Clinical review: The role of the intensivist and the rapid response team in nosocomial end-of-life care

In-hospital end-of-life care outside the ICU is a new and increasing aspect of practice for intensive care physicians in countries where rapid response teams have been introduced. As more of these patients die from withdrawal or withholding of artificial life support, determining whether a patient is dying or not has become as important to intensivists as the management of organ support therapy itself. Intensivists have now moved to making such decisions in hospital wards outside the boundaries of their usual closely monitored environment. This strategic change may cause concern to some intensivists; however, as custodians of the highest technology area in the hospital, intensivists are by necessity involved in such processes. Now, more than ever before, intensive care clinicians must consider the usefulness of key concepts surrounding nosocomial death and dying and the importance and value of making a formal diagnosis of dying in the wards. In this article, we assess the conceptual background, reference points, challenges and implications of these emerging aspects of intensive care medicine.


Introduction
Life-support technology has profoundly changed medi cine and has contributed to improved life expectancy in resource-rich nations. However, the technology has also substantially aff ected the circumstances, mode and timing of death in such societies. In this context, cessation of breathing or of the heartbeat no longer necessarily signifi es the arrival of death [1] because doctors can substitute these vital functions with artifi cial organ supports in patients who would otherwise rapidly die without them. Such technology may facilitate recovery, and is therefore of potential benefi t to the individual. In a signifi cant proportion of cases, however, artifi cial life support will not confer an enduring benefi t to the patient and will be associated with physical, emotional, spiritual and fi nancial burdens to patients, their families and society [2].
Owing to the uncertainty of predicting death at the onset of clinical deterioration, a proportion of ICU admissions inevitably involve patients who receive artifi cial life support, only to die despite provision of such support, or, more often, only able to die when these are removed. In this environment, death may then become a programmed or orchestrated event, where the custodians of life-support technology may substantially infl uence when and how death will happen [3,4]. Th e management of dying and the provision of nasocomial end-of-life care (EOLC) have become ubiquitous and intrinsic to the daily practice of intensivists.
Furthermore, because the intensivist's role has also recently expanded to that of rapid response team (RRT) leader [5,6], EOLC decisions are frequently made out of hours in a limited amount of time, in a non-ICU environment, in patients with whom the intensivist may not be familiar and with limited documentation of advanced care planning ( Figure 1). In this article, we review and discuss the characteristics, conceptual and ethical implications, and practical consequences of the rapidly expanding developments of RRTs [5,[7][8][9][10][11][12][13][14][15][16][17] in the practice of this aspect of intensive care medicine.

End-of-life care and intensivists
Th ere are several reasons why intensivists are in a unique position to improve nosocomial EOLC. Intensive care physicians have a detailed understanding of key aspects of critical illness and limits of vital organ support. Independent of the patient's admission diagnosis ICU physicians also take into account the eff ect of specifi c physiological derangements on the probability of survival. Prognosis based on severity of organ dysfunction may diff er substantially from that attributed to the admission diagnosis alone. Finally, intensive care physicians

