key: cord-0757377-165oq5yy
authors: Clark, David; Inbadas, Hamilton; Seymour, Jane
title: International transfer and translation of an end of life care intervention: the case of the Liverpool Care Pathway for the dying patient
date: 2020-10-27
journal: Wellcome Open Res
DOI: 10.12688/wellcomeopenres.16321.1
sha: 49f54fcb496e638c4b90d611ff98f1d8aa23c27d
doc_id: 757377
cord_uid: 165oq5yy

We explore how and why the Liverpool Care Pathway (LCP) for the dying patient was transferred to 20 countries beyond the UK, and with what consequences for policy and practice. Our paper synthesises findings from 95 publications contained in a historical narrative literature review on the implementation of the LCP outside the United Kingdom, alongside data from 18 qualitative interviews with 19 key actors involved with the LCP in 14 countries. We use the review to explore the timelines and patterns of development and implementation in the specific countries, to consider what forms of research and evaluation about the LCP were undertaken to establish its effectiveness, and to summarise the resulting findings and their consequences. We use the interviews to gain insights into the elements, processes and dynamics that shaped the transfer and translation of the LCP from one location to another, across national boundaries. Using six questions from the policy transfer literature we then explain who were the key actors involved; what was transferred; from where lessons were drawn; the different degrees of transfer that took place; what restricted or facilitated transfer; and how transfer was related to ‘success’ or ‘failure’. We conclude that the spread of the LCP took place mostly in prosperous countries, and was sustained over around 15 years. It took in differing geographies and cultures, and a variety of linguistic, policy and practice contexts. If it did not succeed in a wider transformational goal, it appears to have been well received and perceived as beneficial in many contexts, largely avoiding accusations of mis-use and harm that had occurred in the UK, and in some cases fostering a sustained international collaboration and ongoing use of local variants, even after withdrawal in its country of origin in 2014.

"… human characters emplotted in a story of (in the early stages) pioneering endeavour and (later) systematic puzzle-solving, variously embellished with scientific dramas,surprisesand'twistsintheplot'" (Greenhalghet al.2005 1 ) ***** "We've worked with the LCP programme not just in England, but in over 20 countries in the world, we have translations in six languages and it's recognised around thatbestcareforthedyingpatient"

It is estimated that 80% of those who die could benefit from palliative care to relieve physical, mental, spiritual and social problems at the end of life. Yet despite demonstrable need, there have been difficulties in identifying appropriate, evidence-based interventions that are scalable across jurisdictions and settings and which might ameliorate suffering when a person is dying. Developed by Professor John Ellershaw and colleagues, the Liverpool Care Pathway (LCP) showed great potential as such an intervention when first formulated in the United Kingdom (UK) in the late 1990s, and where initially it was widely endorsed by policy makers and became a key element in government strategy for end of life care. During the following decade and beyond, this led to further interest in and implementationoftheLCPinternationally.Overthecourseofits development, the LCP went through 12 versions in total. In the light of growing professional and public concern about its widespread use in the UK however, including claims that it denied access to treatment or even hastened death, a government appointed review, led by Rabbi Julia Neuberger, recommended thatitshouldbediscontinued,witheffectfrom2014.

The circumstances surrounding the discontinuation of the LCP make up perhaps the most significant cause celèbre in the field of modern palliative care. At the time of its withdrawal and subsequently, most commentators within the UK addressed themselves to the local situation [3] [4] [5] [6] [7] . Conversely, some commentaries were focussed on the UK context, but were generated elsewhere 8, 9 . In addition, a trickle of UK research studies focussing on the workings of the LCP continued to appear, from the point of its abandonment and after it had ceased to beused 10, 11 .

In this extended paper, we explore one dimension of the LCP story that has attracted no attention from commentators or researchers, but is of considerable significance to policy and practice. We refer to the international spread of the LCP in the years before its withdrawal in the UK and also its continuing saliency in countries other than the UK since that time.This issue is not only a vital component in the ongoing international growth of palliative care, but illustrative of how and with what consequences a specific end of life care intervention can be transferred and translated into a variety of contexts, some of them strikingly dissimilar to its place of origin and initial implementation. It therefore adds to our growing understanding of the reverberations surrounding the LCP and also offers widerlessonsaboutpolicytransferinendoflifecare.

The initial spread of LCP beyond the UK, as we shall see, was mainly dependent on personal contacts and contingent circumstances. Building on this, a European funded initiative, known as OPCARE9 12 . and led by John Ellershaw ran from March 2008 to March 2011 13 . It aimed to 'optimise research and clinical care for cancer patients in the last days of life' 14 . Then, bolstered by the perceived benefits of LCP implementation outside of the UK and the success of the OPCARE9 Collaborative, an LCP International Reference Group (IRG) was launched in March 2011 to support the further development of the pathway internationally. It consolidated a process that had begun over a decade earlier, but it also came in the face of mounting concern about the LCP and the imminent announcement of the Neuberger review. To that end the IRG, meeting in November 2012 to review the implications of the growing debate around the pathway, produced a consensus statementandidentifiedawayforward.Itsetoutabeliefintheneed to improve the care of the dying, along with three specific goals (best possible care, effective communication, robust education and training) and called for more organisational oversight and accreditation of end of life care. But it made no reference to theLCP.

The IRG had members from 13 countries. Despite mounting problems in the UK, confidence in the LCP approach was underpinned by the experience of IRG members, all of whom had been involved in its implementation in their own country, and through which 'the LCP has been shown to be transferable for use in other languages and very different cultural contexts' 14 . Ellershaw and his colleagues further observed that: '… as the debate continues in England, the LCP's country of origin, could an international perspective provide the next steps in improving care of the dying?' Here making no reference to the Neuberger report, they took a positive view of the 'recent media' reporting, concluding that it may yet prove a helpful contributingfactor,ifitweretohelp'driveup'researchandclinical excellence for the care of the dying. This is a well-known policy trope that suggests positive things can emerge from 'scandals' 15 .

In November 2013 the group met again and set out its response to the Neuberger review 16 . It also agreed to establish the International Collaborative for Best Care for the Dying Person 17 , and issued 'The Liverpool Declaration' with a vision 'for a world where all people experience a good death as an integral part of their individual life, supported by the very best personalised care'. The Marie Curie Palliative Care Institute Liverpool would be the co-ordinating centre for the Collaborative 18 . In this manner the impetus that had been generated by the LCP internationally was maintained, albeit by a small number of protagonists, and with moderate published output.

Building on our earlier research scrutiny of the rise and demise of the LCP in the UK 19 , we seek here to stimulate critical reflection on the risks and benefits of disseminating an end of life care intervention across international boundaries, in contexts where there are uncertainties about the original evidence base, and where key actors may not be in a position to anticipate thewiderramificationsofwhattheydo.

To make sense of this issue, we adopt ideas from the concept of 'policy transfer', to shape our research questions, help interpret our findings and also to inform the wider discussion that results. The significance of policy transfer in this context was recognised by Professor Sir Howard Newby, then Vice-Chancellor of Liverpool University in introducing the work of OPCARE9. The work, he stated, was 'about knowledge transfer -not just from the laboratory to the bedside but from one country to another. It is vital that we continue to share our experience and expertise among European colleagues and furtherafieldtohelpimprovecareofthedyingglobally' 20 .

Policy transfer, when first fully articulated as a concept in the 1990s, was largely restricted to political jurisdictions and state actions across boundaries 21 . Over time it broadened to include 'voluntary' as well as more 'coercive' forms of transfer and in particular to capture the involvement of a wider range of participants and settings, including non-state actors, pressure groups, supra-national agencies and advocacy organisations. This wider focus is well described by Benson and Jordan 22 in 2011, who also discuss the motivations of those involved in policy transfer, including a sense of frustration about policy development in specific areas, and attempts to rectify this through persuasion and voluntary transfer from one setting to another. Likewise, they show how the emphasis has shifted from 'hard' forms of transfer involving institutions, policy goals and measureable outcomes, towards 'softer' forms that involve ideas, concepts, and where lines of transfer are horizontal ratherthanvertical.

Ideas about the 'transfer' of policies, processes, systems and actions are closely linked to the question of translation. Challenges in the 'transfer' and 'translation' of palliative care models from one setting to another have been highlighted by Zaman et al. 23 who ask: 'Should we focus on the transfer of palliative care narratives, assumptions, policies and practices from developed to developing countries, or should our emphasis be on the translation of these things in both directions?' (p76).

An exploration of policy translation by Freeman 24 , is helpful in this regard. He notes that commentators often use the phrase 'lost in translation' to indicate that things have gone wrong, but he also suggests that translation may be the 'lubricant' or 'key'totransfer.

When multiple social actors are engaged in the processes of translation, we have found it useful to draw on the idea of the 'boundary object' and have ourselves argued that the LCP can be seen in this light. A boundary object needs to be malleable enough to work in specific local contexts but rigid enough to maintain its integrity across settings. As the originators of the concept of 'boundary object' note, 'protocols are not simply the imposition of one world's vision on the rest; if they are,theyaresuretofail.Rather,boundaryobjectsactasanchors or bridges' (Star and Griesemer 1989 25 , quoted in Freeman 2009 ).

We consider that translation of policy thus conceived, becomes a complex, dynamic, multi-lateral affair, full of pitfalls, potentialities, opportunities, road maps and culs de sac. Taking an idea or practice from one place to another will therefore depend on far more than its integrity or robustness. Such transfer will require negotiation, interpolation, bargaining and flexibility in the rules of encounter. Freeman captures precisely the ground we are interested in here, and which links in turn to our earlierpaperontheLCPintheUK:

Translation is something like a boundary object. It is not an object, of course, but a practice and vocabulary within which the nature of research, policy and practice and the relationship between them is being rethought. It is the means by which an array of actors, including international organizations both public and private, governments, sponsors, researchers, policy makers and practitioners have come to communicate about a problem even in the absence of any fully shared conception of it. These debates about translation are themselves instancesofit(p445).

Two major protagonists in the field, Dolowitz and Marsh (the former interestingly a member of staff at Liverpool University, from which John Ellershaw led his programme of work on the LCP), in a classic paper of 2000, set out their conceptual framework for the analysis of policy transfer 26 . The framework is organised around six questions: 1) who are the key actors involved? 2) what is transferred? 3) from where are lessons drawn ? 4) what are the different degrees of transfer? 5) what restricts or facilitates transfer? 6) how is transfer related to 'success' or 'failure'? Towards the end of this paper, we map these questions on to our data about the international spread of theLCP.

Circling around this central theme of policy transfer, the analysis framework and the role of the boundary object within it, we therefore want to find ways to recognise that making sense of the international spread of the LCP also involves a deeper understanding of knowledge construction and interpretation across differing jurisdictional and cultural settings 27 . Our two main sources for this (published literature and interviews with key actors) point to the relationship between formal knowledge claims and their consequences and implications at a local level, and are bolstered by the kind of theoretical frameworks described here. They help us to make sense of a rich and extensive experiment in the transfer of ideas and practices designed to work in many locations with the goal to improve end of life care, even in the absence of any fully shared conception of whatthatis,toparaphraseFreeman.

Our aim in this paper is to construct a detailed case study of how the LCP, an end of life care clinical intervention that had been developed in the UK, was adopted in 20 other countries and the consequences that resulted from this, including when the intervention was withdrawn from use in its country of origin. We have taken as our research questions, those posed by Dolowitz and Marsh and we weave these in with a number of LCP-specific questions and themes that emerged from our analysis.

We made use of two principal approaches 1) a historical narrative review of published and grey literature relating to the LCP in the international context and 2) qualitative interviews with key actors involved in LCP implementation, research or discussionincountriesoutsidetheUK.

The study was approved by the Research Ethics Committee of the University of Glasgow, College of Social Sciences, on 2 April 2017, project number 400160110. Prior to data collection, the purpose of the study was explained to all participants and written consent for 'on the record' participation in the study was obtained from each participant at the time of interview. Participants were informed that they had the right to withdraw fromthestudyatanypointintime,withoutanyrepercussions.

The historical narrative review of the literature and associated commentaries on the use of the LCP outside of the UK is presented as part of the Underlying data 28 to this paper and is intended to serve as a resource for further study. Our inclusion criteria for the review were: studies (and wider commentaries) on the use of the LCP in countries outside of the UK, including published articles, conference abstracts and presentations, reports and grey literature. Our exclusion criteria were: studies (and wider commentaries) on the use of the LCP within the UK, including published articles, conference abstracts and presentations, reports and grey literature; we also excluded all studies referring to 'pathways' from the UK andelsewhere,whichdidnotreferexplicitlytoLCP.

A baseline English language PubMed 'all text' search for 'Liverpool Care Pathway' was conducted in May 2019. It generated 39/211 outputs, which met the inclusion criteria. Further searches were conducted in January 2020, with the following results:CINHAL5/22;PsychInfo3/53;SCOPUS2/252;Proquest 0/47. The searches identified 49 outputs in total. Hand searching of this material for further relevant references, along with Google searches relating to the use of the LCP outside the UK and personal communication with other researchers, plus examples given to us from those we interviewed, together yielded a further 46 outputs. This process involved judicious use of Google translate as well as assistance from colleagues with particular linguistic skills (Mandarin, Japanese, Norwegian, German, Spanish). Some non-English outputs contained abstractsandsummariesinEnglish.

The total number of outputs contained in this review is therefore 95, covering 20 jurisdictions 29 . We realise of course that this will not be comprehensive and we welcome suggestions for further outputs that might be included in revised versions of the review. We are much in agreement with Greenhalgh and Peacockwhentheystate:

Systematic review of complex evidence cannot rely solely on predefined, protocol driven search strategies, no matter how many databases are searched. Strategies that might seem less efficient (such as browsing library shelves, asking colleagues, pursuing references that look interesting, and simply being alert to serendipitous discovery) may have a better yield per hour spent and are likely to identify important sources that would otherwisebemissed 30 .

Sampling. Purposive and snowballing sampling techniques were used to identify potential participants for qualitative semistructured interviews about the international spread of the LCP.Thisapproachtosamplingfacilitatesthechoiceofrespondents who are strategically located in a situation from where they are able to shed light on the subject of study at hand 31, 32 . The target group included: clinicians in leading roles with experience of LCP implementation, researchers who had studied the LCP outside of the UK, policy makers involved in LCP introduction, and global experts in palliative care with knowledgeofLCPintroductioninparticularnon-UKsettings.

