UCSF
UC San Francisco Previously Published Works
Title
Definition and recommendations for advance care planning: an international consensus
supported by the European Association for Palliative Care.
Permalink
https://escholarship.org/uc/item/7j53x1zp
Journal
The Lancet. Oncology, 18(9)
ISSN
1470-2045
Authors
Rietjens, Judith AC
Sudore, Rebecca L
Connolly, Michael
et al.
Publication Date
2017-09-01
DOI
10.1016/s1470-2045(17)30582-x
Peer reviewed
eScholarship.org Powered by the California Digital Library
University of California
www.thelancet.com/oncology Vol 18 September 2017
e543
Review
Definition and recommendations for advance care planning:
an international consensus supported by the European
Association for Palliative Care
Judith A C Rietjens, Rebecca L Sudore, Michael Connolly, Johannes J van Delden, Margaret A Drickamer, Mirjam Droger, Agnes van der Heide,
Daren K Heyland, Dirk Houttekier, Daisy J A Janssen, Luciano Orsi, Sheila Payne, Jane Seymour, Ralf J Jox, Ida J Korfage, on behalf of the European
Association for Palliative Care
Advance care planning (ACP) is increasingly implemented in oncology and beyond, but a definition of ACP and
recommendations concerning its use are lacking. We used a formal Delphi consensus process to help develop a
definition of ACP and provide recommendations for its application. Of the 109 experts (82 from Europe, 16 from
North America, and 11 from Australia) who rated the ACP definitions and its 41 recommendations, agreement for
each definition or recommendation was between 68–100%. ACP was defined as the ability to enable individuals to
define goals and preferences for future medical treatment and care, to discuss these goals and preferences with
family and health-care providers, and to record and review these preferences if appropriate. Recommendations
included the adaptation of ACP based on the readiness of the individual; targeting ACP content as the individual’s
health condition worsens; and, using trained non-physician facilitators to support the ACP process. We present a list
of outcome measures to enable the pooling and comparison of results of ACP studies. We believe that our
recommendations can provide guidance for clinical practice, ACP policy, and research.
Introduction
Advance care planning (ACP) enables individuals to
make plans about their future health care. Robust
evidence from systematic reviews shows that ACP
increases the completion of advance care directives and
occurrence of discussions about future health care in
clinical practice and improves consistency of care with
patients’ goals in various patient populations, including
oncology.
1,2
ACP can improve the quality of patient–
clinician communication, reduce unwanted admission to
hospitals, increase the use of palliative care, and increase
patient satisfaction and quality of life.
1,2
In 2016, a
systematic review
3
suggested broad support for ACP
among patients with cancer and their health-care
providers. Interest in ACP continues to grow, as indicated
by an increasing number of related scientific publications,
programmes, laws, and public awareness campaigns on
the topic. However, several challenges in ACP require
greater consensus before its potential can be fully realised.
First, the concept and content of ACP substantially
varies. Originally, ACP was conceptualised as only the
completion of an advance care directive, to be used when
the individual’s capacity to indicate preferences had been
lost. More recently, ACP is increasingly considered to be
a complex process that includes personal reflection and
discussion with clinicians about the patient’s wishes, the
appointment of a health-care representative, completion
of an advance care directive, and changes to the health-
care system. These developments have resulted in
growing interest in ACP beyond geriatric study, such as
in oncology.
3
Previous initiatives to define ACP have poor
generalisability because they are mostly restricted to
North America or the UK,
4–7
or to specific patient groups
or disciplines.
6,8
Second, there is a need for guidance
regarding the timing of ACP. For example, introducing
ACP too early could lead to a reluctance to engage in
ACP, whilst engaging in ACP in the face of a crisis or
shortly before dying could be too late.
9
A third challenge
in ACP is that dierences in patient preference,
knowledge, and health literacy could complicate
navigation of ACP by health-care professionals.
10
Finally,
there is an urgent need to determine the most relevant
outcome measures for evaluating ACP.
To date, there is no consensus regarding the definition
of ACP, nor are there any practice recommendations that
are applicable to various cultural settings and personal
values. This lack of agreement hinders the development
of ACP programmes and the evaluation of ACP’s
eectiveness. Therefore, we aimed to develop a consensus
definition of ACP and present recommendations for ACP
that can be used by health-care providers, policy makers,
and researchers across a broad spectrum of patient
populations, disease categories, and cultures.
