Part 6 of the series “When Bioenergetic Interventions Work — and When They Don’t”
Why the same diagnosis can conceal completely different states of physiological load and regulation — and why that may help determine the response to a hypoxic stimulus.
Few areas illustrate the central idea of this entire series as clearly as Long and Post-COVID.
Because experiences with intermittent hypoxia could hardly be more contradictory.
On one side are people who report improved exercise tolerance, reduced fatigue, and, after a series of sessions, the sense that their body is finally responding to training again rather than breaking down under it.
And we now have initial clinical data supporting the potential of this approach.
In a controlled clinical pilot trial published in 2024, 145 patients with Long COVID were studied during inpatient multidisciplinary rehabilitation. The group that received additional Intermittent Hypoxic-Hyperoxic Training (IHHT) showed greater improvements than the standard rehabilitation group in physical performance and several patient-reported outcomes. IHHT was well tolerated in the study setting. The trial was prospective and controlled, but open-label and non-randomized.
On the other side are people who report feeling not stronger, but more exhausted after hypoxic interventions — people whose symptoms worsen or whose bodies appear to require considerably more time to recover from a single session.
And this is exactly where two opposing camps tend to emerge.
One says:
IHHT works for Long COVID.
The other says:
IHHT does not work at all — or it only made things worse.
Based on what we currently know about Long COVID and intermittent hypoxia, both conclusions are too simplistic.
Because the fundamental problem may already lie in the question itself.
We ask:
“Does intermittent hypoxic training work for Long COVID?”
Perhaps the more useful question is:
“In which Long COVID phenotype, at what level of physiological regulation and exercise tolerance, and at what dose can intermittent hypoxia represent a meaningful adaptive stimulus?”
That is an entirely different perspective.
And it may explain at least part of the apparent contradiction.
The Diagnosis May Be the Same — the Biology Does Not Have to Be
Long COVID is not a single clinical manifestation with one uniform physiological mechanism.
Under the same diagnostic label, we find people with fatigue, exercise intolerance, shortness of breath, cognitive symptoms, sleep disturbances, palpitations, orthostatic problems, and very different combinations of these symptoms.
The mechanisms that have been proposed — and in some cases demonstrated — are equally diverse. They include alterations in autonomic regulation, vascular and endothelial dysfunction, immune abnormalities, disturbances in microcirculation, and abnormalities in peripheral oxygen extraction.
Systematic analyses of cardiopulmonary exercise testing (CPET) data also suggest that reduced exercise capacity in Long COVID cannot be explained by deconditioning alone. Reported abnormalities include chronotropic incompetence, dysfunctional breathing, and altered peripheral oxygen extraction or utilization.
At the same time, the evidence is heterogeneous, and these mechanisms are not found in every patient. A systematic review and meta-analysis of 38 studies involving 2,160 participants illustrates exactly this complexity.
What does that mean?
Two people may both say:
“Since COVID, I’m exhausted and I can’t tolerate activity anymore.”
And yet completely different things may be happening inside their bodies.
Imagine two houses in which the lights suddenly begin to flicker at night.
From the outside, the problem looks identical.
In one house, the electrical grid can no longer supply sufficient power.
In the other, the wiring itself is damaged.
The fact that the lights flicker in both houses does not mean they need the same repair.
A symptom tells us what we observe.
It does not automatically tell us why it is happening.
And that distinction is crucial when applying bioenergetic interventions in Long COVID.
Fatigue Is Not the Same as Exercise Intolerance — and Exercise Intolerance Is Not the Same as PEM
One of the most important distinctions involves the response to exertion.
A person may have become severely deconditioned after a prolonged illness. Movement feels difficult, physical performance has declined, muscles have become weaker — but the body can still respond adaptively to carefully progressive activity.
Another person experiences something entirely different.
They may initially tolerate the activity surprisingly well.
The real problem comes later.