The initial sampling frame consisted of those individuals reporting on the use of the LCP in the 2011 LCP handbook 33 , totalling 11 countries (Argentina; Slovenia; India; Norway; Italy; Switzerland, Germany and Austria [the DACH German speaking collaborative]; Sweden; Netherlands; New Zealand). Everyone we approached agreed to take part, with the exception of one person (Slovenia). Following leads from the linked literature review and recommendations from interviewees, we then invited potential interviewees from seven further countries where there was evidence of LCP implementation; people from four countries accepted (Australia, Belgium, Denmark and Japan), whilst three (from Hong Kong, Ireland and Spain) declined to take part. In two instances (New Zealand, Belgium) two people took part in the same interview. For some countries we had interviews with more than one person: in two countries we interviewed two separate individuals (Australia, Japan), and in one country we interviewed the same person twice (Netherlands). We thereby completed 19 interviews with 20 people from 14 countries in total -though these figures are each reduced by one, following the withdrawal of one interviewee from the study. The analysis here is therefore based on 18 interviews with 19 people. The interviews tookplacebetweenAugust2017andDecember2019.

Unfortunately, we had to proceed in this work without the involvement of Professor John Ellershaw, leader of the LCP international initiative. We had hoped he would be our first interviewee, in the manner of an 'index case', setting out his perspective on the international spread of the LCP and guiding us towards others who could assist with our research, but he declinedtoparticipateinourstudy 34 .

We sent introductory emails to potential interviewees, explaining the purpose of the study and enclosing an information sheet. Individuals were invited to a telephone or SKYPE interview at a mutually convenient time.We asked individuals to consider participating in an 'on the record' interview (although this was not mandatory), since interviewees were likely to be easily identified by colleagues in the palliative care field from our resulting reports and publications. Individuals who agreed to take part in an interview were asked to complete a consent form, and to indicate on the latter whether theywerewillingtoparticipate'ontherecord'.Allofthemagreed tothis.

We developed an aide memoire (see Figure 1 ) based on the aims of our project and themes in the literature review. Interviews were audio recorded and transcribed bya specialist agency,boundbyaconfidentiality clause. The aide memoire was adapted according to the context of the interview and the background of the interviewee. Where appropriate, we sometimes included more than one person in the interview. In one case (the Netherlands), we carried out a repeat interview to clarify material and to bring our understanding of research developments up to date. In another case (Belgium), the interview was conducted in two parts because of technical challenges and poor sound quality in the first part of the interview. Interviews ranged in length from 36 to 66 minutes. All three authors were involved withtheinterviews(Inbadas,12;Seymour,5;Clark,2).

Our analysis followed the principles of the framework approach to qualitative data, as described by Gale and colleagues 32 . Framework analysis sits within the broad range of thematic approaches to qualitative analysis and is especially appropriate where interviews have followed a similar structure or line of discussion. Its key feature involves the development of matrices to enable systematic comparison of content between 'cases' or interviews (sometimes called 'charting'). It can complement the use of other more analytical strategies: for example, in this study we also undertook a preliminary coding exercise using NVIVO 11 and, as a step towards interpretation, subsequently developed a detailed narrative 'write up' for each interview.

International literature review Twokeydimensionswereidentifiedinthereview.

First is a story about the international spread and publication activity associated with an end of life care intervention. Its parameters are the countries and timelines and intensity of production (Table 1) . Second is a story about the evidence generated to shed light on the use of the intervention in jurisdictions outside the UK. This relates to the frames of evaluation that were adopted to study the LCP, the principles and designs that were used, the settings in which they were deployed, the results that emerged from these endeavours and the forms of adaptation made to the LCP, contingent upon its use in settings outsidetheUK(Figure2).

The literature search produced outputs from 19 countries, to which the LCP had been transferred in some way. To this can be added a commentary on Slovenia, from the LCP handbook of 2011 33 . If we add the UK to the total, this brings us to 21, or the figure of 'more than 20 countries' that is frequently cited in LCPliterature.

The diffusion from the UK to other countries was most evident in parts of western Europe; it was significant in parts of East Asia and Australasia; confined to one country per region in Eastern Europe, the Indian subcontinent, and South America; and non-existent in Africa the Middle East, the USA, and Canada. LCP was therefore a phenomenon essentially of the Global North, albeit without North America (Table 1) . Only two countries to which is was transferred were in the low and middleincomecategory(IndiaandArgentina).

The LCP first appeared beyond British shores in 2001 in a Dutch translation based on LCP Version 4 that was piloted in three palliative care settings (hospital, nursing home and hospice). In total, 13 publications about the LCP appeared from Dutch researchers between 2003 and 2018, gradually spreading beyond the original three settings to cover the whole of the country. By 2012 the 'Zorgpad Stervensfase', as the Dutch version was known, had been implemented in over 100healthcareinstitutionsintheNetherlands

Belgium, by contrast, whilst working with Dutch and Italian researchers, produced no studies on LCP as such, but conducted work leading to four publications (2015-17) on an intervention that had grown out of the LCP, and in particular was developed in light of the critique of LCP that had arisen in the UK. Implemented in 10 hospitals in the Flanders region, it was focussed on the end of life care of older patients in acute geriatricsettings.

Just one relevant study was identified from Ireland. This may seemsurprisinggiventheproximitytoLCPoriginsinLiverpool, and a shared language. It does appear however that the Hospice Friendly Hospitals programme 35 , which got underway in 2007 and adopted an all-systems approach to improving end of life care in acute and community hospital settings in Ireland, saw LCP and other related pathways as useful resources for hospital end of life care improvement, but was reluctant toadvocatefortheLCPspecifically.

German language clinicians and researchers formed a threecountry group (Germany, Austria and Switzerland) to promote their shared interests in LCP and this resulted in 13 publications between 2007-2017, including one journalistic commentary. Although LCP gained traction in Switzerland and had an influence on federal planning, coverage in the other two countries wasmuchmorelimited.

Among the Nordic countries, Norway was a significant early adopter in 2005-6 (especially in nursing homes) but produced almost no primary published work on LCP, though it did generate wider commentary and review. The 2011 chapter in the LCP handbook refers to a 'flying start to implementation' in Norway, where the Regional Centre of Excellence for Palliative Care in Western Norway took a co-ordinating and facilitating role. Sweden was more active in pursuing research work on LCP adoption and also established a national centre for co-ordination and knowledge exchange based at the research and development unit of Stiftelsen Stockholms Sjukhem. By 2014 over 200 services in Sweden were using the LCP,includingspecialistpalliativecareunits,homecare,hospital wards and nursing homes and that experience was feeding into the development of a national plan for the care of the dying. In Denmark, implementation was localised to one setting and dissemination restricted to conference presentation. Ten outputs emerged from these three Nordic countries, including one of the few controlled studies, which was conducted in Sweden.

Colleagues in Spain and Argentina collaborated with each other on their interest in implementing the LCP into local services, resulting in four outputs, with a fifth that came to our attention after the analysis was compete, consisting in total of three journal articles and two conference abstracts. Again, the useofLCPwaslocalisedinasmallnumberofareas.

Interest in LCP in Australia and New Zealand was evident from 2007 onwards and led to 15 published outputs. In Australia, the work was concentrated in acute hospitals at first and later shifted to include aged care facilities also. In New Zealand, Dedicated Ministry of Health funding for LCP implementation was made available to all 20 Health Boards from 2008, prior to which some Boards had already shown interest in LCP. By these means LCP was adopted to some degreeacrossNewZealand,infoursettings-hospices,hospitals, agedresidentialcarefacilitiesandpatients'ownhomes.

Interest in LCP first appeared in Hong Kong, China, Singapore andJapanfromaround2007-8,whenthefirstof21publications

began to appear, the large majority from Japan, where various studies were conducted in hospital and home settings. In 2010, a special issue of the Japanese Journal of Clinical Nursing was published that focussed entirely on the use of LCP in Japan. Again, specific hospitals were involved in LCP development work, but in the absence of a co-ordinating centre. The LCP initiative in India was endorsed by the national association for palliative care, and a local version was agreed with Liverpool and known as LCP-IICP (Indian Integrated Care Pathway). Enthusiastically supported by collaborators in and outside the country, the initiative was nevertheless small scale, confined to four specific locations (Kerala, in two centres; Chandigarh; and Karnataka) and published only through conferenceabstracts.

The work on LCP in Italy is not only substantial in terms of published outputs, but, as we shall see in the next section, is arguably the most robust in terms of scientific rigour, and includes the only example of a randomised controlled trial (RCT) of the LCP itself. It is therefore very closely tied to the overall LCP narrative, since the results of the RCT appeared after, and not before, a decision was taken to withdraw the LCP in the UK. Twelve papers on LCP by Italian authors were published between 2011-2015, the penultimate one a review of pathways and evidence. The work was mainly conducted by a team of researchers and clinicians associated with a regional palliative care network of the National Cancer Research Institute and located in Genoa, Italy. It was led by Dr Massimo Costantini in a three-year programme of research funded by the Italian Minister of Health from 2006, to determine whether the implementation of LCP in a hospital setting could beeffectiveinimprovingendoflifecareforcancerpatients.

Translation,culturesandsystems The LCP international strategy adopted a well-tested and recognised linguistic translation model, based on the principles used by the European Organization for Research and Treatment in Cancer (EORTC). These include forward translations into the target language, reconciliation, back translations into English, proofreading by an independent linguist, pilot-testing and finalisation of the translation 36 . Several of the reviewed publications make reference to this and to how it was applied to a specific numbered version of the LCP. Occasionally there is referencetosomepointoflinguisticdetailwhereawordsuchas 'secretion' (in the Netherlands) is a verb but cannot be a noun, or in Spain and especially Argentina, there is reference to a cultural translation, which in some unspecified way adapts the LCP to the local culture. In New Zealand and Australia there is also reference to the translation of LCP into the local health care system. Nowhere do these accounts delve more deeply. 'Translation' is therefore essentially seen as a technical process, with occasional cultural dimensions, which can be accomplishedbycarefuluseofprocess.

The LCP handbook is not specific about the health care systems into which transfer would be most effective or needed. In general, it takes a somewhat local perspective, focussing on the specific 'institution', 'organization' or 'local health economy' to which LCP can be applied. Yet this can be contrasted with an approach which encouraged the creation of a nationalorjurisdictional'centraloffice'toco-ordinateanddirect LCPimplementationandliaisewiththe'LiverpoolCentral'.

Some form of implementation research, however basic, was conducted in almost all 20 countries that took up the LCP. But in several jurisdictions, implementation research was un-sustained and lacking in strategy; the exceptions to this were Belgium, Italy, and the Netherlands, and to a lesser extent Australia, Japan, Sweden, and New Zealand.

It was possible to assign all 95 outputs in the literature review to a particular category of research or commentary These are shown in Figure 2 , in which the numbers given for the publications in each category are the original numbers used in the full literature review that appears in the Underlying data 28 tothispaper.Inthissectionwediscussthreecategoriesofevaluation that we identified: 1) audit, 2) implementation analysis, 3) controlled studies, along with 4) related commentaries and other works. It should be noted that the following section includes direct mention of 62 references from the literature review,comparedtothetotalfigureof95.

Most common (n=32) were studies using an audit design and assessing aspects of LCP usage, sometimes with and sometimes without baseline, pre-implementation assessment. Of the 15 countries that undertook these types of study, the most active were the Netherlands (n=7), Japan (n=4).Australia (n=3), and New Zealand (n=3). Some of these audits quickly raised questionsaboutadaptationtolocalcircumstances.

The baseline work was usually undertaken as part of a requirement on the part of 'LCP Central', in Liverpool, which favoured pre-implementation case audits of around 20 deceased patients as part of the process of adopting the LCP. There was a marked tendency in some of the early audit studies to ascribe considerable benefit to what might seem like very small improvements in pre-and post-results or from very small samples. An example of the former occurred in the Netherlands, where one study compared the level of documentation, symptom burden and aspects of communication before and after the introduction of LCP in 220 patients iv . It found 'modest but evident' improvement in the amount of documentation of the patient's dying phase post-implementation, a 'small but significant' reduction in symptom burden, but no difference in relatives' reported views about communication. Nevertheless, the authors considered this to be 'a remarkable result of using a care pathway that mainly introduces a structured registration method, rather than a new intervention or therapy'. Similarly, in Switzerland in 2005, although a hospital pilot revealed the need for considerable support in completing the LCP documentation, the authors reported tangible benefits across the pilot stations: faster switching to comfort therapy; greater recognition of the dying process with a more shared language between staff; fewer oversights, due to a more structured procedure; patients and their relatives receiving more comprehensive care; and despite doctors' scepticism about the value of time spent in the associated round table discussion meetings, the meetings themselves were said to be calmer v . Some substantial claims for improvement were therefore built on a modest platform -in this case of 16 LCPexamples. In addition to palliative care units, some general wards in hospitals in Japan also adopted the LCP for their terminally ill patients. A study by Nobuhisa Nakajima, a doctor in a palliative care unit in Sapporo Minamiseisyu Hospital, unlike other empirical studies described here, used a direct Japanese translation of the LCP to provide care for dying patients on a general ward vii . The care pathway was introduced in two phases. Positive outcomes were gained to some extent in the first phase, although the variance rates were relatively high; this was attributed to practitioners' limited knowledge of symptoms at the end of life and the lack of communication with patients' families. To improve the practice, the team integrated the Support Team Assessment Schedule (STAS) to enhance knowledge exchange and communication between different parties involvedinthecare 37 .Assuch,theresultsfromthesecondphase weresignificantlybetter.

Although audit studies of this type were often of simple design, in some cases where there was significant attention to process issues, it was difficult for us to separate them categorically from studies that had a stronger emphasis on aspects of implementation. We categorised studies as 'audit' when pre and/or post implementation measurement was the main objective;

iii but some studies added other elements to this, such as focus groupswithstafforsurveysofopinionamongprofessionals.

Where this tendency to mixed methods was stronger and focusseddirectlyontheacceptabilityaswellastheeffectiveness of LCP or a local variant, then we classified the study as a form of process or implementation analysis. These studies (n=27) were almost equal in number to the more narrowly executed audit work. The most active of the 10 countries where theytookplacewereNewZealand(n=5)Japan(n=5)Germany (n=4),Italy(n=4),Australia(n=3).

