THE LANCET

The promise of a good death

Ezekiel J Emanuel, Linda L Emanuel

For more than three decades, interest in improving the of a good death is more appropriately a topic demanding
care of dying patients has progressed from being the the health professions’ attention than it might have been
concern of a few health-care professionals to being a a century ago. For this promise, what is the period that
widespread social concern. More than 35 years ago concerns us? While the hours or days before death are
Saunders founded the hospice movement in Britain.’ important, the focus should be the weeks and months
Nearly 30 years ago Kubler-Ross published On Death before death when symptoms and needs increase, and
and Dying2 and the “living will” was first proposed.3 interventions can have a significant impact on the well-
More than 20 years ago the Quinlan decision* made clear being of patients and families. Thus, a good death should
that it was legal and ethical to terminate life-sustaining mean making the last weeks-not minutes-of life
treatments in the USA, and California became the first valuable and meaningful. This is more appropriate for
state to enact a living-will statute. 15 years ago US states causes of death that have a prodrome before dying, such
began recognising the legal authority of families to as cancer, congestive heart failure, and COPD, which are
decide whether to terminate life-sustaining care, and the estimated to cause 50-70% of deaths, than for sudden
Medicare system, which provides health insurance for the deaths caused by stroke, myocardial infarctions,
elderly and &sabled in the USA, began covering the cost accidents, and suicides (table 1).
of hospice care. 8 years ago the US Supreme Court ruled
on the Cruzan case, 5 its first in the end-of-life area,
indicating that there was a constitutional right to refuse Framework for a good death
life-sustaining treatments. More recently, there has been Much of the effort to improve care for the dying,
intense interest in euthanasia and physician-assisted especially in the USA, has concentrated on using
suicide in developed countries,6-8 studies of the care of technology to relieve symptoms and avoiding invasions
dying patients, evaluations of interventions to improve during the dying process.4,5J2 Those two goals are
such care,’ and efforts to improve the training of essential, but they do not encompass the ideal of a good
physicians in this area.‘O The media, which used to shun death. Indeed, when surveyed, the concerns of patients
reporting on death and dying as too negative, now deem seem to be as much, if not more, about loss of dignity,
them suitable for news and features. being dependent, not being a burden, and loss of control,
Despite this attention, a “good death” remains more a than about physical or psychological symptoms.‘3J4
hope than standard medical practice for all patients. In What constitutes a good death? The Institute of
part, this is because the contemporary “reexamin[ation] Medicine’s definition is: “[A] decent or good death is one
of how we approach dying and death” coexists with a that is: free from avoidable distress and suffering for
persistent social attitude that denies death.“J” Youth is patients, families, and caregivers; in general accord with
celebrated as the ideal; longevity is desired, and remains patients’ and families’ wishes; and reasonably consistent
a primary standard for evaluating health-care systems; with clinical, cultural, and ethical standards”.12 One
and when a friend or relative is dying people commonly commentator has argued that “the conceptual
avoid the person feeling that “I don’t know what to do or models...of how death should be approached and what
say”. The focus has been on avoiding problems and constitutes suffering are overly simplistic and woefully
stopping bad interventions rather than on a positive ideal inadequate at the bedside”.15 Indeed, the problem with
of a good death. Thus we face a paradox: increasing such definitions is their vagueness and difficulty in
concern about death and dying, tremendous technical translation into practical guidance. In the figure we
capacities to relieve symptoms and improve care, and provide a framework for understanding and evaluating a
persistent suffering of dying patients, all combined with
the continued denial of death. This makes it opportune
Causes of death (ordered hy cause) Number of deaths in USA Rates per
to explore how to fulfil the promise of a good death. 100 000
Total Over 65 years
Because of the lengthening life-span, lower childhood population USA
of age
mortality, smaller families, and the increaseing use of
All causes 2312.1 1694.3 880.0
supportive and other medical interventions, only health-
Heart disease 737.6 615.4 280.7
care professionals today develop sufficient experience CaW?r 538.5 381.1 204.9
and familiarity with death and dying. Hence, the promise Stroke 158.0 138+3 60.1
COPD 102.9 88.5 39.2
Lancet 1998; 351 (suppl II): 21-29 Accidents 93.3 29.1* 35.5
Pneumonia and influenza 82.9 74.3 31.6
Department of Clinical Bioethics, Warren G Magnuson Clinical
Diabetes 59.2 44.5 22.6
Center, National institutes of Health, Bethesda, Maryland HIV 43.1 NAt 16.4
(EJ Emanuel MD); and Division of Ethical Standards, American Suicide 31.3 NAt 11.9
Medical Association, 515 N State Street, Chicago, Illinois, USA Chronic liver disease and cirrhosis 25.2 NAt 9.6
(LL Emanuel MD) *Among those over 65 accidents are the 7th leading cause of death. IHIV. suicide,
and chronic liver diseases are not among the 10 leading causes of mortality among
Correspondence to: Dr Ezekiel J Emanuel, Department of Clinical
the over 65. In this age group, 8th leading cause of death is Alrheimer’s disease, 9th
Bioethics, Warren G Magnuson Clinical Center, National Institutes is renal failure, and 10th is septicaemia. COPD=chronic obstructive pulmonary
of Health, Building 10 Room lC118, Bethesda, MD 20892, USA disease, NA=not applicable.
(e-mail eemanuel@nih.gov) Table 1: Epidemiology of death
-
Vol351 - May * 1998 SII2 1
THE LANCET