Abstract
In-hospital end-of-life care outside the ICU is a new and increasing aspect of practice for intensive care physicians in countries where rapid response teams have been introduced. As more of these patients die from withdrawal or withholding of artifi cial life support, determining whether a patient is dying or not has become as important to intensivists as the management of organ support therapy itself.
Intensivists have now moved to making such decisions in hospital wards outside the boundaries of their usual closely monitored environment. This strategic change may cause concern to some intensivists; however, as custodians of the highest technology area in the hospital, intensivists are by necessity involved in such processes. Now, more than ever before, intensive care clinicians must consider the usefulness of key concepts surrounding nosocomial death and dying and the importance and value of making a formal diagnosis of dying in the wards. In this article, we assess the conceptual background, reference points, challenges and implications of these emerging aspects of intensive care medicine.
are in a position to explain to patients and families what the limits of the technology of life support might be in a specifi c clinical situation. Intensive care physicians can thus guide colleagues, patients and their families to an informed shared decision regarding the clinical utility and limits of artifi cial organ supports.
Similarly, there are several reasons why non-ICU physicians may be less favorably positioned to optimally deliver nosocomial EOLC in the critically ill. First, the ability of non-ICU physicians to prognosticate in cases of advanced disease has been reported to be poor and typically overly optimistic [18][19][20]. Second, as the task of providing care to critically ill patients shifts to the ICU team, primary physicians may only see such patients for a limited period of time during each day. Th is may impair their ability to make a fully informed judgment on their patients' condition and their appreciation of the degree of discomfort, loss of dignity and family distress being experi enced while treatment is continued. Finally, non-ICU physicians are less frequently exposed to EOLC decisions in a high-technology environment and typically fi nd conversations with families in this setting diffi cult [21,22].
Th e above diff erences between ICU and non-ICU physicians are made more relevant by recent develop ments in healthcare. In particular, an increasing propor tion of deaths are happening in hospitals rather than at home. Furthermore, within the hospital, an increas ing proportion of deaths occur in the ICU instead of on the general wards [23][24][25][26][27][28][29][30][31], perhaps as a consequence of the greater use of technology-based interventions. Finally, the introduc tion of RRTs has resulted in an increased involvement of ICU physicians in the care of dying ward patients [5,8,9].
Th e intensive care physician's unique skills, experience and knowledge of advanced organ support technology has increased their involvement in managing acutely deteriorating ward patients. Consequently, they are now making more frequent and often diffi cult decisions about who should receive treatment escalation and who should receive EOLC in the wards. Such concerns necessitate that all clinicians ought to refl ect more deeply on the concept of death and its management; the uncertainties surrounding the concept of and the diff erential diagnosis of dying; the management of these issues within the context of RRT calls; and the future role of intensive care medicine in these processes. Patients with chronic illness have reduced function and increasing frailty, which increase their risk of a crisis. This leads to hospital admission. As this happens, family observation of the patient's physical changes has an eff ect on their attitudes and perceptions of what might be desirable in terms of medical intervention. While in hospital with an acute admission, discussions with the clinician may also shape decision-making. An acute deterioration then leads to intensivist involvement and, in the absence of a clear documented plan, a partly informed semi-urgent or urgent decision has to be made to either pursue aggressive treatment or comfort treatment alone.