Methods
An international taskforce consisting of 15 recognised
experts from eight countries (Belgium, Canada, Germany,
Ireland, Italy, Netherlands, UK, and USA) designed a five-
round Delphi study to build a systematic consensus on
ACP. The European Association for Palliative Care
(EAPC) Board commissioned this consensus project and
invited JACR and IJK to chair the taskforce on the basis of
their expertise in ACP and previous interdisciplinary and
international comparative work. JACR and IJK invited
well-known experts in ACP to the taskforce with the aim
of forming an international and interdisciplinary group
that included experts from a range of regions, with
clinical experience and with research experience, in the
fields of oncology, palliative care, geriatrics, and ethics.
These experts were identified either through their
Lancet Oncol 2017; 18: e543–51
Department of Public Health,
Erasmus University Medical
Centre, Rotterdam,
Netherlands (J A C Rietjens PhD,
M Droger MSc,
Prof A van der Heide MD,
I J Korfage PhD); Division of
Geriatrics, Department of
Medicine, University of
California, San Francisco, CA,
USA (Prof R L Sudore MD);
San Francisco Veterans Affairs
Medical Center, San Francisco,
CA, USA (Prof R L Sudore); UCD
School of Nursing, Midwifery
and Health Systems, University
College Dublin, Dublin, Ireland
(M Connolly PhD); Department
of Medical Humanities, Julius
Center, University Medical
Center, Utrecht, Netherlands
(Prof J J van Delden MD); School
of Medicine, University of
North Carolina, Chapel Hill, NC,
USA (Prof M A Drickamer MD);
School of Medicine, Yale
University, New Haven, CT,
USA (Prof M A Drickamer);
Department of Critical Care
Medicine, Queen’s University,
Kingston, ON, Canada
(Prof D K Heyland MD);
End-of-Life Care Research
Group, Vrije Universiteit
Brussel and Ghent University,
Brussels, Belgium
(D Houttekier PhD); Department
of Research and Education,
CIRO Centre of Expertise for
Chronic Organ Failure, Horn,
Netherlands (D J A Janssen MD);
Centre of Expertise for
Palliative Care, Maastricht
University Medical Centre,
Maastricht, Netherlands
(D J A Janssen); Palliative Care
Unit, Carlo Poma Hospital,
Mantova, Italy (L Orsi MD);
International Observatory on
End of Life Care, Division of
Health Research, Lancaster
University, Lancaster, UK
(Prof S Payne PhD); School of
Nursing and Midwifery,
University of Sheffield,
Sheffield, UK
(Prof J Seymour PhD);
e544
www.thelancet.com/oncology Vol 18 September 2017
Review
Institute of Ethics, History and
Theory of Medicine,
Ludwig-Maximilians University
of Munich, Munich, Germany
(R J Jox MD); and Geriatric
Palliative Care, Centre
Hospitalier Universitaire
Vaudois, University of
Lausanne, Lausanne,
Switzerland (R J Jox)
Correspondence to:
Dr Judith A C Rietjens,
Department of Public Health,
Erasmus University Medical
Centre, 3000 CA, Rotterdam,
Netherlands
j.rietjens@erasmusmc.nl
publication and citation record, or through contacts from
the professional network of JACR and IJK or that of the
EAPC board. Rounds 1 and 5 used a qualitative
methodological approach, whereas rounds 2, 3, and 4
required quantitative assessment. The figure shows the
number of participants and how the recommendations
were adapted at each stage. As defined by the standard
Delphi process, the structured rounds were characterised
by anonymity (protecting the Delphi results from the
eects of group conformity), iteration (allowing for
a change of opinion), and controlled feedback
(communicating the results of the previous round).
11,12
Round 1
In June 2014, during a two-day meeting at the Netherlands
Institute for Advanced Study (Wassenaar, Netherlands),
the taskforce established two draft definitions and five
core domains: core elements, roles and tasks, timing,
policy and regulation, and evaluation. We opted to
establish an extended definition to be used in, for
instance, research and education of health-care sta, and
a brief definition for practical use. To address each
domain in detail, working groups were set up that
consisted of four to five taskforce members. Within each
domain, recommendations were developed based as
much as possible on evidence derived from the medical
literature and on expert opinion. Done in 2014, and
updated in 2016, we studied the literature in three ways.
