Alternative Hypotheses and Future Directions

One newer avenue of research in relation to DCS pathophysiology that is very active but not concluded as this book goes to press relates to the role of so-called intravascular ‘microparticles’. Microparticles are small fragments of membrane material that are shed from the surface of some formed elements of blood and also endothelium. Their presence in the circulation appears to activate or amplify inflammatory processes and coagulation, and as a result they appear capable of initiating or at least exacerbating vascular injury. Microparticles are increased in a variety of disease states including sepsis, myocardial infarction and vasculitis.

Microparticle numbers also appear to be increased by decompression stress, that is, in the presence of tissue supersaturation with inert gas. This raises the possibility that some of the pathophysiological events in DCS that are currently attributed to circulating bubbles may in fact be caused or exacerbated by microparticles. However, many questions remain unanswered. For example, is microparticle generation in decompression stress secondary to bubble formation, or is there some other unknown consequence of inert gas supersaturation that is responsible? How can the invariably wide distribution of harmful microparticles be reconciled against the selective vulnerability of certain organs such as the spinal cord and inner ear in DCS? Similarly, why does the brain seem relatively resistant to harm when in fact its luxurious perfusion would render it at highest risk of exposure to microparticles?

The answers to such questions are unlikely to be simple. The more effort that goes into researching the microparticle phenomenon, the more complex it seems to become. For example, it seems clear that microparticles are not homogeneous; they have different origins, some appear to have mixed origins and they seem to have variable inflammatory potential. Some larger microparticles even appear to have characteristics of gas micronuclei and may contain a core of gas.

This is an exciting line of research that is in its infancy but that has the potential to modify our pathophysiological paradigm for DCS significantly. Alternatively, it may transpire that microparticles are little more than an interesting epiphenomenon. It is a subject for those interested in the detail of this matter to watch closely.

Pathological Effects of Bubbles

Having reviewed the processes that give rise to bubble formation after decompression, the discussion now turns to consideration of the potential pathological effects of those bubbles. As discussed earlier, bubbles may form from dissolved gas in blood and/or tissue, and this section considers their potential pathological effects within those broad ‘anatomical’ categories. The discussion further subdivides the consequences of bubbles forming in blood into those effects arising primarily because of the presence of bubbles in blood and those arising from their distribution in blood via the circulatory system.

Presence of bubbles in blood

The presence of bubbles in blood appears capable of activating reactive formed elements such as platelets and white blood cells, as well as inflammatory cascades such as the kinin, complement and coagulation systems. The means by which this occurs are not clearly established, but there are several possibilities for interactive activations involving known mechanisms. For example, platelet and coagulation activation can occur through exposure to foreign surfaces such as glass or to collagen beneath damaged endothelium. Bubbles may act in a similar way by acting as a foreign surface or by damaging endothelium, the latter having been demonstrated experimentally. Similarly, white blood cells can be activated by damage to endothelium.

There is rarely any obvious evidence of a role for these inflammatory activations in the pathogenesis of milder forms of DCS, although it has been suggested that constitutional symptoms such as fatigue and malaise may occur as a result. These processes may be important in producing specific serious manifestations. For example, one hypothesis for spinal cord injury that has been substantially demonstrated in an animal model holds that bubbles initiate coagulation in the epidural veins, thereby leading to venous stasis and a venous infarction in the spinal cord. Inflammatory responses may also be important contributors in severe or fulminant DCS. For example, disseminated intravascular coagulation can lead to a coagulopathy, and such patients may also develop severe haemoconcentration and shock secondary to the endothelial ‘leakiness’ that can follow inflammatory activations.

Distribution of bubbles in blood

It has been previously noted that bubbles forming from supersaturated dissolved gas first appear in the veins. In addition to any inflammatory responses and resulting harm that this may elicit (described earlier), these bubbles can be transported in the circulatory system to sites distant from their point of origin.

The most obvious target organ in the distribution of VGE is the lungs. As previously discussed, it seems that for the most part the lungs can tolerate and efficiently remove even grade IV VGE without obvious harm. However, on relatively rare occasions divers may develop symptoms caused by the arrival of VGE in large numbers over a sufficiently short space of time to exceed a poorly understood clinical threshold. The trapping of copious bubbles in the pulmonary vasculature can cause a significant rise in pulmonary artery pressure, a ventilation-perfusion mismatch and the various inflammatory changes described earlier. These events manifest as dyspnea, retrosternal chest pain and cough, a constellation of symptoms sometimes referred to as the ‘chokes’, but more correctly called cardiopulmonary DCS. Symptoms typically occur within the first 30 minutes after surfacing, but the natural history is variable. There may be rapid progression to hypotension, collapse and occasionally death, probably as a result of acute right-sided heart failure. There are also many stories of these symptoms resolving spontaneously, especially if oxygen is breathed, because this would markedly enhance the rate at which bubbles are eliminated from the pulmonary circulation. Not surprisingly, because the substantial numbers of VGE required to produce cardiopulmonary symptoms are predictive of a potential for distribution of VGE in to the arterial system (see later) and also of a strong provocation for bubble formation in tissues, it is not surprising that cardiopulmonary DCS is often followed by symptoms of other organ involvement.

VGE may also distribute into the arterial circulation via pulmonary shunts or a PFO as previously described. Little is known about the factors that promote pulmonary shunting of VGE, although formation of large numbers of VGE appears important, and experiments in non-diving situations suggest that exercise is a likely risk factor. One small study in animals also suggested that use of theophylline (a bronchodilator) reduced the efficacy of the lungs as a filter for VGE; however, whether this effect was via recruitment of shunts or another mechanism is unknown. Because these shunts appear to be dynamic and cannot be investigated easily, our knowledge of their role in DCS is poor, although they are potentially very significant.

The behaviour of PFOs is better characterized. As described earlier, some PFOs shunt spontaneously, and there are data suggesting that such lesions are the most significant in causation of DCS. Other PFOs require provocation to raise right atrial pressure sufficiently to open the communication. This can be provided by something as simple as bending and straightening, straining to lift something or possibly even coughing. Along with the Valsalva manoeuvre, all these provocations have in common the tendency transiently to impede venous return to the right side of the heart, thereby causing a brief exaggerated ‘rush’ of blood into the right atrium on termination of the impediment. This transiently increases the right atrial pressure and may open a PFO that is normally kept closed by the prevalent positive left-to-right pressure gradient.

