Background to current situation

In England and Wales, the case of Tony Bland set the legal and clinical scene.3 It was decided that ‘a medical practitioner is under no duty to continue to treat such a patient where a large body of informed and responsible medical opinion is to the effect that no benefit at all would be conferred by continuance. Existence in a vegetative state with no prospect of recovery is by that opinion regarded as not being a benefit, and that, if not unarguably correct, at least forms a proper basis for the decision to discontinue treatment and care’.

Thus, withdrawal of ongoing medical treatment, specifically ‘clinically assisted nutrition and hydration’ (the current jargon for providing food and fluid through a gastrostomy or nasogastric tube) was legitimised, even though it was entirely predictable that death would follow. It was argued that someone in a permanent state of unawareness could never accrue any benefit from treatment which was therefore futile and that there was no need to consider the person's best interests; there was no interest.

Consequently, the diagnosis of being in a permanent vegetative state has become a crucial matter: if it is shown to be the case, then it follows without further consideration that treatment can be stopped. If, however, there is any suggestion that the person has even minimal awareness for even brief periods, then a full Best Interests decision has to be undertaken. In 1993, it was anticipated that the court would be involved with cases until sufficient expertise and experience accrued for a sound medical process to take over. No medical process has developed and, instead, a legal statement requiring all cases to come to court was issued in 2007 and updated in 2015.2

The process of establishing that a person is in the vegetative state8—being unaware of himself and his environment—has evolved. In 1993, it was based on expert clinical assessment supported by a discussion paper published by the Ethics Committee of the British Medical Association in September 1992.9 In 1994 and 1995, major documents were published on diagnosing the vegetative state.10–12 As it became clear that some people were in a borderline state between unawareness and full awareness, the minimally conscious state was defined and criteria for its diagnosis were published in 2002.13 Various other guidelines were published, including the revised UK guideline published in 2014,4 which covered all aspects of management of people in a prolonged disorder of consciousness; it did not specifically review the diagnostic criteria.

At the same time, much more attention has been paid to the clinical aspects of people with a prolonged disorder of consciousness, particularly when withdrawal of medical treatment is being considered. Assessment tools and protocols have been developed and reviewed,14 all based on the same two sets of diagnostic criteria.10 ,13 There has been extensive research into the clinical aspects of people with a prolonged disorder of consciousness. Clinical assessments in the context of the legal process have become much more detailed. Therefore, a review of current evidence and its implications is overdue.

What is awareness?

The nature of consciousness and awareness is still a matter of debate,15 ,16 which will not be reviewed here. Clinically, it is important to distinguish between (1) a person's experience (eg, what it feels like to be cold) and (2) a person's ability to generate, perceive and use meaning (ie, an attribute of a stimulus or action that does not arise directly from its physical observed nature or characteristics; its significance). It is not possible to determine a person's experience except through their own report, which is absent if they do not have the other type of awareness. Therefore, the clinical diagnosis depends upon evidence of ‘cognitive awareness’.

One approach, used to diagnose brain death, is to establish that the neurological substrate needed is not functioning. For brain death, a set of diagnostic criteria were developed.17 ,18 Their goal was to establish beyond reasonable doubt that the brainstem itself was no longer functioning in any way. The assumption was that anyone with a non-functioning brainstem would inevitably die within a few weeks regardless of continued ventilation. Patients who had no brainstem function were then classified legally as being dead, even though they still had a beating heart.

It is important to note that the tests were not intended to show that someone was dead. They simply confirmed that the person had no known brainstem neurological function. In other words, the tests were not intended to prove or disprove death, but to prove or disprove that a functionally defined part of the brain no longer had any functional activity.