First, we did a meta-review.
13
This meta-review was
conducted by searching PubMed for publications with the
term “advance care planning”, and only included reviews
and meta-analyses. The search was limited to the title and
abstract search fields. This search resulted in 89 reviews
and one meta-analysis, of which their respective reference
lists were also reviewed. These studies were used to
support the initial recommendations. Second, we
searched for existing guidelines of position papers by
searching PubMed for publications with the term
“advance care planning” combined with “guideline” or
“position paper”. We did a similar search in Google, and
checked all identified reviews (including their reference
lists) for references to guidelines or position papers. This
search found five clinical practice guidelines.
5–8,14
Third,
each working group did a specific PubMed literature
search for each domain (ACP definition, core elements,
roles and tasks, timing, policy and regulation, and
evaluation), combining the term “advance care planning”
with relevant keywords for their section.
The definitions of ACP were formulated based on
25 definitions derived from the literature search.
Additionally, the working groups were able to use a
previous study on the definition of ACP and outcomes
ratings that was predominantly done in North America.
4
The draft definitions and recommendations were
discussed and improved eight times by each working
group and the taskforce (by email and in face-to-face and
telephone meetings) over the course of a year. This process
resulted in an extended and a brief definition of ACP and
37 draft recommendations.
Round 2
In September 2015, the extended and brief definition of
ACP and the draft recommendations were presented to an
expert panel through an online questionnaire using
LimeSurvey. In a separate document, we provided the
panellists with the definitions and recommendations,
including the supporting literature references. Potential
panel experts (including patient representatives) were
identified through their publication and citation record or
through the professional networks of the members of the
taskforce and that of the EAPC board. In the selection
process, we aimed for an international and inter-
disciplinary group of ACP experts. The invited panellists
were experts in ACP research, practice, and policy, with
backgrounds in medicine, nursing, palliative care,
psychology, ethics, law, and policy. Panellists also included
Figure: Delphi consensus process on the definition and recommendations of
ACP
ACP=Advance care planning. EAPC=European Association for Palliative Care.
Delphi round 1 (June, 2014–August, 2015)
The taskforce (n=15) drafted two definitions of ACP and 37 recommendations.
Delphi round 2 (September, 2015–April, 2016)
The Delphi panel (n=109 of 144 invited; response, 76%) rated the two
draft definitions and 37 draft recommendations, and provided comments.
Agreement (median) and consensus (IQR) were determined. Adaptations
were made in case of no strong agreement or consensus.
Delphi round 3 (May, 2016–July, 2016)
The Delphi panel (n=103 of 109 who completed round 2; 94%) rated the
adapted set of two draft definitions and 44 draft recommendations, and
provided comments. Agreement and consensus were determined. Adaptations
were made in case of no strong agreement or consensus.
Delphi round 4 (August, 2016–February, 2017)
The taskforce (n=12) was asked for consensus with the adapted set of two
draft definitions and 41 draft recommendations. The set was finalised based on
their feedback and the taskforce provided full consensus.
Delphi round 5 (March, 2017)
The EAPC board members reviewed the full final set of two definitions and
41 recommendations. The EAPC board members were unanimous in their
support and had no suggested revisions.
Three recommendations
removed due to redundancy
Ten new recommendations
were added (including five new
outcomes)
Two recommendations
removed due to redundancy;
one removed due to low score
www.thelancet.com/oncology Vol 18 September 2017
e545
Review
nine patient representatives who were trained members
of the Expert Voices Group of Marie Curie, and who had
first-hand experience with end-of-life care as a relative or
friend. For instance, one participant was a 19-year-old
student who was closely involved in the provision of care
for three close family members. We invited 144 experts
(from USA, Canada, Australia, or Europe), of whom
124 (86%) agreed to participate. Of these experts, 109 (76%)
completed the questionnaire. The appendix presents the
characteristics of the Delphi panellists, who were from
14 dierent countries. Of the 109 panellists, 83 worked in
clinical practice, mostly as a physician or as a nurse. Of
the 51 physicians, 34 worked in oncology or palliative
medicine. The number of years that panellists had worked
in ACP was not asked.
For the definitions and each of the recommendations,
panellists were asked to indicate the extent of their
agreement on a 7-point Likert scale (1=strongly agree;
2=agree; 3=agree somewhat; 4=undecided; 5=disagree
somewhat; 6=disagree; 7=strongly disagree). The panellists
could also provide their feedback on the definitions and on
each recommendation and specify whether there were any
important omissions by writing their remarks in text
boxes. The panellists’ responses were used to calculate the
levels of agreement and consensus.
