Oxygen Toxicity: Other Manifestations of Toxicity

It has been suggested that oxygen, although essential for survival of aerobic cells, should be regarded as a universal cellular poison. All organs and tissues of the body are susceptible to damage from oxygen free radical production. Nevertheless, in other organs receiving a high blood flow such as heart, kidney and liver, no toxicity has yet been detected in humans. It may be that CNS and pulmonary toxicity pre-empt its development in other organs.

Haematopoietic system

Oxygen, in space flight exposures, has been shown to have a deleterious effect on red blood cells that is manifested by abnormal cell morphology and/or a decrease in circulating red blood cell mass. This may be caused by depression of erythropoiesis, inactivation of essential glycolytic enzymes or damage to red blood cell membranes resulting from peroxidation of membrane lipid. Mice studies show irreversible damage to haematopoietic stem cells after 24 hours of exposure to 100 per cent oxygen.

There have also been occasional reports of haemolytic episodes following hyperbaric oxygen exposure, but these seem to be related to individual idiosyncrasies such as specific enzyme defects.


In 1935 Behnke reported a reversible decrease in peripheral vision after oxygen breathing at 3.0 ATA. Lambertsen and Clarke demonstrated a progressive reduction in peripheral vision after 2.5 hours of breathing oxygen at 3.0 ATA that reached about 50 per cent after 3.5 hours. Recovery was complete after 45 minutes of air breathing.

Progressive myopic changes have been well documented during hyperbaric therapy, and they have also been noted in divers. Reversal of this myopic shift usually occurs within a few weeks, but could take many months. Changes are probably related to an effect on the crystalline structure of the lens. Butler and colleagues demonstrated a myopic shift after 15 days of hyperbaric oxygen therapy and approximately 45 hours of diving exposure to 1.3 ATA oxygen14.

Cataract formation has also been reported after extreme hyperbaric exposure (more than 100 treatment sessions), with lens opacities not completely reversible.

A reduction of the intraocular pressure may represent a toxic effect on the ciliary process. Retinal detachments, retinal micro-infarcts, changes in dark adaptation, photoreceptor damage and a decrease in the amplitude of the electroretinogram have all been recorded.

Other ocular effects of oxygen toxicity include retinopathy of prematurity in infants breathing supplemental oxygen. Initial retinal blanching resulting from vasoconstriction is followed by vessel obliteration and fibro-vascular proliferation, which may lead to retinal fibrosis and traction retinal detachments. Animal studies have demonstrated death of visual cells and retinal detachments on exposure to 0.9 to 3 ATA oxygen. Irreversible changes in the cornea and lens of guinea pigs develop after exposure to 3 ATA oxygen for between 4 and 16 hours.


Other disorders that may be affected by the hyperoxic-induced vasoconstriction include Raynaud’s phenomenon, Buerger’s disease and migraine. Risk of closure of the ductus arteriosus has been proposed in the foetus exposed to increased oxygen.


Serous otitis media has been noted in aviators exposed to high concentrations of oxygen. It results from absorption of oxygen from the middle ear. A syndrome related to the middle ear was described in US Navy divers breathing 100 per cent oxygen from semi-closed-circuit and closed-circuit diving equipment. The symptoms were fullness, popping or crackling sensation in the ear and a mild conductive hearing loss. On examination the most common finding was fluid in the middle ear. The syndrome was first noted after rising from a night’s sleep, not immediately after the dive itself, and it disappeared rapidly. There was no suggestion of barotrauma.


Repeated or long-term exposure to high levels of oxygen free radicals could be expected to enhance tumour development. A literature review including human and animal studies failed to support a possible cancer-causing effect of hyperbaric oxygen15.

It is likely that, as more sensitive methods of detection are used, evidence of oxygen toxicity in many other cells and organs will be observed.

Oxygen Toxicity: Pulmonary Toxicity (The ‘Lorrain Smith Effect’)

Clinically obvious pulmonary oxygen toxicity does not manifest in short-duration oxygen diving. It assumes greater importance in saturation and long chamber dives and where a high PO2 is inspired, such as in therapeutic recompression. Prolonged exposures to PO2 as low as 0.55 ATA (e.g. in space flight) have been found to produce significant changes. A PIO2 of 0.75 ATA has produced toxicity in 24 hours.

In animals, pulmonary oxygen poisoning causes progressive respiratory distress, leading to respiratory failure and finally death. The wide variation of tolerance among different species invalidates direct extrapolation of animal studies to humans, but early signs in humans are similar to those in animals. In patients receiving high concentrations of oxygen therapeutically, it is sometimes difficult to distinguish between the conditions for which the oxygen is given and the effects of oxygen itself (e.g. shock lung, respiratory distress syndrome).

Clinical manifestations

As in neurological toxicity, the factors affecting the degree of toxicity are the PIO2, the duration of exposure and individual variation in susceptibility. Exposure to 2.0 ATA oxygen produces symptoms in some normal humans at 3 hours, but the occasional individual may remain symptom free for up to 8 hours.

The earliest symptom is usually a mild tracheal irritation similar to the tracheitis of an upper respiratory infection. This irritation is aggravated by deep inspiration, which may produce a cough. Smoking has a similar result. Chest tightness is often noted; then a substernal pain develops that is also aggravated by deep breathing and coughing. The cough becomes progressively worse until it is uncontrollable. Dyspnoea at rest develops and, if the exposure is prolonged, is rapidly progressive. The higher the inspired oxygen pressure, the more rapidly symptoms develop and the greater is the intensity.

