling 1.2 million people put 6.2 percent with at least one of three symptom clusters three months after a symptomatic infection.

Those are the controlled numbers. They are smaller than the headlines and enormous in absolute terms. A wide review of the mechanisms makes the rest clear: no single agreed cause, several plausible ones, and a very large number of people living inside long COVID.

The autonomic nervous system runs the body below the level of decision

Three terms carry every finding in the long COVID literature: the nerve, the autonomic nervous system, and the paired accelerator and brake that the autonomic system runs.

The first is nerve. A nerve is a living wire. It is a bundle of long thin cells that carry information as small electrical pulses, the way a cable carries a signal. A nerve does not pull on anything. Its whole job is to move information from one place to another and to deliver it fast.

The second is the autonomic nervous system. Most of what your body does, you never order. Your heart rate, the width of your blood vessels, your sweating, your digestion, your pupils, your breathing rate at rest. All of it is run by a division of the wiring that operates below the level of decision. Autonomic means self-governing. It keeps the internal conditions of the body in the range where life works, continuously, while you think about something else.

The third is the pair of opposing controls. The autonomic system runs two lines that pull in opposite directions. One is the accelerator, called the sympathetic system. It speeds the heart, tightens the vessels, mobilizes fuel, and braces the body for effort. The other is the brake, called the parasympathetic or vagal system, named for the vagus nerve that carries most of it. It slows the heart, lowers the effort, and permits rest, digestion, and repair.

Tone is the setting the accelerator and the brake hold together, and the range through which they can move that setting when the moment changes.

Hold that definition, because every long COVID finding turns on it. A healthy system has a wide range. It can throw the accelerator down to sprint for a bus and release it fully a minute later. It can hold the brake deep in sleep. It can go from lying flat to standing without you feeling a thing.

Narrow that range and nothing shows up as damage. There is no torn part. Yet the person now spends more to do less. Everything that follows is a description of a range collapsing.

Sickness is a program the nervous system runs on purpose

Almost nothing you feel during an infection is caused by the virus directly. The fever, the aching, the heavy sleepiness, and the fog are outputs your own nervous system produces deliberately.

The veterinary physiologist Benjamin Hart set this out clearly in 1988, when he asked why sick animals across species behave in the same peculiar way. He argued that the whole pattern is an organized survival strategy, not a collapse. Fever runs the body hot because many microbes do badly at that temperature. Sleep and stillness redirect energy to the fight. Appetite loss withholds iron and other nutrients that bacteria need.

The neuroscientist Robert Dantzer and colleagues later described the signaling in detail. Immune cells release messengers called cytokines. The brain reads those messengers through several routes, including nerves running up from the body, and the brain then generates the sickness behavior as a coordinated response. Sickness is a program the nervous system runs. The virus is what triggers it.

If sickness is a program, then a program can fail to switch off. That single sentence carries most of the weight in the long COVID argument.

In most people it does switch off. The infection clears, the cytokines fall, and the nervous system releases the whole arrangement back toward its ordinary setting. The system had a wide enough range to go all the way out and all the way back.

In a minority, the return does not happen. The person tests negative and there is no live infection to find. Months later they are still living inside a version of the sickness state. That is post-viral illness, and it did not begin in 2020.

A century of illness that follows infection

Long COVID felt unprecedented. It was not. Medicine has watched this exact pattern appear after one epidemic after another for more than a hundred years, and has usually forgotten it between outbreaks.

Encephalitis lethargica, the Royal Free, and Q fever

Start around 1917. Physicians across Europe began seeing a strange neurological illness of profound sleepiness, disordered movement, and altered personality, later named encephalitis lethargica. Many survivors never returned to their previous selves. Historians re-examining the record on the centenary of that epidemic report that the cause remains unsettled. The shape is not in doubt: an infectious era, followed by a lasting neurological aftermath.

Move to London, 1955. An outbreak swept the staff of the Royal Free Hospital and closed it for months. Nurses and doctors developed fatigue, muscle pain, and neurological symptoms that lingered long after the acute phase. The hospital's medical staff published their account in 1957. It is a founding document of what would later be called myalgic encephalomyelitis.

Move to the West Midlands of England, 1989, and a large community outbreak of Q fever, an infection caused by the bacterium Coxiella burnetii. The chest physician Jon Ayres followed the infected and compared them with controls. He found a post-infection fatigue syndrome persisting years afterward, with fatigue, sweats, and mood change well above the rate in the uninfected.

The Dubbo study separated the germ from the outcome

Then comes the study that should have settled the argument. In the rural region around Dubbo, in New South Wales, the psychiatrist Ian Hickie and the infectious diseases physician Andrew Lloyd designed a test of a specific question. Does prolonged illness after an infection depend on which germ caused it, or on the personality and psychology of the patient?