Figure: Framework for a good death

good death developed as part of the Commonwealth- and optimise the care of the dying patient in all areas.
Cummings project on the quality of care at the end of Second, delineating the five interventions emphasises
life. that dying is not just a medical experience and the
This framework includes the evolving insights in the patient’s full social network can significantly influence
end-of-life field and characterises dying as a their experience. While health-care providers and
multidimensional experience.‘J6J7 It synthesises the dying institutions have a major responsibility for improving the
experience as a process with four critical components: 1) care of dying patients in medical and non-medical
the fixed characteristics of the patient; 2) the modifiable aspects, others, from family and friends to religious
dimensions of the patient’s experience, or elements that mentors and community groups, do also. The framework
may respond to events or interventions; 3) the potential delineates potential interventions where appropriate; not
interventions available to family, friends, health-care all interventions will be appropriate in all cases. For
providers, and others; and 4) the overall outcome. The many patients continuing with their personal physician
framework can help in the improvements of care for will be preferable to receiving care from a
dying patients in three ways. multidisciplinary palliative-care team. For other patients
First, the framework emphasises the multifaceted a good death may mean forsaking as many palliative-care
character of the experience of dying patients, expanding interventions as possible and being with their family.
considerations beyond physical and psychological Furthermore, the framework emphasises that caring for
symptoms to include: economic demands and caregiving dying patients occurs within an institutional system that
needs, social relationships, hopes and expectations, and can significantly effect the quality of the dying experience
spiritual and existential beliefs. Each of these dimensions by development of specialised care teams and facilities,
incorporates a vast range of specific concerns that are institutional affiliations, and reimbursement policies.”
illustrated by the examples listed. However, the Third, the framework outlines a systematic mechanism
dimensions are not separate. There are psychological to implement and evaluate two elusive concepts that are
components to many physical symptoms and other fundamental to improving care of dying patients: a good
difficulties can intensify and be manifested as physical death, and suffering. Saunders used the term “total pain”
ailments. Patients who are depressed or are experiencing to denote suffering in four dimensions. Building on her
existential meaninglessness can have a lower threshold idea, this framework suggests a quality-of-life scale
for pain; patients who have spiritual fulfilment or good appropriate for dying patients would be a “death-
mechanisms to handle caregiving needs may report less experience scale” that did not only assess the treatment
bothersome symptoms. The purpose of such a of physical symptoms’8a’9 but also assessed the patient’s
comprehensive list of modifiable dimensions of patients’ condition in each of the six modifiable characteristics.”
experience is to help clinicians to systematically evaluate In addition, the framework suggests a mechanism to

s1122 Vol351 * May- 1998

define and measure suffering. Suffering constitutes either loss and disruption in sleeping patterns, that are
poor conditions and outcomes in all, or nearly all, of the frequently consequences of terminal illness and not
six modifiable dimensions of patients’ experience or indicators of distress. Historically, clinicians have
overwhelmingly bad experience in one or two of the underdiagnosed psychological symptoms.34 This may be
modifiable dimensions.20J1 due to a general uneasiness of non-specialists with
psychological illnesses as well as a common but
Current knowledge about a good death misplaced view that depression is “normal” for dying
To achieve a good death means using-or deciding not to patients. As with many physical symptoms, research
use-relevant interventions. Thus we need to identify would be helpful in developing additional understanding
what we know about the frequency, assessment, and of the prevalence and causes of these symptoms and to
importance of each modifiable dimension of the patient’s develop additional interventions. Nevertheless, through
experience and the use and effectiveness of the various appropriate use of psychological counselling,
interventions. psychostimulants, antidepressants, anxiolytics, and other
drugs, the psychological symptoms of many dying
patients should be manageable.3o
Physical symptoms
Of all the aspects of dying, pain is the most
Economic demands and caregiving needs