Diagnosing and defi ning nosocomial death: basic concepts
Diagnosing death is the fi nal step in the provision of EOLC. However, a universally applicable and acceptable defi nition of death remains elusive [32,33]. Scientifi c or philosophic responses to this problem often propose a functional defi nition: an organism is dead when its integra tive and homeostatic functions are irreversibly lost [32]. Although such a defi nition appears intuitive, it presents a number of problems. First, it is not possible to generally defi ne a level of irreversible functional loss that constitutes death in all species. For example, what defi nes death in a jellyfi sh and in a human is profoundly diff erent.
Second, until recently humans have been deemed dead solely on cardiorespiratory criteria -loss of the fl ow of blood in the circulation or air in the lungs [1,32,33]. Unresponsive loss of consciousness was a recognized associated sign, but cardiopulmonary criteria remained primary. Th e irreversible loss of all three of these key physiologic functions has been, and continues to be, used to defi ne the moment of death at an earlier time than the late confi rmatory signs of bodily decomposition. However, intrinsic irreversible loss of cardiorespiratory function can be replaced by technological supports. In the nosocomial environment, therefore, diagnosis of death can be deferred by a medical decision to externally support cardiorespiratory function. In other words, a medical decision regarding whether to intervene in a deteri orating patient directly impacts on the diagnosis of death [30][31][32][33].
Th e third problem with applying a universal defi nition of death is that humans also function within social, cultural, religious and economic domains. Death not only coincides with a signifi cant change in biological function, but is also a marker of change in an individual's moral, social, spiritual, political, legal and economic status within society. It is at this point that the individual's moral rights are lost; it becomes acceptable for friends and relatives to mourn, organize funerals, pray for the soul of the deceased; corporate or political successions occur; and wills are read and inheritances are distributed. Accordingly, a defi nition of death is accepted not only because it conforms to contemporary understanding of biology, but also because it is consistent with the cultural, social, political norms and exigencies of that period. For example, while a defi nition of death based on the cessation of cardiorespiratory function has biologic plausibility, it is similarly coherent with a western cultural construct based on Christian view that the cessation blood fl ow and associated unresponsiveness identify the moment at which the soul leaves the body.
In summary, a defi nition of death allows, justifi es or legalizes important medical and nonmedical practices.
Th e defi nition must both encompass and be coherent with medical diagnostic and technical capability and practice; and societal norms, values and expectations.
Th is historical and cultural contingency in the defi nition of death is exemplifi ed in the evolution of the concept of brain death. Catastrophic brain injury remains technologically irreplaceable in itself, but for over 40 years it has been possible to provide and continue cardio-respiratory support almost indefi nitely. Logically, in such cases, the patient cannot be declared dead by applying cardio-respiratory criteria even though indepen dent existence without cardio-respiratory support would be impossible. Th e medical, social and legal ramifi cations of this problem lead to the concept of brain death. Irreversible loss of consciousness, brain stem refl exes and spon ta neous breathing have all become accepted criteria for death in many countries [1,31,33]. Th is advance also coincided with the development of solid organ trans plantation -an endeavor with wide social, political and legal endorsement, and which is technically and ethically facilitated by using donors declared dead on neurologic criteria.
Most patients who die in the ICU satisfy cardiorespiratory criteria for the diagnosis of death, but only a minority of these deaths occurs despite the presence of maximal vital organ support. Most patients die because vital organ supports are withheld or withdrawn. Th is practice is justifi ed if providing organ support is not likely to confer an enduring important functional or mortality benefi t, or if future independence from such supports is deemed improbable, with inevitable death from treatmentassociated complications. If despite these considerations there is no decision to withdrawal life-support technology the patient may be perceived as "dying" yet unable to die.
Unfortunately, individual prognostication of patients who appear to be dependent on life support is imprecise and the recognition that a patient is dying when organ support technology is either available or in place can be diffi cult and controversial. Importantly, when such support is in place it may retard, obviate or indeed obfuscate the cascade of inevitable vital organ failures that would result from severe injury or illness. Accordingly, a decision to withdraw or withhold further organ supports is both a medically informed decision and a negotiated act following discussion with family and other health providers [27][28][29][30][31]. When the decision to remove organ supports is accepted by all, the time and manner of death is determined and controlled by the orchestrated withdrawal of organ supports [27][28][29][30][31]. Sometimes, especially when there is controversy, orchestration necessitates gradual and sequential removal of organ supports in order to unmask the severity of underlying organ dysfunction, making this obvious to all those involved in the patient's care; and to allow a time and manner of death acceptable to the patient and family.
In some cases, orchestrated withdrawal of organ support facilitates potential organ donation after cardiorespiratory death [1]. In this setting, the timing of withdrawal is not decided by family and physician exigencies alone, but also by the logistics of organ procurement, and recipient availability and suitability [1].
Most nosocomial deaths occur on the wards and in the absence of ICU organ support [34][35][36][37][38][39]. Death outside the ICU may simply be allowed in a patient who is deemed to have an irreversible terminal disease process or injury that would not be preventable with ICU or non-ICU treatment. For others, although ICU admission is deemed inappropriate, active ward-based therapeutic intervention is continued -possibly delaying death but not preventing it. In either case, the timing and manner of death will be aff ected by the treatments that are continued and by if and when palliative care is provided.
In the ICU or on the wards, doctors can substantially infl uence if and when death will occur, as well as the type and manner of death. Interventions to immediately prevent death are justifi ed by providing a possible cure (irrespective of this probability) or (less commonly) by deferring and transforming an expected death to a time and mode that facilitates potential organ donation. If and when such interventions should occur and the objective of these interventions are contingent on whether death is deemed soon and inevitable. Th is requires the recog nition that the patient is dying, consensus as to this determination, and whether organ donation is being considered.
Several diffi culties remain, however: the failure to recognize or have a consensus that a patient is dying; the failure to have clear decisions or understanding regarding ICU treatment; and the continuation of ineff ective active treatment without eff ective palliative care [40,41]. Th ese diffi culties predispose the patient to continual suff ering that only ends following an "unexpected" death, a death that may involve the indignity of futile cardio-respiratory resuscitation. A crucial issue must therefore be considered: the need to correctly determine whether dying is the primary relevant diagnosis.