They enrolled 253 people at the moment of acute infection with one of three unrelated organisms. Epstein-Barr virus, which causes glandular fever. Coxiella burnetii, which causes Q fever. Ross River virus, a mosquito-borne infection. Everyone was followed for twelve months with interviews, clinical assessment, and laboratory testing.

At six months, 29 of the 253 participants, or 12 percent, had a prolonged illness of disabling fatigue, musculoskeletal pain, neurocognitive difficulty, and mood disturbance. Twenty-eight met the diagnostic criteria for chronic fatigue syndrome. Three findings from that paper deserve to be read slowly.

The syndrome was stereotyped. The same illness, in the same shape, regardless of which of the three organisms caused it. It occurred at a similar rate after each infection, despite a virus, a bacterium, and another virus having almost nothing in common. And it was predicted largely by the severity of the acute illness, and not by demographic, psychological, or microbiological factors.

The specific germ did not determine the outcome. The patient's psychology did not determine the outcome. The intensity of the response the body mounted did.

A coronavirus had already done this

Then 2003, and the first SARS coronavirus. A team in Hong Kong led by the psychiatrist Marco Lam followed survivors for around four years. More than 40 percent reported a chronic fatigue problem, and a substantial subgroup met criteria for chronic fatigue syndrome. A coronavirus had already done this, at scale, seventeen years before 2020.

ME/CFS is diagnosed by pattern, and the pattern is precise

Long COVID has a close relative that has been in the medical literature for decades: myalgic encephalomyelitis, or chronic fatigue syndrome, usually written ME/CFS. Its diagnostic criteria took forty years to converge on the two features where the physiology shows.

The name is unfortunate. Myalgic means muscle pain. Encephalomyelitis means inflammation of the brain and spinal cord, which has never been consistently demonstrated. Chronic fatigue syndrome sounds like ordinary tiredness, which insults the reality of the illness.

The diagnosis is made on a pattern, because there is no blood test for it. That is not a weakness peculiar to this illness. Migraine, most epilepsy, and most psychiatric diagnoses work the same way. A pattern-based diagnosis is still a diagnosis.

Three criteria sets, one direction of travel

The criteria have been rewritten repeatedly, and each rewrite tells you what the field had just learned. In 1994 an international group convened by the epidemiologist Keiji Fukuda published the definition that dominated research for two decades. It required six months of unexplained fatigue plus four of eight additional symptoms. Its weakness was that it treated exertion-related worsening as optional.

In 2011 an international panel led by the physician Bruce Carruthers published the International Consensus Criteria, which dropped the six-month waiting period and put exertion-related worsening at the center of the diagnosis.

Then in 2015 the Institute of Medicine, now part of the National Academies, was commissioned to review the entire evidence base. Their report, Beyond Myalgic Encephalomyelitis/Chronic Fatigue Syndrome, reached a conclusion that changed the conversation. This is a serious, chronic, systemic disease. It is not a psychological condition. And it requires four things for diagnosis: a substantial reduction in function lasting over six months, worsening after exertion, unrefreshing sleep, and either cognitive impairment or symptoms that worsen on standing.

The physician Lucinda Bateman and colleagues later restated how to diagnose and manage the illness for working clinicians. It is the most useful single document for a patient to bring to an appointment.

The last two criteria carry the physiology: worsening after exertion, and worsening on standing. Those two features are where post-viral illness becomes visible and measurable.

Post-exertional malaise is disproportionate, delayed, and prolonged

Post-exertional malaise is the single most misunderstood symptom in medicine, and in the RECOVER analysis of long COVID it was the highest-scoring symptom of all. Understanding it correctly changes everything about how a person should be advised.

Ordinary tiredness is proportional, arrives during or straight after the effort, and improves with rest. Post-exertional malaise breaks all three of those rules.

Disproportionate means the trigger can be small. A shower. A phone call. A trip to a shop. A demanding conversation, because mental effort triggers it as readily as physical effort does.

Delayed means the crash usually arrives many hours later, commonly twelve to seventy-two hours after the exertion. This delay is the reason the symptom is missed so often. The person feels acceptable during the activity and even for the rest of that day. The bill arrives two days later, by which time nobody connects it to the cause.

Prolonged means the crash lasts days, sometimes weeks, and sometimes it does not fully lift. And it is not only fatigue that worsens. Everything worsens. Pain, cognition, sleep, temperature control, orthostatic symptoms, immune-type symptoms such as sore throat and tender glands, all of it, together.

Researchers at the National Institutes of Health ran focus groups with patients to characterize the experience systematically, and three core symptoms emerged consistently: exhaustion, cognitive difficulty, and neuromuscular complaints. The patients described the whole illness worsening as a unit, which is exactly what you would expect if the thing being pushed past its limit is a whole-system regulation rather than a single tissue.

Post-exertional malaise is the feature that ties long COVID to ME/CFS. Everything that follows about exercise, advice, and rehabilitation depends on whether a person has it. This is the fork in the road.