This contrasts to the situation concerning consciousness or awareness. There is no understanding of how consciousness arises, and there is no equivalent knowledge of the neurophysiological basis for consciousness. Thus, there is no restricted set of functions that can be tested, and it is not possible to develop a set of physiological tests that can prove or disprove that the neurological foundations needed to create and sustain consciousness are intact or otherwise. Instead, ‘we can only infer the presence or absence of conscious experience in another person’.11 ,19

Existing guidance10 ,13 has little discussion of what how consciousness (termed awareness) is to be determined. For example, ‘The minimally conscious state is a condition of severely altered consciousness in which minimal but definite behavioural evidence of self or environmental awareness is demonstrated’13 and ‘The vegetative state is a clinical condition of complete unawareness of the self and the environment accompanied by … ’.10 In both the statements, it is largely left to the clinician to decide what constitutes evidence of awareness. The examples given are based around concepts of meaning and purpose where the observer needs to judge whether purpose is present and/or understanding of meaning (verbal or non-verbal) is required.

A lack of consciousness is not synonymous with a lack of responsiveness. The Glasgow Coma Scale20 acknowledges different responses to pain, all within the score range indicative of being unconscious. Clinically, some people diagnosed as being unaware nonetheless show localising responses to touch, noise and light: ‘However, patients in a vegetative state often have inconsistent primitive auditory or visual orienting reflexes, characterized by a turning of the head and eyes towards peripheral sounds or movements’.11 Moreover, many patients show spontaneous movements especially of the face and eyes, and less commonly of the arms, legs and trunk.

Therefore, the observer needs to distinguish between a coincidental occurrence of a movement with or after a stimulus, and a movement that shows purpose and/or a discriminatory response to a stimulus; this is difficult when there are spontaneous movements such as blinking.

Clinical diagnosis

Clinicians prefer unequivocal criteria to judgement alone when making any diagnosis. However, when considering a criterion to determine the presence of awareness in a person, it is important to recognise that specific items used as markers of some other state—surrogate variables—cannot be equated to that state. For example, certain appearances on a mammogram are a marker for a carcinoma, and certain brain imaging characteristics are a marker for a glioma within the brain. In all cases, the observed item has a more-or-less close relationship with the unobserved item being present. This association is initially established through experiments, but in an individual patient the diagnosis always needs confirmation by doing a further direct test (eg, biopsy), because surrogate markers always and inevitably have false-positive and false-negative rates.

In the case of consciousness, the only criterion available is human judgement.11 ,19 Any suggested ‘tests’ or criteria must be validated against clinical diagnosis, and the false-positive and false-negative rate established specifically in the situation where the test is being used (ie, not in contrast to known healthy people). The clinical diagnosis of consciousness depends upon answering the question: ‘To what extent does the observed behaviour require one or both of (a) extracting some specific meaning from a stimulus and/or (b) formulating and acting towards some specific abstract goal?’

Visual fixation and pursuit

Visual tracking and/or fixation—‘pursuit eye movements or sustained fixation that occurs in direct response to moving or salient stimuli’—is one suggested specific clinical criterion to distinguish between being vegetative and being minimally conscious.13 As a phenomenon, tracking is probably most easily demonstrated using a mirror.21 ,22 There are many problems with this criterion.

No evidence shows that visual pursuit (or visual fixation) is associated with awareness (ie, as a surrogate marker), nor is any evidence given that it requires awareness (ie, that it is directly dependent upon being conscious). Moreover, the evidence shows that some patients have some degree of briefly sustained visual pursuit or tracking.11 There is no logical or evidence-based explanation of how a brief episode of visual pursuit does not require consciousness but a more prolonged episode does. Either visual pursuit requires consciousness, or it does not.

Several studies have found that visual pursuit or fixation may occur in people who have been classified as being in the vegetative state. In one study of 10 people clinically in a stable vegetative state, 5 showed sustained visual fixation and 5 did not. There were no other differences on a range of factors studied, including prognosis.23 In another study, 5 of 14 patients in the vegetative state showed visual pursuit, as compared with 12 of 16 patients in a minimally conscious state.24 In a third study of variability in performance, which found that responsiveness was best in the morning, it was found that three of nine people in the vegetative state showed visual pursuit on at least one occasion, and conversely that 5 of 13 people in a minimally conscious state did not.25

In a detailed study using special eye movement recording, a statistically significant difference was found between people classified as being in the vegetative state and in the minimally conscious state in terms of accuracy of visual fixation and pursuit, but nonetheless between 2% and 9% of trials showed on-target fixation in people in the vegetative state.26

These observations should not be surprising. Visual fixation and pursuit are fundamental properties of the visual system and have evolved particularly to identify faces, especially eyes, and to monitor them in order to understand what the other person might be attending to or intending to do. People will automatically look at the faces and eyes of other people. They are likely to be automatic behaviours, not dependent upon conscious control and often outside conscious control.