Physical signs, such as rales, nasal mucous membrane hyperaemia and fever, have been produced only after prolonged exposure in normal subjects.

Pulmonary oxygen toxicity

  • Chest tightness or discomfort.
  • Cough.
  • Shortness of breath.
  • Chest pain.

The measurement of forced VC (FVC or VC) is one monitor of the onset and progression of toxicity, although it is less sensitive than the clinical symptoms. Reduction in VC is usually progressive throughout the oxygen exposure. The drop continues for several hours after cessation of exposure and many occasionally take several days to return to normal. Because measurement of VC requires the subject’s full cooperation, usefulness may be limited in the therapeutic situation. Conversely, it is a useful tool in monitoring repeated exposures in hyperbaric workers. It has been used to delineate pulmonary oxygen tolerance limits in normal subjects – this is shown in Figure 17.3, which relates PO2 to duration of exposure. The percentage fall in VC is plotted. The size of the fall in VC does not always indicate the degree of pulmonary toxicity as measured by other lung function tests, such as other lung volumes, static and dynamic compliance and diffusing capacity for carbon monoxide. Changes in diffusing capacity may be the most sensitive indicator.

Pulmonary oxygen tolerance curves in normal men(based on vital capacity changes in 50% of the subjects)
Figure 17.3 Relationship of partial pressure of oxygen breathed and duration of exposure with degree of pulmonary oxygen damage. ΔVC, vital capacity change.

Exposure at 3 ATA for 3.5 hours caused chest discomfort, cough and dyspnoea in most of 13 subjects. There was no significant change in post-exposure FVC. Maximum mid-expiratory flow rates were reduced, but airway resistance did not change.

Some studies in divers have indicated that the reduction in forced mid-expiratory flow rates may persist for at least 3 years after deep saturation dives and also after shallow saturation dives with the same hyperoxic exposure profile. Forced expiratory volume in 1 second (FEV1) and FVC were not significantly altered.

Some individuals, especially at higher PIO2 (2.5 ATA), demonstrate a rapid fall in VC. The recovery after exposure is also more rapid than that after an equal VC decrement produced at a lower PO2 for a longer time.

Although chest x-ray changes have been reported, there is no pathognomonic appearance of oxygen toxicity. Diffuse bilateral pulmonary densities have been reported. With continued exposure, irregularly shaped infiltrates extend and coalesce.


The pathological changes in the lung as a result of oxygen toxicity have been divided into two types; acute and chronic7, depending on the PIO2.

Pressures of oxygen greater than 0.8 ATA cause a relatively acute toxicity that has been subdivided into exudative and proliferative phases. The exudative phase consists of a perivascular and interstitial inflammatory response and alveolar oedema, haemorrhage, hyaline membranes, swelling and destruction of capillary endothelial cells and destruction of type I alveolar lining cells. (This phase was the type described by Lorrain Smith.) Progression of the disease leads to the proliferative phase, which, after resolution of the inflammatory exudate, is characterized by proliferation of fibroblasts and type II alveolar cells. There is an increase in the alveolar-capillary distance. Pulmonary capillaries are destroyed, and some arterioles become obstructed with thrombus.

A more chronic response usually follows PIO2 between 0.5 and 0.8 ATA for longer periods. It is characterized by hyperplasia of type II cells, replacing type I cells and progressive pulmonary fibrosis, especially affecting alveolar ducts rather than alveolar septa. These features are also found in the adult respiratory distress syndrome (shock, drowning, trauma) for which high oxygen tensions are given. Whether oxygen actually causes the damage in these situations or exacerbates the condition by interacting with the initial pulmonary damage is not clear.

A consequence of these effects on pulmonary physiology is to increase ventilation-perfusion inequality. Obstruction of arterioles results in an increase in dead space.


No specific therapy is available that can be used clinically to delay or modify the pulmonary damage caused by hyperoxia. Intermittent exposure may delay the onset of toxicity. Delay of pulmonary toxicity has been demonstrated in humans. It has been suggested that the rate of recovery is greater than the rate of development of cellular changes leading to toxicity.

When toxicity is evident, the PO2 should be reduced. It is therefore important to be aware of the earliest signs of the syndrome.

Traditionally, the monitoring of VC has been employed as an indicator of toxicity. The maximum acceptable reduction in VC depends on the reasons for the exposure. Although a 20 per cent reduction may be acceptable in the treatment of severe DCS, a 10 per cent reduction would cause concern under operational diving conditions. One concept that has gained some popularity is that of the ‘UPTD’ or units of pulmonary toxic dose. UPTDs allow the expression of different exposures in time and PIO2 related to a ‘standard’ exposure at 1.0 ATA expressed in minutes. Expected UPTDs can be calculated for any planned exposure and that exposure can be modified to keep the decrement in VC within acceptable limits (see Clark and Thom7 for a fuller explanation).

The degree of oxygen toxicity equivalent to a 2 per cent decrease in VC (approximately the decrement predicted for a standard US Navy Treatment Table 6 treatment for DCS) is completely reversible, asymptomatic and very difficult to measure under ordinary circumstances. With the elevated pressures of oxygen used in the treatment of serious diseases, such as severe DCS or gas gangrene, it may be reasonable to accept a greater degree of pulmonary toxicity to treat the patient. The primary requirement of any therapy is that the treatment should not be worse than the disease.

Pulmonary toxicity that produces a 10 per cent decrease in VC is associated with moderate symptoms of coughing and pain in the chest on deep inspiration. This degree of impairment of lung function has been shown to be reversible within a few days. It is suggested that a 10 per cent decrement in VC be chosen as the limit for most hyperbaric oxygen therapy procedures.