Minutes or hours afterward — sometimes not until the following day — fatigue, pain, cognitive symptoms, sleep disturbances, or other symptoms become markedly worse.
This deterioration may persist and can be disproportionate to the preceding activity.
This is the pattern of post-exertional malaise (PEM), also referred to as post-exertional symptom exacerbation (PESE).
And in Long COVID, it is far from a rare side issue.
A 2024 systematic review and meta-analysis including 12 studies and 2,665 individuals estimated a pooled PEM prevalence of approximately 55 percent — although there was considerable variability between studies and differences in how PEM was measured.
That does not mean that 55 percent of all people with Long COVID necessarily have the same ME/CFS-like disease pattern.
But it does mean something very important:
Before adding another adaptive stimulus, we need to understand how that person responds to exertion — not merely how much exertion they can tolerate while it is happening.
That distinction matters enormously.
Why This Matters So Much for Intermittent Hypoxia
In previous parts of this series, we explored in detail what intermittent hypoxia actually represents: a controlled physiological challenge.
That is precisely why it can promote adaptation.
But it is also precisely why it is not physiologically neutral.
And this is where Long COVID becomes particularly interesting.
For a person whose reduced exercise capacity is primarily related to deconditioning, whose autonomic regulation remains sufficiently flexible, and who recovers appropriately after a controlled stimulus, carefully dosed hypoxia may have a very different biological meaning than it does for someone with pronounced post-exertional symptom exacerbation.
The mask is the same.
The device is the same.
The protocol may even be the same.
The organism is not.
Think of two hikers standing in front of the same mountain.
One has not trained for several months. They are out of shape, but physiologically capable of rebuilding capacity. If we choose the route carefully, the mountain can become their training.
The other person is standing in front of the same mountain, but their body responds to exertion according to different rules. What appears manageable during the climb may trigger significant deterioration hours later.
For that person, it is not enough to ask whether they can make it up the mountain.
We need to know what the climb does to their system afterward.
Giving both people the same training plan simply because they are standing in front of the same mountain would not be individualization.
It would be the opposite.
Oxygen Availability Is Not the Same as Oxygen Utilization
With an intervention that revolves around oxygen, one particular misconception is easy to make:
We focus on oxygen saturation.
But a good arterial oxygen saturation value does not tell us everything about what happens to that oxygen once it reaches the periphery.
For oxygen to contribute effectively to aerobic energy production, an entire chain must function: ventilation, pulmonary gas exchange, transport through the bloodstream, circulation, microcirculation, diffusion into tissue, and finally cellular utilization.
Think of that chain as a delivery service.
It is not enough for the package to arrive at the central distribution center.
It still has to travel down the right road, reach the correct house — and someone has to open the door and actually use what was delivered.
In some people with Long COVID, evidence suggests that peripheral steps within this chain may be impaired.
Invasive CPET studies have described reduced systemic oxygen extraction in Long COVID cohorts. A 2024 study involving 47 patients with Long COVID found reduced oxygen extraction despite relatively high cardiac output. The authors discussed metabolic and peripheral mechanisms as potential contributors to exercise intolerance.
Other studies and reviews have also described altered peripheral oxygen extraction, endothelial dysfunction, and possible microvascular abnormalities as components of the heterogeneous Long COVID picture.
These findings describe subgroups and possible mechanisms.
They do not justify the blanket conclusion that everyone with Long COVID has impaired oxygen extraction or mitochondrial dysfunction.
The amount of oxygen present in the blood therefore tells us only part of the story.
Why Oxygen Saturation Alone Does Not Capture the Total Physiological Load
This adds another dimension to the use of hypoxic interventions.
During a session, we monitor SpO₂. We observe how far oxygen saturation falls and how quickly it recovers during reoxygenation.
These are important data.
But they do not fully answer the question of how demanding that stimulus actually was for the organism as a whole.
A pulse oximeter tells us about arterial oxygen saturation.
It does not fully tell us how effectively tissue extracts oxygen.