The previously mentioned association between the presence of a PFO and cerebral, spinal, inner ear, and cutaneous DCS implies that the passage of small VGE into the arteries can injure these organs. How these small bubbles cause injury is uncertain, although there are several possibilities.


It is known that a large bubble or bubbles embolizing the arterial supply to the brain can interrupt flow sufficiently to cause ischaemic injury. Patients embolized by large aliquots of gas in iatrogenic accidents and pulmonary barotrauma events can develop stroke-like syndromes that are often multifocal (see Chapter 6). However, it is uncertain that bubbles in the size range that typically arise in the veins after decompression would cause such injuries. Unlike solid emboli, bubbles are known to redistribute through the microcirculation, and small bubbles are likely to do so very quickly. Therefore, it seems implausible that small VGE passing to the arterial system would cause large cerebral infarctions. When such lesions are seen after diving, there should be a high index of suspicion for arterial gas embolism secondary to pulmonary barotrauma.

Nevertheless, it is plausible that sufficient numbers of these small bubbles entering the cerebral circulation may result in dysexecutive symptoms such as cognitive impairment, memory impairment and confusion, which are often labelled ‘cerebral DCS’. Exactly how small bubbles do this, however, is not certain. One possibility that has some experimental providence is that the passage of small bubbles through the microcirculation causes endothelial disruption with consequent inflammatory changes such as the activation and accumulation of white blood cells (see Chapter 6). This can cause a secondary decline in blood flow with a consequent degree of hypoxia, perivascular infiltration by the formed elements of blood, the release of inflammatory mediators and the development of tissue oedema; all of which may inhibit or injure nearby neurons. The widespread occurrence of this sort of event in the cerebral circulation could explain the ‘global’ cerebral impairment typically seen in cerebral DCS.


Neurological organs other than the brain, such as the inner ear and spinal cord, are almost certainly exposed to fewer arterial bubbles than the brain because of their markedly poorer perfusion, meaning that fewer bubbles will be carried to these organs. Paradoxically, these organs may be more vulnerable to ischaemic injury caused by small bubbles for the same reason.

Modelling studies have demonstrated that after a dive the inner ear, for example, remains supersaturated with nitrogen for much longer than the brain, which washes out excess nitrogen within minutes. Thus, if VGE cross a PFO 15 minutes after a dive and enter the basilar artery, most of them will distribute to the brain, and perhaps a few will find their way into the tiny labyrinthine artery. However, whereas those VGE distributing to the brain will distribute through tissues that are not supersaturated, those entering the inner ear microcirculation will be exposed to supersaturated nitrogen and will likely grow as this gas diffuses into the bubble. Thus, despite being exposed to smaller numbers of bubbles, the inner ear may be selectively vulnerable to ischaemic injury because enlarging bubbles will be more likely to cause flow stasis. This mechanism may help explain the repeated reports of an association between PFO and inner ear DCS (implying that arterialized VGE play a role) and the finding that these patients often present with only inner ear symptoms even though a much larger number of VGE must have distributed to the brain. Although less well studied, it is likely that similar mechanisms can be responsible for spinal DCS, which probably has inert gas kinetics similar to those of the inner ear. Moreover, irrespective of whether the bubbles grow from inward diffusion of supersaturated nitrogen, the spinal cord and inner ear circulations are almost certainly vulnerable to the same pro-inflammatory effects of bubble passage as described earlier for the brain.


Other than the finding that arterialized VGE are implicated, exactly how the arrival of small bubbles in the arterial supply to the skin precipitates rash and itch is uncertain, although it could conceivably involve all the mechanisms (ischaemic and inflammatory) mentioned earlier in relation to the brain and other neurological organs. Indeed, one histopathological study of the cutis marmorata form of skin DCS reported endothelial disruption and perivascular inflammatory infiltrates entirely consistent with bubble-induced vasculitis. Questions remain, however, such as the reason for the typical finding of a localized lesion (often on the trunk) when arterialized VGE must distribute widely to cutaneous vascular territories. As with the selective vulnerability of the inner ear to vascular bubbles, the answer may lie in differing levels of post-dive inert gas supersaturation making some areas of skin more vulnerable than others.

Tissue bubbles

As discussed earlier, bubbles may form in tissues where the supersaturation conditions are favourable. Once formed these bubbles could cause harm through the following mechanisms: direct damage to immediately adjacent tissue; indirect trauma within any tissue displaced by the bubble (e.g. stretching or compression of axons in the spinal cord or a peripheral nerve); ischaemia caused by pressing externally on blood vessels; haemorrhage through disruption of nearby blood vessels; stimulation of pain receptors; and incitement of inflammatory reactions initiated by tissue injury.

The pathophysiological paradigm for DCS described earlier is summarized in Figure 10.7.

Figure 10.7 Summary of key pathophysiological processes in decompression sickness. PFO, patent foramen ovale.
Figure 10.7 Summary of key pathophysiological processes in decompression sickness. PFO, patent foramen ovale.

Multiple disorders in single organ systems

It will not have escaped the astute reader’s attention that in the case of some organs this discussion has proposed more than one way in which they can be injured by bubbles in DCS.

The most conspicuous example is the spinal cord, in which injury may be caused by: bubbles forming in the spinal tissue itself, by bubbles inciting coagulation in the epidural vertebral venous plexus and by bubbles formed in the veins distributing to the cord via a right-to-left shunt and the arterial circulation. It is emphasized that this is not self-contradictory and that in fact the spinal cord may be injured by all these mechanisms under various circumstances. The existence of these various forms of bubble pathology may help explain inconsistent results when divers with spinal DCS are recompressed. For example, failure to elicit any improvement may be explained by coagulation in the epidural veins that would not be expected to resolve during recompression even if the inciting bubbles were removed. In contrast, a brisk improvement soon after recompression could suggest that tissue bubbles pressing on surrounding structures were the main culprit because in this case bubble resolution could be expected to elicit some benefit.