15,16
Agreement was
indicated in two ways: by the percentage of respondents
either agreeing or strongly agreeing with a definition or
recommendation; and, by a median score, which
represents the 50th percentile value of opinions. A smaller
median indicated more agreement—a median of 1
indicated very strong agreement, and a median of 2
indicated strong agreement.
17
Consensus was calculated
using the IQR. The smaller the IQR, the greater the
consensus: an IQR of 0 or 1 indicated very strong
consensus, whilst an IQR of 2 indicated strong consensus.
17
Open-text comments were analysed in detail by the
respective working group of each domain, and by JACR
and IJK. Recommendations were revised if appropriate.
Recommendations that received very strong agreement
and very strong consensus were accepted or underwent
minor edits only. All other recommendations were adapted
with respect to their content, wording, or ordering, or a
combination of these, or were eliminated to reduce
redundancy. Proposals for adaptations were discussed
within the working groups and within the taskforce.
Round 3
To maintain conformity between rounds, only those
panellists who responded to the online questionnaire
in round 2 were asked to respond to revised recom-
mendations in round 3. In the third round (May, 2016),
round 2 respondents (n=109) were given the original set
of two definitions and recommendations, including
median and IQR scores, and the revised set of definitions
and recommendations. Again, panellists could indicate
the extent of their agreement on a 7-point Likert scale
and give their feedback. If recommendations had
received very strong agreement and very strong
consensus in the second round, experts were presented
with a choice between selecting the default option (that
is, the median score of that recommendation in
the previous round) or, alternatively, to rate the
recommendation again. Of the 109 panellists who
responded in round 2, 103 (94%) responded in round 3.
Round 4
Recommendations that received very strong agreement
(a median of 1) and very strong consensus (an IQR of 0 or
1) were accepted or underwent minor edits only. JACR
and IJK adapted the other recommendations based on
the panellists’ comments. The revised set of
recommendations was sent to the 15 members of the
taskforce in August, 2016, who each independently
indicated whether they agreed with the suggested
changes for each adapted recommendation (“yes” or
“no”). If not, taskforce members were asked whether
they could suggest further improvements.
Round 5
The set of recommendations and definitions was adapted
according to the final feedback of the taskforce. The full
set was then sent to the EAPC Board of Directors.
Findings
The panel and table present the definitions and final
recommendations of ACP. In round 2, the extended
definition was given a median rating of 2 (strong
agreement) and an IQR of 1 (very strong consensus), and
the brief definition was given a median of 2 (strong
agreement) and an IQR of 2 (strong consensus). In this
round, 28 (76%) of the 37 recommendations received
very strong agreement and very strong consensus (a
median of 1 and an IQR of 0 or 1).
In round 3, ten recommendations were added and
three were removed because of redundancy (figure). The
extended definition was rated with a median of 2 (strong
agreement) and an IQR of 1 (very strong consensus), and
the brief definition was given a median score of 2 (strong
agreement) and an IQR of 1 (very strong consensus). For
36 (82%) agreement and consensus were very strong.
In round 4, two recommendations were removed—one
for redundancy and one because of a low score (figure).
Of the taskforce’s 15 members, 12 members rated the
remaining set of eight recommendations that did not reach
agreement or consensus in round 3. Of these
eight recommendations, four received agreement
by all members, whilst the other four received agreement
from seven to 11 of the 12 taskforce members. Feedback
mainly concerned minor changes to the phrasing. These
changes were made, eventually resulting in a final set of
recommendations that reached consensus by the full
taskforce. The full final set comprised a brief definition of
ACP, an extended definition, and 41 recommendations
(including 14 ACP outcome measures). The full final set
See Online for appendix
e546
www.thelancet.com/oncology Vol 18 September 2017
Review
was reviewed by the EAPC board members, who were
unanimous in their support and had no suggested revisions.
Definition
The panel shows the extended and brief consensus defin-
itions of ACP. The brief consensus definition contains all
the key elements of the extended consensus definition.
