Barriers and facilitators for general practitioners to engage in advance care planning: A systematic review

Abstract Objective The aim of this systematic review is to identify the perceived factors hindering or facilitating GPs in engaging in advance care planning (ACP) with their patients about care at the end of life. Design Studies from 1990 to 2011 were found in four electronic databases (PubMed, CINAHL, EMBASE, PsycINFO); by contacting first authors of included studies and key experts; and searching through relevant journals and reference lists. Studies were screened, graded for quality, and analysed independently by two authors; those reporting the perception by GPs of barriers and facilitators to engagement in ACP were included. Results Eight qualitative studies and seven cross-sectional studies were included for data extraction. All barriers and facilitators identified were categorized as GP characteristics, perceived patient factors, or health care system characteristics. Stronger evidence was found for the following barriers: lack of skills to deal with patients’ vague requests, difficulties with defining the right moment, the attitude that it is the patient who should initiate ACP, and fear of depriving patients of hope. Stronger evidence was found for the following facilitators: accumulated skills, the ability to foresee health problems in the future, skills to respond to a patient's initiation of ACP, personal convictions about who to involve in ACP, and a longstanding patient–GP relationship and the home setting. Conclusion Initiation of ACP in general practice may be improved by targeting the GPs’ skills, attitudes, and beliefs but changes in health care organization and financing could also contribute.


Introduction
Consistency between a patient ' s wishes about endof-life care and the actual care he/she receives at the end of life is considered an important aspect of both patient-centred care and quality end-of-life care [1 -3]. This implies that patients ' preferences regarding end-of-life care must be known before they lose the capacity to make these decisions themselves [4].
Advance care planning (ACP) is defi ned as a voluntary process of discussion about future treatment and end-of-life care preferences care between an individual, his/her family, and his/her care providers The advantages of the timely initiating of ACP are well known: it facilitates access to palliative care, stimulates communication between the patient, family, and physicians, and results in greater satisfaction for the patient and the bereaved [11]. In an ageing population, more people will die from serious progressive illnesses, making timely initiation of ACP important [12,13]. General practitioners (GPs) are well placed to encourage and engage in ACP [14,15] and the long-term relationship many patients have with their GP may be a good basis for initiating timely discussion [16,17]. Yet previous research has shown that the incidence of ACP discussions and the completion rate of ADs remain low among the general public and in specifi c patient populations [18 -22]. Only 8% of the general public in England and Wales have completed an ACP document of any kind [23]. Surveys conducted in the USA show that only one-third of adults have an AD expressing their wishes for end-of-life care [24] and even among severely or terminally ill patients, fewer than 50% have an AD in their medical record [25]. In Belgium and the Netherlands GPs discussed ACP with terminally ill patients in a third of all cases and documented the discussion in only 8% (Belgium) and 16% (Netherlands) [26]. Although both patients and physicians support the idea of ACP, these results suggest that certain obstacles still prevail [27 -29].
The objective of this systematic literature review is to identify the perceived factors that hinder or facilitate GPs in engaging in ACP with their patients; this has not been studied before, though understanding of these barriers and facilitators is important for the development of interventions and training programmes aimed at facilitating ACP in general practice.

Search strategy for the identifi cation of studies
Four electronic databases were searched for studies published in English, French, or Dutch between 1990 and2011: PubMed, CINAHL, EMBASE, andPsycINFO. A search strategy was developed by ADV and DH for Medline and adapted to each database separately. A combination of controlled vocabulary and free text words was used to search in titles and abstracts: advance care planning, advance directives, advance decision, advance statement, living will, general practice, primary health care, general practitioners, family physicians, primary care, primary practice, and family practice.
The reference list of all identifi ed studies was screened for additional relevant studies. The fi rst author of each included study and known experts in the fi eld of ACP were contacted for more studies. Furthermore, the most recent issues of 10 relevant journals were hand-searched for relevant papers.