VC is a relatively crude measure of toxicity. Forced mid-expiratory flow measurements or the less practical diffusing capacity for carbon monoxide may prove to be more sensitive indicators for repeated or long-term exposures.

Intermittent rather than continuous exposure to high oxygen pressure delays the onset of’ both neurological and pulmonary oxygen toxicity.

Adherence to proposed pressure-duration limits for pulmonary oxygen toxicity is difficult where extended durations and changing PO2 are involved.

Methods for calculating cumulative pulmonary toxicity have therefore been devised (e.g. the UPTD), and they may have a role in prolonged decompression and hyperbaric oxygen therapy.

As discussed in the previous section on prevention of neurological toxicity, many drugs have been shown to be effective in animal experiments. They may have a role in the future in the prevention of pulmonary and other oxygen toxicity, at least for hyperbaric therapy exposure.

Oxygen Toxicity: Central Nervous System Toxicity (The ‘Paul Bert Effect’)

In diving, CNS oxygen toxicity is more likely when closed-circuit or semi-closed-circuit rebreathing sets are used, and it is the factor limiting depth when oxygen supplementation is used. With compressed air, the effect of increased partial pressure of nitrogen (see Chapter 15) usually prevents the diver from reaching a depth and duration at which oxygen will become a problem (although occasional reports can be found). ‘Technical’ diving (see Chapter 62), in which a higher FIO2 than air is commonly used, permits CNS toxicity (Case Report 17.16). High oxygen pressures are used in therapeutic recompression for decompression sickness (DCS) and air embolism (see Chapters 6 and 13).


A 47-year-old experienced underwater cave diver with no significant medical history was diving with two tanks – one containing compressed air, the other a 50 per cent mixture of oxygen and nitrogen (nitrox). Towards the end of the 47-metre, 19-minute dive, he was seen floating head down, unresponsive, with his mouthpiece out of his mouth and ‘his fins [flippers] moving as if he was shivering’ (as reported by another diver to the Coroner). The body was carried up to 15-metre depth and then allowed to ascend freely as the other divers decompressed. Cardiopulmonary resuscitation was attempted, but abandoned after 43 minutes, as there was no response.

Examination of the subject’s diving equipment revealed that he had been breathing the 50 per cent oxygen/nitrogen mixture for most of the dive (at 47 metres, the PIO2 would have been 5.7 ATA × 0.5, or 2.85 ATA). Each tank had a separate first stage connected in an unusual fashion by a two-way switch, which the diver had had made by a local engineering shop. This allowed the diver to switch from one tank to another rapidly. This switch supplied a single second-stage mouthpiece. The two tanks were different colours; the circuit from the black (compressed-air) tank was marked with yellow tape, while the circuit from the yellow (nitrox) tank was unmarked. The regulator had a small tear and a bite mark in the mouthpiece. The diver wore a face mask and separate mouthpiece rather than a full-face mask, which covers eyes, nose and mouth.

The cause of death, as determined by the Coroner, was drowning after oxygen toxicity.

(From Lawrence CH. A diving fatality due to oxygen toxicity during a ‘technical’ dive. Medical Journal of Australia 1996;165:262–263. ©Copyright 1996.The Medical Journal of Australia; reproduced with permission.)

Clinical manifestations

A wide range of symptoms and signs has been described, the most dramatic of which is a grand mal–type convulsion. Consciousness is maintained up to the time of convulsion and there are apparently no changes in the electroencephalogram before convulsion.

In practice, there is no reliable warning of impending convulsions (but see Lambertsen’s description later), and only about half of persons affected describe any premonitory symptoms. However, the list of such manifestations is long, and any unusual symptom should be suspect. The most commonly reported signs and symptoms are nausea, vomiting, light-headedness, dizziness, tinnitus, vertigo, incoordination, sensations of impending collapse or uneasiness (dysphoria), facial pallor, sweating, bradycardia, constriction of visual fields (tunnel vision), dazzle, lip twitching, dilatation of pupils, twitching of hand, muscular twitching elsewhere, hiccups, paraesthesiae (especially fingers), dyspnoea, disturbance of special senses, hallucinations and confusion.

Facial twitching is a common objective sign in chamber exposures to oxygen greater than 2 ATA, and it signifies an imminent convulsion. Lip twitching may be seen if a mouthpiece is being used. Nausea, retching and vomiting are particularly noted after prolonged exposures between 1 and 2 ATA.

Central nervous system oxygen toxicity:

  1. Twitching (especially lips).
  2. Nausea.
  3. Dizziness.
  4. Tinnitus.
  5. Tunnel vision.
  6. Dysphoria.
  7. Convulsions.

Many of these manifestations are associated with other potential causes. Facial pallor is thought to result from the intense peripheral vasoconstriction of hyperoxia and is not necessarily a sign of cerebral toxicity. Similarly, the paraesthesiae in fingers and toes do not necessarily indicate an impending convulsion. They may persist for hours after exposure and may represent an effect of local vasoconstriction on peripheral nerves or simply a tight-fitting wetsuit.

An important aspect of toxicity is the great variation in susceptibility. As well as the wide range of tolerance among individuals, there is marked variation in one person’s tolerance from day to day7. Therefore, in any one diver, the time to onset of symptoms cannot be related to a predictable depth or time of exposure. Despite this variation, the greater the PO2 and the longer the time of exposure, the more likely is the toxicity to develop.