Standing up is a circulatory emergency the baroreflex solves every time

The instant you stand, gravity pulls roughly half a liter of blood down into the veins of your legs and abdomen. Left uncorrected, the pressure feeding your brain would fall until you lost consciousness within seconds. In long COVID, the correction is where trouble shows.

That blood is now below the heart. Less returns to the heart, the heart has less to pump, and the pressure driving blood to your brain begins to fall. You stay conscious because of a loop called the baroreflex, and the loop has three steps.

The baroreflex senses pressure, compares it in the brainstem, and corrects within one heartbeat

Step one is the sensor. In the wall of the carotid arteries in your neck, and in the arch of the aorta above your heart, sit stretch detectors called baroreceptors. When pressure is high the vessel wall stretches and they fire faster. When pressure falls the wall slackens and they fire slower.

Step two is the message and the calculation. Those nerves run up into the brainstem, the stalk where the brain meets the spinal cord, and deliver the pressure reading many times per second. The brainstem compares it with what the moment demands.

Step three is the correction, and it happens within a single heartbeat. The brainstem releases the vagal brake so the heart speeds up. It adds sympathetic accelerator so the heart squeezes harder. It tightens the muscular walls of the vessels in the legs and abdomen so the blood is pushed back up rather than pooling. The physicians Arjan Smit, Wouter Wieling and colleagues laid out the mechanics of this defense, and the failures that follow when the autonomic arm of it stops working.

Orthostatic hypotension, POTS, and neurally mediated syncope are how the standing response fails

Orthostatic hypotension means the pressure itself falls on standing. A joint consensus statement of the autonomic societies sets the threshold. It is a drop of at least 20 points in the top number, or 10 points in the bottom number, within three minutes of standing.

Postural orthostatic tachycardia syndrome, POTS, is the opposite picture. The pressure holds. The heart rate climbs steeply to hold it. The Heart Rhythm Society expert consensus requires a heart rate rise of at least 30 beats per minute on standing, with no fall in blood pressure meeting the hypotension threshold.

The threshold is 40 beats for ages 12 to 19. The earlier international consensus puts the rise within ten minutes and asks for at least six months of symptoms. The full physiology of that syndrome, including its post-viral onset, has its own page in this library.

Neurally mediated syncope, the common faint, is a third pattern. The system holds for a while and then abruptly withdraws its support. Heart rate and pressure fall together and the person passes out.

The tilt table and the lean test

How is any of this measured? The oldest formal method is the tilt table. In 1986 the geriatrician Rose Anne Kenny and the cardiologist Richard Sutton set out to reproduce a faint in the laboratory, in patients whose blackouts had no explanation. They strapped patients to a table and tilted them head-up.

Tilting is the point: it removes the leg muscle pump, so the circulation must hold the pressure on its own. Head-up tilt provoked the faint in a large share of the unexplained cases and in almost none of the healthy comparison subjects. Every tilt table in use today descends from that experiment.

The simpler alternative is the active stand, or the ten-minute lean test. The patient stands with the shoulders resting against a wall while heart rate and pressure are recorded. It is cheap and can be done in any clinic. It is also not identical to tilt. A Vanderbilt group led by Walker Plash compared the two and found tilt produces a larger heart rate rise than standing. A threshold borrowed from one test does not transfer cleanly to the other.

The most common post-COVID picture

This matters for long COVID because the finding is common. A Mayo Clinic group reviewed patients referred for autonomic testing after COVID. In 27 patients they found abnormal sweating function in 36 percent, cardiovagal function in 27 percent, and adrenergic function in 7 percent. The single most common picture, at 41 percent, was orthostatic symptoms with no tachycardia and no hypotension at all.

A prospective tilt study of patients with poor exertional tolerance after COVID found orthostatic intolerance in 23 of 24. Four met criteria for POTS, three had neurocardiogenic syncope and one had orthostatic hypotension. The other fifteen showed intolerance only after nitroglycerin was given to provoke it, which is a drug challenge rather than a plain response to standing.

That most common picture is the one to keep: symptoms on standing, with a normal heart rate and a normal blood pressure. The chart says nothing is wrong. The patient is not able to stand. The brain's own blood supply is where both statements turn out to be true.

Brain blood flow can fall while heart rate and blood pressure stay normal

The brain runs its own blood-flow regulation on top of the systemic circulation, and that second regulator is where a great deal of long COVID lives.

The brain is about two percent of your weight and takes about twenty percent of your blood flow. It stores almost no fuel, and a few seconds of interruption costs you consciousness. So it holds its own flow steady across a wide span of blood pressures by widening and narrowing its own vessels. This is cerebral autoregulation.

Carbon dioxide sets the brain's vessels

The main dial on those vessels is not oxygen. It is carbon dioxide. Carbon dioxide is the waste gas you breathe out, and it is also a powerful signal. More carbon dioxide in the blood widens brain vessels and raises flow. Less narrows them and lowers flow. The physiologists Christopher Willie, Philip Ainslie and colleagues describe how tightly brain blood flow tracks this gas alongside pressure and metabolism.