Inclusion and exclusion criteria
An article was included if it reported (1) primary research, (2) on barriers and facilitators (3), on GPs, (4) on patient involvement in ACP. The inclusion criteria are defi ned as follows: (1) Primary research: Both quantitative and qualitative studies reporting original data that contain a clearly formulated research question or study aim were included. Editorials, narrative reviews, comments, and expert opinion were excluded. (2) Barriers and facilitators are conceptualized as predisposing factors, reported by the GP, that hinder or facilitate their engagement in the process of ACP with their patients such as skills, beliefs, and experiences [30].

Inclusion procedure
Duplicates of the retrieved records were removed. Titles and abstracts of all identifi ed reports were screened independently by ADV and DH using a standardized study selection form. The eligibility of selected studies was independently assessed by ADV and DH. Disagreement was resolved by discussion and a third reviewer (KP) was available for arbitration.

Data extraction
Characteristics of the studies included were extracted to a standardized data-extraction form. ADV and DH independently extracted data under the headings of general information, country, research question, design, participants, barriers and facilitators, and quality assessment scores. Barriers and facilitators were extracted from the individual studies as mentioned in the article. Factors that were found as barriers and as facilitators in the same article were reported both as a barrier and a facilitator. Factors only reported as barriers or only as facilitators in an article were also categorized only as barriers or facilitators. Discrepancies between reviewers were discussed and if consensus could not be reached, a third reviewer (KP) was consulted.

Quality assessment and grading evidence
The quality of studies was appraised and evaluated using the Critical Appraisal Skills Programme (CASP) [31]. Since no CASP tool is available for cross-sectional studies, an additional critical appraisal tool developed by Crombie (21-item list) was used [32]. Total quality assessment scores for both qualitative and quantitative studies are presented as scores on a scale from 0 to10.
In addition, the body of evidence from the multiple studies was graded using the three important elements for grading systems suggested by the Agency for Healthcare Research and Quality: quality, quantity, and consistency [33]. The individual studies were categorized as high-quality studies (scores from 8 to 10), medium-quality studies (scores from 6 to 8), and low-quality studies (scores equal to or lower than 5). Articles with low-quality ratings were excluded from further data extraction. Factors reported in two or more high-quality studies were graded as stronger evidence. Factors reported in one high-quality study or two medium-quality studies were graded as medium evidence and factors reported in one medium-quality study were graded as lower evidence. Consistency of the fi ndings was achieved through the classifi cation of all reported factors as barriers or facilitators, preceding the analysis of the results (Figure 1).

Identifi cation of relevant studies (Figure 2)
From the electronic databases searches 442 records were identifi ed. After removal of duplicates and irrelevant reports, the title and abstract of 320 records was screened and 61 full-text articles were retrieved for detailed evaluation. Contact with the fi rst authors of included articles and known experts in the fi eld, a search in reference lists, and handsearching through relevant journals yielded 42 records. Sixteen articles met all inclusion criteria and were included for data extraction and quality assessment as were nine qualitative studies and seven cross-sectional studies.
Characteristics and quality assessment of relevant studies (Table I) Of the 16 included studies, four were conducted in the USA, four in the UK, two in the Netherlands, two in Australia, and one in Belgium, Canada, Singapore, and Israel. Of the nine qualitative studies, six studies used semi-structured interviews and three studies used both interviews and focus groups. Data # of record that were identified through database searching: PubMed (271) Cinahl (86) PsycInfo (48) Embase (37) Total of 442 articles from 4 databases was retrieved 363 records were retrieved after duplicates were removed 320 records were screened by title and abstract 61 full-text articles were retrieved for detailed evaluation 259 records (titles and abstracts) were excluded because they did not meet the inclusion criteria 16 articles were included for data extraction 45 full-text articles were excluded because they did not meet the inclusion criteria 43 records were excludes because they were not original studies (e.g. congress reports) # of records that were identified through other sources: Reference lists (27) Hand-searching journals (7) First authors/Experts (8) in all quantitative studies were collected through questionnaires. Different types of ACP were addressed in the included studies: eight reported on communication about end-of-life care in general, eight others on more specifi c types of ACP (e.g. ADs).
Quality scores ranged from 4.5 to 8.5 for the qualitative studies and from 4.5 to 9 for the quantitative studies, both on a scale of 10. On the basis of these scores, four qualitative studies were considered as high quality, four as medium and one as low (excluded for further data extraction). Of the seven quantitative studies, we appraised fi ve as high-quality studies and two as medium-quality studies. Notes: 1 Research question and study designs are presented as formulated in the articles. 2 Qualitative studies were appraised by using the critical appraisal tool for qualitative research by CASP (10-item list). Quantitative studies were appraised using the critical appraisal tool for surveys developed by Crombie (21-items list). Both scales were converted to a 10-point scale. Table I (Table II) All reported barriers and facilitators were categorized as GP characteristics, perceived patient characteristics, or health care system characteristics.