A very small minority of studies (the best examples were in Italy and Belgium) drew on the foundations of implementation science, specifically, the principles of complex intervention evaluation. Beyond these, most studies in this group were descriptive in character and concentrated, for example, on issues of LCP acceptability to staff or used the views of lay carers to assess the benefits of the intervention. From the year 2000, the LCP Central Team in Liverpool worked with colleagues from several countries to implement the use of the LCP and focussed on four phases of activity: 1) Induction, 2) Implementation, 3) Dissemination, and 4) Sustainability. In each case there were clear requirements and prescriptions for how the work should proceed. Implementation into pilot sites should ideallyfollowanalgorithmof'plan','do','study','act'-inorder to foster continuous learning and some measure of whether improvement occurred. In total, 80% of local staff should take part in an education programme about LCP before first introduction. Periodic status reports should be supplied to LCP Central. The 'study' component in the algorithm sometimes led to aspects of research that went beyond audit and captured the views of actors involved in the process, the dynamics and day to day realities of using the LCP, as well as the direct measurementofclinicaldata.

In a paper on the Dutch pilot process viii adaptation and translation processes are well described, based on Version 6 of the LCP and adopting principles established by the European Organisation for Research and Treatment on Cancer (EORTC). Within these principles, the 'forward translation' (English to Dutch) was undertaken independently by two native Dutch speakers (a doctor and a nurse) and then a third person (a doctor) compared and reconciled the two versions. All three translators worked in palliative care. This version was then subjected to a process of 'back translation' when two native English speakers (a palliative care nurse and a professional translator) independently translated the provisional Dutch version back into English. These versions were in turn reconciled by a third person who had been involved with the original development of the LCP, and who verified that the goals of care had not been changed in the translation process. The Dutch version that resulted was given the name 'Zorgpad voor de Stervenfase-RotterdamZS-r(lcp)', with the acronym added as a confirmationofitsauthenticity.

A similar level of rigour accompanied the pilot implementation in a university hospital, where there were interviews with staff to evaluate their perceptions of the pathway. Regular meetings were held to review how the document was working and staff were surveyed on their views, one year after the pilot was completed. If the linguistic translation had been relatively straightforward, the transition into practice was not. There were issues around the meaning of 'spiritual' (now replaced by the phrase 'important values'). 'Secretion' was understood in Dutch only as a verb and not as a noun. Adjustments were also required to align the document with procedural practices specific to the Dutch healthcare context, for example concerning information-givingafterdeath.

The follow up hospital questionnaire was administered to a total of 20 nurses and 15 doctors. In total, 22 people responded (63%). The authors reported the item responses in percentage terms and found 72% considered the LCP helpful in structuring patient care and 55% felt the same was true for family and proxy carers. A large majority considered the LCP was helpful in anticipating problems (82%), facilitating multi-disciplinary communication (73%) and contributing to better care in the last days and hours of life. The three Dutch and two English authors could end on an optimistic note: 'In this way, the potential for promoting optimal care of the dying and comparing outcomes across geographical borders is promoted, and the opportunity for continuous quality improvement for care of the dying in an international sense is a tangible prospect' (p.159). It was a claim made, however, on the feedback from just a coupleofdozenhealthprofessionals.

A further study, already mentioned, from the palliative care unit of the department of medical oncology at the Erasmus MC-Daniel den Hoed Centre drilled down into how the LCP was working in the Dutch context, using an anonymous retrospective audit methodology ix . Here the aim was to assess experience in the new setting and compare it with a matched group of patients in Liverpool, cared for using the LCP in a free-standing hospice environment. The choice of contrasting settings (hospital, non-hospital) is not explained, but there were similar results across a number of important dimensions, with most care goals being met for the large majority of patients. LCP was activated however in only 50% of those who died in Rotterdam,comparedto85%inLiverpool.

In Germany, the Ev Hospital in Oldenburg was the first to use the LCP. The setting was a specialist palliative care unit. From May 2007, the LCP was deployed over a period of five months, using a version that had been prepared for use in St Gallen, Switzerland, under the leadership of Professor Steffen Eychmüller. In a paper, which includes Eychmüller as a co-author,Simonet al. x describetheresultsofafocusgroupwith 10 members of staff on the ward, conducted after 24 patients had been cared for using the LCP. During this period, a total of 36 patients died in the palliative care unit; the 12 who were not placed on the pathway all died suddenly. All members of the palliative care team were invited to participate in the focus group (10 nurses, three doctors, a caregiver, a social worker, a physiotherapist, an art therapist), but just seven nurses and the three doctors took part. All the participants had experience oflookingafterpatients(theaveragewasseven)ontheLCP.

The results of the discussion are described at length in the paper. Overall, the participants were extremely positive about LCP. Echoing the discussions in St Gallen, the timing of when to start LCP ('diagnosing dying') was described by the focus group participants as an intense and important process. As 'Nurse 5' put it, the 'moment when we decide to start with the LCP, now that is somehow a very special one and reconsidering it consciously, is something I experience as very positive'. After this decision is made, and based on open exchange in the team, all further measures could then be coordinated together, for example: the discontinuation of investigations or therapies that are stressful and unnecessary for a dying person; the use of on-demand medication for common symptoms; orthesupportofrelatives.

Participants reported unanimously that the LCP enhanced communication between nurses and physicians, which in turn encouraged patient interaction and family caregiving. The structure of the LCP provided reassurance (especially around shift hand-over) that everything was being thought through and essential questions clarified. The flowcharts for drug-related symptom control attached to the LCP were found especially helpful for younger doctors with less experience in the field of dying. Likewise, the schematic structure of the LCP was a positive attribute, as the objectives to be achieved were well explained. Practical hints, such as informing the family doctor about the patient's situation were also described as helpful andassomethingthatcouldoftenbeforgottenindailypractice.

At the same time, there were some concerns that 'dying people would be ticked off' (Doctor 3) and that individualised care would be threatened. Working with the LCP was initially considered more time-consuming but despite that, the staff felt better, 'because everything had been thought of and you were not just drifting' (Nurse 4). Importantly, the participants also felt that the LCP was well-suited for staff on wards that are less likely to care for dying patients, as it provides a checklist to think about everything in this situation, particularly if back-up was also available from a specialist team: 'The LCP provides an opportunity to ensure a certain basic care for the dying'(Physician1).

The authors of the Oldenburg study concluded that the LCP is a helpful and pragmatic tool for implementing palliative care in everyday clinical practice, but must always be supplemented and accompanied by qualitative guidance and palliative care training. There were however some limitations to the study, noted by the authors. First, only one focus group was conducted, with just 10 professionals, which could limit the scope of the results, albeit most of the team took part. Second, the LCP was implemented in a palliative care unit, though the target is the general ward of a hospital. Third, the study provided only the impressions of the health professionals involved in the pilot and did not provide evidence of the measuredeffectsofimplementation.

Colleagues in Argentina and Spain worked together on the use of LCP in palliative care services in the two countries and described the LCP translation and implementation processes and the initial piloting with 60 consecutive patients in two hospitals and one palliative home care setting and then the subjectiveperceptionsofhealthprofessionalsbeforeandafterthe introduction of LCP 'in a Latin American cultural context' (Tripodoro 2013:2) xi . Their focus was on the meanings assigned by professionals to the care of the dying, and on communication, teamwork, documentation, and particular attitudes. Here, LCP (Version 12) had been re-named, as in the Netherlands, but now with a much more culturally specific acronym: PAMPA (Program Asistencial Multidisciplinario Pallium). The study had two components. The first comprised a focussed ethnography within a hospital based palliative care team that had started training in PAMPA. The second comprised a questionnaire survey about professionals' views on the implementation of PAMPAinArgentina by palliative care teams trained in PAMPA. The median range of follow up in the five centres from the beginning of the pathway until death varied from 16 to 178 hours. It was concluded that PAMPA demonstrated its feasibility as a model of end of life care for patients and families, based on international qualitystandards 38 .

Dedicated Ministry of Health funding for LCP implementation was made available to all 20 New Zealand Health Boards from 2008, prior to which some Boards had already shown interest in LCP. By these means LCP was adopted to some degree across four settings in New Zealand -hospices, hospitals, aged residential care facilities and patients' own homes. But as early as 2007, staff at Arohanui Hospice, the recognised lead collaborating centre for LCP, had begun to recognise some inconsistencies in how LCP was being implemented around the country. These included: a lack of consultation with specialist palliative care services, inappropriate and sometimes unsafe symptom management algorithms, the absence of general practice teams from LCP education and training, variability in LCP registrations, and the development and use in some places of modified, non-compliant LCP documents. Thus informed, the Arohanui Hospice made a successful bid to the Ministry of Health to establish a national co-ordinating office to oversee LCPimplementationinNew Zealand,withsupportfrom the Liverpool team. The goal was to develop a robust support infrastructure that would minimise the risk of the kind of ad hoc implementation and dissemination of LCP that would dilute and compromise its effectiveness and sustainability over time. The core approach to achieving this was the 10-step continuous quality improvement programme, developed by LCP Central in Liverpool. New Zealand was thus the first country outside the UK to formally establish a National Office with responsibility for promoting the sustainable implementation of LCP within its own borders. A paper by Mackenzie et al. xii presents the results of a mixed methods study to evaluate the role and value of the New Zealand office, from the perspective of key stakeholders, and also provides useful context on the localadoptionofLCP.

Data collection for the evaluation took place in 2009, just six months after the New Zealand LCP office had been established. Committed to principles of dependability, credibility and trustworthiness, the evaluation was designed to provide useful information to inform development. It drew on the perspectives of a purposive sample of key stakeholders across New Zealand through interviews (n=28) and questionnaire surveys (n=36). The results were positive. The goals of the LCP office were deemed important, the service quality was rated good or very good, its ongoing links with LCP Central were considered important, it was leading to better quality use of LCP by linking closely with local facilitators in ways that connected theory to practice, and it was serving as a voice for palliative care in New Zealand. The authors concluded that the New Zealand office was proving successful in mitigating the risks of LCP implementation in a country 'geographically isolatedandculturallydistinctfromtheUK'(p260).

In Sweden there was a substantial engagement with the LCP initiative, albeit in a series of publications that did not begin appearing until after the withdrawal of the LCP in the UK. Ekeström and colleagues xiii sought to explore family members' experiences in a palliative care unit and in a general geriatric ward in Sweden, before and after implementation of the LCP, which had first been introduced into Sweden in 2007 as part of a national project monitored by a palliative care competence centre, with the documents translated according to EORTC guidelines and implemented in collaboration with LCP Central, in Liverpool. The study places a particular focus on the perceptions of family members relating to LCP, citing only a few examples of this from elsewhere in contrast to the large number of studies on the perceptions of staff. The design was a non-controlled, before-after evaluation of the impact of LCP on family members' experiences in a palliative care unit and in a general geriatric ward, with special attention tothegoalsoftheintervention.

The settings for the study, each of which had introduced the LCP in 2009, were in the urban area of Stockholm and data was collected by means of self-complete postal questionnaire sent to a relative 3-6 months after the patient's death. In total 108 family members agreed to participate (85%) and response rates and the before/after numbers were roughly equal across both clinical settings. Satisfaction with care was high in both settings pre-implementation, and family members were confident that staff had done everything possible to prevent suffering. Satisfaction on measures relating to existential issues and information on bereavement support was lower in the hospital ward, where relatives also considered that the patient had been more likely to experience breathlessness in the last three days of life. Post-implementation, only one aspect of care showed better results and this was in the Palliative Care Unit,wherephysicians'abilitytolistentoquestionsandrequests had improved. But post-implementation family members were more likely to state that the patient was worried or anxious. The authors considered that more information may have made family members more observant of symptoms, hence the increaseinreportedanxiety. (11) (12) ,pp. 1663-1673. dementia in nursing homes. The study made use of the three phases (initial assessment, continuous assessment, and after death follow up) as a framework to sort the documentation. The study did not concern the implementation of LCP, but focussed on 50 sets of nursing records and 50 sets of medical records relating to deceased patients with dementia in two nursing homes. Through the lens of the LCP, it revealed that the great extent of the documentation focussed on physical symptoms, with almost nothing recorded on existential issues or follow up withrelativesafterdeath.

A descriptive qualitative study also emerged from the implementation of LCP in the Skellefteå municipality of Sweden xv . It complemented the main evaluation by examining care professionals' experiences of using LCP in the residential care homes of the municipality. The work was conducted through five focus groups and two individual interviews, comprising a mixture of nurses and nursing assistants working in the care homes, as well as local GPs. The line of questioning focussed on how the participants had experienced using LCP, its influence on their practice, as well as their perceptions of its strengthsandweaknessesandtheirviewsonusingitinthefuture.

The participants considered that using LCP had fostered a team approach that increased individual confidence, and this had been generated through the supporting educational programme and the new way of documenting care. Agreement about whether the resident was dying could prove difficult to establish, and this was exacerbated by the short amount of time spent in the homes each week by the GPs (who took the decision in each individual case about LCP use). But there was also an enhanced sense of individualised care when using LCP. The paper-based document kept in the patient's room provided a focus for this, but was also seen as less practical than the usual computerised record keeping system. LCP was seen to have fostered earlier and greater involvement of family members in end of life discussions with relatives. Care staff were also more attentive to the care environment and gave more attention to creating a sense of comfort and welcome for family members when a resident was dying. Appearing late in the LCP 'cycle', the authors were able to contextualise their study in the findings of similar work from the UK and elsewhere. They concluded cautiously that the LCP might be a useful tool for use in residential care homes and could increase attention to the goals of care, the individual needs of residents, and the involvementofrelatives.

Another Swedish study by Andersson et al. xvi attended to family members' experiences of care of the dying in residential care homes where the LCP was in use. A total of 15 family members of deceased residents took part, drawn from 10 different residential care homes. Interviews were conducted January-March 2014 and began with the question: 'Please tell me about your experiences of your relative's last hours/days of life'. Three themes emerged: 1) being confident in a familiar and warm atmosphere; 2) being involved versus not being involved in end of life care; 3) being consoled by witnessing the health professionals' endeavour to relieve suffering. The results showed that family members (most of whom were daughters of the deceased) had an overall positive experience of the care provided, felt involved in the caring process and found that LCP clarified the decision about moving to end of life care, and gave structure to what was happening. Subsequent to the study the authors note (and in the post-Neuberger context), a new care pathway was developed and implemented 'in the earlyandlatephaseofpalliativecareinSweden,notonlyinEoL care'(p200).