It does not fully capture how the autonomic and cardiovascular systems process the stimulus.
It cannot tell us whether the patient will develop post-exertional symptom exacerbation the following day.
And it cannot tell us how much regulatory work was required to maintain homeostasis during and after the stimulus.
A pulse oximeter is a little like the fuel gauge in a car.
It can tell us very reliably how much fuel is in the tank.
But it cannot tell us whether the engine is running properly, whether the cooling system works, or whether the brakes are dragging.
A good measurement answers the question it was designed to answer.
Not every other question at the same time.
How Can a Positive IHHT Study and Negative Real-World Experiences Both Exist?
This brings us to one of the most interesting questions in this article.
How do we reconcile a positive controlled IHHT study in Long COVID with reports from people who experience hypoxic interventions as burdensome?
First, we need to be very clear:
We do not yet know how frequently worsening occurs after IHHT in people with Long COVID.
We still lack sufficiently large randomized trials with careful phenotyping and systematic assessment of PEM or PESE.
But we can understand why the available observations do not necessarily contradict one another.
The study by Doehner and colleagues examined patients participating in an inpatient multidisciplinary rehabilitation program.
IHHT was added to that rehabilitation setting.
The study was controlled but non-randomized and therefore investigated a specific population of patients who were able to participate in inpatient rehabilitation — not automatically the entire spectrum of Long COVID.
That distinction matters.
People who are able to participate in an inpatient rehabilitation program are not necessarily representative of severely affected patients with pronounced PEM or ME/CFS-like symptoms.
This is why we should derive neither too little nor too much from the study.
Too little would be:
“IHHT does not work for Long COVID.”
The positive findings do not support that conclusion.
Too much would be:
“IHHT is an effective treatment for Long COVID.”
The current evidence does not yet justify such a broad statement either.
The more accurate conclusion lies somewhere in between:
There is initial controlled clinical evidence suggesting that, in a selected Long COVID population capable of participating in structured inpatient rehabilitation, IHHT can be well tolerated and associated with additional clinical improvements.
Which phenotypes are most likely to benefit — and which require particular caution — remains insufficiently understood.
That unanswered question is not a weakness.
It is where the science currently stands.
This May Be Why Success and Failure Can Appear So Close Together in Clinical Practice
If we treat Long COVID as a single disease state and give everyone the same protocol, contradictory experiences are almost inevitable.
Not necessarily because intermittent hypoxia is unreliable.
But because our category may be too broad.
Imagine giving 100 people with the symptom “abdominal pain” the same treatment.
Some have gastritis.
Some have a food intolerance.
Some have gallstones.
Some are experiencing stress-related symptoms.
If the treatment works exceptionally well in 30 people and not at all in others, at some point we would stop asking:
“Does this treatment work for abdominal pain?”
We would begin asking:
“What is causing the abdominal pain in this particular person?”
In my view, bioenergetic interventions need exactly this level of maturity.
Phenotyping Comes Before IHHT
For Long and Post-COVID, this does not mean that we already have a perfect diagnostic algorithm.
We do not.
But we can begin asking much better questions.
Is deconditioning the predominant issue?
Are there signs of orthostatic intolerance or dysautonomia?
How does the person respond to everyday activity?
Is PEM or PESE present?
How long does recovery take after physical or cognitive exertion?
How stable are sleep and daytime energy?
Are there relevant cardiopulmonary abnormalities?
How does heart rate behave during exertion?
And — critically — how does this person respond to small physiological challenges?
These questions do not replace appropriate diagnostics.
But they can help us read the clinical picture more accurately rather than prematurely grouping every abnormal response under the single label of “fatigue.”
This differentiation also reflects the direction in which current Long COVID research is moving. Recent work on clinical trial design explicitly calls for better baseline characterization, longitudinal data, and greater attention to PEM and different clinical phenotypes.
In other words:
We do not only need better treatments.
We need better matching.
The First Carefully Dosed IHHT Session Provides a Stimulus — and Information
This idea is particularly important to me.