studied. Validated and reliable pain-assessment measures
exist.17-‘9,22Extensive pharmacological, neuroablative, and The limited empiric data suggest that caring for dying
patients imposes significant economic demands and
anaesthetic treatments, as well as detailed treatment
caregiving burdens on families. About a third of
guidelines, should ensure adequate pain relief in 95% of
American families experience a significant loss of income
cases.23-25 Nevertheless, among dying patients, pain
and savings because of a terminally ill family member.35J6
remains inadequately treated, with a reported frequency
Dying patients often require substantial nursing and
ranging from 20% to over 70%.18
personal care at home as well as home-making
For some pain syndromes, such as neuropathic and
services.12J7 One report found that among terminal
visceral pain, existing interventions are inadequate and
cancer patients, 42% required assistance with personal
new treatments need to be found. Providing adequate
care and home-making and 10% needed assistance with
pain-control to the vast majority of patients requires
transportation.38 Family members, usually wives and
better implementation of existing therapies through
other female relatives, often provide care to meet these
training of clinicians, use (not just development) of
needs, which frequently require major life-changes,
guidelines, education of patients and families, redesign of
including leaving work. 36 One of the major limitations of
health-care systems, and changes in regulations.1*~24~26
existing data is that they do not explain “the mechanism
What we know about other physical symptoms is of these [economic and caregiving] burdens”, making it
considerably less than our understanding of pain.IZ difficult for providers to effectively address them.36
Besides pain, the most common physical symptoms at Preliminary data suggest that severe physical and
the end of life are fatigue, drowsiness and insomnia, psychological symptoms create high caregiving needs
dyspnoea, and anorexia.‘7~19,27-29 For many of these which, in turn, create significant economic demands on
symptoms there are validated assessment too1s.12J7m19 patients and familiesT9 Interestingly, understanding of
Whilst there are effective interventions for some of these how these economic demands and caregiving needs
physical symptoms (eg, nausea and vomiting), for others affect the patient’s overall dying experience is lacking,
(eg, fatigue), treatments are less effective. Additional although the bias is that these demands and needs would
research is necessary to develop therapies for some make it worse. However, a patient’s caregiving needs
physical symptoms; for others, training is needed to seem to affect adversely the patient’s caregiver and
improve use of assessment instruments and available family, causing depression and other medical
treatments, and improved delivery services are problems.39,*o
necessary.12 These caregiving and economic demands may be
worse in the USA than in developed countries with
Psychological and cognitive symptoms universal health coverage that provides more
Less is known about psychological symptoms than about comprehensive medical services.Yet in at least one study
pain and most physical symptoms.12,30 Whilst there are of Americans, these economic demands occurred despite
controversies about the appropriate clinical criteria for health insurance.36J8 The limited data available suggest
some psychological symptoms, such symptoms are both that significant economic demands and caregiving needs
common and a major problem for dying patients.12~*7-19,27-2~ are common among dying patients and that their impact
More than a third of dying patients may be depressed, on a good death is not defined, but they do adversely
and more than half of patients with advanced cancer feel effect the caregivers of dying patients.
sad, anxious, and irritable.27”0 In addition, suicide rates
are higher for the terminally ill than in the normal -
population, and are thought to be associated in part with Do not Withholding of Advance-care Euthanasia and
greater degrees of depression, psychological distress, and resuscitate Ilfesustaining dlrectives physlcfan-
orders treatments assisted suicide
inadequate pain contro1.3’
USA 79-97%* 84-90%? 20-25%$ Unknown
There are validated assessment instruments for many (patients in KU)
of these symptoms, especially depression, that are Netherlands 35% (all deaths) 30% (all deaths) 3%/,n 3.4%5
relevant to dying patients.17m’9,32,33 Many assessment 60% (in hospital)
instruments for psychological symptoms may be *See references 54 and 57.tSee references 55 and 56. $See references 10, 59, 60,
inappropriate for dying patients because they rely on and 61. §See reference 9 and 21. nPersonal communication (Dick Wilhems).
evaluating changes in vegetative functions, such as weight Table 2: Use of medical interventions