Also in Sweden, Olsson et al. xvii focussed on residential care homes and home care settings and examined the perceptions of nurses on the impact of quality of care resulting from use of the LCP. A total of 142 registered nurses and assistant nurses working in a single Municipality in mid-Sweden, where the LCP had been introduced in the period October 2011 -December 2013 and completed a structured study-specific questionnaire containing 50 items. Implementation of the LCP was considered to ensure systematic assessment and alleviation of patients' symptoms and needs, though assistant nurses were more positive in their views. Both groups considered that communication with patients and families and the information exchange between team members was facilitated. The areas for improvement concerned psychological and existential support, as well as patient and families' participation in care. Publishing their results sometime after the widespread with-drawalofLCP,theauthorsnotethecreationofanewpathwayin Sweden, highlight the importance of education of staff if pathways are to be more than a 'tick box exercise' (p.1596), and in particular draw attention to the complexities involved in diagnosing dying, the need for flexibility and the importance of continuallyassessingthestatusofthepatient.

We identified just one item (a conference poster) on the use of the LCP in Denmark, presented in June 2016, in Australia xviii . Here the intervention was described as the 'Danish modified edition' (mLCP) and was presented as 'a tool to provide relief of bothersome symptoms in imminently dying hospitalized cancer patients'. The primary endpoints of the study were relief of symptoms, and correlation between symptoms and dyingpatients-howgoodarewe?SupportCareCancer24(Suppl1):S1-S249 S159.https://link.springer.com/content/pdf/10.1007%2Fs00520-016-3209-z.pdf use of medication. The chosen method was an audit of mLCP records from 45 cancer patients who died in the integrated Palliative Care Unit, of the Oncology Department of Roskilde University Hospital, Denmark in 2014 -so some time after LCP withdrawal in the UK. In total 77% of the dying patients experienced good and immediate symptom relief, whereas 14% presented partly refractory but manageable symptoms. The most common symptoms were pain (56%) and anxiety (42%). Time on the mLCP was on average 48 hours. There was no correlation between presented symptoms and time spent on mLCP, nor was there a correlation between presented symptoms during the first four hours and during the last four hours before death. There was a significant relief of pain using syringe drivers, and a significant correlation between the use of analgesics and symptom relief. The authors concluded that integrated care pathways for best care of the dying person couldbeavaluabletoolforprovidinggoodsymptomrelief.

Kaori Ichihara, a doctoral researcher at the Medical School of Osaka University, reported evidence of the use of what had come to be known as LCP-J (see below in the Commentaries and other publications section) in two Japanese palliative care units xix . Ichihara's article illustrates a whole process of LCP-J implementation, including training for practitioners, collecting information from patients and operationalising the pathway. Forty nurses were involved in the experiment and more than half considered the LCP-J useful, believing that it could contribute to providing standardised criteria for multi-disciplinary healthcare teams, as well as developing consistent and continuing care and support for patients and their families. Educational outcomesforhealthcareprofessionalswerealsohighlypraised.

Kanno et al. (2015) xx report on a study to examine the burden of LCP-J when introduced onto two wards (oncology and respiratory medicine) in Tohoku University Hospital. Making use of audit data and interviews with two doctors and eight nurses, the study found that in a series of 22 patients placed on the pathway (38% of the total study group), there were no significantdifferencesinthemedicationsusedinthetwogroups,but benefits were seen in a more structured approach to preparation for and care in the dying phase. At the same time LCP-J was felt to increase the burden on professionals in relation to the task of diagnosing the dying phase and the need for associated training. The authors concluded that the requisite support and training should come from the hospital palliative care team.

In Italy, the LCP studies were mainly conducted by a team of researchers and clinicians associated with a regional palliative care network of the National Cancer Research Institute and located in Genoa, Italy. The team was led by Dr Massimo Costantini in a three-year programme of research funded by the Italian Minister of Health from 2006 to determine whether the implementation of LCP in a hospital setting could be effective in improving end of life care for cancer patients. To this end, LCP version 11 was first translated into Italian for hospital usein2007andforhospiceusein2009.

An initial paper by Costantini, Beccaro and di Leo (2011) xxi draws attention to the issue of improving end of life care for patientsdyinginhospital,andtheirfamilies.Itreferstocontinuous quality improvement programmes as a vehicle for achieving this, highlighting the development of care pathways in this context, and singling out LCP as 'the most structured and proficient' (p.229) example of its type. Noting its use in over 20 countries, the authors also acknowledge that evidence for its effectiveness remains weak. Crucially, and in contrast to all previous efforts in this regard, they adopt the MRC Framework for the Evaluation of Complex Interventions 39 as the organising principle for their endeavours, in a focus on cancer patients dyinginthehospital.

The researchers recognised that the LCP is a typical complex intervention, involving multiple components interacting with each other as well as with the local implementation setting. Accordingly, they first conducted a literature review to inform the development of the Italian approach, 'Un percoso integrato per le cure di fine vita in ospedale'. They named their intervention LCP-I Program, and it contained 10 constituent steps that followed the continuous quality improvement programme recommended by the LCP Central team in Liverpool. Leo et al. xxii describe the process. After obtaining consent from the institution, steps 1-3 ('Development') involved evaluating the context, the development of the documentation and a retrospective evaluation of end of life care on the ward. As reported in other studies elsewhere, the translation followed EORTC guidelines. Steps 4-8 ('Implementation') involved a detailed programme of intensive training, the introduction of LCP with added support and coaching, evaluation of the education programme, and establishment of LCP as an indicator of quality for all dying patients. Through steps 9 and 10 ('Sustainability') the intervention was endorsed by the quality improvement programme and discussions were initiated about its wider use regionallyandnationally.

LCP-I was led by an experienced palliative care team comprising two physicians, three nurses and two psychologists. This contrasts with the use of 'facilitators' in the UK. The Italian group placed a very strong emphasis on training, with a mandatory programme of 12 hours for ward staff prior to implementation. The setting was three medical wards (72 beds) and one respiratory disease ward (24 beds) of the Villa Scassi Hospital, in Genoa during 2007. In total over 400 deaths occurred on the four wards each year, about half from cancer.Theimplementationwasassessedusingamixedmethods approach. Pre-and post-implementation focus groups conducted with doctors and nurses showed perceived benefits, particularly in pain management and in communication with families. Process and outcome measures were tested using more quantitative approaches. The results were encouraging and began to demonstrate the possibility of undertaking studies aimed at assessing complexinterventionsinendoflifecare.

Di Leo and colleagues describe the carefully designed staff focus groups held at the Villa Scassi Hospital. Two focus groups with nurses and physicians were conducted both at step 4 and at step 8 and were moderated by a psychologist. A researcher took field notes at each session and audio recordings were made. Although the groups were less well attended than planned, the results indicated that the LCP-I Program may have improved the subjective perception of participants' knowledge on the management of physical symptoms, awareness of emotional problems and information needs in end of life care, and matters of communication between medical and nursing staff. Nurses perceived some resistance to change on the part of medical staff as a barrier to the introduction of LCP-I. All participants became more aware of their own limitations in communicating with patients and families, in ways that heightened their own uncertainties -an outcome seen by the authors as potentially positive and as a first step towards improvement. Overall, there was significant agreement that the implementation of LCP-I had improved and intensified communication between the medical and nursing staff on the ward. The researchers were reassured that, although undertaken in a different culture to the one in which LCP was first developed, LCP-I appeared to be acceptable and valid in the Italian context.

The Genoa team then engaged in careful methodological testing of how the intervention could be evaluated xxiii . This involved an uncontrolled before/after intervention trial within the four hospital wards at the Scassi Hospital and included all patients age 18 and over who died of cancer on the medical wards in the four months before and after the introduction of LCP-I (two months before and after on the respiratory ward). The 'intervention' group included those patients who had been assigned to LCP-I, as well as those that had not (though none of the papers give details of the numbers in each category or distinguish between them in the analysis).This was described as an 'intention to treat' methodology, which means all patients are included and are analysed in the groups to which they were located. The researchers used a quasi-experimental before/after design characterised by two measurement points, one before and one after the intervention, and without any external controlgroup.Theyacknowledgedthatthisdesignhasthepotential to exaggerate the effects of the intervention. The researchers contacted the informal care giver most closely involved with the patient, two months after the death, and invited them to take part in an interview. Quality of care was assessed using a Toolkit of items developed in the USA by Teno et al. 40 and translated into Italian and which measured the extent to which care at the end of life met the expectations and needs of the dying person and the family members; deriving a total score from 0 (poorest) to 100 (excellent). Some items from the Italian version of the post-bereavement survey, first developed intheUKandknownas'VOICES' 41 ,werealsoused.

patients and 33 (69%) of the post-intervention patients. The secondgroupshowedhigherscoresonfouroutofsevendimensions of the Toolkit, relating to: respect, kindness and dignity; family emotional support; family self-efficacy; and co-ordination ofcare.Therewasnoimprovementinscoresrelatingtosymptom control.

The group also explored the potential for LCP-I outside the hospital, in the context of the Italian in-patient hospice xxv . Now an adapted version of the LCP-I, with approval from LCP Central, was introduced into seven hospices from three Italian regions, where the proportion of patients who died on it ranged from 36-89%. The interpretation by staff of its value varied widely across the seven settings: two hospices reported a positive impact, two took the opposite view; in three others, opinions were mixed. There was an overall concern about the lack of knowledge to underpin the use of LCP in the hospice and also about the methods of implementation that hadbeenused.

There was also a comparative assessment of the feasibility of the pre-and post-implementation research design in the two settings of hospice and hospital, reported in a methodological article xxvi . Here the primary aim was to evaluate the feasibility of using a combination of assessment methods, directed at different respondents, to create a measure of the quality of end of life care. The two cluster trials took place in eight hospitals and five hospices. Only cancer patients were included in the analysis. Overall, the method seemed to work effectively in both settings, with high levels of compliance and adherence to the study instruments. But the main reservation related to the use of proxies (rather than patients) as the main data source, 'withallthatthisentails'(p.6).

Another study examined the views of staff involved in the hospital implementation of the LCP-I, and who had shown reservations about it xxvii . It was claimed as the first of its type. Six nurses and five physicians from six out of eight hospital wards that had used the LCP-I were interviewed. The authors themselves confirmed that the eligibility criteria for inclusion in the study were subjective and not insufficiently detailed. They identified 'real' concerns with the pathway but were also said to have identified mistaken interpretations of LCP-I among the respondents. Conducted before the Neuberger review had been reported, the authors took the view that their results were neverthelesssimilartoconcernsraisedbyNeuberger.

Verhofstede and colleagues xxviii broke new ground by addressing the effects of the LCP in older patients and at the same time sought to develop a new programme of care for those dying in acute geriatric wards in the Belgian context. Their work began before the publication of the Neuberger review of the LCP, but in some ways foreshadowed its outcomes. They started by conducting an overview of LCP programmes in the UK, the Netherlands and Italy. This led to the identification of threecommonelements:theLCPdocument(andheretheauthors identify the substantive changes made in the translation process from the relevant English language version into Dutch); the supporting documentation; and the implementation guide. Then a literature review of successful LCP implementation strategies revealed five key factors: the importance of a dedicated training facilitator; the provision of initial and ongoing training; the organization of an audit and feedback opportunities; a central co-ordinating office to support local LCP facilitators; funding and available staff time. This led to an analysis of the concerns raised about the use of the LCP in the UK, specifically 1) improper or poor implementation leading to inadequate care, 2) unacceptable communication with patients and carers, 3) the 'tick box' orientation, and 4) the use of the term 'pathway'. The understanding developed from these actions was then used to model a care programme for the last days of life among older hospital patients. This comprised a care guide, supporting documentation, and an implementation guide. The care guide involved translating LCP Version 12 into Flemish and then comparing it with the Dutch translation. 'Liverpool Care Pathway' became 'Care Guide for the Last Days of Life', thereby dropping the protocol orientation of 'pathway'. Adaptations were made to the care goals, with older hospital patients as the focus. In addition, the Care Guide was made shorter than the LCP. Colour highlights were also introduced to improve readabilityofthecaregoals.

The authors located these elements of activity in stages 0-1 of the MRC Framework for the design and implementation of a complex intervention 43 . The paper concluded with the intention to proceed to phase 2 -to evaluate the feasibility of the implementation. As described, it was a process strikingly different to those which shaped the original development of theLCPintheUK.

We identified just seven studies of this type, from just three countries: three each from Italy and Belgium, and one from Sweden. The earlier work in Italy had provided enough evidence to justify the use of a randomised trial to evaluate LCP-I effectiveness. The team had then published a protocol for their new design xxix . They argued that the 'only feasible method' to adopt was a cluster trial, where hospital wards are randomised to receive (or not receive) the intervention. This constituted a Phase III trial within the MRC Framework. In a departure from their previously highly localised work, they proposed that the intervention and control wards should be made up of 'pairs' from participating hospitals, which would be drawn from regions across Italy. The chosen patient group would remain thosewithadiagnosisofcancer.Inclusioncriteriawerecarefully defined, the primary end points and sample size were described, and a detailed account was given of the administrative and implementation arrangements. The study was approved by the National Cancer Research Institute of Genoa in September 2009.

The key paper, published in The Lancet on 18 January 2014, wasthatwhichpresentedtheresultsoftheRCTinthe16hospital wards, involving 147 patients who had been cared for on the pathway and 161 from control wards, who had received 'standard care'; all had died from cancer xxx . The results showed no differences between the intervention and control groups in relationtotheoverallqualityofcare(theprimaryendpoint).But two of nine secondary outcomes reported by family members showed better results in the intervention group -improvements inrespect,dignityandkindnessandinthecontrolofbreathlessness. The authors acknowledged that the study was under-poweredthey had enrolled just 80% of the planned hospitals and slightly over-estimated the expected number of deaths. There were no differences in the medicines prescribed to the two groups, in the survival time after admission to hospital, or in the managementofsymptomslikepain,nauseaandvomiting.Ingeneral,the beneficial effects were lower than in the phase 2 trials in the Netherlands and in Italy. But the die was cast. Neuberger had reported the previous summer and the verdict was that LCP shouldbediscontinued xxxi .

