Happy hypoxemia is severe hypoxemia (poorly responsive to supplemental oxygen) without dyspnea. This isn’t anything especially new – we have occasionally seen this since time immemorial. However, COVID is causing us to re-think how to manage this physiology.
understanding the paradox of happy hypoxemia
The key to understanding this is thinking about oxygenation and CO2 clearance separately. These problems typically run together, but under certain situations they can become divorced.
Hypoxemia occurs when blood somehow passes from the right ventricle into the left ventricle without being fully oxygenated. This generally occurs in one of two ways:
- Ventilation-perfusion mismatch (V-Q mismatch): Some parts of the lung have excess ventilation, whereas other parts have inadequate ventilation. Blood going to portions of the lung which are inadequately ventilated will be starved of oxygen. Alternatively, blood flowing to areas of the lung with excess ventilation will not be saturated over 100%. Taken together, the net result is hypoxemia. Increasing the concentration of inhaled oxygen generally fixes this problem – because even poorly ventilated areas of the lung will now get enough oxygen.
- Shunt physiology: Blood flows from the right ventricle to the left ventricle without ever coming into contact with oxygenated alveoli at all. Examples include an anatomic abnormality (e.g., ventricular septal defect) or complete dysfunction of parts of the lung (e.g., mucus plugging of one lobe of the lung). A hallmark of shunt physiology is that it is poorly responsive to increased levels of oxygen (because blood isn’t coming anywhere close to ventilated alveoli). Thus, any patient who is desaturating despite high concentrations of inhaled oxygen likely has a shunt.
(2) CO2 clearance & dead space
CO2 clearance depends on the amount of gas entering and leaving the lungs every minute, which removes CO2 from the body.
A key variable here is dead space – which is gas that is inhaled and exhaled but which does not participate in CO2 clearance. Dead space is essentially wasted breath – the body moves gas in and out of the chest, but it achieves no CO2 clearance. Excess dead space can arise at a “micro” level or a “macro” level:
- “Micro” level dead space – scarred alveoli due to ARDS become inefficient at CO2 clearance. For any given tidal volume, there is less effective CO2 clearance.
- “Macro” level dead space – imagine that a patient suddenly develops a segmental pulmonary embolism. Gas still goes in and out of that lung tissue, but no CO2 clearance occurs.
(3) work of breathing, dead space, lung compliance & airway resistance
The work of breathing is most strongly affected by the drive to clear CO2. This is influenced by:
- Dead space – increasing the dead space means that the patient must inhale and exhale more gas every minute to maintain the same CO2 level.
- Lung compliance – if the lungs have more elastic recoil impeding inflation, this will increase the work of breathing.
- Airway resistance – any obstruction to airflow, such as asthma, may increase the work of breathing.
(4) how happy hypoxemia occurs
Refractory hypoxemia with a normal work of breathing can occur if all of the following conditions are met:
- A right-to-left shunt is present.
- There isn’t excessive dead space.
- Compliance and resistance of the lungs are reasonably normal.
The most simple example of happy hypoxemia is a patient with an intra-cardiac right-to-left shunt. Such a patient may have normal lungs and thus no difficulty clearing CO2 (with a normal work of breathing). However, the patient may be quite hypoxemic. This represents perhaps the most pure divorce between oxygenation and CO2 clearance.
Now let’s take things a step further. Imagine two patients, with X-rays as shown below. One patient had a left lower lobe resection, whereas the other patient has complete left lower lobe atelectasis (causing the left lower lobe to collapse into a tiny nubbin hiding behind the heart). In both of these patients, the remaining lung tissue is normal.
Neither of these patients has extra dead space – in both cases, the lung tissue receiving ventilation is normal. Both patients have sufficient lung tissue remaining so that their overall chest compliance is close to normal. Thus, both patients can clear CO2 without difficulty – so they won’t be dyspneic.
However, the patient with left lower lobe collapse will shunt deoxygenated blood through the collapsed lung tissue – leading to hypoxemia. Thus, complete atelectasis can cause a happy hypoxemia phenotype – isolated hypoxemia with a normal work of breathing.
An additional quirk of physiology bears mention here – the ability of hypoxemic vasoconstriction to compensate for areas of the lung which are dysfunctional. For example, if the patient with atelectasis is able to strongly vasoconstrict blood flow to the collapsed lung, this may prevent significant shunting. Different patients seem to have varying abilities to compensate via hypoxemic vasoconstriction (e.g. patients with cirrhosis are often poor at hypoxemic vasoconstriction).
Recent case at Genius General Hospital illustrating happy hypoxemia
A woman presented to Genius General with a chief complaint of a wrist laceration. Further history revealed that she had been falling down repeatedly at home. Vital signs showed that she had a room air saturation of 70%! Despite this, she had absolutely no respiratory complaints.
She had a fever, leukocytosis, lymphopenia, and B-lines bilaterally. Thus, she was diagnosed initially with probable COVID-19. She remained hypoxemic despite 100% high-flow nasal cannula oxygen, leading to ICU admission.