Vo1351 ‘May. 1998 SII23

THE LANCET

associated with a good death. This is an area in which

significantly more research is required.

Spiritual and existential beliefs

By spiritual and existential issues we mean how people
find meaning, purpose, and value in their life; they can
be specifically religious, but even people who have no
religious faith or are not members of an organised
religion have belief systems that give their lives meaning
and purpose. ” Spiritual and existential issues become
especially urgent for people confronting death and the
attendant questions about the purpose of their life.‘“J5J”
The health-care profession has long recognised this by,
for example, having pastoral-care departments in
hospitals. Similarly, attending to spiritual matters has
always been a central part of the hospice philosophy and
clergy are regularly part of the multidisciplinary hospice
teams.‘,‘”
Nevertheless, these efforts have been separate from the
central concerns of medical care; clinicians have not been
trained to explore spiritual and existential issues and
rarely inquire about, discuss, or address them with
patients. Consequently, despite the avowed importance
of spiritual and existential issues, there are few data on
them and their relation to a good death.4*a43 We do not
know: what proportion of dying patients become more
religious, or, conversely, alienated from spirituality; or
what proportion receive comfort from their faith,
whether this comfort is from formal interaction with a
religious mentor or a more informal spirituality. As a
result, there is little information on what interventions-
such as training clinicians in discussing spiritual issues or
closer interaction with clergy-might improve the
spiritual condition of dying patients.
There are efforts to change this situation. Questions
about spirituality and personal meaning are being
incorporated into assessment measures.44 For instance,
the McGill Quality of Life Questionnaire has
incorporated questions about personal meaning,
achieving life goals, and finding life worthwhile.‘7~42,43 And
some training programmes for clinicians on death and
dying are including components on spiritual aspects.‘“J’
Nevertheless, more data will be needed to provide
clinicians with a structured approach to spiritual and
existential issues that can assist in providing a good
death.45

Hopes and expectations

What patients imagine their prognosis to be, and what
Social relationships and support they expect of their future medical care and condition,
can have an important impact on their decisions and
As noted, dying patients rely on their families to meet
experience. Patients who overestimate their life
caregiving needs. However, the understanding of
expectancy or think treatments are more effective than
whether, and to what extent, dying patients receive
proven might put off advance-care planning and
emotional and social support from their family and larger
preparing themselves and their family for the end of their
social network is limited. There are validated instruments lives. Similarly, major personal and family events and
to assess social supports, such as whether patients have milestones shape the quality of dying patients’ experience
someone to talk to, someone to share their hopes and and even the timing of their death.
fears with, someone to have a good time with, etc.“s4’ In Clinicians know that upcoming milestones or
general, these support instruments have not been applied “unfinished family business” are important to dying
to the terminally ill, so we lack basic information on the patients, influence their decisions and experiences, and
types of social supports dying patients rely on, whether might even postpone death. Studies have indicated that
and how the social network beyond the immediate family deaths tend to occur after important family events, such
helps dying patients, the deficiencies of existing social as birthdays or major holidays4” Emerging data also
relationships, and w!hich social relationships are indicate that most <terminally ill cancer patients in the