As in Italy, the work in Belgium included the development and publication of a protocol for an intervention study xxxii . The design was again that of a cluster randomised control trial, to be conducted in 10 hospitals in the Flanders region, each with one or more acute geriatric units. In year one a baseline assessment would take place of usual care, based on completion of a questionnaire by relevant physicians, nurses and family members concerning each patient who died on the unit. Thereafter the hospitals would be randomised to receive the intervention, or not. This cluster randomisation, operationalised at the ward level, was thought likely to reduce the potential for contamination, since the comparisons would be between hospitals with and without the intervention. Subsequently in the intervention hospitals, the new Care Programme for the Last DaysofLifewouldbeimplementedoveraperiodofsixmonths. A post-intervention assessment would be performed immediately after the baseline assessment in the control hospitals and after the implementation period in the intervention hospitals. The primary outcomes to be measured were symptom burdenandfrequencyamongpatientsinthelast48hoursoflife.

A process evaluation was also proposed to assess the quality of the implementation of the new programme, to which the Belgian researchers then turned their attention xxxiii . Their aim was (1) to determine the feasibility of implementing the Care Programme for the Last Days of Life in the acute geriatric hospital setting, and (2) to explore health care professionals' perceptions of the effects of the Care Programme on end-of-life care. They undertook a phase 2 mixed methods study, according with the MRC framework, in the acute geriatric ward of Ghent University Hospital between 1 April and 30 September 2013. The approach included observation, interviews and the use of a quantitative tool, which measured the success of implementation using several indicators, such as whether a steering group was formed, whether and how many of the health care staff were informed and trained, and how many patients were cared for according to the Care Guide for the Last Days of Life. The process evaluation tool showed that implementing the Care Programme for the Last Days of Life in the geriatric ward was successful and thus feasible; a steering group was formed consisting of two facilitators, health care staff of the geriatric ward were trained in using the Care Guide for the Last Days of Life, which was subsequently introduced onto the ward and approximately 57% of all dying patients were cared for in accordance with it. Nurses and physicians experienced the Guide as improving the overall documentation of care, improving communication among health care staff and between health care staff and patient/family, andimprovingthequalityofend-of-lifecare.Barrierstosuccessful implementation of the Care Programme included difficulties with the content of the documents used within the Programme and the low participation rate of physicians in the training sessions and audits. The results were encouraging and suggested feasibility and favourable effects. barriers during the implementation process, the authors were able to make recommendations for future implementation and to further refine the Programme before its deployment in a phase 3 cluster randomized controlled trial for the evaluation of itseffectiveness.

Again echoing the Italian study and with Costantini as a co-author, the results of the trial were published in The Lancet in 2017 xxxiv . The authors included online a 22-page supplementary appendix to the published paper. The Care Programme for the Last Days of Life now had an acronym -CAREFul. The Belgian authors noted that in a Cochrane review of end of life pathways, updated in 2015, the Italian study had been judged to be statistically under-powered and at high risk of bias, mainly because patients were not masked to treatment allocation and there were high levels of attrition. The Belgian study avoided some of these weaknesses, though again the nurses knew about and took part in the CAREFUl intervention with the patients for whom they gave ratings. Nevertheless, the results still seemed equivocal. There were 118 patients in the control group and 164 in the CAREFul group who were eligible for assessment. Of these 92% and 80% respectively were assessed by nurses; but only 19% and 29% respectively were assessed by family members.

Nurse assessed scores were significantly increased by CARE-Ful, but there were no significant differences between the intervention and control group in family members' ratings. The study also found a negative effect on satisfaction with care among family members, described as 'a serious concern that needstobeinvestigatedfurther'.Theauthorsconcludedthattheir results 'suggest' implementation of CAREFul -an intervention based on the LCP -'might' improve care during the last days of life for patients in acute geriatric wards. In an accompanying editorial, Aslakson and Lorenz 44 praise the ambition of the study, but home in on the involvement of nurses as both the unmasked study interventionists and the study assessors, creating a high risk of unconscious bias leading to more favourable reported outcomes. They conclude that the results of the CAREFul study are welcome 'but critical questions remain unanswered about the study itself and the contextual and implementation issues that the experience with LCP revealed' (p98).

A randomised study of an end of life pathway had therefore been conducted in both Italy and Belgium, and published in a worldclassmedicaljournal,butinbothinstancestheresultswere equivocal.

Researchers in Sweden recognised the lack of controlled studies of LCP outside of cancer settings and focussed on its use in care homes and in residential care homes xxxv . In Skellefteå municipality, they assessed the effects of the LCP on patients' symptom distress and wellbeing, when compared to usual care. The design was an exploratory, controlled before and after study. During a 15 month baseline period (June 2009 -August 2010), usual care was assessed in two areas of the municipality, containing 10 and 9 care homes, respectively. In the following 14 months, staff introduced LCP in one area and usual care continued in the other.A translation of LCPVersion 11 was used, in collaboration with the Swedish LCP co-ordination centre based at Stockholm Sjukhem, and the process included a structured education programme to support the implementation. In both areas in the pre-and post-intervention periods, care was assessed through the completion by relatives after the resident had died of two structured instruments (Edmonton Symptom Assessment System and Views of Informal Carers -Evaluation of Services, known as VOICES),sentbypost.Thestudycoveredresidents(referredto by the authors as 'patients') who died in all 19 residential care homesinoneSwedishmunicipality.

A total of 837 patients died during the whole study period. The LCP was completed for 132 (60%) of those who died in the intervention area; two thirds of the remainder died suddenly in the care home or died in hospital. Cardiovascular disease and dementia were the main causes of death. The total response rate for completion of the questionnaire was 46.2%, there was a significantly higher response rate in the intervention area. The main result was a reduction in two symptoms among residents in the intervention group -shortness of breath and nausea. Reductions in both of these symptoms were reported by the symptom assessment measure and for shortness for breath only by the carer evaluation tool. On such a basis, and following a detailed reflection on the practical and technical limitations and potential biases of the study, the authors felt able to conclude that the LCP may be a useful tool for providing end of life care for elderly people in non-cancer settings. Interestingly they note that, based on the positive experience of conducting the study (as much as its outcome), the municipality involved made a policy decision to implement the use of LCP in all its residentialcarehomes.

Beyond these audit, implementation and controlled studies we identified 20 commentaries of various kinds and also nine miscellaneous pieces that could not easily be classified -almost a third of the total number of publications on the use oftheLCPoutsidetheUK. processes being adopted, including the translation-back translation approach xxxvi-xxxviii . By 2012, Geijteman, Dekkers and Zuylen xxxix could observe that 119 institutions, including 35 nursing homes were working with the LCP in the Netherlands and that a digital version of the LCP was under construction, which would make it possible to integrate it in electronic patient files, as well as serving the purposes of quality assurance andscientificresearch.

A paper by Raijmakers et al. xl acknowledges the withdrawal of the LCP in England and the rest of the UK and also gives a detailed account of the wider 'roll out' of LCP in the Netherlands. This was built on the perceived positive research results in the pre and post implementation Dutch studies. Critical to the scaling up was endorsement of the LCP by the Comprehensive Cancer Centre of the Netherlands (CCCN). This led to 'roll out' in 66 regional palliative care networks -groups covering specific geographical areas and committed to 'intensive collaboration and synchronisation' (p.260). This had involved advocacy for the LCP, training programmes of two days'durationforprojectleaders,andsupportonimplementation throughahelpdesk.

Raijmakers et al. also note that implementation research to bridge the gap between evidence and practice in palliative care is not widespread. They decided to address this through a study of LCP implementation by interviewing CCCN consultants involved in LCP training in each of its eight regions. These people were asked to nominate organisational examples of successful implementation, 10 of these led to follow on interviews with project leaders from 25 examples given. Perspectives from one palliative care network from each of the CCCN regions were also obtained and all interviewees in the study were invited to take part in a summative focus group by way ofconclusion.

No studies of LCP or similar pathway implementation were identified for Ireland, but one case note review, given ethics approval by the Beaumont Hospital, Dublin, focussed on dementia patients during acute hospital admission, and used LCP as a standards benchmark for quality palliative care in a comparison with patients who did not have dementia xli . Data were collected between January and June 2008 on a total of 50 patients, a condensed version of the LCP was used to focus on nine aspects of good quality care -rationalization of medicine, discontinuation of unnecessary invasive interventions, prescription of palliative drugs, referral to specialist palliative care, documentation of resuscitation decisions, communication with primary care, supporting caregivers in bereavement, and assessment of religious and spiritual needs. The researchers found that whilst both groups of patients had similar levels of invasive treatment, those with dementia were less likely to be referred to palliative care, were prescribed fewer palliative medications, and had less involvement of relatives in dialogue about treatment. The authors drew no conclusion about the value of LCP implementation, in a study where it was used simply as partoftheresearchdesign.

As late as 2012, a reflective piece by a Swiss internist on experiences of palliative care concluded that, among a list of shortcomings, there was still a lack of standardization across settings of care in the final phase of terminal care, and used comparison with the guidelines contained in the LCP as an indicationofthis xlii .

Around the time of publication of the Oldenburg study, in 2009, the German journalist Florian Rötzer, wrote a critical article for Heise Online about the LCP, entitled 'Sentenced to death by palliative care? Physicians warn of a guideline for the care of dying patients' xliii . Rötzer could see the value of clinical guidelines for end of life care.These would avoid acting arbitrarily or criminally, provide security to medical staff and advice on what to look for while helping the dying to leave life in a dignified and pain-free way. But taking his cue from controversies emerging in the UK, he argued that the LCP also has the potential to be dangerous in a context where the matter of diagnosing dying can be imprecise. This could mean nothing short of the potential for a 'national crisis' in which relatives witness the discontinuation of treatment, apparently driven by economic expediency. The remarks neatly anticipated the subsequent eventsanddiscussionsinEngland.

It was then four years before further work was published on the use of the LCP in Germany, in a textbook chapter byVoltz, Nübling and Lorenzl xliv on care of the dying for neurologic patients, which set out the rationale, development and perceived benefits of LCP and asserted that whilst LCP had been withdrawn that year in the UK, it remained a good model for structuredendoflifecare.

Meanwhile, commentary on LCP emerged from Austria in 2013 in the form of a published undergraduate dissertation in advanced nursing practice, by Zinner xlv . Her focus was on how the LCP could be implemented in hospitals in the German-speaking area and how the quality of life of the dying can be preserved or improved by using the LCP. Zinner identified a nursing textbook chapter on the LCP by Müller xlvi that describes LCP version 11 as a well-functioning, multidisciplinary procedure and a very useful tool in caring for the dying. Addressing repeatedly voiced fears that guidelines for care at the end of life run the risk of 'standardising' dying, Müller asserts that any guideline or tool can only be as goodasitsusers(Müller,p.102,quotedinZinner2013).

After the withdrawal of the LCP in England, senior clinicians from Australia commented on the implications for the continuing use of end of life pathways in their country (Chan et al., 2014) xlvii . They acknowledged that, across Australia, several end of life care pathways had been adapted from the LCP, though with no precise numbers of the institutions involved. In particular, they highlighted the recommendation for a national roll out of an integrated end of life care pathway through primary, acute and aged care sectors, that had been included in theAustralian National Palliative Care Strategy of 2010. Despite the widespread experimentation with and policy endorsement of the end of life pathway in Australia however, the authors took the view that (as elsewhere) there had been insufficient evaluation of its net effect. The withdrawal of LCP in the UK had created a dilemma for Australia. The authors made a plea for more rigorous, randomised, studies of end of life pathway outcomes and implementation, but acknowledged that this would take time. Meanwhile they recommended that the shortfalls and adverse effects identified in the Neuberger review should be carefully assessed in the Australian context, and concluded that 'if the LCP is to be replaced, there needs to be systematic measurement of the benefits and harms generatedbysuchaprocess'(p573).

Norway's part in the LCP narrative includes reactions to it which took place after the recommended withdrawal in the UK in 2013, though it was not until 2016 that the commentary emerged in a rapid review produced by the Norwegian Institute of Public Health and requested by the Norwegian Directorate of Health xlviii . The authors explained the goals and background to the LCP and noted its ongoing use in Norway since 2005, though with no details of extent or setting. The aim of the review was to assess the effectiveness of the LCP and also its relevance to patients with dementia. Only two studies met the inclusion criteria (from Sweden, in 2014 and Italy, in 2015). The results (as we have seen above) showed small or no differences between LCP and standard palliative care at the end of life. The available evidence indicated that LCP possibly can improve breathlessness in dying patients, however there was no evidence that LCP was superior to standard palliative care with respect to quality of care, management of most of the dying patients' physical symptoms, and prescription of palliative medications. At the same time the authors had very low confidence in the evidence they reviewed, mainly due to a high risk of bias in the studies, which also enrolled few participants and contained shortcomings with regard to the implementation of the intervention. For example, in the Italian study, only 34% of cancer patients in the intervention wards received LCP, and in the Swedish study, only 60% of patients in the intervention homes received LCP. The authors therefore noted that the various weaknesses of the available studies limit confidence in the then current evidence and restrictedthepotentialtodrawfirmconclusions.

The following year a systematic review of the use of LCP in nursinghomeswasproducedbyateaminNorway xlix .Itincluded the provocative subtitle 'discarded in cancer patients but good enough for dying nursing home patients?' and described the history and purposes of the LCP, emphasising that the approach was designed for use with cancer patients and 'presumed an open and timely communication between the treating physician, nursing staff, patient and relatives' (p1). It noted that LCP was in use in 17 countries outside the UK, including Norway. Describing the media and public concerns that had arisen about LCP and which led to the Neuberger review, the authors observed that in was therefore to assess the evidence for the use of LCP, but specifically in the nursing home context and in relation to people with dementia. It focussed on how LCP had been validated and tested in these contexts, with which study designs and methods, the implementation strategies in use, and how theyhadbeendescribed,alongwiththemainoutcomes.

Twelve papers, one purely methodological, were selected for inclusion in the Norwegian review, based on nine clinical studies; seven studies were from outside the UK and are discussed elsewhere in the present paper. None of the selected studies used randomized, blinded or prospective designs. Only one was controlled. The reviewers concluded that evidence for the use of LCP in nursing homes was virtually absent and when present was weak and the results not definitive. Acknowledging the problems of randomised designs in these populations, the authors were nevertheless critical of the absence of alternatively designed studies that document the development and testing of the instrument by including elderly, multi-morbid patients and those with dementia. They concluded that the LCP had not been adapted to the individual needs of people in nursing homes and those with dementia. In Norway almost 50% of deaths take place in nursing homes, where 80% of patients have dementia. Husebø, Flo and Engedal close by stating: 'After changing the name of the procedure, the LCP is still in use in many countries, as a low-cost camouflage of the real need for education and competence in nursing homes'(p.12).

