Mechanisms of Hypoxaemia — V/Q Mismatch, Shunt, Diffusion, Hypoventilation

Overview & definition

Hypoxaemia is a low arterial oxygen tension (PaO2); hypoxia is low oxygen delivery to the tissues. There are five mechanisms of hypoxaemia: low inspired oxygen, hypoventilation, ventilation-perfusion (V/Q) mismatch, shunt, and diffusion impairment. Two bedside tools — the alveolar-arterial (A-a) oxygen gradient and the response to 100 per cent oxygen — distinguish them and direct treatment.[1][1]

The oxygen cascade and the alveolar gas equation

Oxygen partial pressure falls in steps from the dry inspired gas to the mitochondrion:[1][1]

• Dry inspired PO2 = FiO2 x 713 = 0.21 x 713 = about 150 mmHg (sea level, room air).
• Humidified tracheal PO2 = 150 - 47 (water vapour) = about 150 mmHg (water vapour is subtracted in the equation).
• Alveolar PO2 (PAO2) — given by the alveolar gas equation: PAO2 = FiO2(Patm - PH2O) - PaCO2/R, where R is the respiratory quotient (about 0.8). On room air: PAO2 = 150 - 40/0.8 = 150 - 50 = 100 mmHg.[1]
• Arterial PO2 (PaO2) = about 95 mmHg (the small A-a difference).

The A-a gradient (PAO2 - PaO2) is normally under about 15 mmHg in a young adult, rising with age (roughly age divided by 4, plus 4). A normal A-a gradient means gas exchange is intact and the hypoxaemia is from low inspired oxygen or hypoventilation; a high A-a gradient means the lung itself is the problem.[1][1]

The five mechanisms

1. Low inspired oxygen

Altitude, a low-delivered FiO2, or a hypoxic gas mixture. The A-a gradient is normal; the problem is what is being breathed. Corrected by increasing the inspired oxygen.[1]

2. Hypoventilation (type 2 respiratory failure)

Alveolar ventilation is inadequate, so CO2 accumulates and PAO2 falls (the alveolar gas equation: as PaCO2 rises, PAO2 falls).[1][1]

• The A-a gradient is normal — gas exchange is intact, the problem is ventilation.
• For every 1 mmHg the PaCO2 rises, the PaO2 falls by about 1.25 mmHg.
• Causes: reduced central drive (opioids, brainstem injury, obesity-hypoventilation), neuromuscular weakness (Guillain-Barre, myasthenia, myopathy), chest-wall disease (kyphoscoliosis, obesity), and airway obstruction (COPD).
• Treatment: ventilation (non-invasive or invasive) and reversal of the cause — oxygen alone improves PaO2 temporarily but does not stop the CO2 from rising.[1]

3. Ventilation-perfusion (V/Q) mismatch — the commonest cause

Mismatch between ventilation and perfusion across the lung. Low-V/Q units (perfused but under-ventilated, as in consolidation, oedema, atelectasis, bronchospasm) act as a partial shunt; high-V/Q units (ventilated but under-perfused, as in embolism or emphysema) act as dead space. The net effect in disease is dominated by the low-V/Q units that drag down the PaO2.[1][1]

• It is the commonest mechanism of hypoxaemia in lung disease (pneumonia, asthma, COPD, pulmonary oedema, pulmonary embolism).
• The A-a gradient is high.
• It is largely correctable by 100 per cent oxygen — even poorly ventilated units can be recruited to oxygenate their blood when the inspired oxygen is high.[1]

4. Shunt — does NOT correct with 100 per cent oxygen

Blood bypasses ventilated alveoli entirely, so it never equilibrates with alveolar gas. Because the shunted blood contributes deoxygenated blood to the arterial circulation, raising the inspired oxygen barely helps — the defining feature of true shunt.[1][1]

• Causes: an anatomical right-to-left shunt (patent foramen ovale, pulmonary arteriovenous malformation, hepatopulmonary syndrome) and a physiological shunt from completely unventilated but perfused lung (severe ARDS, dense consolidation, complete atelectasis).
• The A-a gradient is high.
• Treatment: PEEP and alveolar recruitment (to re-open collapsed alveoli and convert shunt back to low-V/Q units), and treatment of the anatomical shunt. Oxygen alone is insufficient.[1]