Factors lowering threshold to central nervous system toxicity:

  • Severe exercise.
  • Immersion in water rather than in air (e.g. a chamber).
  • Hypothermia.
  • Increased arterial carbon dioxide from any cause.

Exposure in water rather than in dry chambers markedly decreases the tolerance to oxygen. Many of the previously listed clinical features are much less apparent in the water, where convulsions are more often the first manifestation. A convulsion is much more dangerous under water because of the added complications of drowning and pulmonary barotrauma. Therefore, most authorities have set a maximum safe depth for pure oxygen diving at about 10 metres. Short dives may be safe at greater depths and prolonged ones at shallower depths (Table 17.2). Exposure in a compression chamber is considered to be less hazardous for an equivalent depth-time profile. Current decompression procedures, if performed in chambers, prescribe oxygen exposures at 18 metres (2.8 ATA), and seizure is rare.

US Navy oxygen depth time limits in water

Exercise has also been shown to hasten the onset of symptoms. Shallower maximum safe depths have been set for ‘working’ as opposed to ‘resting’ dives on oxygen. This observation is also of importance when oxygen is used to shorten decompression times in the water. Divers undergoing decompression should be at rest, e.g. supported on a stage – not battling swell, current and buoyancy to maintain constant depth. Hypothermia is likely to hasten the onset of symptoms.

CO2 build-up during exercise has been suggested as a potentiating factor in producing convulsions. Increased inspired PCO2 may develop with inadequate absorbent systems and in poorly ventilated helmets and chambers, thus rendering the diver more susceptible to oxygen convulsions.

The danger of convulsions prevents divers breathing 100 per cent oxygen deeper than 8 to 10 metres of sea water.

The role of inert gas in the exacerbation of oxygen toxicity needs to be more fully elucidated.

The frequency of presenting symptoms in ‘wet’ divers resting and working is shown in Table 17.3 from Donald’s work2,3. In all cases exposure continued until the first symptoms developed (‘end-point’).

Incidence of symptoms resulting from exposure of divers to ‘end-point’ in water

Convulsions, which may be the first manifestation of toxicity, are indistinguishable clinically from grand mal epilepsy (Case Reports 17.2, 17.3 and 17.4). A review of neurological toxicity in US Navy divers showed that convulsions were more likely to be the presenting feature in inexperienced divers breathing oxygen, compared with trained divers. It is inferred that some of the so-called premonitory symptoms may result from suggestion rather than oxygen. Of 63 divers, 25 had convulsions as the first clinical manifestation, 10 had focal twitching, and 13 more progressed to convulsions despite immediate reduction of PO2. A more recent study revealed nausea as the most common manifestation, followed by muscle twitching and dizziness.


BL, a 20-year-old trainee naval diver, was taking part in air diving training to a depth of 21 metres. He was using surface-supply breathing apparatus (SSBA), which consists of a demand valve and a hose to the surface connected to large cylinder via a pressure regulator adjusted according to the depth. After approximately 20 minutes, he was signalled with a tug on the hose to return to the surface because the cylinder was running low. He remained in the water at the surface while his hose was connected to another cylinder and then recommenced his dive. Some 12 minutes into the second dive, BL’s surface attendant noted that there were no surface bubbles. The instructor told the attendant to signal BL via the hose tug system. There were no answering tugs on the line. The standby diver was then sent into the water to check BL. He found BL floating a metre off the bottom with his demand valve out of his mouth. He was brought rapidly back to the diving boat and cardiopulmonary resuscitation was commenced using a portable oxygen resuscitator. After some time, probably about 15 minutes, the small oxygen cylinder of the resuscitator was noted to be low, so one of the group was instructed to connect the resuscitator to the emergency large oxygen cylinder. The oxygen cylinder was then found to have a line already attached to it – BL’s SSBA! BL failed to respond to intense resuscitation carried on for more than 2 hours (inspired oxygen tension at 21 metres would have been 3.1 ATA).

Diagnosis: death resulting from central nervous system oxygen toxicity (presumably convulsions).


AM and his buddy, both military divers, were practising night-time underwater ship attack while using closed-circuit 100 per cent oxygen rebreathing sets. While approaching the ship, they exceeded the maximum safe depth to avoid being spotted by lights and had to ascend to the ship’s hull (depth 9 metres). During their escape from the ship, AM had difficulty in freeing his depth gauge and, when he finally did examine it, discovered he was at 19 metres. He started to ascend and remembers ‘two to three jerkings’ of his body before losing consciousness. The buddy diver noted that AM ‘stiffened’ as he lost consciousness and then started convulsing, which continued while AM was being brought to the surface. Total time of dive was 28 minutes. At the surface, AM was pale with spasmodic respirations, and the lug on the mouthpiece had been chewed off. Artificial respiration was administered. AM was incoherent for 20 minutes and vomited once. A headache and unsteadiness in walking persisted for several hours after the incident. An electroencephalogram 3 days later was normal. (The buddy diver was exhausted on surfacing, felt nauseated and was unable to climb into the boat but recovered quickly.)

Diagnosis: near drowning secondary to central nervous system oxygen toxicity.


TL was using a semi-closed-circuit rebreathing apparatus rigged for 60 per cent nitrogen. After 17 minutes at 22 metres, he suddenly noted a ringing noise in his head. He flushed through his set, thinking his symptoms were the result of carbon dioxide toxicity. He then noted that his surroundings were brighter than usual and decided to surface.