sr124 Vol351 . May - 1998

 THE LANCET

USA significantly overestimated their survival time, and units and with cancer die without resuscitation.4R~49 In
this overestimate was associated with preferences for addition, in medical intensive-care units, as many as
more medical interventions.47 Such denial of prognosis 90% of patients -who die have other life-sustaining
may impede communication and planning about death, treatments withdrawn, with an average of almost four
and may lead to unrealistic requests for treatment. For interventions being stopped per patient4” This is a
some patients such denial may be important as a dramatic increase in the withholding or withdrawing of
psychological coping mechanism to maintain hope. treatments over the past decade.48 In the Netherlands, in
Whether denial improves or impairs a patient’s overall about a third of all deaths, treatments are withdrawn or
quality of dying is unknown. withheld, with a tenth involving stopping respirators and
25% involving stopping artificial nutrition and
Interventions for a good death hydration.8~‘3 In 1998, physicians stoping, or not starting,
Of all the possible interventions, the ones that have been life-sustaining treatments is routine. Indeed, there are
the most thoroughly studied are: withdrawal of life- now guidelines about how to withdraw life-sustaining
sustaining treatments, advance-care planning, hospice, treatments properly.50
and euthanasia and physician-assisted suicide. We shall
briefly review the importance of each of these Advance-care planning
interventions (table 2). When originally proposed 30 years ago, advance
directives were short statements by patients wishing to
avoid “heroic measures”.‘J2J1 Such directives have
Withdrawal of life-sustaining treatments evolved into three types of advance directives: 1) living
The withdrawal or withholding of life-sustaining wills in which patients specify their preferences for--or
treatments is intended to remove inappropriate medical to refuse-specific medical treatments; 2) proxy-
interventions.4,5J2 Over the past decade, physicians have designation forms in which patients appoint a surrogate
become more comfortable stopping treatments and who has the authority to make medical decisions for
“letting nature takes its course”. In the USA and the them if they become incompetent; and 3) values histories
Netherlands, most inpatients now die without in which patients indicate their values and what gives
resuscitation, and up to 90% of patients in intensive-care their lives meaning. Some forms, such as the so-called

Vol351 *May. 1998 ~25

 THE LANCET

Medical Directive, are a combination of all three types5’ hospitals; in the USA, the emphasis is on hospice care in
At least in the USA, there is a constitutional right to have the patient’s home.
these directives honoured.S Since 1983 hospice has been covered by Medicare in
Advance-care directives have been most thoroughly the USA; 80% of employees in large and medium
developed, disseminated, and evaluated in the USA. So American businesses are covered by hospice, and 4 1 state
far, experience with them has been disappointing. Polls Medicaid programmes also offer hospice benefits.56 There
repeatedly demonstrate that more than four-fifths of the are about 3000 hospices in the USA that cared for more
population endorse advance directives;l’ and advance- than 450 000 Americans (about a fifth of the Americans
care directives have been widely endorsed by medical who died) in 1997. Of hospice patients in the USA, more
organisations, courts, the media, and others. Despite this than 60% had cancer, 6% had heart disease, and 3% had
tremendous support, only ZO-25% of the public and AIDS.56 Indeed, about half of all patients dying of cancer
patients have completed such documents.S1a5Z And most each year in the USA receive hospice care. The average
interventions-education efforts, legally mandated length of stay in hospice is about 2 months, but almost
information about advance directives on hospital 55% of patients receive such care for less than a month.
admission, hospital nurses to discuss advance-care About 15 years ago, comparisons of the quality of
planning with patients and families-have not greatly hospice care with that of conventional care revealed that
increased the use of advance-care directives.12 However, while there were some small but statistically significant
some data suggest that a high proportion of terminally ill differences in the proportion of patients who had pain
patients, between 40 and 50%, do have advance-care and satisfaction with care, in general there were no
directives. Whilst there is widespread social acceptance of consistently observable, large differences in the quality of
the idea of advance-care directives, there remains life of hospice and non-hospice patients.57a58 The lack of
widespread reluctance to actually use them. observed difference in outcomes between hospice and
Attempts to implement advance-care directives to conventional care in these studies might be attributable
guide the care of patients have revealed other problems. to assessment instruments that were inappropriate for
Such directives do not necessarily facilitate comm- and insensitive to the conditions of dying patients.59
unication between patients and families about care at the Families provided more direct care for hospice patients
end of life. Indeed, patients and family members and and such hospice patients were more likely to die at
proxies may not communicate about care at the end of home. Importantly, the caregivers of hospice patients
life; family members and proxies do not accurately know reported more stress and social disruption, probably
patients’ values, goals of care, or specific preferences for because they did more direct patient-care; however, they
treatment.53 Therefore, relying on proxies and families to did not report higher rates of health problems.57 Current
make decisions for patients does not ensure that patients’ data collected by hospice organisations suggest
tremendous satisfaction by caregivers.
wishes will be implemented and respected. Another
problem is that completed advance-care directives may
Euthanasia and physician-assisted suicide
not be forwarded to the hospital record; physicians and
patients might not discuss the specific preferences in the Over the past few years there has been growing interest
in the possibility that euthanasia and physician-assisted
advance-care directive; clinicians may be unaware of the
suicide might provide a “good death”.6 The potential
patients’ wishes; and stated preferences may simply be
effects of these interventions on a good death are limited;
ignored. ‘w*
they might provide some psychological reassurance or
Such studies make it clear that advance-care planning
alleviate additional suffering by ending a life early.
is not a document but a complex process involving many
Importantly, euthanasia and physician-assisted suicide
parties over time.‘Z,55 Patients must understand their own
are not interventions aimed at the modifiable dimensions
preferences, discuss their preferences with their physician
of the patient’s experience, since they are used when
and proxy, complete an advance-care directive, the
efforts to improve symptoms, economic demands and
directive must be placed in the medical record, all
caregiving needs, spiritual and existential beliefs, and
clinicians caring for a patient must know there is a
social relationships and supports have been exhausted.
directive and understand the patient’s specific wishes,
Data from the Netherlands suggest that even when
and when appropriate clinical situations arise, the legal sanctions are removed from euthanasia and
directive must be available and its directions physician-assisted suicide, only a small minority of
implemented. While there are many good suggestions on patients use these methods: 3.4% of decedents die by
how to achieve these steps, they are not all performed euthanasia or physician-assisted suicide.*J3 Most (80%)
routinely, and patients, as well as families and physicians, of patients receiving these interventions had cancer.8 On
cannot currently rely on them.53z54 Attention by patients, average, patients’ lives are shortened by 2-3 weeks
physicians, and health-care institutions is needed to (nearly three-fifths “lose” less than a week) .8,L3
make effective advance-care planning the standard of Pain is not a major factor motivating requests for these
care. interventions.8J3J4,60,61 Rather, patients who are depressed,
have psychological distress, and have substantial
Hospice caregiving needs are significantly more likely to desire
The first modern hospice, St Christopher’s in Sydenham, euthanasia or physician-assisted suicide.*,13,14,39~6”,6’This
UK, was founded in 1967, and the first hospice in the suggests, although there are no confirmatory data, that
USA, the Connecticut Hospice, opened in 1974.‘,16 providing better care in dimensions of psychological
Hospice advocates a multidisciplinary approach that symptoms and caregiving needs may reduce interest in
emphasises symptom control, attending to spiritual as euthanasia or physician-assisted suicide. In addition,
well as physical needs, and care of the family unit.1,1*a16 whatever psychological reassurance some patients receive
There are residential ihospices and hospice units in from the availability of euthanasia or physician-assisted