The other important example is the inner ear, which, as previously described, may be injured either by bubbles forming within the tissues or by bubbles that have shunted from the veins and arrive in the arterial system. As explained, if these arterial bubbles arrive while the inner ear remains supersaturated with nitrogen, then they may expand as tissue nitrogen diffuses into the bubble. It is generally presumed that cases of inner ear injury arising during decompression from deep dives are most likely to be caused by bubbles forming within the inner ear tissue itself, particularly if symptom onset is temporally related to a switch from a helium-based breathing mix to air or nitrox. It is impossible to distinguish among disorders in cases arising after surfacing, but the strong association between such cases and the presence of a large PFO suggests that a significant proportion of them are caused by arterial bubbles.

Fate of Bubbles After Formation

The pathogenicity of bubbles is described later, but it is interesting to reflect on what happens to a notional bubble in tissue over time if only to discuss the concept of the so-called oxygen window because this will arise again in several subject areas.

We must begin by assuming that when a diver surfaces from a dive, a bubble forms from dissolved supersaturated nitrogen in a tissue. While the tissue remains supersaturated with nitrogen the bubble will grow, but eventually the PN2 in the tissue will come to equilibrium with that in the blood and alveoli for an air breathing subject at 1 atm. The bubble will not grow any more, and one may then ask ‘What is to stop the bubble, once formed, from simply sitting in the tissue unresolved for long periods?’ There are two reasons.

First, the pressure inside a spherical bubble is always likely to be greater than 1 atm, creating a driving force for the inert gas contained therein to diffuse into tissue, thence to blood and alveolus. This is because some pressure will be generated by surface tension at the bubble–tissue fluid interface and because some pressure will be generated by surrounding tissue that has been displaced by the bubble. These factors are summarized by the following equation, whose middle term describes pressure resulting from surface tension and whose final term describes pressure resulting from tissue displacement:

where Pbub is the pressure inside the bubble, Pamb is the ambient pressure, σ is the surface tension of the fluid, r is the bubble radius, Vtis is the volume of tissue affected by bubble displacement, and B is a term describing the bulk modulus of elasticity of the tissue.

Second, even during air breathing, there is a small partial pressure gradient for nitrogen diffusion from bubble to tissue created by the oxygen window. This arises primarily because of the solubility difference between the oxygen consumed and the carbon dioxide (CO2) produced by metabolism. The partial pressure of oxygen (PO2) in alveolar gas during air breathing is approximately 100 mm Hg, and after exchange with the blood, the PO2 in arterial blood is about 95 mm Hg. Oxygen is carried to the tissues, where a given number of molecules are consumed through metabolism and replaced with a similar number of molecules of CO2. Removal of these oxygen molecules drops the PO2 from 95 mm Hg in arterial blood to 40 mm Hg in venous blood. However, because CO2 is much more soluble, the addition of the same number of molecules of CO2 to the venous blood only raises its partial pressure to 46 mm Hg (from 40 mm Hg in arterial blood). The PO2 in the tissues where the oxygen is actually being consumed is slightly lower than the venous PO2, but this is difficult to measure, so we have to speculate a little. Relevant data are summarized in Table 10.2.

Approximate gas partial pressures in millimetres of mercury at various sites during air breathing at a surface pressure of 1 ATA

Note from Table 10.2 that for the purposes of this discussion it is assumed that a tissue bubble has an internal pressure equivalent to ambient (760 mm Hg). As discussed earlier, the typical internal pressure of a bubble in tissue is probably higher, which would actually enhance the effect described here, but for the purposes of illustrating the oxygen window, we will assume that the internal pressure is same as ambient. The gas contained within the bubble will be composed of water vapour at a pressure equivalent to the saturated vapour pressure for water at 37°C (47 mm Hg), and oxygen and CO2 in equilibrium with the tissue pressures of those gases. The balance of the bubble gas must be nitrogen, and by Dalton’s Law of partial pressures, the PbubN2 must be given by this equation:

where PbubN2 is the pressure of nitrogen inside the bubble, Pbub is the pressure inside the bubble that in this example we are considering to be the same as ambient pressure (760 mm Hg), PbubO2 is the partial pressure of oxygen in the bubble, PbubCO2 is the partial pressure of CO2 in the bubble, and PH2O is the saturated vapour pressure for water at 37°C.

Table 10.2 shows that this resolves to a PN2 of approximately 637 mm Hg which is about 64 mm Hg greater than the PN2 in the tissue, venous blood and alveoli (573 mm Hg). This difference, which creates a gradient for diffusion of nitrogen from the bubble to the tissue, is referred to as the ‘oxygen window’. We reiterate that it is created by the dissolved gas partial pressure difference that arises from removing relatively insoluble oxygen from solution and replacing it with very soluble CO2.

As discussed again later, the oxygen window could be further enhanced by breathing oxygen. Although this markedly elevates the alveolar and arterial PO2, it has a much smaller effect on venous and tissue PO2 because the small amount of extra oxygen dissolved in the arterial blood will be preferentially removed and metabolized, thereby dramatically dropping the PO2 back down to near normal levels. The venous PO2 (and therefore the venous oxygen saturation) may be marginally elevated. Because the same amount of oxygen is consumed and the same number of molecules of CO2 is produced, there will be virtually no effect on tissue or venous PCO2. Thus, the PbubN2 as calculated earlier will change very little, while at the same time the alveolar, arterial and tissue PN2 will fall markedly; potentially to zero if 100 per cent oxygen is breathed for long enough. The difference in PN2 between bubble and surrounding tissue will be correspondingly exaggerated, and nitrogen will diffuse out of the bubble more quickly. The same is true if the bubble is compressed. In this case, the Pbub in the foregoing equation is elevated, whereas bubble oxygen, CO2 and water vapour are little affected, even if oxygen is breathed during the compression.

The existence of the oxygen window, even during air breathing at 1 atm, at least partly explains why bubbles of nitrogen cannot exist in a stable condition in tissues. It also explains why bubbles involute even more quickly during oxygen breathing, especially when combined with recompression.