This has an immediate consequence that most people have never been told. Breathe a little too fast or too deeply for any length of time and you blow off carbon dioxide. The level in your blood falls. Your brain vessels narrow in response. Brain blood flow drops. Your heart rate and blood pressure can be entirely normal throughout.

The Doppler measurements during tilt

Now the measurements. A Dutch cardiology group, C. Linda van Campen and Frans Visser, worked with the Johns Hopkins pediatrician Peter Rowe. They asked whether the brain genuinely receives less blood when these patients stand. They used Doppler ultrasound on the neck arteries, which measures actual flow, in 429 people with ME/CFS and 44 healthy controls during a thirty-minute head-up tilt.

Flow at the end of tilt fell by 7 percent in the healthy controls. In the ME/CFS group it fell by 26 percent. The critical detail is in the subgroups. Among patients whose heart rate and blood pressure responses to tilt were completely normal, flow still fell by 24 percent, and 82 percent of them were outside the normal range. Ninety percent of the whole cohort showed an abnormal fall. The same group later reported the same pattern in long COVID patients with orthostatic symptoms.

The chart was normal. The brain was not getting its blood. Both statements were true at the same time, and only one of them was being measured.

Over-breathing and the underfed brain

The neurologist Peter Novak at Brigham and Women's Hospital pursued the breathing side of this. He had previously described hypocapnic cerebral hypoperfusion, a pattern where low carbon dioxide from over-breathing drives brain blood flow down during upright posture, in patients with orthostatic intolerance and normal conventional tests.

He then applied the full battery to people with fatigue, brain fog, and orthostatic intolerance after a mild COVID infection. Autonomic testing, brain blood flow velocity monitoring, capnography to measure exhaled carbon dioxide, and skin biopsies.

Compared with healthy controls, brain blood flow fell about 20 percent on standing against 3 percent, in patients with and without orthostatic tachycardia. Small fiber neuropathy was present in 89 percent, and carbon dioxide was low both lying and standing. The cohort was nine patients, which is small, and other groups have since reported the same pattern.

So a person with long COVID can have a normal heart rate, a normal blood pressure, and a measurably underfed brain. The brain's supply is set by a second regulator that the standard bedside numbers do not report.

Low blood volume, and the hormones that fail to defend it

Blood pressure is not held by the vessels alone. It is held by how much fluid is in the system, and in post-viral orthostatic intolerance the body's own volume defense reads the shortage wrong.

The defense runs through the kidney, and it is simple to follow. The kidney contains sensors that detect falling pressure and flow. When they do, the kidney releases an enzyme called renin. Renin starts a cascade that produces angiotensin, which tightens vessels, and aldosterone, a hormone that tells the kidney to hold on to salt. Salt holds water. Water is volume. Volume is pressure. This is the renin-angiotensin-aldosterone system, and it is the body's long-term fluid defense.

Many people with orthostatic intolerance are running on low plasma volume. Less fluid in the circuit means less returns to the heart on standing, so the accelerator has to work harder to compensate, which is felt as a racing heart.

The obvious question is why the hormonal defense does not simply fix it. A Vanderbilt group led by the cardiologist Satish Raj measured this and found something odd. In POTS patients with reduced blood volume, renin and aldosterone were inappropriately low for the degree of volume depletion. The system that should have been shouting was quiet. They called it the renin-aldosterone paradox.

That is a regulation failure. The fluid apparatus is intact. The reading of the situation is wrong.

Salt and compression help without fixing

What happens if you supply the volume from outside? A randomized crossover trial led by Emily Garland compared six days of high sodium intake against six days of low sodium in POTS patients. High sodium raised plasma volume, lowered standing norepinephrine, and reduced the heart rate rise. It also did not normalize them. Standing heart rate remained well above healthy controls on the high salt diet.

Mechanical support tells a similar story. A Calgary group led by Kate Bourne tested compression garments during tilt in thirty POTS patients. Full abdominal and lower body compression cut the standing heart rate from 109 to 92 beats per minute and improved symptoms, in a dose-dependent way, by keeping stroke volume up.

Supporting the fluid side of the circulation helps, measurably and reproducibly. It does not restore the system, because the fluid is one coupled part among several, and what is off in long COVID is the arrangement they hold together.

Small fiber nerves, and how they are counted

The thinnest nerves in the body carry pain, temperature, and a large share of the autonomic traffic, and in a documented fraction of long COVID patients they are measurably reduced. This is an objective physical finding in an illness patients are routinely told is not physical.

Not all nerves are the same size. Large nerve fibers are wrapped in a fatty insulating sheath and carry fast traffic: touch, vibration, position sense, and the commands that move muscle. These are the fibers a standard nerve conduction study measures, which is why that test can come back perfectly normal in a person who is clearly unwell.