The one paper we located on LCP in China l took the form of a professional commentary and looked at the potential of LCP in relation to the development of hospice nursing in mainland China. The article explains how LCP was developed, reviews the research on its implementation and benefits, but also acknowledges the critical commentary on LCP that had occurred in the British press and from some clinical commentators. With reference to the experiences of the LCP in other countries, the authors argue for strategies to improve Chinese nurses' knowledge of palliative care and also for considera-tionofhowtheadvantagesoftheLCPandothermodelsofcare, could be used to develop a more appropriate model of palliative carefortheChinesecontext.

In a 2008 paper on the status of palliative care in Japan, a group of authors li commented that standardised frameworks, guidelines and clinical pathways for end of life care, available in the English-speaking world, had not been developed or were inadequately disseminated in the Japanese context. They indicated that a validation study of several such interventions was underway to modify the originals and make them suitable for Japanese culture. The list included the Liverpool Care Pathway.

Two years later, in 2010, a special issue of the Japanese Journal of Clinical Nursing was published that focussed entirely on the use of LCP in Japan lii . Edited by Mitsunori Miyashita,aprofessorinpalliativenursingatTohokuUniversity, the special issue aimed to provide a comprehensive picture of the introduction and implementation of the LCP in Japan, at that time. The editor introduced the issue with some back-groundonendoflifecareinJapan,addressingtheimportanceof promoting and adapting the LCP to the Japanese context, and pre-viewing the main articles in the special issue liii . This one special issue contributed eight of the 17 articles on LCP in Japan within our review, though most were commentaries of onesortoranother.

Karoi Fukuta from the Seirei Mikatahara Hospice offered an insightintoLCP-Jbyreportinganindividualcaseofaterminally ill cancer patient liv . The implementation is recorded in great detail by showing how the patient and her family were supported at different stages of the dying process. By referring to the criteria of the LCP, the author provides a clear picture of how the patient's symptoms were assessed and how the tech-niquesoftheLCP-Jwereemployedtosupporthervariousneeds and those of family members. The author concludes that the LCP-J improved the quality of palliative care for the patient and herfamily.

The last article in the special issue pays attention to an evidence-based booklet produced for family members of dying patients lv . Hiroyuki Otani, from the Kyushu Cancer Centre, argues that the booklet can help family members to better understand end of life care, reducing anxiety and empowering them to give improved support to their dying family member. The author also points out the benefits of using the booklet withtheLCP.

A short paper by Yoshikazu (2016) care in Japan. It acknowledges the usefulness of the LCP as a guideline for end of life care practices and for the education of medical professionals. It then reviews how the LCP was introduced and developed in Japan from 2003, leading to the production of LCP-J which began in the same year. LCP-J and its manual were completed in 2009, and the home-care version, LCP-H was introduced in 2012. While highlighting the increasing influence of the LCP in Japan, the paper also documents the wider debates in the UK on the effectiveness and appropriateness of the LCP. Given the criticisms and the withdrawal of the LCP in the UK, the author suggests that the promotion of the LCP in Japan should stop and an alternative pathwaythatfitsJapanesecontextsneedstobedeveloped.

In a further paper from 2016, Chinone lvii , a palliative care doctor in charge of developing the Japanese version of LCP-J, also sets out the wider historical background of the LCP, its contents and some examples of its implementation elsewhere, before describing the reasons for introducing the LCP to Japan, why it should be adapted to the Japanese context, and the continuingpossibilitiesforitsimplementation.

Tanaka and Satoshi lviii in 2017 also provide an overview of end of life care in Japan, with a particular focus on developments in the UK, including the LCP, and their relevance in the Japanese context. Reviewing the debates surrounding LCP in the UK and its eventual withdrawal, the authors suggest that end of life care in Japan needs to 1) pay more attention to individualised care and communications, 2) develop education programmes as well as institutional and home-based services, and also 3) enhance the involvement of local authorities and communities forthebenefitofpatientsandcarers.

The Italian researchers Costantini and di Leo lix went on to describe the debate that took place in Italy, following the recommendations of the Neuberger report. They used the term 'malpractice' to describe some of the cases that found their way into the mass media in the UK, described and summarised the Neuberger report process and findings and noted the subsequent creation of the Leadership Alliance for the Care of Dying People in England, its aim of developing policies and processes to ensure high quality, consistent care for people in their last days, and its production of five priorities for care at the end of life. They observe that 'The story of the LCP is also Italian' (p266) and describe its development in the Italian context, as detailed here in our own paper. Early resultsinGenoahadbeenpositiveandsupportedthedesignofa cluster randomized Phase III study, to assess the effectiveness of the programme in improving the quality of care of end of life care for patients dying of cancer in Italian hospitals. Despite the standardized implementation process and carefully planned support from the palliative care teams, the findings of this study were less encouraging than the Phase II studies and didnotreachsignificancefortheprimaryoutcome,butdiddemonstrate some secondary benefits. In light of all these factors, over50cliniciansfromsixItalianregionscametogethertoshare opinions and experiences about LCP-I. A list of 12 strengths and 13 weaknesses of the LCP programme in Italy was generated. There was a strong emphasis on coping with the risks associated with the intrinsic limitations of LCP and its inappropriate dissemination. The group took the decision to 'phase out the LCP in Italy' (p266-67) and noted that the same approach would be taken in all countries where the LCP had been disseminated. But di Leo and Costantini observed '... the problem remains. The quality of care for the dying patients in hospital is suboptimal'. The LCP studies had demonstrated the feasibility of conducting high quality research in this field and thisstrategywouldcontinue.

In the same year, 2014, Costantini, Alquati and di Leo published a review on the evidence for pathways in end of life care lx . They acknowledged that two separate Cochrane reviews, last updated at June 2013, did not find any studies of end of life care pathways that met the inclusion criteria, though their unique Italian study, albeit with its negative results, post-dated this. In a somewhat ambiguous sentence in the abstract, they concluded that 'the overall amount of evidence supporting the dissemination of end-of-life-care pathways is rather poor' (p1741). Two major drawbacks could be seen in the analysis of the quantitative studies on end of life pathways. First, poor internal validity of design: uncontrolled before and after studies have intrinsic weaknesses and are vulnerable to changes in the external environment that might lead to improvement, they are at risk of the Hawthorne effect, regression to the mean and are likely to overestimate the effects of interventions. Second: the piecemeal character of the studies meant they lacked a comprehensive strategy for research, in line with the MRC framework for the evaluation of complex interventions. The pool of studies undertaken in Italy, and proceeding through the three phases of the MRC framework, seems to be the only example of such a comprehensive strategy, though the authors note developments in Belgium and Sweden (at the time unpublished, but described above here) that might merit inclusion in a future Cochrane review of end of life pathways. In noting the outcome of the Neuberger review, the Italian authors also observe that none of the published studies report any adverse effects, relating to patients, family members or involved professionals. They called for more appropriately designed studies of end of life pathways, as well as studies taking account of the LCP 'debate' and the outcomesoftheNeubergerreport.

Conclusions to literature review. We identified 95 publications covering the period 2003-09 and referring to 20 countries in which LCP interest and implementation had been considered ortried.

Just over half (n=11) of the countries identified in the literature review produced three published outputs or less: Slovenia, Ireland, Switzerland, Spain, Singapore, India, Denmark, China, Argentina, Norway. These countries generated 18 outputs in total, making just 19% of the total reviewed. It is difficult to see how most of these countries reveal a high-level commitment to LCP transfer, with the exception of Switzerland, which was part of a three country German language group and Norway, where there was significant in-country take up of LCP. Beyond those, publications from these countries arose mainly from small scale developments, only occasionally based on wider collaborations and which had little impact on wider transfer or diffusion (for example colleagues in Spain working with others inArgentina). However, this is not to undermine the value of this work to the actors involved, who gained local experience of the translation of an intervention into a new context, who sometimes adapted it with imagination and flair to local cultures and healthcare systems, and who used it to auditandbenchmarktheirownendoflifeservices.

The remaining nine countries, just under a half of those identified in the review, produced 4-17 publications each: (Japan, Netherlands, Italy, New Zealand, Australia, Sweden, Belgium,Austria, Germany). This accounted for no less than 77 outputs, making 81% of the total. These were countries where, in some cases, LCP transfer gained significant momentum, sometimes operating at scale and in some instances being subjected to rigorous testing in robust research studies. This in turn produced impactful publications in international journals, greater visibility in professional conferences, and in the case ofJapan,ajournalspecialeditiondevotedtoLCP.

The authors of the papers reviewed were overwhelmingly clinicians, some with a strong research orientation and holding senior academic positions. The leading authors were mainly medical, and several of them took part in our interviews (see below). There were few papers that involved nurses or other health professionals, and almost none that included social scientists or implementation experts. This is reflected inthecharacterofthepublishedoutputs.

The largest single group of outputs, making one third of the total and completed in 15 countries, comprised descriptive audit studies, drawn from clinical records. These had often been undertaken following the guidance received from LCP Central in Liverpool. Some were baseline, pre-implementation assessments, others compared outcomes pre-and post-implementation.

After these, comprising 28% of the total and from 10 countries, were mixed methods studies that examined acceptability and effectiveness of LCP, drawing on perspectives from process orimplementationanalysis.

Just three countries, making up only 7% of the outputs, producedcontrolledstudies,ofwhichonlytwowereRCTs.

The remaining one fifth of the outputs comprised commentaries or descriptive pieces of some kind, all without a research design.

The nine countries with the most outputs, were also those that produced the most robust research results, albeit with only three of these conducting the sort of controlled studies that would stand up to critical review and be published in leading journals. The 95 outputs we reviewed were therefore long on commentary and the sharing of experience, or focussed mainly on descriptive audit methodologies, sometimes with added processmeasurestoshedlightonacceptabilityandfeasibility.

Only a tiny minority used rigorous designs, albeit with some flaws, and equivocal results. In Sweden the key study had modestresults-areductionintwosymptomsintheintervention group -shortness of breath and nausea. In Italy, in an underpowered RCT, there were no overall differences in quality of care between the intervention and control group, though the latter, as described by family members, showed better results in the intervention group on two counts: improvements in respect, dignity and kindness and in the control of breathlessness. In the key Belgian study, there were generally no significant differences between the intervention and control group in family members' ratings but a negative effect on satisfaction with care among family members was observed in the interventiongroup. Table 2 lists the 19 interviews we conducted with 20 people from 14 countries. One interviewee subsequently withdrew from the study and is not included in the analysis. Six countries with some measure of LCP activity identified in the literature review were not included in the interviews. In Spain, Ireland and Hong Kong those we approached declined to participate. In Slovenia our desired interviewee was unavailable for interview. In China and Singapore, we failed to track down potential interviewees who could be approached. All interviewees agreed to be interviewed 'on the record' after signing the informed consent form and approving the full analysis of the interviews (Extended data 28 ) for this part of the study; they are thereby identifiable in our reporting. Interviews ranged in length from 36 to 66 minutes. Only three interview participants (Boughey, Douglas, Medicus) were not among the authorshipofpapersweidentifiedfortheliteraturereview.

We set out six dimensions that resulted from our analysis of the interviews: 1) context and motivation for engaging with the LCP; 2) translating and adapting the LCP for a new context; 3) deployment and diffusion of the LCP within countries; 4) perceived benefits of the LCP; 5) challenges and drawbacks associated with the LCP; and 6) perspectives on the withdrawal ofLCPintheUKanditsconsequences. 

India (1 interview A similar issue occurred in Japan, where for cultural and linguistic reasons the term 'Mitori' -which has broad resonance inJapanandevokesthenotionofbeingwithadyingperson-was used instead of a literal translation of the word 'dying'. Translation and adaptation was further complicated in Japan because the flow chart for pain management in the UK version of LCP was considered unsuitable for Japanese practice. Accordingly, in Japan an emphasis was placed on the concept of the LCP, rather than on the detail. The solution was to encourage clinicians working in different areas to develop their own flow chartbasedontheLCPexample: …I'd say it was difficult to agree with the translation, especially on this algorithm … (about) the usage of medicine. But at that time we did not have clinical guidelines of pain management (and) … they could not agree with this flow chart … After I became principal investigator, we didn't emphasise the flow chart. (We said) this is simply just an example.

(ProfessorMitsunoriMiyashita,Interview10)

In the Netherlands and in New Zealand careful attention was paid to the suitability of all the goals of care in the UK version of the LCP, and some revisions were made for cultural reasons.

In the Netherlands this became necessary after the changes made in the UK version LCP 12 introduced what was perceived as an unnecessary focus on clinically assisted hydration and nutrition:

We had a second version and that's based on version number 12 in the UK. But the difficulty between version number 11 and version number 12 is that therewasalreadyalotofproblemsintheUKandtherefore there were two new goals about fluid and feeding. It is a discussion we can't understand ... really understand in the Netherlands. … so we didn't make it two different goals … we put it together with our judgement about, for example, oxygen and antibiotics, so it's part of another goal and not one itself. (Dr Lia van Zuylen,Interview9)

In New Zealand the view was that cultural and spiritual care goalsshouldbeseparated:

In consultation with the Liverpool team, we had another goal added to the list of goals of care on the LCP and that was around cultural support and cultural care. Here in New Zealand we felt that putting spiritual care and cultural care together wouldn't be appropriate … There were some language differences ... and that did have an impact on how we taught people to use the documentation. So it did have some limitations around some of the language. But the [Liverpool] team allowed us to modify some language, but not all language and that did cause some confusion at times, I think. (Dr Simon Allan and Bridget Marshall,Interview2,BMspeaking)

In Italy and Belgium, a very different set of circumstances prevailed. The research-based approach taken there to the development, adaptation and trialling of a complex intervention based on the LCP meant that considerable changes and adaptations were made to the original LCP programme, and thesewentfarbeyondastricttranslationofthecoredocument.