Bubble Formation: Tissue Bubbles

Bubble formation from dissolved nitrogen in tissues is plausible wherever the supersaturation conditions are favourable. However, this is the least understood and documented of the processes likely to contribute to DCS mainly because, unlike bubbles moving in blood, tiny bubbles in tissue are difficult to detect with current technology, and studying small pathological bubbles in tissue post mortem is notoriously difficult.

The spinal cord is the most scrutinized organ in this regard. The 1990s saw a number of studies published that demonstrated ‘non-staining space-occupying lesions’ 20 to 200 micrometres in diameter and presumed to be bubbles in the spinal cord white matter after provocative decompressions. These experiments determined that the supersaturation threshold was moderately high for formation of these bubbles and that significant bubbling in the spinal cord white matter was unlikely unless dives were deeper than 25 metres. They also determined that tissue bubbles tend to form in the spinal cord early, because as supersaturation declines quickly after surfacing the probability of bubble formation also rapidly declines.

The brain tissue is not considered a likely site for bubble formation because of its luxurious perfusion, which prevents significant or prolonged supersaturation after most plausible decompressions. However, nitrogen is eliminated more slowly from the inner ear than the brain, and there is some experimental evidence for bubble formation in inner ear tissue itself. The inner ear is also uniquely vulnerable to enhancement of local tissue supersaturation by a process frequently referred to as ‘isobaric counterdiffusion’ (IBCD). This can arise during decompression from deep dives in which the diver makes a switch from a breathing mix containing helium to one containing nitrogen (see Chapter 62). Such switches are undertaken in the belief that they accelerate decompression because helium in the tissues will diffuse into blood faster than the nitrogen in the blood will diffuse into the tissue. The inner ear has a unique and relevant anatomy. There are relatively large unperfused reservoirs of helium (in the perilymph and endolymph) that can eliminate accumulated helium only via the vascularized labyrinthine tissue. This maintains an elevated partial pressure of helium in the vascular labyrinth after the switch to a nitrogen-based mix, while at the same time this tissue is also exposed to high pressures of nitrogen diffusing inward from the blood stream. This ‘counterdiffusion’ process may transiently enhance any pre-existing supersaturation of helium and result in bubble formation.

In addition to clinically relevant and largely proven tissue bubble formation in the spinal cord and inner ear, there is strong circumstantial evidence that tissue bubble formation is the cause of musculoskeletal pain in DCS. Specifically, the lack of any association between the presence of PFO and musculoskeletal pain in DCS suggests that in situ bubble formation is the most likely cause, rather than bubbles arriving in the arterial blood. The exact location of tissue bubbles responsible for musculoskeletal pain is unknown, but there are multiple possibilities including tendons, ligaments, periosteum and marrow. Similar reasoning has led to the hypothesis that bubbles may form in peripheral nerve tissues. Patchy paraesthesiae in a non-dermatomal distribution are common symptoms that have not been linked to the presence of a right-to-left shunt, and it follows that tissue bubble formation is the likely cause. It is plausible that bubbles could form in the myelin of a peripheral nerve, or elsewhere within the perineurium, and cause neurapraxia through a mass effect. However, this mechanism is not substantively proven.

Finally, one presumed location for bubble formation most appropriately categorized as ‘tissue’ is the lymphatic system. The infrequent occurrence of discrete regional areas of oedematous soft tissue swelling, often accompanied by other symptoms of DCS, has led to the assumption that bubbles may form in lymphatic drainage channels and cause stasis.

Bubble Formation: Arterial Bubbles

Bubbles almost certainly do not form from dissolved nitrogen in the arterial circulation because once the venous blood passes through the lungs, the dissolved PN2 in the blood and alveoli should have equilibrated, and the arterial blood leaving the lungs will no longer be supersaturated. Bubbles can be introduced into the arterial circulation by pulmonary barotrauma (see Chapter 6) and also by left-to-right (venous-to-arterial) transfer of VGE. This can occur via several recognized ‘shunt’ pathways.

The lesser known and least researched of these pathways are pulmonary ‘shunts’. The existence of such shunts has been known for some time. They can be detected in some subjects at rest and in many subjects during exercise. Indeed, their physiological role may be to unload the right side of the heart to some extent during heavy exercise. The increasing use of echocardiography in the observation of bubble behaviour after diving, and also in saline contrast tests for patent foramen ovale (PFO; see later), has revealed that VGE can sometimes be seen emerging from the pulmonary veins into the left side of the heart. Thus, these bubbles have crossed the pulmonary circulation rather than an intracardiac shunt. It is notable that one small study failed to detect pulmonary shunting of VGE in divers exercising after a dive. Nevertheless, pulmonary shunts may contribute to the development of the DCS syndromes known to be associated with right-to-left shunting of VGE (see later), especially where there is no PFO to explain the mechanism.

The better known and most widely researched pathway is an intracardiac shunt, usually a PFO. The foramen ovale is a communication through the atrial septum that during fetal life allows blood arriving at the right side of the heart in the inferior vena cava to be directed straight across the septum into the left atrium, thereby by-passing the right ventricle and pulmonary circulation. With the haemodynamic changes that occur at birth, the foramen ovale closes in a valve-like manner, with the higher left atrial pressures tending to keep it shut. In the majority of people, the tissue pads that close the foramen ovale become ‘healed’ in the shut position, but in a minority (some 25 to 30 per cent), the foramen ovale remains open, or at least able to open should pressures in the right atrium exceed those on the left for any reason. This is referred to as a ‘patent’ foramen ovale (PFO).

As implied earlier, a PFO can be found in 25 to 30 per cent of adults who are unaware they have one, and most go through life suffering no ill effects. However, there are now multiple case-control studies that collectively demonstrate associations between the presence of a PFO and DCS involving the brain, spinal cord, inner ear and skin. In the various relevant studies, these associations are established by a substantially higher prevalence of PFO among cases of DCS than found among control divers who have not suffered DCS. If we cautiously accept that causation can be inferred from this association, the clear implication is that VGE that become ‘arterialized’ across a PFO are important in the pathophysiology of these forms of DCS. The way these small arterial bubbles may cause harm is discussed in more detail later. Another unsurprising and consistent finding among the relevant studies is that the size, or more correctly the shunting behaviour, of the PFO seems important. Thus, a grade 1 (see later) or ‘small’ PFO is likely to represent little if any risk, whereas a grade 3 or spontaneously shunting PFO almost certainly imparts extra risk for the relevant forms of DCS.