Small fibers are the thinnest in the body, thinly insulated or bare. They carry pain and temperature. They also carry a large share of the autonomic traffic. That includes the nerves that make you sweat. It also includes the nerves that set the tension in the smallest blood vessels, which decide whether blood pools in your legs or is pushed back to your heart.

Losing small fibers therefore produces a strange combination. Burning or odd sensation, poor temperature regulation, abnormal sweating, and orthostatic symptoms, all with a normal standard neurological examination.

The skin biopsy, and what it found after COVID

They are counted with a punch skin biopsy. A circle of skin about three millimeters across is removed under local anesthetic, usually from the lower leg. The tissue is stained so the nerve endings show, and a pathologist counts how many fibers cross into the outer layer of skin per millimeter. That count is compared against age-matched and sex-matched normal values. European neurological societies published the guideline that standardized this procedure, which is why counts from different laboratories can be compared at all.

The neurologist Anne Louise Oaklander at Massachusetts General Hospital applied this to patients with long COVID referred for neuropathy evaluation. Among seventeen patients with no prior neuropathy history or risk factors, 59 percent had at least one test confirming neuropathy, including ten of sixteen skin biopsies.

Symptoms had usually begun within a month of the infection. Over an average of 1.4 years of follow-up the group improved by about half on average, and none reported complete resolution. Eleven of the seventeen received immunotherapy during that stretch, so the improvement is not a clean picture of natural history.

Small fiber loss is present in only some long COVID patients. The model expects that unevenness: a disturbance spread across coupled channels is loudest in a different channel in each person.

The vagus mostly listens, and heart rate variability reads one part of what it runs

Most vagal fibers run upward, carrying information from the organs to the brain rather than commands downward. The brain holds a continuous readout of the heart, the lungs, the gut, and the immune system, and regulation requires that information first.

The vagus nerve is usually described as the body's brake, which is true and incomplete. The more useful fact is that the vagus is mostly a listening cable. The organ speaks as much as it is spoken to.

The inflammatory reflex

The neurosurgeon and immunologist Kevin Tracey showed that this listening has teeth. He described the inflammatory reflex: the nervous system detects inflammatory signals through vagal afferents, and vagal outflow then damps the production of inflammatory cytokines directly. Inflammation is under neural control, in a loop, like blood pressure.

The human test came later. A team led by Frieda Koopman implanted vagus nerve stimulators in patients with rheumatoid arthritis, a disease driven by inflammatory cytokines. Stimulation reduced cytokine production and lowered disease activity. That was a small, open-label study, and it demonstrated the principle in living people: change the neural signal, and the immune output changes.

What heart rate variability can and cannot say

Heart rate variability is the number a smartwatch offers for all of this. Your heart does not beat like a metronome. The gap between beats varies slightly from beat to beat. That variation is heart rate variability, and its fast component is produced mainly by the vagus, which can adjust the timing of the very next beat.

A joint Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology published the measurement standards in 1996. Those standards still define how the metrics are calculated and what they may be said to represent.

The boundary is precise. Heart rate variability is a validated index of autonomic state. That is established science. Reading it as a window on tone is this model's interpretation, and it is not an established equivalence. Anyone who tells you that your heart rate variability is your nervous system is skipping a step. It is one measurable output of the arrangement this model calls tone.

The practical limits go further. Consumer wearables estimate the variability, usually from light absorption at the wrist rather than the electrical signal of the heart. The numbers move with posture, breathing, alcohol, illness, sleep, and the manufacturer's algorithm. Comparing your figure with a friend's is close to meaningless. Comparing it with your own baseline over weeks is defensible.

And the research in long COVID is genuinely mixed. A systematic review of heart rate variability after COVID found diverse and inconsistent effects on autonomic balance across studies, varying with severity, timing, and method. Measured through one channel, a disturbance spread across many channels produces exactly that: a real effect, inconsistently sized.

What the exercise test measures, and what happened on day two

The deconditioning argument about long COVID has a test that settles a great deal of it, and the test has been run. A deconditioned but otherwise healthy person reproduces their exercise result the next day. These patients did not.

Start with the test itself. A cardiopulmonary exercise test puts a person on a bike or treadmill with a mask that measures every breath. Two numbers matter here.

The first is peak oxygen uptake, VO2 peak. It is the maximum rate at which your body can take in oxygen and use it during hard effort, and it is the standard measure of aerobic capacity.

The second is the ventilatory threshold, sometimes called the anaerobic threshold. Below it you are working mostly aerobically and could keep going. Above it, lactate accumulates, breathing climbs steeply, and the clock starts. It marks where sustainable effort ends, which describes daily life better than the peak does.

The deconditioning hypothesis, tested a day apart

The deconditioning hypothesis starts from a real fact: bed rest wrecks aerobic capacity. Studies of bed rest show substantial falls in maximal oxygen uptake within weeks, along with reduced blood volume and cardiac changes. A person who has been ill and inactive for months will be deconditioned. That is real and it is part of the picture.