GP characteristics
Socio-demographic characteristics of GPs. There was medium evidence that the GP being younger was signifi cantly and positively associated with the proportion of patients with whom they discussed endof-life decisions [34].
Knowledge. Medium evidence was found for the reported lack of GP knowledge about ACP as a barrier to involving patients in ACP [35 -37].

Skills.
There is stronger evidence that GPs perceive their own lack of skill in dealing with patients ' vague requests, and their diffi culties in defi ning the right moment for initiating discussion, as barriers to engaging in ACP [14,37,38]. Medium evidence was found that they perceive their lack of skill in dealing with a patient ' s changing preferences and with the emotional impact or discomfort of having ACP discussions as barriers [37,39,40]. Lower evidence supported the perceived lack of skill in advising patients on expressing their wishes, and the prognostic uncertainty for chronically ill patients, as barriers [15,41]. Addressing patient initiation, accumulated skills, and foreseeing health problems in the near future were factors reported as facilitators for which stronger evidence was found [14,15,39,41 -43].
Experience. Medium evidence was found for the length of their experience as a GP and having a living will themselves as perceived facilitators [37,40]. Lower evidence supported positive experiences with end-of-life conversations in the past as a facilitator [15]. Attitudes. The attitude that GPs should initiate ACP was perceived as a facilitator for which stronger evidence was found [37]. There is medium evidence that a conviction that it is their job to cure people whereas other healthcare professionals should initiate ACP prevents GPs engaging in ACP [14,39]. Doubts regarding the content and practical availability of living wills are hindering factors as well [39].

Perceived patient characteristics
Perceived patient-related obstacles can hinder GPs in initiating ACP. The GP holding the following beliefs is perceived as a barrier and supported by lower evidence: patients lack knowledge of ACP, patients have a fear of upsetting their families, and patients are reluctant to think about future health care problems [35,37,42,44]. Medium evidence supports that a patient ' s denial of his/her terminal illness makes talking about preferences for end-of-life care very diffi cult [14].
Anticipated adverse outcomes. Fear of depriving a patient of hope or damaging the GP -patient relationship were cited as factors that keep GPs from engaging in the process of ACP, for which respectively stronger and medium evidence was found [15,34,39].
Personal convictions about who and who not to involve in ACP and when . When asked who should be approached about ACP, GPs designated terminally ill patients and healthy patients in medium-quality studies [15,35,36,45]. GPs reporting that competent patients and cancer patients are more involved in ACP is supported by stronger evidence [26,34,38]. Medium evidence was found that three events trigger discussion between GPs and patients: admission or discharge of patients from hospital, when end-of-life decisions are estimated by the GP to shorten patients ' life by more than one week, and when patients receive treatment aimed at palliation in the last week of life [15,26,34,45].

Health care system characteristics
Related to the GP practice . Stronger evidence supported a longstanding patient -GP relationship as a perceived facilitator for ACP [14,38]. GPs also considered it advantageous if talking about ACP could take place in the home setting [26,38]. There is medium evidence for the time available, and the chances of reimbursement, being facilitators [14,45]. The limited resources available in primary care were perceived as a barrier [40].
Related to other healthcare providers . There is lower evidence that lack of collaboration with secondary care is perceived as an impediment to the process of ACP [15,44]. Consultation with other healthcare professionals and hospital policy supporting or requiring the use of ADs was considered as a facilitator, supported by medium evidence [34,45].
Related to legislation . GPs reported that legislation supporting the use of ADs as well as protecting GPs who follow them would encourage them to offer ADs to patients, which is supported by medium evidence [35,45].