Not surprisingly, divers may request testing for the presence of a PFO. The issues of which divers should be tested for a PFO and what should be done when a PFO is found are discussed later in Chapter 12; however, the testing process is described briefly here because some of the terminology that arises is relevant to discussion of the pathogenicity of arterial bubbles. The process involves performing echocardiography while introducing agitated saline (which somewhat paradoxically contains many small bubbles) into a peripheral vein. The arrival of the bubbles in the right side of the heart often causes its virtual opacification on echocardiography, and the left side of the heart is then monitored to see whether bubbles cross the interatrial septum. Release of a Valsalva manoeuvre causes a temporary rise in right atrial pressure and is used to unmask a PFO that remains closed most of the time but that can open if right atrial pressure rises. The results are often crudely graded as follows: 0 = no bubble shunting; 1 = few bubbles shunted even during a Valsalva manoeuvre; 2 = moderate numbers of bubbles shunted during a Valsalva manoeuvre; 3 = spontaneous shunting of bubbles without a Valsalva manoeuvre.

There has been debate about whether echocardiography should be transthoracic (less expensive and less invasive) or transoesophageal (better-quality imaging) for these tests. In general, it is agreed that in expert hands and provided good views can be obtained, transthoracic echocardiography is ideal for this purpose. In fact, it is more likely to result in accurate studies because patients are typically better able to cooperate with provocative manoeuvres such as Valsalva manoeuvres than during transoesophageal echocardiography, when patients may be uncomfortable or sedated. If transthoracic views are poor, then a transoesophageal investigation should be considered. Another variant of the ‘PFO test’ is the use of carotid or transcranial Doppler imaging to detect bubbles in the respective arteries after injection of saline contrast and a Valsalva manoeuvre. These tests detect a right-to-left shunt, but they do not definitively distinguish among the lesions that are potentially permitting it (e.g. PFO, pulmonary shunt, atrial septal defect).

Bubble Formation: Venous Bubbles

The incidence of decompression sickness associated with venous gas emboli grades (Kisman-Masurel system) following human dive trials at the Canadian Defence and Civil Institute of Environmental Medicine

Of all the sites where bubbles form from supersaturated dissolved gas, most is known about bubbles in the veins. This reason in no small part is that Doppler technology has made it relatively simple to detect bubbles moving in the venous system. Doppler ultrasound systems can detect flow in blood vessels and generate an audible flow signal. The passage of a bubble through the ultrasound beam causes a characteristic ‘chirp’ over and above the background ‘whooshing’ sound of the flowing blood. More recently, high-quality echocardiography has become feasible with the use of small portable devices. This allows direct visualization of bubbles as they arrive in the right side of the heart. Much research about the quality of decompression has been conducted using quantitative estimates of venous bubble ‘grades’ as the outcome measure.

It has become clear that when bubbles form from supersaturated dissolved gas in blood they first appear in the veins (as opposed to the arteries, which are discussed later). This is logical because these bubbles almost certainly develop in the capillary beds of supersaturated tissues, and they can be expected to distribute with the flow of blood into the venous system. Bubbles may be present minutes after a dive, but their detection typically peaks around 30 minutes after surfacing and then may be sustained for several hours. Studies in animals suggest that these bubbles vary in size between 19 and 700 micrometres. This variability is difficult to interpret because once formed, bubbles may coalesce to create larger bubbles. Nevertheless, when it is considered that capillary diameter is somewhere around 5 to 10 micrometres, it is clear that even the smallest of these bubbles are likely to interact physically with any tissue bed that they enter.

Bubbles forming in the veins, often referred to as ‘venous gas emboli’ (VGE), pass to the right side of the heart and thence to the lungs, where they encounter a capillary network for the first time. Given that the pulmonary circulation is a low-pressure system, it is perhaps not surprising that the lungs are an efficient filter for VGE. Removal of most incoming VGE by the lungs has been documented in numerous experiments in both animals and humans. In most cases this does not appear to be associated with obvious harm, although harm can occur (see later) and this is probably dependent on the number of bubbles and their rate of arrival.

The formation of VGE after diving has been extensively studied. These studies have required a standardized means of quantifying the detected bubbles. There are two bubble classification and scoring systems commonly in use: the methods described by Spencer, and by Kisman and Masurel.

Spencer’s grading system monitors the precordium for bubbles with the subject sitting quietly:

Grade 0 – No bubble signals on Doppler.

Grade I – An occasional bubble but with most cardiac periods free.

Grade 2 – With many, but less than half, of the cardiac periods containing Doppler signals.

Grade 3 – Most of the cardiac periods containing showers or single bubble signals, but not dominating or overriding the cardiac motion signals.

Grade 4 – The maximal detectable bubble signals sounding continuously throughout the heart cycle and overriding the amplitude of the normal cardiac signal.

The Kisman-Masurel classification system was designed with the aim of easily incorporating the bubble signal into a computer program. The bubble signal is divided into three separate categories – frequency, percentage of cardiac cycles with bubbles/duration of bubbles and amplitude. Each component is graded separately and then a single-digit bubble grade is awarded. The primary site for monitoring is the precordium over the right side of the heart. Monitoring is conducted at rest and after a specified movement, typically a deep knee bend (squatting up and down in a continuous fashion). Although this system appears more complicated than the Spencer system, it is the preference of many researchers and is easily learned.

Both systems suffer from interobserver error with some difficulties in subjectivity. Although computer-based counting programs are being developed, human observers are still more accurate than the automated models.