The return of visual pursuit in patients in the recovery phase often does precede eventual recovery of consciousness.27 It seems likely that its value as a prognostic indicator of a possible recovery of consciousness has been conflated to it being a surrogate marker of actual consciousness. It is worth noting that another visual phenomenon, a response to a visual threat, is not indicative of awareness and has no prognostic value.28

Other specific tests

A second widely reported ‘test’ supposed to detect consciousness is functional MRI (fMRI).29 Strong philosophical and other arguments have been put forward to suggest that findings in experiments using fMRI cannot be interpreted as demonstrating consciousness.30 ,31

Furthermore, the original study on 54 people with a prolonged disorder of consciousness found that only 1 of 31 people in a minimally conscious state showed an fMRI response in comparison with 4 of 23 people in a vegetative state.29 This is the opposite of what one would expect if the fMRI responses did actually reflect consciousness, strongly suggesting that it is an invalid measure of the presence of consciousness.

A third, small study in 16 patients using EEG32 has also suggested wilful modulation of brain activity in three patients. However, the analysis and interpretation have been questioned, and the status of ‘covert consciousness’ is still unclear.33

Those who argue that technologically based investigations of cerebral activity can be used to detect consciousness need to demonstrate test specificity and sensitivity against agreed clinical criteria. It has been argued that demonstrating test specificity is not a valid requirement, because the tests of cerebral activity are themselves a part of the whole diagnostic picture and should not be considered as single diagnostic tests.34 This argument would be reasonable if these newer tests were demonstrably more likely to be positive in people who were more conscious clinically, but they show the opposite.

Structured assessment protocols

A clinical diagnosis should rarely be made on the basis of a single observation; in almost every situation, the final diagnosis depends upon observations covering several or many variables. There are many assessment schedules that investigate consciousness and awareness. These have been reviewed.14 The three best were recommended in the UK guideline:4 the Coma Recovery Scale—Revised (CRS-R), the Wessex Head Injury Matrix (WHIM) and the Sensory Modality Assessment and Rehabilitation Technique (SMART).

The CRS-R35 was specifically designed to monitor recovery in prolonged disorders of consciousness, with an emphasis on the vegetative state and minimally conscious state. Indeed, it was used in a prospective study which also identified the prognostic importance of visual tracking,36 where 8 (73%) of 11 people in the vegetative state with visual tracking eventually regained some consciousness but only 20 (45%) of 44 without visual tracking did so.

It was considered the best single assessment schedule.14 However, it is not diagnostically certain. In a study37 on 103 people, the clinical consensus diagnosis (conscious or not conscious) was compared with the diagnosis given on the basis of structured assessment using the CRS-R. In 44 patients clinically considered to be in the vegetative state, 18 were considered to show signs of minimal awareness using the CRS. Unfortunately, the researchers did not investigate how many patients thought clinically to be minimally aware were diagnosed as being in the vegetative state using the CRS. As there is no reason to favour one method of diagnosis above the other, it could be that the 18 patients clinically considered minimally aware were not actually aware.

The WHIM38 was also developed to measure recovery from prolonged coma, and it consists of 62 behavioural observations. It contrasts with the CRS-R in being based primarily upon observing spontaneous behaviour though in response to naturally occurring stimuli. The main difficulty is the ambiguity of many of the instructions. For example, ‘Vocalises to express mood or needs’—the behaviour of making a noise has to be interpreted by the observer. Or ‘Frowns, grimaces etc to show dislike’—many patients frown or grimace spontaneously, and one cannot know if the grimace is showing dislike.