The primary diagnosis of dying
Th e dying process can be interrupted at any stage of its development by the application of technology, thus masking a primary overarching diagnosis that the patient is dying. Consider the following example.
An 80-year-old patient with Alzheimer's disease just coping at home with his wife's assistance presents with pneumonia, septic shock and renal failure following a fractured neck of femur following a fall at home. Th is patient could be considered to have sepsis-induced multiple organ dysfunction syndrome necessitating vasoactive medications, renal replacement therapy and mechanical ventilation. Conversely, this frail patient with dementia could simply be considered as having the primary diagnosis of dying. Th erapeutic interventions may postpone death but not prevent it. Even if short term ICU survival is achieved it will be followed by increased risk of hospital mortality, and further functional disability and loss of independence necessitating discharge to nursing home care. Th is poorer quality of life will precede a probable death within 12 months (Figure 2). In this setting, the diagnosis of dying should be seriously considered in hierarchical preference to that of septic shock due to the presence of chronic, cognitive decline and probable outcome trajectory. Th e patient still has septic shock with multiple organ failure, and dementia is not a direct contributor to this state. However, in this context although ICU survival is possible, the probability of an enduring survival benefi t, and functional recovery with an acceptable quality of life is very small. Furthermore, cognitive decline might render any benefi t unrealisable to the patient, but would not preclude the possibility of suff ering. Th e primary and relevant diagnosis of dying would thus be justifi ed and management should center on discussing such issues with the family and the primary clinician and recommending compassionate EOLC.
Th is approach would be completely diff erent from that applied to a person with the same physical conditions of pneumonia, septic shock and renal failure following a fractured neck of femur from a motorbike accident in a 40-year-old man (Figure 3). In such a person, given the expected survival and functional recovery trajectory after critical illness, the diagnosis of dying appears much less likely, and advanced organ support technology should be initially applied.
Diagnosing dying is much more complex than the normal diagnosis of disease (whether the patient does or does not have a fractured neck of femur complicated by pneumonia and septic shock), where the only immediately relevant dimension is medical. Th e elevation of dying to the primary position in the hierarchy of patient diagnoses is a medically informed, technology-mediated, social decision.
Th e primary diagnosis of dying should be formally considered in the diff erential diagnostic list of all patients who present with severe illness and injury, especially but not exclusively patients with multiple co-morbidities, advanced age or poor physical performance, although its recognition will be infl uenced by the social context. Th e epidemiology of dying in hospitals from diff erent countries reveals these social, cultural and resourcerelated dimensions. In the USA a signifi cant proportion of hospital deaths occurs in the ICU [24][25][26], indicating that the diagnosis of dying is either not formally considered or is delayed until an advanced stage in the course of the patient's illness. In the UK the percentage of patients dying in the ICU is much lower (approximately 6%) [25], suggesting that this diagnosis is considered earlier and/or more commonly pursued outside the ICU, or that a lack of ICU resources and acute bed rationing leads to an inability to move some of these patients to the ICU, or both. In any of these scenarios, the social structure appears to aff ect the primary diagnosis of dying in hospital patients outside the ICU.