On surfacing he was noted to be conscious but pale and panting heavily. He moved his mouthpiece and while being brought on board ‘went into a convulsion, where his whole face changed shape, his eyes rolled up into his head, his face turned a dark colour and his body began to cramp’. He recovered within 3 to 4 minutes, and 30 minutes later there was no abnormality on clinical examination.

Equipment examination revealed that the emergency oxygen cylinder was nearly empty, i.e. that he had used 64 litres of 100 per cent oxygen in addition to approximately the same amount of 60 per cent oxygen. The oxygen in his breathing bag would therefore have approximated 80 per cent and the inspired oxygen tension 2.4 ATA. The carbon dioxide absorbent was normal.

Diagnosis: central nervous system oxygen toxicity.

The following description of a typical convulsion has been given by Lambertsen8, who performed much of the original work in the United States on this subject.

Localized muscular twitching, especially about the eyes, mouth and forehead usually but not always precedes the convulsion. Small muscles of the hands may also be involved, and incoordination of diaphragm activity in respiration may occur. After they begin, these phenomena increase in severity over a period, which may vary from a few minutes to nearly an hour, with essentially clear consciousness being retained. Eventually an abrupt spread of excitation occurs and the rigid tonic phase of the convulsion begins. Respiration ceases at this point and does not begin again until the intermittent muscular contractions return. The tonic phase usually lasts for about 30 seconds and is accompanied by an abrupt loss of consciousness. It is followed by vigorous clonic contractions of the muscle groups of head and neck, trunk and limbs, which become progressively less violent over about 1 minute. As the uncoordinated motor activity stops, respiration can proceed normally. Following the convulsion, hypercapnia is marked due to accumulation of carbon dioxide concurrent with breath holding. Respiration is complicated by obstruction from the tongue and by the extensive secretions, which result from the autonomic component of the central nervous system convulsive activity. Because the diver inspired a high pressure of oxygen prior to the convulsion, a high alveolar oxygen tension persists during the apnoea. The individual remains well oxygenated throughout the convulsion. Due to the increased arterial carbon dioxide tension, brain oxygenation could increase the breath-holding period. This is in contrast to the epileptic patient who convulses while breathing air at sea level.

The latent period before the onset of toxic symptoms is inversely proportional to the PIO2. It may be prolonged by hyperventilation and interruption of exposure and shortened by exercise, immersion in water and the presence of CO2.

The ‘oxygen off-effect’ refers to the unexpected observation that the first signs of neurological toxicity may appear after a sudden reduction in PIO2. Also, existing symptoms may be exacerbated. The fall in PIO2 is usually the result of removing the mask from a subject breathing 100 per cent oxygen in a chamber. It may also occur when the chamber pressure is reduced or when the diver surfaces. It has been postulated that the sudden drop in cerebral arterial PO2 in the presence of persisting hyperoxic-induced cerebral vasoconstriction results in cerebral hypoxia in a brain already impaired by oxygen poisoning.

The risk of oxygen toxicity in the presence of a diagnosis of epilepsy is unclear. The conservative assumption that such individuals are at increased risk is widely held, and epilepsy is one of the absolute contraindications to diving. Yet although there are some reports of seizures during and after hyperbaric therapy in these patients, no formal association has been described, and many patients with epilepsy are routinely treated in hyperbaric chambers without incident.

Some animal experiments have shown that older and male animals are more susceptible to oxygen toxicity, but this has not been conclusively demonstrated in humans.

The major differential diagnoses of neurological toxicity are cerebral arterial gas embolism (CAGE) and hypoglycaemia. The timing of symptoms and signs in relation to oxygen exposure, measurement of blood glucose and the subsequent recovery from oxygen toxicity are usually enough to distinguish among these entities.

Finally, there are anecdotal reports of a syndrome of fatigue, headache, dizziness and paraesthesiae in operational divers exposed to repeated oxygen at depth on a daily basis. This has not, however, been reported in patients having daily hyperbaric oxygen therapy at 2.4 ATA for 90 minutes daily.


No pathological changes in the CNS directly attributable to oxygen toxicity have been observed in humans. Animal experiments with intermittent or continued exposure cause permanent neurological impairment with selective grey matter and neuronal necrosis (the John Bean effect)9. Changes have been reported on both light and electron microscopy in rats after exposure to 8 ATA oxygen. Lesions are found in specific areas, such as in the reticular substance of the medulla, the pericentral area of the cervical spinal grey matter, the ventral cochlear nuclei, the maxillary bodies and the inferior colliculi.

Pharmacological control of convulsions and pulmonary oedema does not alter the findings. Severe exposure eventually leads to haemorrhagic necrosis of the brain and spinal cord, but even single exposures (30 minutes at 4 ATA) produce ultrastructural changes in anterior horn grey matter.


Predictable prevention of cellular changes by administration of drugs is not yet feasible. In animal experiments, many different pharmacological agents have been shown to have a protective effect against toxic effects of oxygen. The agents include disulfiram (also described in animals to potentiate toxicity), glutathione, lithium, iso-nicotinic acid, hydrazide, GABA and sympathetic blocking agents. None of these agents is in prophylactic clinical use at present. Prevention of convulsions by anaesthetics or anticonvulsant agents removes only this overt expression of toxicity, and damage at the cellular level will continue.

The only current safe approach is to place limits on exposure. These limits depend on the PO2, the duration and the conditions of exposure (e.g. ‘wet dive’ or in a dry chamber, at exercise or rest, intermittent or continuous).