A central element of the definitions is that ACP is
considered to be a process that includes the identification
of values and defining goals and preferences for future
medical treatment and care and discussion of these factors
with the patient’s family and health-care providers. ACP
can include the docu men tation of preferences and the
appoint ment of a proxy decision maker. These preferences
should be regularly reviewed. Other key points are that the
scope of ACP is broader than the physical domain alone
and can include concerns across the psychological, social,
and spiritual domains. Furthermore, ACP is not limited to
specific patient groups but should concern individuals with
decisional capacity. Both final definitions were rated with a
median of 2 (strong agreement) and an IQR of 1 (very
strong consensus) in round 3. Overall, 91 (88%) panellists
(versus 90 (83%) in round 2) indicated that they agreed or
strongly agreed with the extended definition, and 92 (89%)
(versus 71 [65%] in round 2) with the brief definition. In
total, the panellists provided 97 comments with suggestions
for improvement regarding the extended definition, and
88 comments regarding the brief defin ition. Adaptations of
the extended and brief definitions predominantly con-
cerned the addition that individuals must have decisional
capacity to engage in ACP, the inclusion of the social
domain, and the importance of reviewing preferences.
Recommendations
The table shows the 41 consensus recommendations for
ACP, along with their respective agreement and median
scores, IQRs, and the number of comments provided by
the panellists. The appendix provides an overview of
agreement and consensus scores of the 41 recom -
mendations, indicating that median scores and IQRs
were skewed towards very strong agreement and con sen-
sus ratings. Of the 41 recommendations, 36 (88%)
received very strong consensus and very strong agree-
ment, three (7%) received strong agreement and very
strong consensus, and two (5%) received strong agree-
ment and strong consensus. Of the five domains of ACP:
12 recommendations were related to elements of ACP,
six on the roles and tasks, three on timing, five on policy
and regulation, and 15 on evaluation.
Recommendations that received very strong agreement and very
strong consensus
Recommendations relating to the elements of ACP
concern the exploration of the individual’s current
understanding of ACP and the adaptation of the process
to a patient’s readiness to engage in the ACP process.
Furthermore, it is recommended that ACP should include
the exploration of an individual’s personal values and
goals for future care. Where appropriate, ACP should
include the provision of medical information (eg, about
diagnosis and prognosis) and the clarification of goals
and preferences for future medical treatment and care
(including a discussion of whether these are realistic).
Additionally, ACP should involve discussing the option of
completing an advance care directive and of appointing a
personal representative, along with determining their
role, as per local legal jurisdiction. ACP should also
encourage individuals to provide family and health-care
professionals with a copy of the advance care directive.
With regards to the roles and tasks domain, it was
recommended that health-care professionals tailor the ACP
conversation to the individual’s health literacy, style of
communication, and personal values. Health-care
professionals need to have the necessary skills and show
openness to discuss ACP and to provide individuals and
their families with clear and coherent information.
Furthermore, it is recommended that a trained facilitator
who is not a physician supports an individual in the ACP
process and that the initiation of ACP can occur within or
outside of a health-care setting. For medical elements of
ACP (such as discussing diagnosis, and exploring the extent
to which goals and preferences for future medical treatment
and care are realistic), health-care providers are needed.
For the timing of ACP, it was recommended that indi-
viduals can engage in ACP at any stage of their life, but that
the ACP content should be more targeted when the
individual’s health condition worsens or as they age.
In these circumstances, ACP conversations and documents
should be updated regularly because values and preferences
can change over time. It is further recommended that
public awareness of ACP should be raised.
For policy and regulation, it was recommended that
advance care directives have both a structured (ie, check-
box) and an open-text format. Health-care organisations
are encouraged to develop triggers for the initiation of
ACP, and set up reliable and secure systems to store
copies of advance care directives in a patient’s medical
Panel: Consensus definitions of advance care planning
Extended definition
Advance care planning enables individuals who have decisional capacity to identify their
values, to reflect upon the meanings and consequences of serious illness scenarios, to define
goals and preferences for future medical treatment and care, and to discuss these with family
and health-care providers. ACP addresses individuals’ concerns across the physical,
psychological, social, and spiritual domains. It encourages individuals to identify a personal
representative and to record and regularly review any preferences, so that their preferences
can be taken into account should they, at some point, be unable to make their own decisions.
Brief definition
Advance care planning enables individuals to define goals and preferences for future
medical treatment and care, to discuss these goals and preferences with family and
health-care providers, and to record and review these preferences if appropriate.