Dr Costantini in Italy recalled that while translation of the paperwork was relatively straightforward, the development of a detailed implementation manual -which he saw both as essential and lacking in the original UK model -was more complex. He describes how such a manual was developed based on the core principle that the implementation of the LCPinItalymustbeledbyaspecialistpalliativecareteam:

We started with the idea of implementing the LCP as part of a research framework, as part of a research project. The first thing we did was the translation of the document, it was required by the international LCP group. They revised our translation and they accepted our adaptations. To be honest it wasn't completely different from the original. In my opinion a care pathway is not just the document but also the way to implement the document. We didn't change the document a lot just small adaptations, which I have put in context, but we probably changed a lot [in] the way we implemented the LCP. First we wrote a manual for its implementation. We did all these things before receiving the answer from the Ministry of Health. The manual for implementation was based on the idea that the LCP had to be implemented by a specialised palliative care team. The palliative care team was responsible for the process of implementation and the appropriateness of the procedures, of training about end-of-life of care and correct application of the LCP. So in our approach the specialised palliative care team was a necessary condition for doing that. (Dr Massimo Costantini, Interview19)

This approach was mirrored in Belgium, where the research team closely collaborated with the Italians, as well as the UK and Dutch teams, in the development of an intervention for use in geriatric wards. As in Italy, the main challenge was seen asthedevelopmentofaguideforimplementation: So we started by talking to people in the Netherlands and also to Massimo Costantini in Italy, we knew he was doing a big trial in hospital with cancer patients so we went to Italy to talk to him and we went to the Netherlands to talk to those people and we also went to Professor Ellershaw's group in the U.K. We started with the various documents and the programme and thought, well, what can we use, what is suitable, do we need to adapt this and how can we make this work in Belgium? I think the main changes were in wording, but the main challenge was working out step-by-step the implementation guide. For the document itself, we did not make very many changes. We started from the LCP process ('Zorgpad voor de Stervenfase-RotterdamZS-r(lcp)') used in the Netherlands, and we mainly adapted the language and added some extra things for older people and some caregoals.(DrTinneSmetsandDrKimBeernaert,15, part1,TSspeaking)

In all countries except Belgium, deployment of the LCP was initiated by specialists in palliative care. However, there was considerable variation according to where the LCP was first used, and whether this was solely in a specialist palliative care context, a generalist care context, or in both. Similarly, there was variability in the patient population targeted: in some countries (for example the Netherlands) the primary target was oncology, in others (such as Belgium) the target extended into broader categories of patients approaching the end of life, such as older people with end stage frailty or dementia. There was also variation according to whether the use of the LCP remained confined to one or two small local areas, or was diffused regionally or nationally, and the extent to which this was organicandunplanned,orstrategicandsystematic.

In this section we look at interviewees' accounts of the care setting they targeted when they first starting using the LCP and the associated patient groups with which they sought to employ the LCP. We also examine their accounts of diffusion within the countries where LCP was deployed. Insights into the factors and mechanisms influencing the different levels and types of diffusion emerge from these accounts and are summarised in Table 3 They include the presence or absence of: some form of nationalised health care system, a national policy for palliative care into which the LCP (or a version thereof) could be inserted, and the extent to which there was some level of integration of palliative care services into mainstream health care; funding for a programme of research on the LCP or its implementation; and a wider quality control or governance structureontowhichtheLCPcouldbegrafted.

In those countries where there was no 'national' or centralised mechanism for the uptake of health care interventions, including elements of LCP in professionally endorsed guidelines for palliative care was seen as the most effective way to encourage its use. In these cases, not only were numerous versions of the LCP forthcoming, contrary to the UK originators' intent that LCP should retain a standardised format, but the 'spread' of the LCP was relatively organic in form. Moreover, in these cases, LCP uptake was dependent upon clinicians to make an active choice to 'opt in'. There was no element of compulsion, as was the case in those places where its use was mandatedunderanationalstrategy.

At one end of the continuum in terms of target of use and level of diffusion were Denmark and Austria, where in each case use of the LCP was narrowly confined to the initiator's own specialistpalliativecareunit.

In Denmark, Dr Otteson described local implementation of the LCP, focused first in his own specialist palliative care unit, where it became a standard protocol for the care of the dying Dr Medicus went on to describe how she tried to collaborate with staff in an Austrian nursing home once she became aware that they were using the LCP for residents with complex long-term conditions, but the link did not develop and she was unawareoftheextentofanywideruseinAustria. Limitedregionaldiffusion Limited regional diffusion was described by interviewees fromIndia,ArgentinaandJapan.

India and Argentina were the only low and middle-income countries represented in our study and in each case interviewees described attempts to introduce the LCP into both hospices and general palliative care settings (with an emphasis on oncology patients), but in contexts where these efforts were geographically limited. The introduction of the LCP to India in 2006 was described as an initiative prompted by a UK doctor who had an elective in India, and as a collaboration between the Institute of Palliative Medicine (IPM) in Calicut In Argentina, a dearth of funding and of a wider infrastructure for palliative care similarly circumscribed efforts to 'spread' the PAMPA programme beyond its original site in Buenos Aires. This was despite considerable efforts to engage practitioners from all over Argentina and from other countries in South America, through education and training about endoflifecareandtheuseoftheLCP: So we have professionals, students, professionals from Argentina, the different provinces and also other countries from South America. And when we teach around this programme they like to implement it in their institutions. They like it very much but don't have the structure, or the political decision or the way to implement these kinds of programmes. Because again, the palliative care … probably they don't have a team, multidisciplinary team or it's just a doctor and a nurse for a whole province. So it's difficult to think at this time to implement this kind of programme … Wehadatthetimetheillusionthatitwouldbeamore spread programme. But till now we have one more hospital, one more home based care team or programme, and a hospice.The last to join is a hospice, a little hospice. But we have now five institutions involved in the programme. (Dr Vilma Tripodoro, Interview3) Dr Tripodoro's colleague Dr Gustavo de Simone also gave an account of the process of spread in Argentina, describing how it was linked to the education of physicians and referring to a plan to expand the implementation of the LCP to Patagonia, though at the time of the interview this had not yet occurred.

In Japan, limited regional use of the LCP was also reported. One of our two Japanese interviewees, Professor Miyashita, referred to a pilot study carried out following earlier piloting of the LCP in two specialist palliative care units, and which drew attention to a number of problems associated with use of the LCP in general oncology wards. He reported that the 'explanation and training were insufficient' -medical staff did not wish to use it, nurses were worried about recognising whether someone was dying and there was a lack of necessary resource to support staff and sustain systematic implementation. The work was abandoned and modifications were considered, but then the LCP was withdrawn in the UK and everything cametoahalt.

Professor Miyashita estimated that around 20-30 individuals from 'maybe four or five hospitals' had used the LCP at some point in Japan, although it was difficult for him to be sure of the extent of wider use. There was also a circumscribed attempt to develop a home care version of the LCP. This was described by Dr Ai Oishi, a Japanese general practitioner who collaborated with Professor Miyashita. Dr Oishi had spent time in the UK and, on her return to Japan, found information about the Japanese version of the LCP on the internet, prompting contact with Professor Miyashita to request a copy of the document. Dr Oishi's intention to adopt the LCP in general practice in Japan came to partial fruition, as she gained experience of itsuseinthecourseofherhomecaretraining: Well the challenge was to implement it with nurses and then other professionals [but] without support from other people … I knew I couldn't implement it.

(DrAiOishi,Interview11)

Professor Miyashita (LCP 10) described how he was asked by a member of the Japanese government about the potential use of the LCP nationally to improve end of life care but had advised against its incorporation into government policy because, in his view 'it's immature … You cannot incorporate it in national governmental policy'. (Professor Mitsunori Miyashita,Interview10).

Extensiveregionaldiffusion More extensive regional diffusion was reported from Switzerland,Germany,Belgium,andItaly.InSwitzerland,extensive regional spread followed the work of Professor Steffen Eychmüller (Interview 17) of the Cantonal Hospital of St Gallen and founder of the DACH collaborative, who recalled his intent from the outset was to use the LCP to improve the care of the dying in general palliative care contexts. The translation of the LCP, and the development of an associated training programme, proceeded with this goal in mind. An opportunity then came to establish a quality improvement initiative in oncology, in whichcareofthedyingbecameoneofthestandards: I used to work in St. Gallen in the beginning and we decided to become like a collaborative centre for the German speaking regions. We translated everything, the whole document and we also established the training programme in German for health professionals. We had also, in early 2003 I think we started, we got the opportunity to establish in the whole hospital a programme for oncology improvement and one of the seven standards became care for the dying. So this was, for us, something like a lap to establish and to test the dissemination programme with a tertiary hospital ... and also to use the quality management circle that has been established in many surroundings in our German speaking world as a vehicle to improve care for the dying … if you combine quality management, quality improvements in hospitals together with such a topic, it works. (Professor SteffenEychmüller,Interview17)

The subsequent broader regional diffusion that occurred in Switzerland was dependent on voluntary collaboration or 'opt in' by other providers of palliative care. Uptake was encouraged through work led by Eychmüller to include elements of the LCP in national guidelines for palliative care, which clinicians were encouraged to use to develop their own versionoftheLCP:

In our country, we worked out and developed together with the other language regions [in Switzerland], a national guideline for how to deal with people and their family carers during the time of dying. This was mainly based on the competencies and knowledge we had from the Liverpool Care Pathway but it served more as a framework document so that different institutions, from hospitals to community care to nursing homes, could derive their own versions from this framework version. (Professor Steffen Eychmüller,Interview17)

In Germany, through the DACH initiative, initial use of the LCP also took place in specialist palliative care units. Professor Raymond Voltz reported that although some spread then occurred in general palliative care settings in Germany (especially in hospitals) its extent was limited, as in Switzerland, by the lack of centralised organisation of the health care system and no precedents for national implementation projects of a similar type. He observed that the recommendations from the Liverpool team for national implementation would 'never have worked' and that its use depended on the personal initiative of 'active' individuals. As in Switzerland, diffusion wasdependentontheinclusionofaspectsoftheLCPinGerman nationalguidelinesforpalliativecare: Well, at the moment it's implemented in several … as far as I know in individual institutions around very active people. And once they leave the Liverpool Care Pathway is also dying in its use. I would say it's not more than maybe 30 [hospitals] or something like that. So it's not really implemented by institutions I wouldsay.Becausewedon'thavethatnationallevelat all as in the UK … It's very fragmented and regional over here, and so it would have never worked … We have national … our national guidelines for palliative care. One of the components is on the care of the dying ... our national guidelines are built around the content of LCP. So actually this is the best as we can get on the national level. (Professor Raymond Voltz,Interview4)

In Belgium, implementation of the CAREful programme, based on the LCP, was reported to have extended to around 70% of hospital geriatric wards in Flanders, following the completion and reporting of a cluster randomized trial. Implementation in each case was preceded by the requirements of registration and attendance at a two-day training programme. The process was formally supported as an implementation and evaluation programme by the National Cancer Society, as described by our Belgianinterviewees: So we aimed for 30% coverage but now we are already at 70 or 80% coverage. Just hospitals.You see it's not a funded programme it's an implementation project funded by the Cancer Society … all 'control' wards after the intervention was finished also got the training, and could use the implementation. (Dr Tinne  SmetsandDrKimBeernaert,Interview15) A similar process took place in Italy, where the LCP was evaluated in a programme of research that culminated in a cluster randomised trial on general medical and respiratory hospital wards and was followed by a time-limited period of broader implementation. This took place in hospitals in a number of regions in Italy. Implementation in hospices in one region, Liguria, also occurred after the formal trial. As we have seen, the research and implementation programme was led by Dr Massimo Costantini from his workplace in Genoa, under the auspices of the palliative care network of the Italian National Cancer Research Institute. Dr Costantini had gained funding from the Italian Ministry of Health and describes here the early daysoftheresearch: When we did phase two we implemented LCP in three medical wards of Genoa and one respiratory disease ward in Genoa, in four wards, and we assessed the impact of the LCP before and after the implementation of the LCP. We published three papers in Palliative Medicine. One methodological, one with the results of before and after and one where … we interviewed the professionals, physicians and nurses, before and after the implementation about expectation and the perceived efficacy of implementation, and problems of course. The goal of phase two, as [reported] in these three papers, was to decide if we could start with a randomised trial. In the meantime, we received a positive answer from the Minister of Health and the project was funded. In phase two at the end of implementation we decided to slightly change the programme. We started with phase three butthestructurewasthesame.(DrMassimoCostantini, Interview19)

A distinct characteristic of the process of wider implementation in Italy was that Dr Costantini strongly advised against use of the LCP in those circumstances where a specialist palliative care team was not in place in the hospitals requesting use of the intervention: …our LCP was different from other LCPs. The document was the same but I stressed a lot the way thatacarepathwayisnotjustdefinedbythedocument you use but also the way you introduce the document because it can make the difference. For example, after the publication of phase two we received requests for LCP documentation from different hospital wards in Italy. In New Zealand and Norway, national co-ordinating offices were set up which enabled emulation of processes of implementation that had occurred across general care contexts in the UK. In New Zealand, introduction of the LCP into a hospice in Palmerston North following participation in a meeting of an international interest group for the LCP in 2008 and sparked a process of national diffusion, marked by the establishment of a nationalLCPofficein2011: …we managed to persuade the Ministry of Health that this was a good and useful, positive assist for good dying, if you like, in all settings and I suppose the question of quality … rang a good sound with them because there was some Ministry support for leadership and palliative care at the time. We argued strongly from our hospice that we should set up a national coordinating office as the best means of gettingaunifiedandbench-markableprocessacrossallof NewZealand.Forfouryearsorsowemanagedtosucceed in that process … (Dr Simon Allan and Bridget Marshall,Interview2,SAspeaking)

New Zealand had an extensive existing infrastructure for palliative care and employed nurse practitioners sourced from district nursing services or specialist palliative care services to liaise with GPs in the community to enable the use of the LCP in residential care settings and the domestic home. They calledthisthe'palliativecarepartnership': … and it was through that mechanism of partnership and leadership from general practice that we were able to get a good uptake by general practitioners working with our specialist nurses from hospice and the district nursing service to apply a lot of pathways to home death as well. 

Here we report on the most notable challenges discussed by interviewees. In addition, some practical issues were also highlighted mainly to do with the difficulty of adapting the paper-based LCP so it could be used in electronic records, or in the day to daymanagementofthedocumentsinclinicalsettingswherestaff wereunfamiliarwiththeLCP.