~26 Vol351 . May * 1998

 THE LANCET

suicide, an equal number of patients will find such symptoms of dying patients and instituting appropriate
possibilities threatening.60 Because these interventions treatments for those symptoms. In addition, patient-care
can be justified only after all other efforts have been conferences should routinely include presentations and
tried to improve the modifiable dimensions of a patient’s discussions of the care of the dying patient. Since much
experience, they do not substitute for any of the learning in health care is experiential (“seeing one”),
other interventions. training programmes also need to have role models
in the care of dying patients who can serve as mentors
Initiatives to make a good death the standard for students, housestaff, and fellows and teach by
of care their example. Finally, withdrawing life-sustaining
To help bring the goal of a good death to all patients, interventions is routine but involves more than “pulling
making it the standard of care, initiatives need to be the plug”.5o All clinicians who care for dying patients
undertaken in four areas: research, training, system must have formal training in the proper method of
design, and reimbursement (panel 1). As noted, empiric withdrawing life-sustaining treatments.
research is needed on almost every aspect of death and For well-trained clinicians with good intentions, the
dying. There are six high-priority areas for research. delivery of the best care requires institutional settings
First, research that could help define the characteristics that facilitate the provision of good care to dying patients
of those dying patients at high risk for suffering and and proper reimbursement arrangements. All specialised
doing poorly is needed. This would permit clinicians and cancer centres as well as other health-care institutions
the health-care system to target resources on those with should have palliative-care specialists to provide
the greatest need for interventions. Second, the consultative services for difftcult cases and ongoing
development of a reliable validated assessment- education and training of staff.lZ A pain service that is
instrumennthat covers the important dimensions of the competent to address only one physical symptom is
dying experience and is simple enough to use in clinical insufficient. Further, continuity and coordination of care
practice is important. There are good instruments for given by different providers in different settings is
research,12~17~‘9~28~43~5a but most of the instruments suggested essential. The evolving vertical integration of physicians,
for practice have not been tested for reliability and hospitals, nursing homes, home care, and other providers
validity.44 Having validated instruments would facilitate into organised delivery systems should facilitate the
efforts in continuous quality improvement in the care of continuity and coordination of care if attention is
the dying.12 Third, research on psychological symptoms, devoted to the needs of dying patients. Also, provisions
especially to determine the most effective screening must be made to facilitate a team-based coordination of
mechanisms and interventions for dying patients is care between hospital, doctors’ offices or clinics, and
necessary.3o Fourth, we need research on caregiving patients’ homes.*2 Because the cost of caring for dying
needs-defining predictors to determine which patients patients is so large and because such coordination of care
have significant caregiving needs and what the most is likely to produce efficiencies and financial savings,
effective interventions are to meet their needs.36’39 Fifth, there is likely to be a “win-win” situation in which
some research on patients’ spiritual and existential beliefs systems have an incentive to coordinate care better,
is important, especially defining patients’ spiritual and which will benefit dying patients and families.
existential concerns, predictors for identifying patients Finally, there are reimbursement issues. Costs of care
who lack spiritual and existential meaning, and for the dying are substantial. How services are paid for can
interventions that can improve the spiritual and greatly affect the care these patients receive. Capitation
existential care of dying patients.4”43 Sixth, because a rates that are not adjusted for the significant costs
substantial proportion of care for the dying occurs within generated by proper care of dying patients, by mechanisms
institutions, health-services research is necessary to such as stop-loss coverage or separate reimbursement
determine the best organisational structures and pools for the dying, can create incentives for providers to
practices to ensure optimum care.lz avoid patients at high risk for dying or to reduce important
Training initiatives must focus on ensuring physicians care. In addition, reimbursement rules intended to limit
and other clinicians are more competent in care of the services, such as Medicare’s attempt to limit use of home
dying, and that this knowledge is not haphazardly health-services by increasing patients’ co-payments, can
acquired by chance interactions with mentors and lead to restricted use of these services, making it difficult
patients but is an integral part of a systematic for patients to get the caregiving services they need and
educational programme.1G12~62 Like the general public, stay at home. Further, dividing payments to each provider
physicians often feel awkward in the presence of (physicians, hospices, hospitals, etc) may inhibit
dying patients and tind it difficult to talk with them appropriate referrals and undermine coordination of
about death. Clinicians must receive training care.12 For instance, physicians may be reluctant to refer to
in communication, not general communication skills hospice if this reduces their reimbursement for managing
but skills specific to the concerns of dying patients. the care of dying patients. Each of these reimbursement
Clinicians must be as comfortable in talking to patients issues is complex and requires careful solutions. But
about the dying experience as they are taking a history of government and insurers must review their
angina. 9,103 This training must include structured reimbursement rates and policies to be sure that they do
approaches-sometimes called steps or talking points- not adversely effect the care of dying patients eve.n while
for clinicians to use in common aspects of caring for potentially generating good outcomes in other ways.
dying patients, such as delivering bad news, advance-care
planning, addressing spiritual concerns, and giving Advice to the clinician caring for dying
condolences.1G’2’45*55 Training programmes also need patients
to educate physicians about both using standardised What should the clinician who seeks to ensure his or her
assessment techniques for physical and psychological dying patients receive a good death do? First, clinicians