Most studies of VGE formation have had one or both of two interrelated goals in mind – first, to establish the relationship between appearance of VGE and DCS; and second, refining decompression strategies. In regard to the first goal, it has become abundantly clear that VGE are commonly detectable after recreational dives that do not result in DCS. Indeed, even the presence of high-grade bubbles is often not associated with the appearance of symptoms, and the positive predictive value of VGE grades for DCS is thus poor. Large studies do report a correlation between VGE grade and risk (Table 10.1), but this is not precise enough for VGE grades to be used in diagnosis of DCS.

The incidence of decompression sickness associated with venous gas emboli grades (Kisman-Masurel system) following human dive trials at the Canadian Defence and Civil Institute of Environmental Medicine
The incidence of decompression sickness associated with venous gas emboli grades (Kisman-Masurel system) following human dive trials at the Canadian Defence and Civil Institute of Environmental Medicine

The lack of a ‘diagnostic correlation’ between VGE and DCS symptoms is perhaps not surprising when it is considered that some manifestations of DCS are almost certainly not directly caused by VGE (see later). However, even the weak correlation demonstrated in Table 10.1 suggests that numbers of VGE are at least indicative of the probability of significant bubble disorders in other sites, and this has encouraged the use of VGE grading as an admittedly imperfect tool for assessing and refining decompression strategies. The ongoing use of VGE grading in this regard is also influenced by the lack of readily acceptable alternative outcome measures. For example, performing manned dive trials using DCS as the outcome requires extremely large studies and may encounter ethical objections. It should be obvious from this discussion that although VGE counts are a useful tool in decompression research, great care must be taken over the conclusions that are drawn in relation to the risk of DCS.

Bubble Formation: Gas Micronuclei

Depiction of a gas micronucleus resident in a crevice on a blood vessel wall.

It is noteworthy that relatively small gas supersaturations seem capable of provoking bubble formation in vivo yet huge supersaturations are required to provoke bubble formation in pure solutions. This is almost certainly because of massive pressures caused by surface tension forces at the fluid-gas interface of an evolving bubble. These are inversely proportional to bubble radius, and a small bubble nucleating de novo from supersaturated dissolved gas would need to overcome them. The most popular and widely accepted explanation for in vivo bubble formation from relatively small supersaturations is the postulated presence of tiny gas micronuclei in blood and possibly tissues. These micronuclei are hypothesized to act as seeds for the inward diffusion of supersaturated gas after ascent from a dive, thus causing them to grow into bubbles.

Both the source and nature of these micronuclei are uncertain, but one suggestion is that they are created by tribonucleation in tissues where movement creates momentary areas of depressurization within the tissue fluid or blood. Bubbles created under such circumstances are usually tiny and rapidly involute in response to the high pressures caused by surface tension mentioned earlier that force the gas back into solution. However, micronuclei theory holds that some of these bubbles acquire a stabilizing outer layer of surface active molecules (these being extremely common in vivo) that reduce surface tension and increase the life span of the bubble, which now becomes a micronucleus.

Evidence supporting the existence of micronuclei is largely circumstantial, beginning (as described earlier) with the mere fact that bubbles can form in vivo following relatively small supersaturations. In addition, in vivo experiments in which a short ‘spike’ to particularly high pressure was imposed before a decompression demonstrated less bubble formation than expected; a finding implying that at least some micronuclei had been crushed out of existence during the high-pressure spike. Consistent with the notion (mentioned earlier) that tissue movement may result in regeneration of micronuclei, one famous experiment using shrimps showed that ‘exercise’ following the high-pressure spike partially reversed the spike’s prevention of subsequent bubble formation.

Variations and parallel theories exist. For example, it has been proposed that micronuclei could reside in imperfections or crevices in capillary walls and that when the underlying tissue becomes supersaturated, nitrogen molecules will diffuse into the micronucleus, thus causing it to grow and bud off bubbles in an analogous manner to the steady stream of bubbles that can be seen issuing from specific points on the wall of a glass containing a carbonated beverage. The essential features of this notion are illustrated in Figure 10.6.

Depiction of a gas micronucleus resident in a crevice on a blood vessel wall.
Figure 10.6 Depiction of a gas micronucleus resident in a crevice on a blood vessel wall. In the unsaturated state (a), the nucleus is quiescent. When the underlying tissue becomes supersaturated (b), gas diffuses into the nucleus and causes it to grow until such time that a bubble breaks off (c), after which the cycle will be repeated.

Another theory invoked to explain bubble formation within tissues themselves is that if tribonucleation on movement or impact creates tiny bubbles in a tissue that is supersaturated at the time, the supersaturated gas will diffuse into the bubbles and not only prevent their early involution, but also cause them to grow. This is analogous the act of shaking an open soda bottle and could explain bubble formation even in the absence of persistent gas nuclei.

Decompression Sickness: Uptake and Elimination of Inert Gas

A tissue depicted as a single well-stirred perfusion-limited ‘compartment’.

Because it is believed that DCS is primarily caused by bubble formation from dissolved gas taken up into tissues during a dive, it is appropriate to examine process of gas uptake and elimination in some detail.

Most diving is performed using air as the respired gas. Dives to depths beyond the recreational diving range are performed using mixed gases that contain helium (see Chapter 62). The key point is that one or more inert gases are breathed in all but very shallow dives, where oxygen rebreathers may be used. For the general purpose of this discussion, the assumption is that the breathing gas is air and therefore that the relevant inert gas is nitrogen.

Underwater breathing apparatus supplies the diver with air at ambient pressure. Thus, the inspired (and alveolar) pressure of nitrogen (PN2) is directly proportional to depth. Exposure to an elevated alveolar pressure of nitrogen will result in its uptake into the arterial blood, as predicted by Henry’s Law, and its distribution to the tissues. Unlike highly soluble anaesthetic vapours, the equilibration of alveolar and arterial PN2 occurs very quickly, and this process can largely be ignored in consideration of inert gas kinetics. In contrast, the exchange of dissolved nitrogen between blood and tissues is influenced by several factors.

There are multiple ways in which blood-tissue gas exchange can be conceived, but the simplest and most common approach in decompression modelling is to consider a tissue to be a well-stirred perfusion-limited compartment (Figure 10.1).