A recent study has validated the WHIM against other criteria and changed the order of items to reflect a better concordance with clinical judgement.39 The data also illustrate dramatically the considerable overlap of individual items between the vegetative state and two minimally conscious states (MCS−, MCS+).

The SMART40 was developed to assist in rehabilitation. Its specific feature is a series of structured stimulation protocols used to investigate responsiveness. The scoring criteria given lack of logic—for example, why are five consecutive observations spread over many days considered more relevant than the same number spread over an equivalent time scale but not consecutively? Moreover, there is no clear guidance on how to interpret chance association between a stimulus and a movement. The original study on 60 people did not validate it against a clinical diagnosis.

Variability in awareness

These single tests and test batteries provide a snapshot of awareness over a short time (up to 30 min). One consequence of the legal and ethical debate over withdrawal of clinically assisted nutrition and hydration has been a much greater level of observation and scrutiny of people with prolonged disorders of consciousness. This has drawn attention to the instability of patient behaviour over time. For example, close repeated observations in people well into the permanent vegetative state using the CRS-R showed sufficient variability to classify some patients as being in the vegetative state at one time, and in the minimally conscious state at another time;41 usually, patients score best in the morning.25 ,41 This should not be a surprise, because changes in level of consciousness were the primary reason for developing the Glasgow Coma Scale.20

In addition to relatively frequent but minor fluctuations in behaviour, there are larger but more infrequent fluctuations. Speaking has been reported in people otherwise considered to be in the vegetative state.42 In one case I have seen, a farmer who was clinically in the vegetative state was reported as saying quite clearly, ‘Aylesbury duck’, in response to a question posed by a visitor (to another visitor) concerning the particular name for white ducks. This was his only complex behaviour in over 4 years; he died naturally a year or so after this reported event. Another patient I know, closely observed over 2 years with no evidence of awareness, was on one occasion noted to move his leg in response to a command; the observer was experienced and had little doubt that it was in response to command. No similar response has been seen since.

Lastly, there are reports of late recovery from the apparently permanent vegetative state. It is currently accepted, in the UK, that anyone in the vegetative state 6 months after metabolic brain damage (hypoxia or hypoglycaemia) and 12 months after any other acute brain damage can be considered permanent.4 However, in one case series of 50 patients entering a permanent vegetative state, 6 were reported to recover consciousness.43 A recent case report records recovery from the vegetative state after 7 years.44 A case diagnosed in 2004 and reported in 2007 was reported to start recovering at 19 months after severe subarachnoid haemorrhage.45 Another report said to record late recovery actually reports recovery starting at 10 months, before the accepted cut-off of 12 months.46 In all cases, the person remained extremely severely disabled, though with some awareness.

Conclusions on diagnosis and categorisation

One cannot rely on single clinical tests when determining whether someone with a prolonged disorder of consciousness is aware or is not aware. None have been validated either as directly indicating awareness or as being a surrogate marker. Visual pursuit (visual tracking) does not definitively prove that someone is aware, but it seems probable that visual pursuit is a behaviour that is near the borderline between consciousness and unconsciousness. Its return during the recovery phase often predicts eventual return of awareness, and the accuracy of visual pursuit probably increases as consciousness emerges. Other proposed tests such as fMRI, functional EEG and visual fixation equally cannot be used to determine the presence or absence of consciousness.

The various assessment batteries proposed, which collect data over a range of behaviours, also lack good evidence concerning their diagnostic validity and lack evidence on the rate of false positives and negatives. They specifically do not consider how to overcome the problem of chance association between spontaneous movements and an external stimulus.

This difficulty in categorising an individual was highlighted in a systematic review of studies.47 It compared the responses of people who were clinically diagnosed as being in the vegetative state or minimally conscious state with responses to several different assessments and tests (eg, fMRI, EEG, evoked potentials). Often, the responses to these ancillary tests did not differ between the two groups. The authors concluded: ‘Overall, there was no combination of variables that allowed reliably discriminating between VS and MCS. This pattern of results casts doubt on the empirical validity of the distinction between VS and MCS’ (VS, vegetative state; MCS, minimally conscious state).