The question is whether it is the whole picture. There is a clean way to test that, and it turns on a simple fact: fitness does not fall overnight.

So investigators ran the test twice, twenty-four hours apart. In a study led by the exercise physiologist Betsy Keller, twenty-two ME/CFS patients did two maximal tests a day apart. On the second test peak oxygen uptake fell by 13.8 percent and work at the ventilatory threshold fell by 21.3 percent. The respiratory exchange ratio confirmed maximal effort on both days, so this was not reduced trying. Healthy and deconditioned populations do not do this.

Their sustainable work capacity dropped by a fifth, one day after a single bout of exercise. That is post-exertional malaise, rendered as a number.

The same protocol has been tried in long COVID, and there the picture differs. Fifteen long COVID patients, four in five of whom met criteria for post-exertional malaise, did the two-day test at Harbor-UCLA. The investigators found no difference between day one and day two in peak oxygen uptake, peak work rate or the gas exchange threshold.

They concluded that post-exertional malaise in this group is not explained by a failure to recover exercise capacity overnight. That is one small study, and so far it is the only published two-day cardiopulmonary exercise test in long COVID. The day-two drop is established in ME/CFS and unconfirmed in long COVID, and the rest of the case does not rest on it.

Oxygen delivered and not extracted

An invasive version of the test goes further. In invasive cardiopulmonary exercise testing, catheters are placed in the arteries and the veins. That lets researchers measure the pressure filling the heart. It also lets them measure how much oxygen the muscle actually pulls out of the blood as it passes.

A Yale group led by the physicians Inderjit Singh and Phillip Joseph ran this in ten patients with exercise intolerance after COVID and no heart or lung disease. Peak oxygen uptake was markedly reduced and cardiac output was preserved. What was abnormal was systemic oxygen extraction. The blood reached the muscle carrying oxygen and came back still carrying it. Ventilation was also excessive for the carbon dioxide produced, with no dead space to explain it.

The same group then compared ME/CFS and post-COVID patients directly and reported that the two share more physiology than not, including impaired oxygen extraction and low filling pressures.

And at the tissue level, a Dutch team led by the physiologist Rob Wüst biopsied muscle before and after an exercise challenge that induced post-exertional malaise. They found worsened muscle abnormalities after the crash, including metabolic disturbance and severe exercise-induced muscle damage. They also found amyloid-containing deposits in the muscle. Those deposits increased after exercise in the healthy comparison group too, so that particular finding is not specific to long COVID.

So the deconditioning hypothesis survives as a contributor and fails as an explanation. Deconditioning does not leave oxygen unextracted in the blood. It does not damage muscle after a modest effort. And in ME/CFS it does not fall apart on day two.

The exercise guidance that harmed people, and why it changed

For roughly two decades the standard advice for chronic fatigue syndrome was graded exercise therapy: activity rising on a fixed schedule regardless of symptoms. In 2021 the body that recommended it withdrew it.

The reasoning behind graded exercise was internally consistent. Assume the acute illness has resolved. Assume the disability comes from deconditioning and from a fear of activity that keeps the patient inactive. It follows that activity should rise on a fixed schedule, and that symptoms are signals to work through rather than to respect.

PACE, and what the reanalysis found

The largest trial of that approach was PACE, published in 2011 and led by the psychiatrist Peter White. It randomized 641 patients to graded exercise therapy, cognitive behavioral therapy, adaptive pacing, or specialist medical care alone. The published reports concluded that graded exercise and cognitive behavioral therapy moderately improved outcomes. Guidelines around the world adopted the conclusion.

Two problems then emerged. The trial's outcome definitions were changed after the protocol was published, in a direction that made improvement and recovery easier to reach. And the primary outcomes were self-reported questionnaires in a trial where patients knew which treatment they had received.

A reanalysis by the psychologist Carolyn Wilshire and colleagues, using data obtained through a freedom of information request, applied the originally specified procedures. On the protocol-specified outcomes, the exercise and therapy groups did not significantly outperform the control group after correcting for the trial's own planned comparisons. Recovery rates were low and not significantly different across groups. The effects that remained were confined almost entirely to self-report, and they did not last beyond two years.

Meanwhile patients had been reporting harm for years, and were dismissed for years. The two-day exercise physiology explains what was happening. If one maximal effort drops sustainable work capacity by a fifth the next day, a program that adds load weekly regardless of symptoms is not building fitness. It is pushing a system past a limit it has, repeatedly.

Pacing replaced graded exercise

In 2021 the United Kingdom's National Institute for Health and Care Excellence rewrote its guideline. NICE guideline NG206 states that people with ME/CFS should not be offered any program based on fixed incremental increases in physical activity, and should not be offered graded exercise therapy. It also directs clinicians to help patients stay within their energy limits.

The old advice was wrong for this population, it harmed people, and it has been withdrawn by the body that once recommended it.