When we begin with a very low, controlled dose in a complex system, we are not only delivering a therapeutic stimulus.
We are also learning something about how that system responds.
That does not turn the first session into a formal diagnostic test.
But in an individually managed intervention, the response to a cautiously selected initial dose provides clinically relevant information that can inform subsequent dosing.
How quickly does oxygen saturation fall?
How does heart rate respond?
How does the person feel during the session?
How quickly do the monitored parameters normalize?
And then comes the part I consider at least as important:
What happens afterward?
Not only ten minutes later.
But that evening.
During the night.
The next morning.
And — particularly when PEM may be relevant — 24 or even 48 hours later.
A delayed response can easily be missed if we look only at the device display.
Think of a lake.
If I throw a stone into the water and look away after three seconds, I know that a wave was created.
But I do not know how long the water needed to become calm again.
For regulation, that return may be at least as informative as the initial reaction.
“Start Low, Observe Carefully” — in Vulnerable Systems, the Response Matters
There is currently no validated, phenotype-based IHHT dosing algorithm for Long COVID based on a “start low, go slow” principle.
However, as a clinical precaution, a conservative starting point may be reasonable in vulnerable systems: apply a stimulus, observe the immediate response, consider recovery, and only then decide on the next dose.
Especially in the presence of PEM or PESE, progression should not be formulaic.
We apply a stimulus.
We observe the response.
We observe recovery.
Only then do we decide on the next stimulus.
The protocol does not lead the patient.
The patient’s response leads the protocol.
That is a fundamental difference.
And it may be one of the most important reasons why the same technology can produce excellent outcomes in one clinical setting while repeatedly overwhelming patients in another.
The difference may not necessarily be the device.
It may be the quality of clinical decision-making and stimulus control.
HRV Can Help — but It Cannot Make the Decision by Itself
Heart rate variability can provide valuable information about autonomic regulation and its trajectory over time. Changes relative to an individual’s baseline, responses to stressors, and subsequent recovery may all be informative.
But HRV is not a clearance test for a hypoxic intervention.
Long COVID-related dysautonomia cannot be reduced to a single HRV value either.
Evidence for the treatment of autonomic dysfunction in Long COVID remains heterogeneous and limited. A recent scoping review identified several approaches that have been studied — including HRV biofeedback, inspiratory muscle training, pharmacological interventions, and multimodal approaches — but many of the available studies were small and frequently uncontrolled.
The current evidence therefore does not provide a simple standardized treatment algorithm.
Recent HRV research also argues for differentiation.
A 2026 systematic review and meta-analysis of 11 studies involving more than 1,100 participants found a trend toward lower HRV in people with Post-COVID Condition. Effect sizes were small to moderate, statistical heterogeneity was substantial, and nine of the eleven included studies were judged to have a high risk of bias.
The authors therefore concluded that an association between Post-COVID Condition and altered neurocardiac autonomic regulation may exist, while the clinical meaning of individual HRV changes remains insufficiently established.
That distinction is highly relevant to our question.
It does not mean HRV has little value.
It means we need more than HRV.
If I want to understand whether someone may be able to process an additional hypoxic stimulus, I want to see the broader regulatory picture.
How do heart rate and blood pressure respond when the person stands up?
Do they experience dizziness, palpitations, or orthostatic symptoms?
How does oxygen saturation behave?
What is their functional capacity in everyday life?
What happens after physical or cognitive exertion?
And most importantly:
How long does the system need to return to its baseline afterward?
When clinically indicated, orthostatic heart rate and blood pressure measurements, active stand testing, and further autonomic assessment such as tilt-table testing are among the established tools that can help evaluate these symptoms. Standardized instruments such as COMPASS-31 can also help characterize autonomic symptom burden.
Think of it as the dashboard of a car.
HRV is an important gauge.
But no one would decide whether a car is ready for a demanding mountain road by looking at a single gauge.