A tissue depicted as a single well-stirred perfusion-limited ‘compartment’.
Figure 10.1 A tissue depicted as a single well-stirred perfusion-limited ‘compartment’.
Nitrogen uptake in a hypothetical perfusion-limited tissue compartment during a dive to 30 metres (4 ATA) using air.
Figure 10.2 Nitrogen uptake in a hypothetical perfusion-limited tissue compartment during a dive to 30 metres (4 ATA) using air. Pamb is the ambient pressure in atmospheres (atm). The inspired pressure of nitrogen and the alveolar pressure of nitrogen rise to ~3.1 atm (not depicted in the figure), and the arterial pressure of nitrogen (PaN2) ‘immediately’ equilibrates. In contrast, the pressure of nitrogen in the tissue (Ptiss) is slower to equilibrate. The arterial-tissue difference decays in a mono-exponential fashion. (Adapted with permission rom a figure by Dr David Doolette.)

Within such a model, the arterial-tissue PN2 difference declines mono-exponentially (Figure 10.2), at a rate determined substantially by tissue perfusion, but also by the blood-tissue partition coefficient for nitrogen. Tissues with luxurious perfusion will take up inert gas quickly, whereas poorly perfused tissues will take up gas slowly. Similarly, tissues in which nitrogen is less soluble than blood will equilibrate quickly, whereas tissues with high solubility for nitrogen will equilibrate slowly. Thus, for the purposes of conceptualizing nitrogen kinetics (and for decompression modelling), the body is often regarded as a set of parallel compartments with differing kinetic properties for nitrogen exchange (Figure 10.3).

Figure 10.3 Conceptual depiction of body tissues as parallel compartments with differing kinetic properties (1 ‘fastest’ to 5 ‘slowest’ – see Figure 10.4) determined largely by perfusion and tissue composition. Pa, arterial pressure of inert gas. (Adapted with permission from a Figure by Dr David Doolette.)
Figure 10.3 Conceptual depiction of body tissues as parallel compartments with differing kinetic properties (1 ‘fastest’ to 5 ‘slowest’ – see Figure 10.4) determined largely by perfusion and tissue composition. Pa, arterial pressure of inert gas. (Adapted with permission from a Figure by Dr David Doolette.)

A notional depiction of differing nitrogen kinetics for each of the five hypothetical tissues in Figure 10.3 is illustrated in Figure 10.4. It can be seen that by the end of the period at depth the inert gas tension in the ‘faster’ tissues has equilibrated with arterial blood, whereas tensions in the slower tissues have not. Those tissues in which equilibration occurs are said to be ‘saturated’ with inert gas for that depth. If sufficient time were spent at depth, then all tissues would eventually saturate, a principle used in so-called saturation diving in which divers live under pressure for extended periods in the knowledge that no further inert gas uptake can occur (unless they venture deeper), and after becoming ‘saturated’ there is no increment in their decompression obligation if they remain under pressure longer. In contrast, the pattern of tissue equilibration with inert gas at the end of the period at depth on a typical recreational dive will look more like that depicted in Figure 10.4. Some of the faster tissues may be saturated and the slower tissues will not be. It should be self-apparent that the deeper the dive, the higher the tissue nitrogen pressures will be when the tissue comes into equilibrium with the arterial tension. Similarly, the longer the diver remains at depth, the further toward saturation the various tissues will progress.

Nitrogen uptake in five hypothetical perfusion-limited tissue compartments during a dive to 30 metres (4 ATA) using air.
Figure 10.4 Nitrogen uptake in five hypothetical perfusion-limited tissue compartments during a dive to 30 metres (4 ATA) using air. Pamb is the ambient pressure in atmospheres (atm). The inspired pressure of nitrogen and the alveolar pressure of nitrogen rise to ~3.1 atm (not depicted in the figure), and the arterial pressure of nitrogen (PaN2) immediately equilibrates. The tissue pressures of nitrogen are slower to equilibrate, with only tissues 1 and 2 approaching saturation within the duration of the exposure depicted. (Adapted with permission from a figure by Dr David Doolette.)

The principles discussed earlier in relation to nitrogen uptake remain relevant to nitrogen elimination during and after ascent from a dive. Ambient pressure decreases as the diver ascends. As ascent begins, the alveolar and arterial pressures of nitrogen will decline, and this will immediately create a diffusion gradient for nitrogen elimination from those tissues that equilibrated with arterial PN2 at the bottom. As the ascent proceeds, similar outward diffusion gradients will be established in increasing numbers of slower tissues that had absorbed less nitrogen during the bottom time. Because the kinetics of nitrogen elimination in these tissues is slower and the ascent is relatively fast, the tissues will tend to develop a state of ‘supersaturation’ in which the sum of dissolved gas pressures in the tissue exceeds the ambient pressure. This is depicted for a single tissue in Figure 10.5. Depending on the rate of ascent, the very fastest tissues may avoid this condition because they eliminate nitrogen extremely quickly, but tissues with slower kinetics are likely to become supersaturated at some point in the ascent.

Nitrogen uptake and elimination in a hypothetical tissue compartment during and after a dive to 30 metres (4 ATA) using air.
Figure 10.5 Nitrogen uptake and elimination in a hypothetical tissue compartment during and after a dive to 30 metres (4 ATA) using air. The tissue does not equilibrate with the arterial nitrogen pressure (PaN2) (become saturated) during the bottom time, but because nitrogen elimination does not match the rapidly declining ambient pressure (Pamb) early in the ascent, the tissue becomes supersaturated; that is, the sum of dissolved gas in the tissue (of which most is nitrogen under these circumstances) exceeds the ambient pressure. The supersaturation pressure (Pss) is represented by the vertical distance between the tissue pressure (Ptiss) and ambient pressure. (Adapted with permission from a figure by Dr David Doolette.)

Supersaturation of a tissue establishes a diffusion gradient for gas to pass from tissue to blood to alveolus, thus facilitating inert gas elimination. However, it is also the pivotal condition required for dissolved gas to separate into the gas phase, that is, to form bubbles.