The practical and possibly philosophical question is whether it is possible to generalise from a limited set of observations to place a patient in a fixed category which is not expected to change thereafter. Is there any specific amount (number, duration of observation, period over which observations made) of observation that allows one to conclude that someone is in the permanent vegetative state? The evidence suggests not. There will always be a degree of uncertainty which arises from several factors:

• awareness is not a dichotomous state but covers a range from heightened awareness to complete unresponsiveness;

• the absence of any agreement on what clinical observations can definitively prove the presence of awareness within the spectrum of responsiveness, and specifically the absence of any unique criterion;

• the resultant dependence upon human judgement and interpretation;

• the natural variability in levels of responsiveness seen even in people with a prolonged disorder of consciousness, the natural sleep–wake cycle and hour-by-hour variability and also possibly a longer term variability such that infrequent brief episodes of low awareness may occur;

• the reported cases of a very few people who, well after the current time when permanence is assumed, seem to recover at least some awareness.

Categorisation: is it necessary, is it possible?

The original Bland case was based on the assumption that he was in a specific state where he was not only unaware at the times he was assessed or observed but also he was never aware. It was also assumed, reasonably, that his state would not change. It was then argued that, because he never had and never would have conscious awareness, he had no interest in continued medical treatment and so it was legally permissible to stop (withdraw) medical treatment without considering his Best Interests. In contrast, for a person who is deemed to be in a minimally conscious state, however minimal that consciousness is in terms or duration and extent, the court (if involved) must decide the patient's Best Interests.

Thus, legally there is a chasm between the appropriate decision-making process in someone who is judged to have even minimal evidence of consciousness (the Best Interests process) and someone who is judged never to be conscious (no need for a Best Interests process). Over the last 20 years, the court process has focused on determining whether or not someone can be classified as being in the vegetative state. The clinical evidence does not support this categorical distinction. It is not possible to identify the vegetative state as an obviously different state from the (lower end of) minimally conscious state. The boundary has moved; as Mr Justice Hayden remarked in the recent case where I was an expert witness ‘Dr Wade, is it true to say that, if the vegetative state were an exclusive club, it would have become increasingly difficult to join over the last twenty years?’.

To illustrate the distinction, consider the chemical, water (H₂O). The temperature of water can vary from −200°C up to +200°C, smoothly and along a spectrum. However, at 0°C and 100°C it changes state, from solid ice to liquid water, and from liquid water to a gas (steam). These are state changes. I am suggesting that there is no state change in consciousness; instead the person is more or less responsive to internal and external changes, and exhibits more or less complex actions and behaviours.

Anyone who argues that patients in the vegetative state are in fact in a distinct, separate state has to account for the observation that patients determined to be in the vegetative state at 4 months after traumatic brain injury slowly become more responsive and emerge into a state of full awareness by 8 months. The process is seamless, with no sudden step change. Anyone suggesting that patients can enter a categorically unique state rather than simply existing at the extreme end of a spectrum needs to explain how this arises and how it can be determined.

Therefore, in the absence of evidence to the contrary, it is best to consider that all people are somewhere on a spectrum between normal awareness and total absence of awareness, and that all people show natural variation in their level of awareness in all time scales from minute by minute to week by week. People with a prolonged disorder of consciousness show a similar variability, but only within a much reduced range and at a much lower level; some people cross the boundary rarely or never. Even if one does argue that there is a separate state, it is not possible at present to identify that a particular person is in that state because we have no means of doing so.

The problem and solutions

The disjunction between the assumption made by the legal system and by many clinicians and members of the public that it is possible to state with great certainty that someone is and will remain totally unaware at all times and the clinical impossibility of doing so has led to several problems.