The response to the diagnosis of dying
Dying has powerful social and cultural features outside the above system-related aspects. Some cultures tend to be more pragmatic and accept death as a component of the lifecycle, as a good outcome under circumstances where quality of life is valued above quantity of life. In such cultures, the primary diagnosis of dying is sometimes perceived as helpful in preventing unnecessary, burdensome, painful, futile and undignifi ed interventions. In other cultures, the prevention of death is seen as more important than the prevention of loss of dignity and quality of life. In these diff erent cultural and social environments, doctors respond to dominant social attitudes [42][43][44][45][46].
Th ere are also powerful psychological and family-based features to the diagnosis of dying [47,48]. Th ey involve the expressed wishes of the patient (if known) and the psychological state of the family [49,50]. Loving individuals or families that have shared a common life-path have also often discussed these issues well, often have advance care directives and are more likely not only to accept the diagnosis of dying but also to report satisfaction with the outcome. On the contrary, families or individuals with the opposite psychological and social characteristics are more likely to dispute the diagnosis of dying and reject any alternative to active, technologybased intervention [2,23,[51][52][53]. Once again, physicians respond to these psychological situations and pressures in their diagnostic approach.
However, clinical characteristics can remain compelling despite specifi c patient, family and physician social and psychological context. For example, the combined presence of a massive dominant hemispheric stroke, age >80, aspiration pneumonia, septic shock, acute renal failure and hypoxemic respiratory failure provide powerful and almost incontrovertible clinical evidence that the primary diagnosis is dying. In this patient, the brain computed tomography results together with blood lactate >25 mmol/l and pH 6.98 provide strong objective data that the patient's primary diagnosis is dying. Of course, the response to this still occurs within a valueladen context [51]. Although the primary diagnosis may not be disputed, a social context that emphasizes mere and unlikely survival over palliative care is likely to provide maximal technological support and defer the moment of death.

Dying and key human values
Th e primary diagnosis of dying also requires consideration to several related and value-based concepts. Although such concepts have a subjective quality and are impossible to quantify, they have value in social and medical discourse and need to be explicitly considered. Th ese concepts are futility, dignity, and compassion. The patient has a limited life expectancy due to age and ICU admission may be a major trigger for loss of independence and subsequent nursing home admission and progressive and more rapid decline. These possibilities need to be clearly and honestly discussed with patient and family as well as the primary clinician to develop a crisis response plan that is both patient-centered and realistic. Futility refers to the concept that a given intervention will not result in the desired outcome [54]. Futility has multiple dimensions and is dynamic in nature, and the desired outcome similarly changes with time, person, perspective and response to intervention. Unsurprisingly, futility has been the subject of much debate in medicine [54] with both support for and antagonism towards it as a concept. Despite such controversy, the word is commonly used in medical conversations dealing with EOLC decisions and should be discussed when dying is a diff erential diagnosis.
Dignity is a similarly elusive concept with strong subjective components and yet is supported as useful by clinicians [55]. For many individuals, having an endotracheal tube, a urinary catheter, a rectal tube and being the subject of multiple invasive interventions is not an acceptable price for a few extra months of life, especially if its quality is poor. When asked, very few intensivists or intensive care nurses who understand these actions well and infl ict them upon patients every day indicate they would be happy to receive such interventions unless the survival and quality-of-life gain was clear and likely.
Compassion is another key aspect of EOLC [56]. Compassion is an important key to the interpretation of futility and dignity and, when perceived to be present by the family, a major enabler of decision-making. Care must be explicitly given and seen to be given. As patients at risk of dying experience signifi cant symptoms [57], a shift away from technology-based life-support systems must always be accompanied by an interventional plan that targets comfort, pain control, explanation, kindness, support, peace and understanding for both patient and family.
Th ese key terms of futility, dignity and compassion when explored and specifi ed in a particular clinical context, are important for determining whether dying is the primary diagnosis and its subsequent management. In the previous example of the 80 year old with hemispheric stroke, and sepsis with multiple organ failure, invasive organ support is technically possible and immediate death preventable. However, conferring an enduring benefi t in terms of independent natural survival is statistically improbable, and attempts to do so are very likely to be associated with loss of dignity and suff ering. When faced with this clinical scenario a compassionate stance is likely to result in the primary diagnosis of dying, with patient care directed towards the relief of suff ering and preservation of dignity. Such consideration helps distinguish between what can be done (a technological decision limited by available resources) and what should be done (a moral decision shaped by the social and cultural context).