We would want to know how the engine is running, whether the cooling system is functioning, how much fuel remains, whether warning lights are on — and how the vehicle responds when we actually press the accelerator.
That is how biological regulation should be viewed.
Not as a number.
But as the behavior of a system under changing conditions.
And that brings us back to the hypoxic intervention.
The critical information is not only what a person looks like before a session or which values they bring with them.
At least as important is what happens during the stimulus — and what happens afterward.
How quickly does the patient desaturate?
How do heart rate and subjective symptoms change?
How does reoxygenation occur?
And then perhaps the most important question:
Does the system find its way back?
An adaptive stimulus should not merely generate a reaction. It should be part of a process after which regulation and functional stability can be restored.
This is where the image of the stone in the lake becomes practically relevant:
We should not only watch how high the wave rises.
We need to observe how long the lake takes to become calm again.
And in Long and Post-COVID, this second half of the response may be crucial.
A patient who appears to respond uneventfully during an IHHT session but experiences a significant decline in functional capacity for hours or the following day may be giving us more clinically relevant information than the most reassuring oxygen saturation curve on the device.
That does not automatically mean IHHT is fundamentally inappropriate for that person.
But it may mean that the dose, timing, or even readiness for that type of physiological challenge needs to be reassessed.
And in people with pronounced PEM or PESE, this level of caution becomes particularly important. Increasing physiological load should not follow a standardized progression when post-exertional symptom exacerbation is present.
Perhaps this is ultimately the greatest difference between a protocol and a therapy:
A protocol knows what is supposed to happen next.
A good clinician first looks at how the organism responded to the last step.
That is why, in Long and Post-COVID, it is not enough to simply combine a diagnosis, an IHHT protocol, and a target oxygen saturation.
Between those three points lies the real clinical decision-making space: the patient’s phenotype, their response to exertion, their autonomic and cardiopulmonary profile, and, above all, their ability to return to functional stability after a physiological challenge.
Only when we include this level of assessment does a technically correct hypoxia protocol become an individually guided intervention.
In the End, the Method Is Not What Decides
Perhaps this is the most important insight to take away from this entire series:
A good method remains a good method.
But even the best method can only be as appropriate as the context in which we use it.
Intermittent hypoxia is an excellent example.
Hypoxia can provide a biologically meaningful adaptive stimulus. It can activate signaling pathways, challenge the organism, and, under appropriate conditions, initiate adaptive processes.
But that does not mean that more hypoxia produces more benefit.
And it certainly does not mean that every exhausted person will benefit from the same protocol.
Long and Post-COVID make this particularly clear.
Two people can carry the same diagnosis and describe the same exhaustion while having completely different physiological prerequisites for an additional training stimulus.
One may primarily have lost functional capacity and may benefit from a carefully dosed adaptive challenge.
In another, autonomic dysfunction, orthostatic intolerance, or pronounced post-exertional symptom exacerbation may be part of a considerably more vulnerable clinical picture.
Long COVID is clinically heterogeneous. Autonomic abnormalities and PEM or PESE have been described in relevant subgroups.
And this may be precisely why success and failure with IHHT in Long and Post-COVID can appear so close together.
Not necessarily because the method works in one case and fails in another.
But because under the same diagnostic label, we may be treating very different biological starting points.
The controlled pilot trial of IHHT conducted within an inpatient Long COVID rehabilitation program is therefore an important positive signal. In the population studied, the addition of IHHT was associated with greater improvements in outcomes including functional capacity and was well tolerated.
At the same time, the study was non-randomized and examined a specific population capable of participating in inpatient rehabilitation within a multimodal program.
It therefore does not provide a basis for broadly extrapolating IHHT to every Long COVID phenotype or to ME/CFS-like presentations.
Perhaps we should therefore stop evaluating bioenergetic interventions solely by asking what they are capable of doing.
The more interesting question is:
What can this organism do with this stimulus right now?
Because good bioenergetic medicine does not begin when we switch on the most powerful device.