5. Diffusion impairment

A thickened alveolar-capillary membrane slows oxygen transfer so that venous blood does not fully equilibrate with alveolar gas before leaving the capillary.[1]

• Causes: interstitial lung disease, pulmonary fibrosis, and the early exudative phase of pulmonary oedema.
• It worsens with exercise — the capillary transit time shortens as cardiac output rises, so there is less time for equilibration, and the PaO2 falls.
• The A-a gradient is high, and it is correctable by 100 per cent oxygen (a higher alveolar PO2 drives faster diffusion).[1]

Two bedside discriminators

The mechanisms collapse into a practical approach:[1][1]

So: a hypoxaemic patient with a normal A-a gradient and a high CO2 has hypoventilation (ventilate them); a hypoxaemic patient with a high A-a gradient that corrects with oxygen has V/Q mismatch (give oxygen and treat the cause); a patient who does not correct with oxygen has shunt (PEEP, recruitment, and treat the shunt).[1]

Clinical application

• The commonest cause of hypoxaemia in the ICU is V/Q mismatch, which responds to oxygen.
• The patient who is hypoxaemic despite a high FiO2 has shunt — the ARDS picture — and needs PEEP and recruitment, not more oxygen.
• The patient with a high CO2 and a normal A-a gradient has hypoventilation and needs ventilatory support, not just oxygen.
• Hypercapnia is not one of the five mechanisms of hypoxaemia — it is type 2 respiratory failure, a separate entity, although it causes hypoxaemia through the alveolar gas equation (a high PaCO2 lowers the PAO2).[1][1]

[1]

A-a gradient — calculation and worked examples

The alveolar-arterial oxygen gradient (A-a gradient, PAO2 − PaO2) is the single most useful calculation for working out the cause of hypoxaemia. It compares the PAO2 the lung should generate (from the alveolar gas equation) with the PaO2 it actually achieves (from the ABG).[1][1]

[1]

The normal A-a gradient rises with age because of progressive closure of dependent airways and a small increase in baseline venous admixture — about 3 mmHg per decade, captured by the bedside formula age/4 + 4 (so a 20-year-old has a gradient of about 9; an 80-year-old about 24). A gradient of 25 mmHg is reassuring in an 80-year-old but pathological in a 20-year-old — always interpret the number for age.[1]

Worked example 1 — hypoventilation (opioid toxicity). A post-operative patient on an opioid PCA: PaO2 55, PaCO2 70, on room air. PAO2 = 0.21 × (760 − 47) − 70/0.8 = 150 − 87.5 = 62.5 mmHg. A-a gradient = 62.5 − 55 = 7.5 mmHg (normal). The low PaO2 is entirely explained by the high CO2 — this is hypoventilation. Treatment is ventilation (NIV or reversal with naloxone), not chasing a lung problem.[1][1]

Worked example 2 — V/Q mismatch (pneumonia). PaO2 55, PaCO2 32, on room air. PAO2 = 0.21 × 713 − 32/0.8 = 150 − 40 = 110 mmHg. A-a gradient = 110 − 55 = 55 mmHg (markedly elevated). The lung is the problem — V/Q mismatch from consolidation. Oxygen will help substantially; treat the pneumonia.[1]

Worked example 3 — severe shunt (ARDS). A ventilated patient on FiO2 0.8: PaO2 60, PaCO2 40. PAO2 = 0.8 × 713 − 40/0.8 = 570 − 50 = 520 mmHg. A-a gradient = 520 − 60 = 460 mmHg (massive). The gradient is enormous and the patient is refractory to a high FiO2 — this is shunt, needing PEEP, recruitment, proning, and possibly VV-ECMO. Pushing FiO2 higher will not fix it.[2][3]

Worked example 4 — altitude (low inspired PO2). A climber at 2500 m (Patm about 560 mmHg): PaO2 50, PaCO2 34, breathing room air. PAO2 = 0.21 × (560 − 47) − 34/0.8 = 0.21 × 513 − 42.5 = 108 − 42.5 = 65 mmHg. A-a gradient = 65 − 50 = 15 mmHg (normal). The whole cascade is shifted down by the low atmospheric pressure, but the lung transfers gas normally — this is low inspired PO2. Corrected by oxygen (or descent).[1]

