Oxygen Toxicity & the Hazards of Oxygen Therapy

Overview

Oxygen is a drug, and more is not better. It is one of the most frequently administered therapies in the hospital, and because it feels intuitively safe it is often started at a high flow or a high FiO₂ and then forgotten. That reflex causes harm. Hyperoxia — excess oxygen in the blood and tissues — carries specific, well-described hazards, and the modern standard of care is to titrate to a target SpO₂ rather than run a high FiO₂ out of habit.

The harms fall into five buckets: pulmonary (and, at high pressures, CNS) oxygen toxicity, absorption atelectasis, oxygen-induced hypercapnia in chronic CO₂ retainers, retinopathy of prematurity in neonates, and fire. Critical-care trials comparing liberal with conservative oxygen have found no benefit — and signals of harm — from running oxygen high, which is exactly why titration to a target replaced the “flood it with oxygen” approach.

Key Concepts

Five distinct hazards explain why oxygen is titrated rather than maximized. Each has a different trigger and mechanism, and recognizing them at the bedside is what keeps oxygen in its safe window.

Pulmonary oxygen toxicity & absorption atelectasis

Pulmonary oxygen toxicitydevelops when a high FiO₂ — generally above 0.60 — is sustained for many hours to days. The excess oxygen generates reactive oxygen species faster than the lung's antioxidant defenses can neutralize them, and those radicals injure the alveolar-capillary membrane. The earliest sign is tracheobronchitis; with continued exposure the injury progresses to diffuse alveolar damage that resembles ARDS — the Lorrain Smith effect. The defense is simple: minimize time at a high FiO₂ and accept a lower-but-adequate SpO₂ rather than leaving a patient on a prolonged FiO₂ of 1.0. (At hyperbaric pressures oxygen also injures the CNS, but that is the domain of dive and hyperbaric medicine rather than routine bedside care.)

Absorption atelectasis is the paradox where giving more oxygen worsens the hypoxemia you are treating. Nitrogen makes up most of room air and is poorly absorbed, so it lingers in the alveolus and acts as a splint that keeps it open. Breathing a high FiO₂ washes that nitrogen out. The oxygen that replaces it is absorbed into the pulmonary blood faster than ventilation can refill the alveolus, so poorly ventilated units collapse. The collapse increases shunt, and the rising shunt deepens the hypoxemia — a vicious cycle that a more measured FiO₂ avoids.

Oxygen-induced hypercapnia in COPD

In COPD and other chronic CO₂ retainers, giving too much oxygen can raise the PaCO₂. The classic teaching — that oxygen simply knocks out the hypoxic drive and the patient stops breathing — is real but is the least important of three mechanisms. Understanding the order matters, because it explains why the answer is to target oxygen rather than withhold it.

1. Release of hypoxic pulmonary vasoconstriction (dominant). Diseased lung normally diverts blood away from poorly ventilated regions by constricting their vessels. Excess oxygen relaxes that constriction, so perfusion floods back into poorly ventilated lung. The result is worse ventilation-perfusion mismatch and more dead space — CO₂-laden blood passes lung that cannot clear it, and the PaCO₂ climbs.
2. The Haldane effect. Oxygenated hemoglobin carries less CO₂ than deoxygenated hemoglobin. As you saturate the blood, CO₂ that was bound to hemoglobin is displaced into the plasma, raising the measured PaCO₂.
3. A modest reduction in minute ventilation. The blunted hypoxic drive contributes a small fall in ventilation — a genuine effect, but a minor one compared with the V/Q mechanism above.

The fix is not to withhold oxygen from a hypoxemic patient — hypoxemia kills faster than a rising PaCO₂. The fix is to target an SpO₂ of 88-92% in patients at risk of hypercapnia: treat the hypoxemia, just to a controlled ceiling, and watch the PaCO₂ on a blood gas rather than guessing.