The Royal Navy and Royal Australian Navy place a limit for pure oxygen diving of 9 metres for a resting dive and 7 metres for a working dive. The US Navy relates the maximum duration of exposure to the depth (see Table 17.2).

The US Navy formerly required divers to undergo a test exposure of 30 minutes at 60 feet breathing 100 per cent oxygen to eliminate those divers who are unusually susceptible10. This does not take into account the marked variation in individuals from day to day or the marked influence of the exposure environment. Some individuals may be excessively oxygen sensitive, but the variability makes this uncertain and difficult to screen for. The oxygen tolerance test probably has no value in assessing normal dive candidates.

An awareness of levels at which toxicity is likely, and close observation for early signs such as lip twitching, should reduce the incidence of convulsions. If early signs are noted, the subject should signal his or her companion, stop excessive exertion and hyperventilate. However, premonitory symptoms or signs are unlikely in the working diver in the water.

In chamber therapy, most therapeutic tables do not prescribe 100 per cent oxygen deeper than 2.8 ATA. Periods of air breathing are used to interrupt the exposure to high levels of oxygen and thus reduce the likelihood of toxicity. Animal studies have demonstrated that interrupted exposure delays the onset of CNS toxicity by up to 100 per cent.

Electroencephalographic monitoring has not proved useful in predicting imminent convulsions.

When treating serious cases of DCS where the risks of oxygen toxicity are acceptable, drugs such as diazepam may be used to reduce the effects of toxicity. If available, most practitioners would prefer the option to use helium-oxygen mixtures to reduce the risk of toxicity.

Oxygen-breathing divers are advised to avoid the following:

  1. Exposure while febrile.
  2. Drugs that increase tissue CO2, e.g. opiates, carbonic anhydrase inhibitors.
  3. Aspirin, steroids, sympathomimetic agents.
  4. Stimulants such as caffeine (e.g. coffee).
  5. Fluorescent lights.


The initial aim of treatment is to avoid physical trauma associated with a grand mal convulsion. A padded tongue depressor to prevent tongue biting may be useful in a chamber. In contrast to epilepsy, hypoxia is not a concern, at least initially.

In the water, the traditional advice was that the diver should be brought to the surface only after the tonic phase of the convulsion ceased, but this practice has been challenged and is no longer the recommendation of the Diving Committee of the Undersea and Hyperbaric Medical Society11. The same action is often indicated in compression chambers, but with allowance made for decompression staging. If it is against the interests of the patient to ascend, it is usually a simple matter to reduce the oxygen in the breathing mixture.

Anticonvulsants may be used in exceptional circumstances. Phenytoin has been successfully used to stop convulsions in a patient with cerebral air embolism who was treated with hyperbaric oxygen. Diazepam is also very effective, as is the induction of anaesthesia with propofol.

The management of oxygen toxicity in the recompression chamber

Oxygen Toxicity: Aetiology

The precise mechanism of oxygen toxicity is unknown. Oxygen is a highly reactive element and has wide-ranging, dose-dependent effects in the body, including the regulation of blood flow, tissue oxygenation and energy metabolism in the brain. These effects are pressure dependent and are involved in the development of toxicity. There are a great many sites at which oxygen acts on metabolic pathways or on specific cellular functions. These sites may involve cell membranes, ‘active transport’, synaptic transmission, mitochondria or cell nuclei. Rather than causing an increase in metabolism, as suggested by early workers, hyperoxia has been demonstrated to depress cellular metabolism.

Many enzymes are inactivated by high PO2, particularly those containing sulphydryl groups (-SH). It is postulated that adjacent -SH groups are oxidized to form disulphide bridges (-S-S-), thus inactivating the enzyme (this may be important in the development of cataracts). Enzymes containing -SH groups, and known to be susceptible, include glyceraldehyde phosphate dehydrogenase (a key enzyme in glycolysis), the flavoprotein enzymes of the respiratory chain and the enzymes involved in oxidative phosphorylation.

The oxygen free radical theory of toxicity is widely accepted as an explanation at the molecular level. The production of a range of free radicals is a normal consequence of aerobic metabolism, and for this reason, aerobic organisms (e.g. ourselves) have developed antioxidant mechanisms to cope with molecular oxygen exposure. In the presence of hyperoxia these mechanisms may be overwhelmed, leading to the formation of excess reactive oxygen forms and direct cellular toxicity through enzyme inactivation and structural damage (e.g. lipid peroxidation). These radicals are intermediates formed in many cellular biochemical enzyme catalyzed reactions and are the result of the reduction of the oxygen molecule by electrons. Superoxide anion (O2−) is formed when oxygen accepts a single electron and hydrogen peroxide (H2O2) two electrons. The final reaction is the acceptance by oxygen of four electrons to form water or a stable hydroxyl anion. Superoxide and peroxide can react to form the hydroxyl radical OH-. All these species of oxygen, referred to as oxygen radicals, are highly oxidative.

Cells have a system of enzymes to scavenge these radicals called the tissue antioxidant system. Two of these enzymes, superoxide dismutase and catalase, are involved in maintaining adequate supplies of reduced glutathione (containing sulphydryl groups) to deal with the free radicals. Hyperoxia may cause this system to be swamped, and the excess free radicals may then produce cell damage. Examples of unwanted oxidation reactions are peroxidation of lipid in cell membranes and protein oxidation in cell membrane and cytoplasm. Both have been demonstrated in rat brain after hyperoxia4. Aerosolized (recombinant human manganese) superoxide dismutase preserves pulmonary gas exchange during hyperoxic lung injury in baboons5. Antioxidants such as glutathione have also been shown to offer some protection.