What replaced it is pacing. Pacing means identifying the ceiling above which the crash reliably occurs, and then living below it deliberately. Activity is split into short blocks. Rest comes before exhaustion rather than after it. Cognitive and emotional effort count as real expenditure. A consensus document written by clinicians and researchers including Ellen Goudsmit sets out what pacing is and is not. Pacing does not cure the illness. It prevents the repeated crashes that keep entrenching it.

One boundary matters. Not everyone with post-COVID symptoms has post-exertional malaise. People whose limitation is deconditioning, lung damage, or orthostatic intolerance without the crash may do well with supervised, symptom-titrated activity. Screening for post-exertional malaise before prescribing exercise is the step that separates helpful care from harm.

A findable cause must still be found, every time

Fatigue, exercise intolerance, and cognitive difficulty are the least specific symptoms in medicine. They are produced by many treatable conditions that will be missed if everything is filed under long COVID.

The clinical guidance summarized by Bateman and colleagues is explicit: a diagnosis of exclusion means the exclusions have to be done.

Post-COVID organ damage is a separate category and a genuine one. Some people have measurable injury to the lungs, the heart, the kidneys, or the blood vessels from the acute infection. That is not a regulation problem and it needs its own investigation and its own care.

None of this is a reason to skip a workup. The argument here begins after the workup has been done and has come back clean, which is where a very large number of people are currently standing with no explanation at all.

One infection, many outcomes: input meets tone

Separately, the long COVID findings look like a pile of unrelated abnormalities. Assembled, they look like one thing: a single organization disturbed everywhere it can be measured.

Set the measured findings side by side. Brain blood flow falling on standing, sometimes with a normal heart rate and a normal blood pressure. Carbon dioxide too low from a breathing pattern that is over-driven. Low plasma volume with the hormones that defend volume inappropriately quiet. Small fiber nerves reduced in some patients. Oxygen delivered to the muscle and not extracted. Muscle that deteriorates after exertion. Heart rate variability shifted, inconsistently, across studies.

Medicine's instinct is to ask which one of those is the cause. That question has been asked for seventy years across every post-viral outbreak on record, and it has never produced an answer, which is itself information.

The Unified Model of Tone answers differently. It reads them as one disturbance appearing at once in every medium the nervous system regulates through: electrical, neural, vascular, fluid, chemical, mechanical. The same organizing information rides all of them and is held in step. What you measure in any one channel is a chord sounded by all the coupled voices together, never a note one channel plays alone.

Health is the width of the range the whole coupled arrangement can move through. Post-viral illness is that width collapsed, which is why every channel reads slightly wrong and no channel reads catastrophically wrong.

Why the tests come back normal

Every test in the standard panel is built to find damage in a part, and no part is damaged in long COVID. The organization across the parts has narrowed, and no assay measures organization. Tone within its healthy range is health, because it keeps the flexibility to adapt. Tone that drifts or distorts outside that range is what manifests as illness, and this illness is the drift with the lesion removed.

Why the same syndrome follows different organisms

Hickie's Dubbo cohort found a stereotyped illness after a virus, a bacterium, and a second unrelated virus, at similar rates. The model reads this directly. The germ is the input. The syndrome is the state the system settled into after the input. Different inputs, one system, one characteristic failure mode.

Why severity predicted who stayed ill

Severity of the acute illness predicted prolonged illness in Dubbo, while psychology and microbiology did not. Severity is the size of the response the nervous system mounted. A larger excursion out of the normal range demands a larger return, and the return is what fails.

Why only some people are affected

This is the model's central claim. An infection is an input, and an input never has a fixed effect. It meets a nervous system that already had a tone, wide or narrow, and the outcome is a product of the two.

The same viral load meets one person with a wide range and produces two weeks of illness and a full return. It meets another whose range was already narrowed, by chronic sympathetic load, poor sleep, prior injury, or unrelieved stress. That system cannot make the return. It settles into the sickness program, and the program becomes the new setting.

Why the research is messy

Half the studies find reduced heart rate variability and half find something else. Some patients have small fiber loss and some do not. Salt helps and does not fix. Compression helps and does not fix. The lesion model must treat these as contradictions, or dismiss them as poor studies.

The model predicts them. If the disturbance is spread across a coupled arrangement, it will be loudest in a different voice in each patient. In one it is the vascular voice. In another it is the fluid, or the small fiber nerves, or the breathing.

Study a mixed population through a single channel and you will get exactly what the literature contains: a real effect, inconsistently sized, in some of the people, some of the time. Support one voice and the system improves partially, then routes around your support, because the other coupled voices are still holding the narrowed arrangement in place.

This is where an idiopathic illness stops being idiopathic. Nothing about post-viral illness is unexplained under this reading. It is a regulation that went out to meet a threat and did not come back, in a system that was already closer to its edge than anyone had measured.