Retinopathy of prematurity & fire risk

Retinopathy of prematurity (ROP) is the neonatal face of hyperoxia. In a premature infant the retinal vasculature is still developing, and excess oxygen can injure those vessels and lead to abnormal regrowth that threatens vision. This is why neonatal SpO₂ targets are deliberately conservative — commonly about 90-95% — and why oxygen in the NICU is titrated with particular discipline. The neonatal and pediatric hub covers ROP and its target ranges in more detail.

Fire riskis the environmental hazard that has nothing to do with the patient's lungs and everything to do with chemistry: oxygen vigorously supports combustion. An ignition source that would smolder in room air can flare in an oxygen-enriched environment. The rules are concrete — no smoking near oxygen, keep open flames and heat sources away, and secure cylinders so they cannot fall and shear a valve.

The evidence for restraint

The shift from liberal to targeted oxygen is not just mechanistic theory — it is backed by trial data. Studies of conservative versus liberal oxygen in critically ill patients, such as ICU-ROX, found no benefit from running oxygen high, alongside signals of harm. That evidence reinforces the same bedside conclusion the physiology points to: titrate to a target rather than maximize.

The practical targets that follow are a general-ward SpO₂ of 92-96% for most acutely ill adults, and 88-92% for patients at risk of hypercapnia. Premature neonates sit lower still at roughly 90-95% to limit ROP. In every group the discipline is identical: pick the target, then titrate the oxygen up or down to hold it.

What the RT does with it

At the bedside, treating oxygen as a titrated drug comes down to a handful of habits:

• Titrate to the SpO₂ target, not to a fixed FiO₂. Set the target first — typically 92-96% on the general ward — and adjust flow or FiO₂ up and down to hold it rather than running oxygen high by reflex.
• Use 88-92% for hypercapnia-risk patients. In COPD and other chronic CO₂ retainers, target 88-92% — and still treat their hypoxemia. Do not withhold oxygen out of fear of CO₂; target it and confirm the PaCO₂ on a blood gas.
• Minimize time at an FiO₂ above 0.60. When a high FiO₂ is unavoidable, treat it as a temporary bridge and wean it as soon as oxygenation allows to limit pulmonary oxygen toxicity.
• Watch for absorption atelectasis. If a patient on a high FiO₂ is deteriorating with worsening shunt, consider that the oxygen itself may be collapsing poorly ventilated lung.
• Protect premature neonates. Hold conservative SpO₂ targets (about 90-95%) to limit retinopathy of prematurity.
• Enforce fire safety. No smoking near oxygen, keep ignition sources away, and secure cylinders.

Common Pitfalls

• Equating more oxygen with safer. Hyperoxia has real harms; the goal is an adequate SpO₂, not the highest one.
• Withholding oxygen from a hypoxemic COPD patient. Fear of CO₂ retention is not a reason to let a patient stay hypoxemic — target the SpO₂ to 88-92% rather than withhold oxygen.
• Blaming hypercapnia solely on loss of hypoxic drive. V/Q mismatch from released hypoxic pulmonary vasoconstriction and the Haldane effect dominate; the blunted drive is the minor contributor.
• Ignoring absorption atelectasis. A high FiO₂ can cause worsening shunt by collapsing alveoli — it is not always the disease progressing.
• Leaving a patient on an FiO₂ of 1.0 too long. Wean the FiO₂ once oxygenation allows; a prolonged FiO₂ of 1.0 invites pulmonary oxygen toxicity.

Board Exam Pearls

FAQ

Related Resources

Sources

1. Kacmarek RM, Stoller JK, Heuer AJ. Egan's Fundamentals of Respiratory Care. 12th ed. Elsevier; 2021. Hazards of oxygen therapy.
2. O'Driscoll BR, Howard LS, Earis J, Mak V; BTS Emergency Oxygen Guideline Group. BTS guideline for oxygen use in adults in healthcare and emergency settings. Thorax. 2017;72(Suppl 1):ii1-ii90.
3. ICU-ROX Investigators and the ANZICS Clinical Trials Group. Conservative oxygen therapy during mechanical ventilation in the ICU. N Engl J Med. 2020;382(11):989-998.