Hypoxemia / Low SpO₂

SpO2 is a non-invasive estimate of arterial oxygen saturation measured by pulse oximetry. It measures the percentage of hemoglobin that is bound to oxygen. PaO2 is the partial pressure of dissolved oxygen in arterial blood, measured by arterial blood gas. These are related but not interchangeable.

The relationship between SpO2 and PaO2 is described by the oxyhemoglobin dissociation curve — which is sigmoidal, not linear. The curve has a flat upper portion and a steep lower portion. In the flat portion (SpO2 above 90%), large changes in PaO2 produce small changes in SpO2. An SpO2 of 95% may correspond to a PaO2 of 70–80 mmHg; an SpO2 of 88% may correspond to a PaO2 of 55–60 mmHg. This means that SpO2 is relatively insensitive to moderate hypoxemia — a patient can have a PaO2 of 65 mmHg (below the threshold for supplemental oxygen) with an SpO2 that appears reassuring at 93–94%.

The curve shifts with temperature, pH, and 2,3-DPG levels. Acidosis, fever, and elevated 2,3-DPG shift the curve rightward (lower affinity — oxygen is released more readily to tissues). Alkalosis, hypothermia, and reduced 2,3-DPG shift the curve leftward (higher affinity — oxygen is held more tightly by hemoglobin, potentially impairing tissue delivery despite normal SpO2).

Pulse oximetry is unreliable in several important clinical settings. Poor peripheral perfusion — from hypotension, vasoconstriction, hypothermia, or peripheral vascular disease — reduces the pulsatile signal and produces inaccurate readings. Motion artifact is common in agitated or shivering patients.

Carbon monoxide poisoning is a critical limitation. Carboxyhemoglobin (hemoglobin bound to CO) absorbs light at the same wavelength as oxyhemoglobin and is read as oxyhemoglobin by standard pulse oximeters. A patient with severe CO poisoning may have an SpO2 of 98–99% while being profoundly hypoxic at the tissue level. Co-oximetry on arterial blood gas is required to measure carboxyhemoglobin directly.

Methemoglobinemia causes SpO2 to read approximately 85% regardless of the true saturation — the SpO2 converges toward 85% as methemoglobin levels rise. This can cause both overestimation (in a truly hypoxic patient) and underestimation (in a patient with normal oxygenation). Co-oximetry is required to measure methemoglobin.

Skin pigmentation affects pulse oximetry accuracy. Studies have demonstrated that pulse oximeters overestimate SpO2 in patients with darker skin pigmentation, potentially masking clinically significant hypoxemia. This bias is clinically important and should prompt lower thresholds for ABG measurement in patients with darker skin when clinical concern exists.

Nail polish (particularly dark colors), artificial nails, and severe anemia can also affect accuracy. When clinical concern is high and SpO2 seems inconsistent with the clinical picture, an ABG with co-oximetry is the definitive test.

Hypoxemia results from five mechanisms: hypoventilation, V/Q mismatch, shunt, diffusion impairment, and low inspired oxygen (altitude). Understanding the mechanism guides treatment.

Hypoventilation reduces alveolar oxygen tension by allowing CO2 to accumulate. The alveolar gas equation shows that as alveolar CO2 rises, alveolar O2 must fall. Hypoventilation causes hypoxemia with a normal A-a gradient — the lungs are not being ventilated, but gas exchange is intact. Causes include opioid or sedative overdose, neuromuscular disease, and severe obesity hypoventilation. Treatment is to increase ventilation.

V/Q mismatch is the most common cause of hypoxemia in clinical practice. Lung units with low V/Q ratios (ventilated poorly relative to perfusion) contribute deoxygenated blood to the pulmonary veins, lowering the overall PaO2. V/Q mismatch causes an elevated A-a gradient and responds to supplemental oxygen — increasing FiO2 improves oxygenation in the poorly ventilated units. Causes include COPD, asthma, pneumonia, pulmonary edema, and PE.

Shunt occurs when blood passes through the lungs without contacting ventilated alveoli — either through anatomical shunts (intracardiac defects, pulmonary arteriovenous malformations) or physiological shunts (completely collapsed or fluid-filled alveoli, as in ARDS or lobar pneumonia). Shunt causes an elevated A-a gradient and does not respond well to supplemental oxygen — the shunted blood never contacts the oxygen-enriched alveolar gas. A poor response to high-flow oxygen (FiO2 above 0.6) should raise concern for significant shunt physiology.

Diffusion impairment occurs when the alveolar-capillary membrane is thickened or the transit time of red cells through the pulmonary capillary is too short for complete equilibration. It is most apparent during exercise (when transit time is reduced). Causes include interstitial lung disease and pulmonary fibrosis. It causes an elevated A-a gradient and responds to supplemental oxygen.

The alveolar-arterial (A-a) oxygen gradient compares the calculated alveolar PO2 (PAO2) with the measured arterial PO2 (PaO2). A normal A-a gradient indicates that gas exchange is intact — hypoxemia is due to hypoventilation or low inspired oxygen. An elevated A-a gradient indicates impaired gas exchange — V/Q mismatch, shunt, or diffusion impairment.

The alveolar gas equation: PAO2 = (FiO2 × [Patm − PH2O]) − (PaCO2 / RQ). At sea level breathing room air: PAO2 ≈ 150 − (PaCO2 / 0.8). The A-a gradient = PAO2 − PaO2. Normal A-a gradient is approximately 5–15 mmHg in young adults and increases with age (approximately 2.5 mmHg per decade) and with supplemental oxygen.

A normal A-a gradient with hypoxemia confirms hypoventilation or low inspired oxygen as the mechanism. An elevated A-a gradient with hypoxemia confirms impaired gas exchange. This distinction is clinically useful: a patient with opioid overdose and hypoxemia will have a normal A-a gradient (hypoventilation), while a patient with PE and hypoxemia will have an elevated A-a gradient (V/Q mismatch).

Hypoxemia is low oxygen in the blood (low PaO2 or SpO2). Hypoxia is inadequate oxygen delivery to tissues. These are related but not synonymous — and the distinction has clinical implications.

A patient can be hypoxic without hypoxemia. Severe anemia reduces oxygen-carrying capacity — the PaO2 and SpO2 may be normal, but the total oxygen content of the blood is reduced. Low cardiac output reduces oxygen delivery even with normal oxygen content. Carbon monoxide poisoning produces a normal SpO2 but impairs oxygen delivery because carboxyhemoglobin cannot carry oxygen. Cyanide poisoning impairs cellular oxygen utilization — oxygen delivery may be normal but cells cannot use it.

Conversely, a patient can be hypoxemic without significant tissue hypoxia if compensatory mechanisms are sufficient: increased cardiac output, increased oxygen extraction (widened arteriovenous oxygen difference), and rightward shift of the oxyhemoglobin dissociation curve (acidosis, fever) all increase oxygen delivery and extraction to compensate for reduced arterial oxygen content.

The clinical implication: treating the SpO2 number without assessing the adequacy of oxygen delivery misses the full picture. In a patient with severe anemia and a normal SpO2, the problem is not hypoxemia — it is reduced oxygen-carrying capacity. In a patient with CO poisoning and a normal SpO2, the problem is not hypoxemia — it is impaired oxygen delivery despite apparently normal saturation.