Restoring versus masking in long COVID, and how to tell them apart

The Unified Model of Tone makes one prediction about long COVID that no drug makes, and it separates an input that restores regulation from one that manages a number. Existing instruments run the test.

The difference the model insists on comes first. Masking an output means acting on the number the system is producing. Restoring tone means widening the range the system can produce numbers within.

Two POTS trials draw the line

Two POTS trials make the difference concrete, and both are legitimate medicine. Ivabradine slows the heart directly by acting on the pacemaker current. A randomized crossover trial led by the cardiologist Pam Taub found it lowered standing heart rate and improved quality of life in hyperadrenergic POTS. Compression garments, as above, cut the standing heart rate mechanically.

Both help, and both push in one direction. Take the drug away and the number returns, because the arrangement that produced it was never addressed. For a person who cannot get through a working day, that is worth having.

Compare a Texas study led by the physiologist Qi Fu. Twenty-seven POTS patients were assessed with autonomic testing, cardiac MRI, and blood volume measurement. Nineteen completed a three-month recumbent-first exercise program designed to avoid upright stress early on. Left ventricular mass and blood volume rose, standing heart rate fell, quality of life improved, and ten of the nineteen no longer met POTS criteria. The boundary on that result matters: this was POTS, and the population was not screened for post-exertional malaise.

Something structurally similar shows up in breathing work. Heart rate variability biofeedback trains a person to breathe at the frequency where the baroreflex loop resonates. The psychologists Paul Lehrer and Richard Gevirtz describe how it appears to work by exercising the reflex itself rather than by forcing a value.

That is the shape of a restorative input. It gives the regulator its range back, and the numbers follow.

The bidirectional prediction

The model stakes itself on one prediction.

A genuine restoration of tone will move a dysregulated value toward the healthy middle from either side. What is too high should trend down. What is too low should trend up. In the same population, under the same input.

A drug cannot do this, and does not pretend to. A beta blocker lowers heart rate in the person whose rate is too high and lowers it in the person whose rate is already too low. That is what one-directional means. If an input genuinely restores the regulator rather than overriding it, the direction of change has to depend on where the person started.

This is testable today. Take a population after infection. Measure resting heart rate, standing heart rate rise, brain blood flow on tilt, end-tidal carbon dioxide, plasma volume, and heart rate variability. Split each measure into those running high and those running low. Apply the input. Then report the two groups separately rather than pooling them, which is the step trial design usually skips.

The model predicts convergence toward the middle in both directions, with the largest movement in whoever started furthest out. If instead every participant moves the same direction regardless of baseline, the input is masking rather than restoring. It helps whichever group it happens to point at and carries the other group further from the middle. That reading applies to any input, including anything offered under the banner of restoring tone.

What can be measured today

More can be measured in long COVID today than most people are told. The standing heart rate rise over ten minutes. The blood pressure response to standing. End-tidal carbon dioxide, if the clinic has capnography. Brain blood flow on tilt with Doppler, in centers that do it.

Skin biopsy fiber counts where neuropathy is suspected. Two-day exercise testing, with the warning that it can provoke a severe and prolonged crash. And at home, a heart rate variability trend read against your own baseline over weeks, understood as one output of the arrangement rather than the arrangement itself.

None of that is a cure. There is no proven cure for post-viral illness today. What exists is a coherent reading of why the tests are clean and why the illness is real. It explains why the same shape follows so many different infections. And it explains why the answer was never going to be found by looking harder at any single part.

How long COVID relates to the rest of the library

Long COVID is the library's clearest case of one input meeting many differently organized systems, and each neighboring page carries one part of that claim.

Three foundations of tone do the heaviest work here.

• Load is the cost of holding a state, and the Dubbo finding that acute severity predicted chronicity is load written as an outcome.
• Coupling is whether separate systems stay in step, and an underfed brain above a normal blood pressure is coupling measurably broken.
• Time course is how a problem changes with age, and a two-week sickness program still running at month six is its central exhibit.
• The sickness state that becomes the new defended setting belongs to set point, and the over-driven breathing that narrows the brain's vessels belongs to gain.

The condition pages divide the territory.

• POTS owns the standing heart rate in depth: why the 30-beat criterion is a measurement of coordination, and what each treatment class actually does.
• Dysautonomia carries the full anatomy of the baroreflex and the autonomic testing battery that post-COVID clinics now run.
• Fibromyalgia is the neighboring pattern diagnosis where pain rather than fatigue leads, with the same clean scans.
• Unexplained symptoms states the general case: a regulation failure leaves no lesion, so the workup comes back clean while the person stays ill.
• Sleep carries the unrefreshing-sleep criterion of ME/CFS: sleep that arrives, lasts a full night, and restores nothing.

Two more pages complete the instrument set. Heart rate variability is the instrument behind the wearable numbers, and its page explains why the post-COVID literature disagrees with itself. The vagus nerve carries the inflammatory reflex in full: the loop through which the nervous system reads and damps the immune response that starts the whole story.

Frequently asked