The 100 per cent oxygen test — V/Q mismatch versus shunt

The definitive bedside discriminator between V/Q mismatch and true shunt is the response to 100 per cent oxygen. Give FiO2 1.0 for about 15 minutes (long enough to wash out nitrogen from even poorly ventilated units), then re-measure the PaO2.[1][1]

[1]

The ~350 mmHg threshold is pragmatic: on pure oxygen, a lung with only V/Q mismatch (no shunt) should achieve a PaO2 well above 350 mmHg, because every perfused alveolus — even a poorly ventilated one — is now filled with 100 per cent oxygen. A PaO2 stuck below 350 mmHg means a meaningful fraction of the cardiac output is bypassing ventilated alveoli altogether. This single test redirects the whole management: a correctable patient gets oxygen and treatment of the cause; an uncorrectable patient gets PEEP, recruitment, proning, and consideration of ECMO. Do not keep escalating FiO2 in a patient who fails the 100 per cent test — that is the definition of refractory shunt.[1][3]

The P/F ratio and the Berlin ARDS definition

The P/F ratio (PaO2 divided by FiO2) is the simplest oxygenation index and the backbone of the Berlin ARDS definition. It is measured on a minimum of 5 cmH2O PEEP (or CPAP). A normal P/F is above 400 mmHg; a P/F below 300 (with a qualifying clinical insult and bilateral imaging not fully explained by cardiac failure) defines ARDS.[2]

[1]

The P/F ratio is not perfect — it depends on FiO2 (the ratio improves then paradoxically worsens as FiO2 rises, because of the shunt effect), on PEEP, and on chest-wall compliance — but it is universally available and is the entry criterion for virtually every ARDS trial. It frames the mechanism: a low P/F on a high FiO2 is shunt-dominant disease, which is exactly why ARDS management is built around PEEP and recruitment rather than FiO2.[2]

Oxygenation indices — P/F, A-a gradient, and the Oxygenation Index

Three indices rank the severity of oxygenation failure. Each captures something slightly different.[2]

[1]

The Oxygenation Index (OI) is underused in adult practice but is the gold-standard severity marker in paediatric ARDS and is part of many ECMO referral criteria: OI above 40 for more than 4 hours is a classic VV-ECMO trigger in children. The principle is the same at any age — the more pressure and oxygen you must apply to achieve a given PaO2, the sicker the lung.[2]

The shunt equation and venous admixture

The shunt equation (Qs/Qt) formally quantifies the fraction of cardiac output that perfuses non-ventilated alveoli — the strict definition of "how much shunt is there?" [1]

Qs/Qt = (CcO2 − CaO2) / (CcO2 − CvO2) [1]

Where CcO2 is the end-capillary oxygen content (the theoretical maximum, calculated from the PAO2), CaO2 is the measured arterial oxygen content, and CvO2 is the mixed venous oxygen content (from a pulmonary artery catheter). Normal Qs/Qt is below 5 per cent; above 10 per cent is significant; in severe ARDS it can exceed 30 to 50 per cent. Because it requires a mixed venous sample from a PA catheter, it is rarely calculated at the bedside — but the concept is examined, and the bedside 100 per cent oxygen test is its practical surrogate.[1]

Venous admixture is the unifying concept: the total amount of mixed venous blood that effectively mixes with oxygenated end-capillary blood and drags down the arterial content. It has two components — true shunt (V/Q = 0, blood perfusing completely unventilated alveoli) and low-V/Q units (blood perfusing poorly ventilated alveoli that are not fully equilibrated). The crucial practical point: venous admixture worsens as mixed venous oxygen (CvO2) falls. A falling CvO2 — from low cardiac output, fever, shivering, or high extraction — magnifies the effect of any existing shunt, because the venous blood dumped into the arterial tree is more desaturated. This is why improving cardiac output, haemoglobin, and reducing oxygen demand (analgesia, sedation, paralysis, fever control) can improve PaO2 in a shunt patient without touching the lung — you raise the CvO2 and dilute the shunted blood. Conversely, the same shunt looks far worse in a shocked, extracting patient.[1][3]