Diagnostic uncertainty
Th e concept of diagnostic uncertainty applies to the primary diagnosis of dying just as it does to any other medical diagnosis [58]. Th ere is, however, a unique dimension to this uncertainty [59]. While the diagnosis of pneumonia may or may not be correct, the risk that true pneumonia will develop is never increased by its diagnosis. In contrast, the risk of dying is increased by the primary diagnosis of dying. Th is self-fulfi lling prophecy eff ect requires caution in making the diagnosis of dying. Objective prognostic models for hospitalized patients [60] should be carefully applied, the limitations of these considered, and the consequences and degree of uncertainty of the diagnosis openly discussed [61]. Such uncertainty demands a careful discussion and consul tation process with colleagues within the ICU and outside the ICU, a strong attempt to understand the patient's wishes, and a desire to, within reason, respect them. Respect for a patient's wishes and a family's wishes does not imply always, only and simply asking the family what to do. Th e ability of surrogate decision-makers to predict the patient's wishes may be just as limited as the doctor's ability to predict prognosis [62]. A surgeon cannot be made to operate on a patient who he/she feels will die anyway. An intensivist should not logically place a patient who he/she believes will die anyway on invasive mechanical ventilation: to do so appears misleading and unethical because it confers no enduring benefi t and is not in the patient's best interests.

Distributive justice
Th e intensive care physician usually cares for many patients but physical and human resources are often limited. While it is unwise for the physician to make resource considerations his or her primary concern, it is not uncommon for intensivists to have to make decisions that are infl uenced by the availability of resources. Th e wise use of these resources requires consideration of distributive justice, an aspect of the dying process [63] that is not often appreciated. Primarily, it is important to identify the dying patient in the ICU or on the ward in order to change the focus of care from actual or potential technologic interventions to palliative care so that the patient's best interests can be served through the specifi c allocation of appropriate resources directed towards patient comfort. But, undeniably, a dying patient receiving futile treatment with therapeutic intent in the intensive care occupies and consumes a resource that cannot be made available to other patients. Th is may prevent, delay or dilute the allocation of intensive care resources to patients who may obtain a therapeutic enduring benefi t from intensive care organ supports.

Collegiality, family management and ICU team leadership
Th e diagnosis of dying and its management is often extremely time consuming, emotionally burdensome and challenging. Such diffi culties often impede consideration of the diagnosis of dying and implementation of an EOLC strategy. Negotiations with medical and/or surgical colleagues are often diffi cult. Primary physicians who have cared for a patient for many years are loath to let go.
Intensivists should tackle such issues with openmindedness and, to a degree, welcome such advocacy. Some patients with the primary diagnosis of dying will die even though the primary physician insists on aggres sive care; yet others will live even though the intensivist would have preferred EOLC. When such disagreement exists, collegiality, private discussions away from the bedside, courtesy, negotiation, involvement of senior colleagues from the same and/or diff erent specialties, the setting of agreed indicators, and so forth, are all important means to achieve better identifi cation and management of the dying process. Family discussions can be at least as challenging. Yet the same principles apply. Families typically respond to a physician who clearly wants to respect the patient's wishes, shows care and com passion, and yet guides the family to an under standing of why the diagnosis of dying is being considered and why it has now become the major relevant diagnosis. Although most patients are unable to participate in decision-making with regard to dying, when this is possible the patient's wishes should be considered, discussed and respected.
Th e diagnosis of dying also requires management of the ICU team, which also poses major challenges. ICU nurses and junior medical staff who have to care for a dying patient are often distressed by the length of the decisionmaking process and by prolongation of life in a setting where they feel the patient is suff ering for no good reason [64]. Th ey often wish to become involved in the decisionmaking process and to understand why particular decisions are made. In our unit nurses and junior medical staff often participate in family discussions. Th is helps in understanding the many and often confl icting issues involved, thereby providing comfort and understanding for the agreed meeting outcomes. Importantly, they are seen by the family to be part of a team eff ort directed towards the provision of best care.
After acceptance that dying is the primary relevant diagnosis, then a therapeutic transition should take place toward palliative care. However, such transition should not occur with undue haste as the dying patient and family should be treated with due respect. Adequate time should be provided for the family to be present at the bedside, adjust to the therapeutic transition and witness a peaceful death. Th e goal is to generate a recollection of the events that emphasizes understanding, peace and a sense of freedom from suff ering. Of course all of these considerations may be seen as a series of commonsense statements, which is correct. But deliberate refl ection on each of the above aspects of dying is helpful in providing better care. Increasingly, societies such as the Australian and New Zealand Intensive Care Society require advanced trainees to undertake formal training in these aspects of intensive care practice.
Th e above aspects and challenges of EOLC are now an established part of the daily lives of intensivists within the ICU, and in the hospital in general. However, another development is changing this landscape further in almost all English-speaking countries, in Scandinavian countries, in the Netherlands and in several hospitals in Europe and worldwide: the introduction of RRTs.