It begins when we understand a person well enough to know which challenge may help their system grow — and which challenge is still too much.
And sometimes the most intelligent intervention is not to apply a stronger stimulus.
It is to first create the conditions that allow a stimulus to become training again.
Marion Massafra-Schneider
References and Further Reading
- Navarrete-Opazo A, Mitchell GS. Therapeutic potential of intermittent hypoxia: a matter of dose. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology. 2014;307(10):R1181–R1197. A foundational review on the dose dependence of intermittent hypoxia and the distinction between potentially therapeutic and pathological exposure patterns.
- Semenza GL. Hypoxia-inducible factors in physiology and medicine. Cell. 2012;148(3):399–408. A foundational review of cellular oxygen sensing and HIF-dependent adaptive mechanisms.
- Doehner W, Fischer A, Alimi B, et al. Intermittent Hypoxic-Hyperoxic Training During Inpatient Rehabilitation Improves Exercise Capacity and Functional Outcome in Patients With Long Covid: Results of a Controlled Clinical Pilot Trial. Journal of Cachexia, Sarcopenia and Muscle. 2024;15(6):2781–2791. doi:10.1002/jcsm.13628. A prospective, controlled, open-label, non-randomized pilot trial evaluating IHHT as an adjunct to inpatient multidisciplinary rehabilitation for Long COVID.
- Marques KC, et al. Reduction of Cardiac Autonomic Modulation and Increased Sympathetic Activity in Long COVID. 2022. An investigation of autonomic alterations in a Long COVID cohort.
- Twomey R, et al. Chronic Fatigue and Postexertional Malaise in People Living With Long COVID. Physical Therapy. 2022;102(4). An investigation of fatigue and post-exertional symptom exacerbation in people living with Long COVID.
- Dani M, Dirksen A, Taraborrelli P, et al. Autonomic dysfunction in ‘long COVID’: rationale, physiology and management strategies. Clinical Medicine. 2021;21(1):e63–e67. A review of autonomic dysfunction and orthostatic intolerance in the context of Long COVID.
- Zhang Q, et al. Intermittent Hypoxia Conditioning: A Potential Multi-Organ Protective Therapeutic Strategy. 2023. A review of potential adaptive effects of controlled intermittent hypoxia and the importance of dose and exposure pattern.
- Durstenfeld MS, Sun K, Tahir P, et al. Use of Cardiopulmonary Exercise Testing to Evaluate Long COVID-19 Symptoms in Adults: A Systematic Review and Meta-analysis. JAMA Network Open. 2022;5(10):e2236057. doi:10.1001/jamanetworkopen.2022.36057. A systematic review and meta-analysis addressing exercise intolerance, chronotropic incompetence, dysfunctional breathing, and abnormalities in peripheral oxygen extraction in Long COVID.
- An Y, Guo Z, Fan J, et al. Prevalence and measurement of post-exertional malaise in post-acute COVID-19 syndrome: A systematic review and meta-analysis. General Hospital Psychiatry. 2024;91:130–142. doi:10.1016/j.genhosppsych.2024.10.011. A meta-analysis of 12 studies involving 2,665 participants examining the prevalence and assessment of PEM in post-acute COVID-19 syndrome.
- Schoene D, Deckert S, Barlinn K, et al. Neurocardiac Autonomic Dysfunction in Patients With Post-COVID-19 Condition: A Systematic Review and Meta-Analysis. European Journal of Neurology. 2026;33(3):e70561. A systematic review and meta-analysis of 11 studies examining neurocardiac autonomic dysfunction and HRV in Post-COVID Condition.
- Treadwell JR, Wagner J, Reston JT, et al. Treatments for Long COVID autonomic dysfunction: a scoping review. Clinical Autonomic Research. 2025;35(1):5–29. doi:10.1007/s10286-024-01081-w. A scoping review of the available evidence on treatment approaches for autonomic dysfunction in Long COVID.