This is her X-ray:
This doesn’t look like a COVID chest x-ray at all. Based on this X-ray, a repeat focused chest ultrasound was performed, which revealed a densely consolidated right lower lobe with dynamic air bronchograms – suggestive of a lobar bacterial pneumonia (she had already been started on empiric ceftriaxone and azithromycin, so her antibiotic coverage was quite adequate).
(Incidentally, this reveals the strength of chest X-ray during a COVID epidemic. Yes, lung ultrasound is very good. But a chest X-ray allows you to check your work. Especially if you’re moving quickly and blinded by availability bias, a chest X-ray will help prevent you from missing things. So, to all the ultrasonography purists out there – I’m sorry, but I still like admission chest films.)
So, even in the middle of a COVID pandemic, this patient presented with a happy hypoxemic presentation due to lobar pneumonia! She was shunting blood through the consolidated lung tissue, causing hypoxemia. Her remaining lung tissue was essentially normal, allowing her to clear CO2 without much difficulty (hence the lack of dyspnea). This is essentially the same as the patient with complete lobar collapse discussed above.
Based on our evaluation, we were increasingly suspicious that she had a lobar bacterial pneumonia (rather than COVID). The next question arose – how to manage her hypoxemia?
Before the COVID pandemic, I would have intubated this patient due to bacterial pneumonia with refractory hypoxemia. Without thinking twice. That was the textbook management of this condition. In 2018, there would have been 100% consensus that this patient required intubation.
But… this is April 2020. And she looked really good. Her saturation was dipping into the low 80s on 100% high-flow nasal cannula, but she was quite comfortable. We decided to try awake proning, and this caused her saturations to rise into the 90s. She continued to be comfortable. We weaned back the FiO2 to 90%.
Over the next day, her saturation continued to rise and fall, at times dipping into the low 80s. Her overall trajectory was one of recovery, however. She gradually improved and didn't require intubation.
Eventually her labs showed pneumococcal bacteremia. Her COVID PCR came back negative. So she did indeed have a classic, timeless presentation – lobar pneumococcal pneumonia. Treated with awake proning and permissive hypoxemia. What’s old is new again. (I have used awake proning for occasional patients with viral pneumonia or interstitial lung disease since 2016, but not usually for lobar pneumonia).
Approach to hypoxemia: avoiding a paint-by-the-numbers strategy
Whenever possible in medicine, triage and treatment decisions should be based upon an understanding of the physiology and disease process (rather than isolated numbers). For example:
- Saturation of 85% in a patient with asthma on high levels of FiO2 suggests severe hypercapnia or mucus plugging – which is extremely concerning and may predict deterioration.
- Saturation of 85% in a patient with a lobar shunt due to pneumococcal pneumonia who is on appropriate antibiotics is less worrisome – the natural history of this disease is often gradual resolution. As long as the patient is comfortable and defending their CO2, intubation is less likely to be required.
- Saturation of 85% may be a chronic feature in a patient with advanced congenital heart disease or interstitial lung disease.
Thus, the saturation number (85%) means little in isolation. Clinical context is required to understand what this means. Approaches to intubation or patient triage based solely on saturation or ABG values are doomed to failure.
What is the pathophysiology of happy hypoxemia in COVID-19?
So, the physiology of happy hypoxemia in COVID-19 likely involves the following elements:
- Intra-pulmonary shunting
- Relatively preserved lung compliance
- Lack of excessive dead space
- (Possibly also: dysfunctional hypoxemic vasoconstriction)
This is consistent with experience in mechanical ventilation of intubated patients with early COVID-19. Specifically, the lung compliance seems to be relatively normal. Additionally, there doesn't seem to be increased dead space (so CO2 clearance isn't a problem).
The precise mechanism behind this physiology is currently unknown (because autopsy studies typically involve patients who have been intubated for a while, leading to superimposed ventilator-induced lung injury). My best guess is that the pathophysiologic mechanism might involve:
- Most of the lung tissue functions normally.
- Some alveoli receive no aeration, causing them to function as a shunt. My guess is that this could result largely due to atelectasis (given that early APRV, proning, or CPAP seem to improve oxygenation substantially).
- Lack of substantial pulmonary embolism or microvascular thrombosis (these can occur, but seem to be a delayed feature of COVID-19).
This is generally compatible with the concept of malignant atelectasis explored earlier:
- Happy hypoxemia (severe hypoxemia without dyspnea) can be generated by a combination of shunt physiology, preserved lung compliance, and lack of dead space. This may result from any lung disease which causes a limited amount of shunt, while preserving the remainder of the lung (e.g., lobar consolidation or atelectasis).
- Happy hypoxemia has existed forever, but these patients presented only occasionally. COVID has forced us to re-think our approach to treating this physiology.
- When considering a patient with hypoxemia, the underlying pathology and physiology of hypoxemia is often more important than the exact saturation. Clinical context predicts the likelihood of deterioration or improvement.
- Reconceptualizing oxygen saturation will change the way we practice critical care – even long after the epidemic has passed.