Rapid response teams and end-of-life care
Th ere is an increasing trend for intensivists to respond and review patients who are deteriorating outside the ICU. Such review takes the form of RRTs, critical care outreach teams, and patient at-risk teams. RRTs are activated when patients fulfi ll objective criteria for deterioration that are typically based on derangements in vital signs (for example, blood pressure, respiratory rate, heart rate and conscious state) [5,6].
RRTs evolved because of a variable and inconsistent recognition of ward patient deterioration, and the mismatch between the needs of deteriorating patients and available hospital ward resources. Th e original intent of RRTs was thus to prevent unplanned ICU admissions, serious adverse events, cardiac arrest and unexpected death [5,6].
With expansion of these services to many hospitals, and with their increasing utilization within hospitals, there is emerging evidence that RRTs have become important in the provision of EOLC in hospital wards. Th is is perhaps not surprising, as patients at the end of life often display abnormalities in their vital signs. In some cases, ward doctors may not appreciate that the patient is dying, and advanced care directives may not be in place. In a 1-year study of 713 RRT calls, the RRTs determined that the patients should have been designated 'do not resuscitate' in 23% of calls [65]. In another study of 900 RRT patients, investigators found that 21.7% of such patients already had a documented do not resuscitate order [66].
In some cases of RRT review, the team members, in conjunction with the treating clinicians, form an opinion that the patient is not likely to benefi t from cardiopulmonary resuscitation and/or intensive care therapies. In two single-center studies, the RRT members newly implemented limitations of medical therapy (LOMT) in 10% of RRT calls [67,68].
A recent international prospective observational study specifi cally examined the role of the RRT in EOLC planning in seven hospitals. Th e study revealed that there were 652 RRT calls occurring in 518 patients during 1 month, of which 31.1% had LOMT [8]. Such LOMT orders were instituted in 20.3% of patients before and in 10.8% of patients after the RRT call, respectively. Compared with patients who did not have LOMT during the call, those with LOMT were older, were more likely to be admitted with a medical (rather than a surgical) condition and were less likely to be admitted from home. In addition, they were also less likely to be discharged home and were more likely to die in hospital. Calls in patients with LOMT tended to occur more often out of hours. Amongst the 78 patients who died with LOMT, the RRT call was most likely to occur on the day of their death (29.5% cases), with a further 28.2% having their last RRT call within 2 days of their death [8].
Combined these studies suggest that approximately one-third of RRT calls involve patients with an EOLC issue, and that one-tenth of calls involve the RRT implementing LOMT. Th e RRT therefore appears to be used as an intermittent agent for the provision of palliative care, and in one-tenth of cases appears to be essentially making the primary diagnosis of dying.