Vol351 - May. 1998 ~1127

 THE LANCET

cannot take action to ensure a quality dying experience Conclusion

unless they know the conditions, concerns, and problems In 1998, we have the capacity to make a good death the
of their patients. Therefore, clinicians should adopt their standard of care in developed countries. Increasingly,
own simple but comprehensive assessment tool to be society is focused on ensuring a good death; physicians
used on all their patients.12~‘7-19,*8,44 The framework (figure) have more powerful medications and other interventions
should help make this assessment systematic. Panel 2 to alleviate pain than ever before; clinicians are
provides a suggested list of questions that cover each recognising the multidimensional aspects of dying and
dimension, although not every category of the the importance of attending to concerns other than pain;
dimensions. Because they are to be used in practice, not hospice is widely available and increasingly used;
to collect publishable data, clinicians can modify the advance-care planning is strongly endorsed; and medical
precise wording of these questions to conform to their schools, hospitals, and professional organisations are
own clinical style. committing themselves to training physicians to improve
Second, just as in any other area of medical care, the care of the dying. In the developed world, a good
clinicians need to develop their knowledge base and skills death is within our grasp, yet it is not a promise we can
in palliative care and keep them current through fulfil for all dying patients. The framework provided here
appropriate continuing medical education.11,6z Caring for can help implement our ideas about a good death and
dying patients requires knowledge and skills in symptom guide the additional research, training, design of health
management, communication, and coordination of care. systems, and reimbursement necessary to ensure the
Third, clinicians must acknowledge that they alone promise becomes a reality.
cannot provide all the care dying patients require. Their
role is to provide the care they can and to solicit and, Acknowledgements
when necessary, coordinate care from others. They must Marion Danis, Jane Lunney, Balfour Mount, Paul van der Maas, and
Dick Wilhems, Lousia Smith provided invaluable criticisms of this
forge close working-relations with palliative-care manuscript.This review was partly supported by grants from the
providers who can consult on patients when appropriate, Commonwealth Fund and the Nathan Cummings Foundation.
with hospice and home-care agencies who can provide
appropriate nursing, homemaking, and hospice care in References
patients’ homes, with clergy who can assist in the Saunders C. Care of the dying. London: Macmillan, 1959.
spiritual care of patients, with social workers who can Kubler-Ross E. On death and dying. LondomTavistock, 1970.
help define entitlements to care, and with others. While Kutner L. Due process of euthanasia: the living will-a proposal.
many if not most patients will not require these services, Indiana LawJ 1969; 44: 539-54.
In re Quinlan 70 NJ 10,1976.
they must be available for the dying patients who do Cruzan v Director, Missouri Department of Health 110 S. Ct. 2841,
need them. Further, when it will improve the care of the 1990.
dying patient, clinicians must facilitate the involvement Emanuel EJ. Euthanasia: historical, ethical and empiric perspective.
of family members, friends, and other members of the Arch Intern Med 1994; 154: 1890-901.
Kinsella TD,Tate PA. Attitudes among NHS doctors to requests for
patients’ social network. Not only can this disperse the euthanasia. BMJ 1994; 308: 1332-34.
caregiving and other burdens, but it can also provide Van der Maas PJ, van derWa1 G, Haverkate I, et al. Euthanasia,
essential social support and interaction to reduce the physician-assisted suicide, and other medical practices involving the
end of life in the Netherlands, 1990-1995. NEngZJMed 1996; 335:
isolation some patients may experience. 1691-705.
Fourth, in caring for dying patients, clinicians need Association for Palliative Medicine and the Royal College of General
to remember that a death affects a wide social Practitioners. Palliative medicine curriculum, Southamptom:
network.35-37J9 The families of patients experience Association for Palliative Medicine for Great Britain and Ireland, 1993.
10 American Board of Internal Medicine. Caring for the dying:
significant caregiving and economic burdens; they are identification and promotion of physician competency. Philadelphia:
also preparing for loss and must cope with loss after the American Board of Internal Medicine, 1996.
death. More than in most other aspects of health care, 11 Aries I? The hour of our death. NewYork: Alfred A Knopf, 1981.
12 Institute of Medicine. Approaching death: improving care at the end of
clinicians must not just care for their dying patients but life. Washington, DC: National Academy Press, 1997.
minister to the patients’ families and friends. This entails 13 Van der Maas PJ, van Delden JJM, Lijnenborg L, Looman CWN.
inquiring whether families understand the patient’s Euthanasia and other medical decisions concerning the end of life.
clinical situation, and preparing the family for what to Lancet 1991; 338: 669-74.
14 Back Al, Wallace JI, Starks HE, Pearlman RA. Physician-assisted
expect. It requires bringing families into the process of suicide and euthanasia inwashington state.JAMA 1996; 275: 919-25.
advance-care planning and discussing future care.5LJ5 15 QuillTE. A midwife through the dying process: stories of healing and
Further, because most families are not trained in health hard choices at the end of life. Baltimore: Johns Hopkins Press, 1996:
care, clinicians should instruct the family on how to care 202.
16 Stoddard S. The hospice movement: a better way of caring for the
for the patient at home when appr0rpiate.l’ It can require dying. NewYorkVintage, 1974.
more substantial involvement such as inquiring about the 17 Cohen SR, Mount BMM, Strobe1 MG, Bui F. The McGill quality of
families’ burdens and what additional assistance they life questionnaire: a measure of quality of life appropriate for people
with advanced disease: a preliminary study of validity and acceptability.
might require. Because families of dying patients P&at Med 1995; 9: 207-19.
experience higher rates of illness, including psychological 18 Portenoy RK,Thaler HT, Kornblitb AB, et &The Memorial symptom
distress, clinicians must also ensure families get the assessment scale: an instrument for the evaluation of symptom
health care they need.39a40 In addition, physicians should prevalence, characteristics and distress. EurJ Cancer 1994; 30:
1326-36.
never forget to express their condolences to bereaved 19 Bruera E, Kuehn N. Miller MJ, Selmser I’, MacMillan K.The
families.s* And, in stressful cases clinicians may be Edmonton symptom assessment system (ESAS): a simple method for
required to provide counselling themselves, or secure the assessment of palliative care patients.JWlliat Care 1991; 7: 6-Q.
20 Cassel E. The nature of suffering and the goals of medicine. NewYork:
professional counselling, for family members after the Oxford University Press, 199 1.
patient’s death. This may be especially necessary when 21 Mount BM, Hamilton I?When palliative care fails to control suffering.
the dying patient has y&ng children or young siblings. J Palliat Care 1994; 1~~24-26.