The characteristic feature of chronic pulmonary oxygen toxicity is pulmonary fibrosis (see later). In animal studies, paraquat, bleomycin and ozone have all been noted to produce pulmonary fibrosis. These agents are known to produce oxygen free radicals.

Gamma-aminobutyric acid (GABA) is a transmitter at CNS inhibitory nerve synapses. One of the demonstrated consequences of enzymatic changes induced by hyperoxia is a reduction in the endogenous output of GABA that results in the uncontrolled firing of excitatory nerves and the development of convulsions. Agents that raise brain levels of GABA appear to protect against convulsions. Lithium (useful in the treatment of bipolar disorder) has proved to be effective in inhibiting convulsions in rats. It was also shown to prevent the decrease in brain GABA that normally precedes the convulsions. In the rat lung lithium inhibits the development of oedema.

Exercise, hypoventilation and CO2 inhalation predispose to convulsions, whereas hyperventilation may be protective. CO2 may play a role in lowering seizure threshold at the cellular level, but more likely by influencing cerebral blood flow and hence the ‘dose’ of oxygen delivered to the brain.

At greater than 3 ATA PIO2, oxyhaemoglobin is not reduced on passing through capillaries and so is not available for the carriage of CO2 as carboxyhaemoglobin. Therefore, this route cannot eliminate CO2. The resultant increase in brain CO2 tension (PCO2) has proved to be small (2.5 to 6 mm Hg). An equivalent rise is caused by breathing 6 per cent FICO2 and does not cause convulsions in the presence of a normal PIO2. It does, however, appear that the slight rise in PCO2 reduces the cerebral vasoconstrictive effects of hyperoxia.

In contrast, CO2 retention is unlikely to contribute to pulmonary toxicity, although related changes in acid-base balance may modify the syndrome via neurogenic and endocrine mechanisms. Very high levels of inspired CO2 may actually protect against pulmonary damage.

Atelectasis (collapse of alveoli so they are no longer ventilated) results from absorption of oxygen during 100 per cent oxygen breathing and has been suggested as a contributory mechanism to oxygen toxicity in divers. Although atelectasis has been demonstrated, it is not an initiating factor, and toxicity also develops in the presence of inert gas. If the inert gas is at narcotic levels, it may actually enhance the onset of toxicity.

Human studies show no difference in the progression of pulmonary oxygen toxicity when comparing pure oxygen and diluted oxygen at the same PO2. Rat studies indicate that the risk of CNS toxicity is enhanced by the presence of even small amounts of inert gas in the inspired mixture.

Endocrine studies show that hypophysectomy and adrenalectomy protect against hyperoxia. Adrenocorticotropic hormone (ACTH, corticotropin) and cortisone reverse this effect and, when given in normal animals, enhance toxicity. Adrenergic-blocking drugs, some anaesthetics, GABA, lithium, magnesium and superoxide dismutase have a protective effect. Adrenaline, atropine, aspirin, amphetamine and pentobarbital are among a host of agents that augment toxicity.

Light, noise and other stressful situations also affect CNS tolerance. Thus, the general stress reaction, and more specifically adrenal hormones, may have a role in enhancing CNS (and pulmonary) toxicity. Several observations suggest a role of the autonomic nervous system in modifying the degree of toxicity. Convulsions have been shown to hasten the onset of pulmonary oxygen toxicity in some animal studies. This may be related to an activation of the sympatho-adrenal system during convulsions.

Table 17.1 contains a list of factors that increase oxygen toxicity.

Factors increasing oxygen toxicity

Oxygen Toxicity: History

Oxygen was ‘discovered’ in the latter half of the eighteenth century and immediately excited interest in its possible therapeutic effects. Following a series of experiments, in 1775 Priestley was among the first to suggest that there may be adverse effects of ‘dephlogisticated air’ – that is, air that was free from ‘phlogiston’, a substance thought to be released from a burning object1. Priestley had observed the rapid burning of a candle and speculated that ‘the animal powers be too soon exhausted in this pure kind of air’. In fact, in 1772 Carl Scheele had already postulated the existence of a substance he called ‘fire air’ that supported combustion (later to be called ‘oxygen’ by Anton Lavoisier) following a similar experiment. Later, in 1789, Lavoisier and Sequin demonstrated that oxygen at 1 ATA does not alter oxidative metabolism but did note a damaging effect on the lungs.

In 1878, Paul Bert published his pioneer work La pression barometrique, in which he presented the results of years of study of the physiological effects of exposure to high and low pressures. He showed that although oxygen is essential to sustain life, it is lethal at high pressures. Larks exposed to air at 15 to 20 ATA developed convulsions. The same effect could be produced by oxygen at 5 ATA. Bert recorded similar convulsions in other species and clearly established the toxicity of oxygen on the CNS, also known as the Paul Bert effect. He did not report respiratory damage1,2.

In 1899, the pathologist J. Lorrain Smith noted fatal pneumonia in a rat after exposure to 73 per cent oxygen at atmospheric pressure. He conducted further experiments on mice and gave the first detailed description of pulmonary changes resulting from moderately high oxygen tensions (approximately 1 ATA) for prolonged periods of time. Smith was aware of the limitations that this toxicity could place on the clinical use of oxygen. He also noted that early changes are reversible and that higher pressures shortened the time of onset. Pulmonary changes are also called the Lorrain Smith effect.