Discussion
We found numerous perceived barriers and facilitators infl uencing GP engagement in ACP with patients. All reported factors were related to three groups: GP characteristics, perceived patient characteristics, and health care system characteristics. Stronger evidence was found for lack of skills to deal with vague requests, diffi culties with defi ning the right moment, the attitude that patients should initiate ACP, and fear of depriving them of hope as perceived barriers. The perceived facilitators for which stronger evidence was found were accumulated skills, the ability to foresee health problems in the future, skills in addressing patient initiation of ACP, cancer patients, patients capable of decisionmaking, a longstanding patient -GP relationship, and a home setting.
To our knowledge, this is the fi rst study to provide a systematic overview of the perceived barriers to and facilitators for GPs engaging in ACP. All required methodological steps to complete a systematic review were implemented and performed separately by two reviewers. This review adds to the knowledge in this fi eld by also including studies on ACP discussions, whether or not these discussions result in written advance directives [46].This review also has limitations. Given the variation in how ACP is implemented and documented and the variation in GP practice, our fi ndings may not be generalizable to all countries and health care systems. Second, only barriers and facilitators reported by GPs were considered although understanding the barriers and facilitators for patients is equally important and deserves research. Third, we retrieved only qualitative research and observational studies, though in our opinion such research designs provide the best way of addressing the research question. As the studies used different methods, it was not appropriate to combine data across the studies for meta-analysis [47,48]. However, the methodological quality was assessed and, in addition, the body of evidence was graded. This approach allows for provision of a conclusion that incorporates both the results and quality of the studies [49].
Stronger evidence was found for the GP attitude that patients should initiate discussions being a barrier and for having the skill to address a patient ' s initiation of discussion as a facilitator. Remarkably, many studies show that patients believe it is the physician ' s responsibility to initiate ACP, which suggests that there is a gap in expectation between patients and GPs. This difference has been pointed out in previous studies and may explain why ACP consultations are often initiated tardily when end-of-life decisions need to made [50 -53].
Most of the perceived barriers for GPs were classifi ed as a lack of skills; it is recognized that physicians are less likely to initiate ACP when they believe they lack the skills required [54]. The perceived lack of skills to deal with a patient ' s changing preferences and to address vague requests and diffi culties with defi ning the right moment to initiate ACP were also found in other health care settings [28,55]. Many of the same barriers could also be found in the literature on communication at the end of life in general and may cover the same ground, since ACP is often initiated at the end of life [56 -58].
According to GPs, cancer patients are more involved in the process of ACP than non-cancer patients. As they often have a more predictable disease course, defi ning the right moment to initiate ACP might be easier. Research has shown that one of the reasons ACP was not initiated with patients with chronic obstructive pulmonary disease (COPD) was because of physicians ' lack of understanding that COPD is a life-threatening disease [21]. Not only physicians but also chronically ill non-cancer patients often have a poor understanding of their illness [59,60]. It is possible that non-cancer patients initiate ACP less often because of a reduced awareness of their diagnosis and prognosis. Most patients and professionals agree that talking about ACP should take place around the time of diagnosis of a lifethreatening illness, but fear of depriving patients of hope is a barrier preventing GPs from initiating ACP for which stronger evidence was found [61,62].
The facilitators identifi ed were often related to health care system characteristics. Previous research showed that conversations about ADs averaged 5.6 minutes and physicians spoke for two-thirds of this time, making patient ' s values and preferences rarely explored [63]. Financial compensation for the time spent on ACP could possibly encourage GPs to make ACP a current practice but it could also acknowledge the importance of this aspect of care.
Understanding the barriers and facilitators is important for the development of interventions aimed at facilitating ACP in general practice. Initiation of ACP in general practice may be improved by targeting GP-related barriers and facilitators, but changes in health care organization and fi nances could also contribute. Training programmes are necessary to change skills, attitudes, and beliefs preventing GPs from initiating ACP and to provide good role models for the diffi cult task of initiating communication about end of life in a helpful and empathetic way.