These bubbles are generally accepted to be the pathological vectors in DCS, and the means by which they cause harm are discussed in detail in the following section. In general terms, a greater degree of supersaturation will drive more bubble formation with a correspondingly higher risk of developing DCS symptoms. Not surprisingly, most dive planning algorithms, used by divers to control their time/depth exposures and decompression procedures, invoke some means of calculating and controlling supersaturation during decompression as a core function. This is also discussed in more detail later.

Decompression Sickness: Historical Perspective

Von Guericke developed the first effective air pump in 1650. This permitted pressurization of gas for respiration at elevated ambient pressure. Shortly afterward, in 1670, Robert Boyle exposed experimental animals to the effects of increased and decreased pressures. His reports of these experiments included the first description of the presumed pathological vector of DCS, a bubble, in this case moving to and fro in the aqueous humour of the eye of a viper. The snake was ‘tortured furiously’ by the formation of bubbles in the ‘blood, juices and soft parts of the body’.

There was a long hiatus before related effects were recognized in humans. In the 1840s Colonel Pasley noted rheumatism and excessive fatigue in divers employed on the wreck of the Royal George. The divers were presumably suffering from DCS, not surprising because their bottom times exceeded the accepted limits by a factor of three.

A French mining engineer, Jean Triger, described cases of DCS in humans in 1841. He designed and constructed what came to be known as a ‘caisson’; a pressurized vertical shaft sunk into sites that would otherwise be flooded such as pylon sites in bridge construction or mine shafts extending below the water table. The air pressure excluded water from the shaft, and men could work at the bottom, albeit under whatever pressure was required to maintain a dry environment. Triger himself suffered knee pain on several occasions after pressure exposure in caissons, and as their use expanded there were fatalities and episodes of paralysis. There were almost certainly manifestations of DCS. Two physicians, Pol and Watelle, in 1854, published a report indicating the nature of the disease, together with case histories to demonstrate the relationship between decompression and onset of symptoms.

Hoppe-Seyler repeated the almost 200-year-old Boyle experiments, and in 1857 he described the obstruction of pulmonary vessels by bubbles and the inability of the heart to function adequately under those conditions. He suggested that some of the cases of sudden death in compressed air workers were the result of this intravascular liberation of gas. He also recommended recompression to remedy this.

Le Roy de Mericourt, in 1869, and Gal, in 1872, described an occupational illness of sponge divers that was attributed to the breathing of compressed air and equated this with the caisson workers’ disease. Although some informed physicians were postulating a role for bubbles, a host of other imaginative theories were proposed during the nineteenth century to explain the aetiology of this disorder.

In 1872, Freidburg reviewed the development of compressed air work and collected descriptions of symptoms of workers given insufficient decompression after exposure to high pressure. He compared the clinical course of severe and fatal cases of DCS with that of the venous air embolism occasionally seen in obstetrics and surgery. He believed that rapid decompression would be responsible for a rapid release of the gas that had been taken up by the tissues under increased pressure. He suggested that the blood was filled with gas bubbles that interfered with circulation in the heart and lungs.

Smith described ‘caisson disease’ or ‘compressed air illness’ in 1873 as a disease dependent on increased atmospheric pressure, but always developing after reduction of the pressure. It was characterized, he noted, by moderate or severe pain in one or more of the extremities and sometimes in the trunk as well. There may or may not be epigastric pain and vomiting. In some cases, there may be elements of paralysis that, when they appear, are most frequently confined to the lower half of the body. Cerebral symptoms, such as headache, vertigo, convulsions and loss of consciousness, may be present.

Paul Bert, in 1878, demonstrated in a most conclusive manner that DCS is primarily the result of inert gas (nitrogen in the case of compressed air divers and caisson workers) dissolved in blood and tissues during pressure exposure and then released into the gas phase during or following decompression. He used various oxygen concentrations to hasten decompression, demonstrated the value of oxygen inhalation once experimental animals developed DCS and proposed the concept of oxygen recompression therapy.

Andrew Smith, a surgeon from the Manhattan Eye and Ear Hospital in New York, noted in the 1870s the origin of the term ‘bends’. Because pain in the hips and lower extremities was generally aggravated by an erect position, the victims often assumed a stooping posture. Sufferers among the workers on the Brooklyn Bridge caissons in New York were the objects of good-natured ridicule by their comrades, who likened their angular postures to a fashionable stoop in walking, termed the ‘Grecian bend’, which was practiced by sophisticated metropolitan women at the time. He was aware of the value of recompression, but this was unacceptable to some of his patients. Instead he used hot poultices, ice packs, hot baths, ergot, atropine, whiskey and ginger – or morphine if the others failed. He constructed the first specialized treatment chamber.

Moir, in the early 1890s, working on the Hudson River caisson tunnel in New York, reduced the DCS death rate from 25 per cent of the work force per annum to less than 2 per cent by the use of more careful decompression from caisson exposures and by recompression therapy for symptomatic cases. However, problems were not eliminated, and in projects requiring exposure to relatively high pressures to construct tunnels under New York’s East River in the early 1900s there remained a significant problem with mortality and morbidity, as reported by Keays in 1909.

Not surprisingly, during the early part of the twentieth century there was considerable controversy regarding the speed and manner in which divers and caisson workers should be decompressed. An English physiologist, John Scott Haldane, proposed equations to describe tissue inert gas kinetics and proposed that limiting inert gas supersaturation (see later) during decompression was the key to preventing DCS. His decompression tables, first published in 1908, resulted in a remarkable reduction in the incidence of DCS in both diving and caisson work, and variants of Haldane’s methodology are still in use for planning decompression today. These issues are discussed in Chapter 12.

Decompression Sickness: Pathophysiology


Decompression sickness (DCS) is a disease caused primarily by bubbles formed from dissolved gas in blood and/or tissue following a reduction in ambient pressure. It is most commonly seen in divers after surfacing, but it may also occur in aviators ascending to altitude in unpressurized or semi-pressurized aircraft or in astronauts decompressing for space walks. DCS is a puzzling, variable and to some extent unpredictable disorder with a wide range of potential presentations. At one extreme it may be rapidly fatal, whereas at the other it may manifest with mild, non-specific symptoms. Perhaps not surprisingly, it is the most widely recognized medical complication of diving and the subject of much interest among divers and diving physicians alike.