Advantages and disadvantages of the rapid response team providing end-of-life care
Th e RRT initially evolved because of a mismatch between the needs of deteriorating patients and the ward resources available to respond. Th e initial aim of this model of care was to identify, review and treat deteriorating patients in order to avoid preventable morbidity and mortality. Th e studies outlined above suggest that the RRT is frequently required to review ward patients who have urgent EOLC issues, and such reviews are requested urgently. Th is suggestion indicates a disparity between the resources on the ward to provide EOLC and the needs of the patients. Th ese studies may also indicate a need for improved clarity of documentation by the parent medical team with respect to the extent and aggressiveness of medical interventions in deteriorating patients. A further interpretation of this data is that parent unit medical staff may not recognise dying patients or their ensuing EOLC needs. Th ese aspects of EOLC increasingly involve intensivists and require investigation and improvement. Provision of prompt EOLC by the RRT probably prevents unneces sary suff ering, distress or pain. In addition, 50% of RRT patients with documented LOMT survive to hospital discharge, suggesting that the RRT is not purely palliating such patients. A single-center study revealed that intro duction of a RRT resulted in improvement of EOLC in ward patients [69].
Despite the advantages of this approach, there are several disadvantages. First, the RRT may not know the patient, which may compromise the validity of EOL decisions. Th is compromise may result in confl ict with the parent unit, or withholding of potentially benefi cial therapy from the deteriorating patient [70]. However, a single-center study reported a high degree of concordance with decisions by the RRT and subsequent opinion of the treating team [71]. Second, the RRT may be the patient or family's fi rst therapeutic encounter, thereby limiting rapport with RRT. A further problem is that many RRT EOLC calls occur out of hours when the parent unit may not be present. As the patient is often too unwell to participate in discussions, the calls often involve expe dited decisions with the patient's family members and parent unit over the telephone. Such calls are stressful to families, parent units and RRT staff . Th e calls also remove staff from the ICU, often out of hours when medical staffi ng levels are lowest. Th ese observations suggest that intensivists should be strong advocates for the widespread use of advance care directives in order to improve the quality of EOLC decisions.

The importance of advanced care directives
Although expedited provision of EOLC by the RRT is better than allowing a patient to suff er in discomfort, LOMT should ideally be documented prior to patient deterioration. Th is approach allows the patient to be informed of and involved in treatment options, as well as the potential burden/discomfort and likelihood of success for such therapies. With this approach, patient wishes and escalation plans could be clearly articulated to the arriving RRT when the patient deteriorates, avoiding unnecessary escalation in cases where palliation is indicated. A recent single-center randomized controlled trial assessed the eff ects of advanced care planning of EOLC for older patients [36]. Patients who received the intervention were more likely to have their end-of-life wishes known and followed than those in the control group. In addition, family members of the intervention group experienced less stress, anxiety and depression than those of the control patients. Importantly, patients in the intervention arm did not experience an increased mortality rate.

Conclusion
EOLC is a growing aspect of intensive care practice within the ICU, and now outside the ICU. Understanding the social, cultural and psychological dimensions of death should become part of the training for intensive care physicians. Formally considering the primary diagnosis of dying in all patients with critical illness either in the ICU or on the wards is also an important step toward delivering appropriate care and formulating a management plan that delivers outcomes that more closely satisfy individuals and families and avoid unnecessary suff ering. Th e development of RRTs has added a new challenge in the fi eld of such nosocomial EOLC by taking the intensivist's diagnosis of dying and its management to the general wards and requiring decisions to be made with more time constraints and less information. Th is development poses new challenges and requires training and refl ection.
Finally, although there are a myriad of individual variations in the diagnosis and management of dying that make it, at times, more of an art than a science, there are also many universal aspects and values that can and should be studied using scientifi c methodology and controlled trials. We owe it our patients, their families, our colleagues and ourselves to conduct such studies, however diffi cult they might prove. Th ere is one more reason to do this: the future will deliver even more of these challenges as the population ages, the organ-support technology advances, costs escalate, the resources available to pay for them proportionally diminish, the quality and quantity of life gained with aggressive intervention directed at preserving life decrease, and society is faced with an epidemic of older patients with dementia. As the custodians of this technology, we must be prepared for these challenges and educate the community to be prepared for them as well. Nosocomial thanatology will only become a bigger and more complex component of intensive care medicine in the future.