Although numerous animal studies were performed, evidence of the effect of high-pressure oxygen on humans was sparse until the 1930s. In 1933, two Royal Naval Officers, Damant and Philips, breathed oxygen at 4 ATA. Convulsive symptoms were reported at 16 and 13 minutes. Behnke then reported a series of exposures to hyperbaric oxygen. Exposure at 4 ATA terminated in acute syncope after 43 minutes in one subject and convulsions at 44 minutes in the other. At 3 ATA no effects were seen after 3 hours, but at 4 hours some subjects noted nausea and a sensation of impending collapse. At that time it was believed that 30 minutes of exposure at 4 ATA and 3 hours of exposure at 3 ATA were safe for men at rest. That the dose was important was confirmed in 1941 when Haldane reported a convulsion in less than 5 minutes at 7 ATA oxygen.

Meanwhile, at lower pressures, in 1939 Becker-Freyseng and Clamann found that 65 hours of exposure to 730 mm Hg oxygen at normal atmospheric pressure produced paraesthesiae, nausea and a decrease in vital capacity (VC).

At the beginning of World War II, some unexplained episodes of unconsciousness were noted in divers using closed-circuit rebreathing oxygen sets at what were considered safe depths. This prompted Donald, in 1942, to commence a series of experiments on oxygen poisoning (Figure 17.2)3. His observations in more than 2000 exposures form the basis of current oxygen diving limits. Unfortunately, many of his experiments were performed using rebreathing equipment, without CO2 measurement. Among the more important findings were the marked variation of tolerance and the aggravating effects of exercise and underwater exposure. Donald suggested a maximum safe depth for oxygen diving of 8 metres.

In 1942 and 1943, Donald carried out extensive testing for oxygen toxicity in divers.
Figure 17.2 In 1942 and 1943, Donald carried out extensive testing for oxygen toxicity in divers. The chamber is pressurized with air to 3.7 bar. The subject in the centre is breathing 100 per cent oxygen from a mask. (From Donald KW. Oxygen poisoning in man: I, II. British Medical Journal 1947;1:667–672, 712–717.)

Research since the 1980s has been primarily directed at elucidation of the mechanism of the toxicity. Workers have looked at such factors as the role of inert gas and CO2, blockage of airways and atelectasis, changes in lung surfactant, changes in cellular metabolism, inhibition of enzyme system and the role of the endocrine system. Also, further efforts to delineate the pulmonary limits of exposure have been undertaken. This has become increasingly important with saturation diving involving prolonged stays under increased ambient pressure and the use of oxygen mixtures to shorten decompression time.

Oxygen Toxicity: Introduction

Oxygen toxicity is not encountered in routine scuba diving using compressed air. It is a consideration when higher partial pressures of oxygen are used in the inspired gas. Increased oxygen concentration and increased ambient pressure lead to higher partial pressures of oxygen. Divers may use high-oxygen gas compositions to reduce inert gas narcosis, reduce decompression obligations or prolong underwater time.

Central nervous system toxicity, manifested by convulsions, is potentially lethal in the diver.

Pulmonary toxicity is more likely in the longer exposures of saturation chamber diving or hyperbaric oxygen therapy.

Oxygen also has a major role in therapy of many diving disorders.

Toxicity can be avoided by controlling the inspired partial pressure of oxygen and/or the duration of exposure. It can be delayed by intermittent exposure.

The normal partial pressure of oxygen (PO2) in air is approximately 0.2 ATA. Although essential for survival, oxygen may become toxic at an elevated PO2, and the complex systems we have for defending ourselves from oxygen toxicity are a testament to the evolutionary pressure to use this highly reactive molecule. A rise in the inspired oxygen fraction (FIO2), an increase in the environmental pressure or a combination of both will elevate the inspired PO2 (PIO2).

High PIO2 has several physiological effects on the body. Although increased PIO2 has no direct effect on ventilation, there is a decrease in alveolar and arterial carbon dioxide (CO2) buffering tension caused by the reduction in CO2-carrying capacity of haemoglobin. Other physiological responses to high oxygen include vagally mediated bradycardia and vasoconstriction of intracranial and peripheral vessels. There is a small rate-dependent fall in cardiac output.

High PIO2 is known to be associated with retinopathy of prematurity in pre-term infants and lung damage, convulsions, red blood cell suppression, visual defects, myopia and cataracts in adults. In vitro, toxic effects on cells of many other organs have also been demonstrated.

In diving, toxic effects on the central nervous system (CNS) and lungs are of prime importance, and only these are discussed in detail. The CNS threshold is above 1.5 ATA, and the pulmonary threshold is 0.55 ATA. At 1.6 ATA oxygen, pulmonary toxicity is the limiting factor regarding duration of exposure, whereas at higher pressures neurological toxicity is of prime concern.

In both the CNS and lungs there is a latent period before the onset of detectable toxicity. This delay enables high PO2 to be used for increasingly short periods as the PIO2 rises (Figure 17.1).

Predicted human pulmonary and central nervous system (CNS) tolerance to high-pressure oxygen
Figure 17.1 Predicted human pulmonary and central nervous system (CNS) tolerance to high-pressure oxygen.

In diving and diving medicine, oxygen toxicity is possible in the following situations:

  1. Closed-circuit and semi-closed-circuit rebreathing equipment.
  2. Use of high FIO2 mixtures.
  3. Saturation diving.
  4. Situations in which oxygen is used to shorten decompression times.
  5. Oxygen therapy for diving disorders (pulmonary only, and with prolonged use).
  6. Therapeutic recompression.
  7. Respiratory failure (e.g. near drowning).