Mixed venous and central venous oxygen saturation — SvO2 and ScvO2

Oxygen saturation in venous blood reflects the balance between oxygen delivery (DO2) and oxygen consumption (VO2) — captured by the Fick principle, VO2 = CO × (CaO2 − CvO2). Rearranged, the venous oxygen content (and hence SvO2) is what is left over after the tissues have taken what they need, so it is a sensitive, real-time index of whether delivery is keeping up with demand.[1]

• SvO2 — mixed venous oxygen saturation, sampled from a pulmonary artery catheter (the true mixed venous, blending SVC, IVC and coronary sinus blood). Normal 65 to 75 per cent (mean about 73).
• ScvO2 — central venous oxygen saturation, sampled from the superior vena cava / right atrium via a central line (no PA catheter needed). Normal about 70 per cent (65 to 75). [1]

[1]

The ScvO2 − SvO2 gradient. In a healthy, well-perfused patient the two values are close (within 2 to 3 per cent). In shock the gradient widens and reverses: ScvO2 reads higher than SvO2, by up to 8 to 10 per cent, because the splanchnic and renal beds (drained by the IVC) and the myocardium (drained by the coronary sinus) extract oxygen aggressively and their desaturated blood enters the mixed venous sample but not the central venous sample. The practical message: ScvO2 is a reasonable trend surrogate, but in established shock it will systematically overstate how well the patient is really perfused — a normal ScvO2 does not exclude ongoing tissue hypoxia.[1]

Interpreting a low SvO2 / ScvO2

A low SvO2 (below 65 per cent) or ScvO2 (below 70 per cent) means delivery is not keeping up with consumption. By the Fick equation there are only four variables to interrogate — the same four you optimise in any shocked or hypoxaemic patient:[1]

[1]

So when the SvO2 or ScvO2 is low, work through the four columns in order: is the pump failing? is the haemoglobin or saturation low? is demand too high? Most low-SvO2 states in the ICU correct with fluid, inotrope, transfusion, oxygen, and fever/agitation control. [1]

Interpreting a high SvO2 / ScvO2 (above 80 per cent)

A high venous saturation is not reassuring — it usually means the tissues cannot extract or use the oxygen being delivered:[1]

• Sepsis and septic shock — cytopathic (mitochondrial) dysfunction impairs oxidative phosphorylation; SvO2 is high while lactate rises and the patient remains shocked. This is the classic "high SvO2, high lactate" paradox of impaired oxygen utilisation.
• Cyanide toxicity — blocks cytochrome c oxidase; extraction collapses and SvO2 climbs, often with a rising lactate and a normal-appearing SaO2.
• Left-to-right intracardiac shunt — oxygenated blood recirculates into the venous side and artefactually raises the mixed venous saturation.
• Over-resuscitation / very high DO2 with low VO2 — deep sedation, hypothermia, high-flow oxygen in a paralysed patient can push SvO2 above 80 per cent simply because delivery far exceeds demand. [1]

The teaching point: a normal or high SvO2 does not equal adequate tissue oxygenation. Always read it alongside the lactate, the capillary refill, the mottling score, and the clinical picture — a high SvO2 with a rising lactate is a red flag for impaired utilisation, not a sign of success. [1]

The clinical approach to the hypoxaemic patient

[1]

The three steps collapse the five mechanisms into a single pathway: the gradient tells you whether to ventilate or image; the image tells you where the problem is; the mechanism (and the 100 per cent oxygen test) tells you whether the patient needs oxygen or PEEP. The whole framework is designed to stop the two commonest errors in hypoxaemia — escalating FiO2 in a shunt patient who needs PEEP, and chasing a lung problem in a hypoventilating patient who needs ventilation.[1][1]

[1]

Red flags

Prognosis

Clinical pearls

Key trials and evidence

[1] [1] [1]

References

1. [1]Petersson J, Glenny RW. Gas Exchange in the Lung Semin Respir Crit Care Med, 2023.PMID 37816345
2. [2]ARDS Definition Task Force, Ranieri VM, Rubenfeld GD, et al. Acute respiratory distress syndrome: the Berlin Definition JAMA, 2012.PMID 22797452
3. [3]Mancini M, Zavala E, Mancini J, et al. Mechanisms of pulmonary gas exchange improvement during a protective ventilatory strategy in acute respiratory distress syndrome Am J Respir Crit Care Med, 2001.PMID 11704594