The lack of any distinct, absolute boundary between being vegetative and not has resulted in an increasing reluctance, legally and clinically, to accept that someone is in the vegetative state. There are increasing debates both about the observations made and about their interpretation. This leads to delays in making a decision, and often a reluctance or refusal to make a decision. Furthermore, the legal guidance given on clinical decisions for people with prolonged disorders of consciousness2 has unfortunately led to clinicians and their legal advisors avoiding decisions for fear of breaking the law, which distresses families.6 ,7 Often, probably usually, no meetings are held to consider what is in the person's Best Interests,48 even though this is required (in England and Wales) by the Mental Capacity Act 2005 and is recommended in guidance.4 Active treatment is just continued, usually with no firm legal basis (ie, Best Interests meeting).

This problem ultimately arises from the inevitable uncertainty associated with decision-making in people with a prolonged disorder of consciousness. This uncertainty encompasses determination of both the patient's usual level of consciousness (ie, over a 24-hour period), and the speed and final extent of recovery. Similar uncertainty is present in almost all clinical situations. There is no reason why people with a prolonged disorder of consciousness should be managed in a different way to all other patients.

One solution is for all clinical management of all people with a prolonged disorder of consciousness to revert to the Best Interests process. Some might argue that the court should be involved because the decisions may lead to a patient's death, but this is a weak argument because decisions that may lead to the death of a patient are commonplace in health. Others might argue that these cases are complex, but they are no more complex than most other similar cases faced on a daily basis. It might be argued that clinical teams lack the expertise. This may be true and it arises because the decisions have been removed from clinical practice into a legal domain; moreover, patients are moved into long-term care away from expert review. However, the legal process is slow, expensive and limited, and is inappropriate for this purpose.

Others have already come to the same conclusion, on other grounds. A recent review on the effects of the law on clinical practice in people with a prolonged disorder of consciousness considered all the possible benefits and many of the disadvantages and concluded that the courts should not be involved routinely.52 A chapter based on the experience of families came to the same conclusion.53

Returning primary responsibility to clinical teams will require them to use proper clinical processes. As with any other rare and complex condition, specialist services with expert clinicians need to be involved, to ensure correct and proper management occurs. The clinical processes have been laid out clearly4 although the parts that appertain to the legal guidance would need revising. The particular focus must be the proper use of the Best Interests process,1 ,49–51 and this requires considering Best Interests as soon as a person loses their capacity to make decisions. Further, this is true both for acute and for progressive conditions.

For acute onset conditions, it should start at the time of admission and in every case a Best Interests meeting should be held before 4 weeks have elapsed. For slowly progressive conditions such as multi-infarct dementia, multiple sclerosis and Parkinson's disease, it is less easy to specify a time to start but once someone is totally dependent, even for feeding, there should always be a Best Interests meeting. There should be no requirement or expectation that any decisions in people with prolonged disorders of consciousness should involve a legal process. The court should only be used in exceptional circumstances, as occurs in all other healthcare.

The specific important clinical steps are:

• An accurate diagnosis of the underlying cause of a person's reduced level of consciousness. This is a mandatory starting point.

  • Any treatable secondary complications should also be considered and investigated if needed.

• A determination of the level of consciousness should occur, especially in the early phases, but

  • no attempt need be made to determine specifically whether a person is totally unaware or only minimally conscious.

• Critical consideration of interventions that may improve consciousness.

  • By far the most common is to reduce and stop drugs that sedate and that are either totally unnecessary or only have questionable benefit.

  • Amantadine could be considered in the first 3 months.54

• Establishment of prognosis, in as far as this is known.

  • The precision of any prognosis increases with time.

  • There is no specific change occurring at 6 or 12 months after onset; the change is simply a slight further decline in the chance of further improvement and not a categorical change.

• Consideration of Best Interests1 ,51 ,55

  • One conclusion that might arise is that it is no longer in the person's best interests to continue with active medical treatment. One of the active medical treatments might be clinically assisted hydration and nutrition.

• At all times, but especially if withdrawing any treatment, it is critical to ensure that the person does not suffer pain and distress.

  • Appropriate palliative care should be given especially if hydration is stopped. In practice, this can easily be undertaken by local services, and it is rarely a significant problem.