sn28 Vol351 * May * 1998

 THE LANCET

22 Cleeland CS. Measurement ofpain by subjective report. In: Hill CS, 43 Cohen SR, Mount EM, Bruera E, Provost M, Rowe J,Tong K.
Fields WS (eds.) Advances in pain research and therapy.Vol. 11. New Validity of the McGill quality of life questionnaire in the palliative
York: Raven Press, 1989: 391-403. care setting: a multi-center Canadian study demonstrating the
23 Agency for Health Care Policy and Research. Clinical practice importance of the existential domain. Palliat Med 1997; 11: 3-20.
guideline number 9: management of cancer pain. Rockville: AHCPR 44 VITAS Healthcare. Missoula-VITAS quality of life index. Missoula,
Publication, 1994. Montana:VITAS, 1995.
24 World Health Organization. Cancer pain relief and palliative care. 45 WelchT. Existential and spiritual concerns. In: Billings JA (ed).
Geneva:WHO,1990. Outpatient management of advanced cancer: symptom control,
25 Foley KM.The treatment of pain. NEngZJMed 1985; 313: 84-95. support and hospice-in-the-home. Philadelphia: JB Lippincott, 1985.
26 Von Roenn JH, Cleeland CS, Gonnin R, et al. Physician attitudes and 46 Phillips DP, King EW. Death takes a holiday: mortality surrounding
practice in cancer pain management: a survey from the Eastern major social occasions. Lancet 1988; ii: 728-32.
Cooperative Oncology Group. Ann Intern Med 1993; 119: 121-26. 47 Weeks JC, Cook EF, O’Day DJ, et al. Relationship between cancer
27 Seal C, Cartwright A.The year before death. Aldershot: Ashgate patients’ predictions of prognosis and their treatment preferences.
Publishing, 1994. JAMA (in press).
28 Portenoy RK,Thaler HT, Kornblith AB, et al. Symptom prevalence, 48 PrendergastTT, Lute TM. Increasing incidence of withholding and
characteristics and distress in a cancer population. Qual Life Res 1994; withdrawal of life support from the critically ill. Am J Respir Crit Care
3: 183-89. Med 1997; 155: 15-20.
29 Donnelly S, Walsh D. The symptoms of advanced cancer. Senzin Oncol 49 Faber-Langendoen K. A multi-instutional study of care given to
1995; 22: S67-72. patients dying in hospitals: ethical practices and implications. Arch
30 BreitbartW, Bruera E, Chochinov H, Lynch M. Neuropsychiatric Intern Med 1996; 156: 2130-36.
syndromes and psychological symptoms in patients with advanced 50 Brody H, Campbell ML, Faber-Langendoen K, Ogle KS.
cancer. J Pain Symptom Manage 1995; 10: 131-41. Withdrawing intensive life-sustaining treatment: recommendations for
31 Allebeck P, Bolund C, Ringback G. Increased suicide rate in cancer compassionate clinical management. N Engl J Med 1997; 336:
patients: a cohort study based on the Swedish cancer register. J Clin 652-57.
Epidemiol1989; 42: 611-16. 51 Emanuel LL, Emanuel EJ. The medical directive: a new
32 Berwick DM, Murphy JM, Goldman PA, et alPerformance of a five- comprehensive advance care document.JAMA 1991; 325: 3288-93.
item ment I health screening test. Med Care 1991; 29: 169-76. 52 Emanuel LL, Barry MJ, Stoeckle JO, Ettelson LM, Emanuel EJ.
1 Advance directives for medical care-a case for greater use.
33 Kohout TJ,.Berkman LF, Evans DA, Cornoni-Huxley J. Two shorter
forms of the CES-D depression symptoms index. JAgixg Health NEnglJMed 1991; 324: 889-95.
1993; 5: 179-93. 53 Emanuel EJ, Emanuel LL. Proxy decision-making for incompetent
34 Schulberg HC, Saul M, McClelland M, Ganguli M, Christy W, patients: an ethical and empirical analysis. JAMA 1992; 267:
Frank R. Assessing depression in primary medical and psychiatric 2067-071.
practices. Arch Gen Psychiatry 1985; 42: 1164-70. 54 Danis M, Southerland LI, Garrett JM, et al. A prospective study of
35 Siegel K. RaveisVH, Houts P, MorV. Caregiver burden and unmet advance directives for life-sustaining care. N Engl J Med 199 1; 324:
patient needs. Cancer 1991; 68: 1131-140. 882-88.
36 Covinsky KE, Goldman L, Cook EF, et al.The impact of serious 55 Emanuel LL, Danis M, Pearlman RA, Singer PA. Advance care
illness on patients’ families. J&Z4 1994; 272: 1839-44. planning as a process: JAm G&m Sot 1995; 43: 440-46.
37 Brook DB, Foley DJ. Demography and epidemiology of dying in the 56 National Hospice Organization. Hospice fact sheet. Arlington:
U.S., with emphasis on deaths of older persons. HospiceJ (in press). National Hospice Organization, July, 1997.
38 Houts PS,Yasko JM, Harvey HA, et al. Unmet needs of persons with 57 Greer DS, MorV, Morris JN, Sherwood S, Kidder D, Birnbum H. An
cancer in Pennsylvania during the period of terminal care. Cancer alternative in terminal care: results of the national hospice study. J
1988; 62: 627-34. Chronic Dis 1986; 39: 9-26.
39 Emanuel EJ, Fairclough D, Slutsman J, Omundsen E, Emanuel LL. 58 Kane RL, Wales J, Bernstein L, Leibowitz A, Kaplan S. A randomised
Predictors and outcomes of significant caregiving needs and economic trial of hospice care. Lancet 1984; i: 890-94.
burdens among terminally ill oncology patients: results from the 59 Mount BM, Scott JS.Whither hospice evaluation? J Chronic Dis 1987;
Commonwealth-Cummings Project. J Clin OncoZl998; 17: 1628 36: 731-36.
(abstr 4229) 60 Emanuel EJ, Fairclough DL, Daniels ER, Clarridge BR. Euthanasia
40 Patrick C, Padgett KD, Schlesinger HJ, Cohen J, Burns, BJ. Serious and physician-assisted suicide: attitudes and experiences of oncology
physical illness as a stressor: effects on family use of medical services. patients, oncologists, and the public. Lancet 1996; 347: 1805-10.
Gen Hasp Psychiatry 1992; 14: 219-27. 61 BreitbartW, Rosenfeld BD, Passik SD. Interest in physician-assisted
41 Sherbourne CD, Stewart AL.The MOS social support survey. Sot Sci suicide among ambulatory HIV-infected patients. Am J Psychiatry
Med 1991; 32: 705-14. 1996;153: 23842.
42 Cohen SR, Mount BM,Tomas J, Mount L. Existential well-being is 62 Billings JA, Block S. Palliative care in undergraduate medical
an important determinant of quality of life: evidence from the McGill education: status report and future directions. JAA&4 1997; 278:
quality of life questionnaire. Cancer 1996; 77: 576-86. 733-38.

-
VOI 351 * May. 1998 ~129