Interviewees perceived that aspects of the wider societal and clinical understanding of palliative care constrained the extent to which it was possible to introduce the LCP.This was difficult in all of the countries, but especially so in resource poor settings. For example, in India, Dr Stanley Macaden emphasised that neither patients nor clinicians had a clear grasp of palliative care principles, thus making the introduction of LCP very difficult: … the main thing is palliative care is not well understood by our own colleagues …(and) …a lot of times patients don't know what palliative care is, they think it's another way of some cure, so they're willing to grabatanystraw. (DrStanleyMacaden,Interview1) At the other end of the continuum, most of the resource rich countries had seen a high profile given to palliative care across policy, practice and public spheres. This created fertile terrain into which to introduce the LCP. As Professor Dagny Faksvåg Haugen, from Norway put it: '… palliative care has had a high focus in Norway for many years and we have done a lot to improve skills and knowledge and influence attitudes'. These wider understandings influenced in turn the extent of the training and education challenge perceived to be associated with introducing the LCP. However, whether respondents came from resource poor or resource rich countries, they regarded this as both the most important determinant of scale or level of implementation and the most difficult aspect to sustain, with hospital settings identified as the most challenging environment. For example, in New Zealand, lack of confidence and training among hospital clinicians in communication skills related to end of life care were described as a 'core challenge' tothewholeprojectofintroducingtheLCP:

When it came to hospital settings, the challenge there was...and I'm now looking retrospectively to some extent, it really was hitting to the core challenge of communicationarounddeathanddyingandthelackof desire by clinicians to go there, the lack of confidence and training in that area and really reflecting the poor way in which death and dying was done and to some extent still is done in a hospital setting. (Dr Simon AllanandBridgetMarshall,Interview2,NewZealand, SAspeaking)

Turnover of staff was identified as a problem in many settings, shownhereincommentsfromArgentinaandIndia:

The challenge, first of all, is the way to train the team, the different teams. And when the training is done I think the challenge is to train more people the next year, or the next time when people change and new doctors come or a new nurse comes, and we have tostartagain.(DrVilmaTripodoro,Interview3) This is where the problem is because junior staff and nursing staff also, doctors and nurses at the junior or the middle level they keep changing and they keep going. And unless you are very knowledgeable about that and aware that this will happen, just because you've trained, done a fantastic training for one set of people you have to do the same thing again. Once you do that it's a regular thing then you can get results. Training is key in this. (Dr Stanley Macaden, Interview1)

Where funding was forthcoming from governmental sources, it was possible to ameliorate this challenge by creating facilitator networks, establishing dedicated funded posts for nurse practitioners or nurse consultants, or using the model of practice 'champions'. As we have seen, it was sometimes also possible to capitalise on quality improvement programmes already in place that had a much broader focus than end of life care.Australiawasacaseinpoint: One of the things that we did over time, was that I did identify a champion for Care of the Dying [in each ward]. We would pull together those individuals from the different wards once a month and provide in-depth education. Then, they would go back to provide that education to their nurses, because it's justnotpossibletoprovideeducationtoeverynurseon the wards. That works very well. (Associate Professor CarolDouglas,Interview8)

[The] acute health environment already had a very strongnational qualityframework builtin,eventhough it didn't have specific criteria for end of life care or dying. We were quite familiar with having national quality cycles, a four-yearly cycle of quality improvement and introduction of documents and care plans, pathways needing ongoing review. So, for the rollout and education and the orientation and adoption of these things … systems were already in place.(DrMarkBoughey,Interview12)

Involving senior medical staff in education and training initiatives was found to be difficult by all interviewees, even when funding was available for the purpose.An example of this was manifest in the large scale research-based implementation andevaluationofaversionoftheLCPinBelgium: I think the main challenge was getting the physicians involved and especially in the training …The PhD student that was working on the trial had sleepless nights over it -she had had enough of it after four years … That's always the hardest part, to find physicians. Nurses and other staff, they are mostly motivated, but to find the physicians to take two days' time to come to a training, it's often more difficult (Dr Tinne Smets and Dr Kim Beernaert, Interview 15part1,KBspeaking) Tensionsbetweenstandardisationandvariation Some interviewees reported the tension that arose between the need they recognised to alter the LCP (so that it made sense in their context and culture or in the light of their experience) and a desire for standardisation. The latter came partly from their concern to align themselves with international colleagues by use of a standard LCP 'tool' and partly from the concern of the Liverpool 'home' team to maintain 'quality control' over the translation and adaptation process. Professor Raymond Voltz captured this tension in his recollection of the development anduseoftheLCPintheGermanspeakingcountries: It's very formalised and it used to be very UK dominated and driven. And so initially it was not possible to change any single word. I would say this is a real hindrance to using it as an instrument. It could not be locally adapted. You had to get registered in a very strict form. And so following all these steps. And sometimes I had the feeling that the emphasis of the groupwasmoreontheformalisedtechnicalaspectsthan on the content, and improving content. And that was personallyformyself,butalsoformanypeopleIknow, itstillisverycounterintuitive…Ifthiswasusedinan openway,justeverybodycoulduseit,andthenwecould collect and grow and learn from each other. It was not meant to learn from each other, the experience, this was just meant to get distributed 100 per cent as it is. (ProfessorRaymondVoltz,Interview4) The withdrawal of the LCP in the UK and the associated freedom from Liverpool copyright requirements led to quite considerable adaptations of the original documentation. In some cases, it was clear that variations of the LCP pre-dated the UK withdrawal. For example, in Victoria, Australia, Dr Boughey recalled that as the implementation process unfolded, extensive use of the core LCP idea was employed locally by service providers to develop a variety of end of life care plans (especially in residential aged care) that were relevant and useful: … the LCP was pinched and rebadged and reimaged a little bit by a lot of services for their own usage. A couple of the tools that were developed were really the LCP, but, you know, pared down or a modified form.(DrMarkBoughey,Interview12) Similar situations occurred in New Zealand and in Switzerland.

In New Zealand, implementation spread from one locality through clinical networks to several areas. Our respondents described how each area made 'their own mark on it… we had various documents that resembled LCP but I would thinkitwouldbefairtosaytherewereatleastadozeninoperation in New Zealand' (Dr Simon Allan and Bridget Marshall, Interview 2, SA speaking). In Switzerland, the development of regional and local versions of the LCP was encouraged, provided these addressed the key principles and competences of a palliative care guideline produced by the DACH collaborative and inspired by the LCP. In this way, national spread was encouragedinspiteofaregionalgovernmentstructure:

The adaptation to local factors and local behaviours and local guidelines they may already have in place, this needs to be offered. So if we [were to] come in with a national standardised document and [say] everybodyneedstodoit,thisdoesnotcorrespondwith our Swiss idea of building up competencies. It's not a national health systems approach. It's a very much a local regional approach (Professor Steffen Eychmüller, Interview17)

The enthusiasm to adopt and adapt the LCP in many different settings was also matched by a concern, identified by our interviewees, about the strength of the evidence base for LCP anditswiderrollout.

In Australia, Dr Mark Boughey recalled that while the rapid spreadoftheLCPthroughoutVictoriaandotherStateswasassociated with the expression of some misgivings about a lack of research evidence, these were quickly overwhelmed as '… in practical terms, it was filling a gap that people recognised, demonstrating good care at the end of life'. Other respondents, such as those from New Zealand, acknowledged that if there wasalackofclinicalevidencefortheLCPfromtheirowncountries, they were reassured by awareness of research taking place elsewhere:

We had done some research around the implementation of the LCP in New Zealand initially. However, that wasn't actually an issue in New Zealand in terms of robust evidence. I think because it was new and the evidence was emerging and there were still studies, and there was the Italian study as well that was going on. So we knew there were studies, yes, it wasn't a big issue, there wasn't an issue raised here.

(Dr Simon Allan and Bridget Marshall, Interview 2, SAspeaking)

Professor Dagny Faksvåg Haugen, described her perception of a critique that had emerged in Norway, concerning the use of LCP in nursing homes for people with dementia, focusing on the debate or conflict as she understood it about whether one category of dying people is similar to another, and describing anattempttobuildaconsensuspositionaroundthisissue: … but we've had some challenges from … well, what should I say, especially one physician and a small group of physicians later on. And that has really been our very main challenge. That said, they think that the LCP is not suited for persons with dementia. They are not opposed to the LCP and support its use in care homes for patients. But say that the plan has not been sufficientlyvalidatedinpersonswithdementia.Andthe leader is a former nursing home physician but now she is a researcher and head of a research centre for nursing home medicine …. But, then we had taken this, well, I don't know whether I should call it conflict, to the Directorate of Health. And we had a national meeting with this other group and with us, and we had a lot of discussion and the conclusion was there is really no reason to warn against the use of the LCP in persons with dementia. Because a dying patient is a dying patient. All dying patients need care.And we think that in the dying patient individual differences are really much greater than differences based on diagnosis. (Professor Dagny Faksvåg Haugen.Interview5)

As we have seen, respondents from Belgium, the Netherlands and Italy, adopted a primarily research-based rationale as the motivation to develop the LCP for use in their own countries, recognising from the start that research evidence was a necessary precondition for use of the LCP. However, in each case, they came to the realisation from their studies that the processes of staff training in palliative care and in implementation of the LCP were just as important as any other types of data they mightgather: … you should really make them follow a training programme and I think the main issue in the UK was people can just use it without any training or implementation process to follow whereas here [in Belgium] people have to register to get all the materials and we educate them how to implement it. (Dr Tinne Smets andDrKimBeernaert,Interview15,KBspeaking) I was always, I was very keen on saying [in the Netherlands] please be careful. It's not about using a care pathway … it is about caring for people who are dying.Anditisnotaboutthatyouhavetotick…ticking (the) box has to be done, (but) you have to know whatareyoudoing.(DrLiavanZuylen,Interview14) Dr Costantini from Italy described how he was aware of the problem with regard to lack of guidance and knowledge aboutimplementationfromthestartofhisworkwiththeLCP: I didn't understand very well the way the LCP was implemented in the UK … So the risk of the cooking manual, can you understand what I mean with 'cooking manual'? The risk of the cooking manual is that 'it's very easy, it's not a problem' -just reach for the drugs, and so on. The risk was very high … I always asked for the manual for implementation but I realised that the UK group, the Liverpool group, didn't have a structured manual for LCP implementation. Dr Costantini recalled his realisation that the process of training the specialist palliative care team (he describes how he expected them to 'drive the car' of the process) in order to lead the implementation in the general wards was going to be something both lengthy and complex. It was this realisation (together with the additional implications highlighted by the UK wide withdrawal of LCP) that was a key determinant in the subsequent recommendation to withdraw the LCP from general use in Italy. Dr Costantini drew a stark contrast between the Italian insistence on the close involvement of the specialist palliative care team and the lack of emphasis that he perceived had been placed either on the detail of the implementation process or on the relationship between specialist palliative care teamandgeneralcarecontextintheUK: When we contacted a centre for the implementation of the LCP we had two kinds of contact.The first one with a palliative care team that we expected to 'drive the car' for implementation and the second contact for the hospital team for implementation, so two different subjects.We gave the palliative care team the document for implementation of the LCP of course but also, I can't remember the number of hours, three fulltime days of training with slides about training a team to be familiar with implementation. We gave them the document for collecting information about the structure of the hospital team, a document for collecting information about palliative care professionals approaching the team and also the most important, the guide for implementing the LCP. This manual is the manual that we gave to the palliative care team and it required well-structured rephrasing, with training without LCP implementation.At the end of the training introducing LCP into the hospital ward, sort of intensive support to the hospital team so it includes for example revising together all patients' data received into the ward. Then the second part of sub-intensive support and the last part of consolidation. And it's a mix of evidence based, and consensus, and expert, and experience. And we still consider the LCP to be an important document … But of course, it's not perfect, at all. It's true that we should improve our knowledge and approach in a topic that is not so easy to perform research, you know. (DrGustavodeSimone,Interview6).

Professor Carl Johan Fürst from Sweden explained that trying to gather robust and comprehensive clinical evidence for a complex intervention like the LCP was both disproportionate and showed a lack of understanding of the context of practice: But, to create evidence for a very complex intervention, so that the evidence is robust for positive effect of the whole thing … it's so difficult and so resource consuming, that … I mean, it's almost impossible. To demand that is, I think, a bit out of context. (ProfessorCarlJohanFürst,Interview7)

Our interviewees were asked how they had felt when they heard about the withdrawal of the LCP in the UK following the recommendations of the National Independent Review, and what their reflections were on the implications for the use of the LCP in their own country. Many interviewees were deeply troubled and angered about the turn of events in the UK, for two main reasons. First, some felt that the conclusions of the independent review were irrational because they covered territory that was much broader than the scope of the LCP on the last days of life. Second, others took the view that Professor Ellershaw had been badly let down by his own colleagues, who sought to take advantage of the ballooning critique in the UK to further their own research portfolios. Professor Voltz in Germany and Professor Fürst in Swedenexpressedparticularlystrongviews: Absolutely shocked and not understanding what was going on in the crazy UK. Because of course we thought 'oh wow, this is a great thing, that a palliativecaretoolisnowusednationally.Great.Andthisis an advancement of palliative care using it nationally, thank God that the UK has such a national healthcare system that they can do these things. We would love to have the same thing'. This was the initial approach. And when we heard, we were shocked. We were just shocked. Did of course not understanditatallwhathappenedthere.AndpoorJohn (Ellershaw) and his group, I mean they were really devastated and we felt very, very, very sorry for them. Anditwasjustveryunfairwhathadhappenedtothem.

And it was very unfair how it was treated, even from other groups within palliative care … they used it for argumentation of their own research projects, which was not fair at all. So [we] felt very, very sorry for the Liverpool group and this was absolutely unfair.

(ProfessorRaymondVoltz,Interview4) What the hell is going on, what are they doing, what is this? I didn't understand, at all. I understood, very well, the debate, the media and all that stuff. But, for me, it was like a blame game, where you had to blame somebody and that blame was on the LCP. I think that was very, very bad. I don't understand how you can blame something that is made to promote good care of the dying and, if it is misused, you blame it. I mean, if morphine is misused, you don't blame morphine, you blame the clinician or something like that. If an operation is done too much, it's not the problem with the operation, it's a problem with the surgeon. This is the same, I think. So, I was very upset and I didn't understand how this was going on, how it was … that it was sort of accepted by the medical community and even, you know, driven by parts of the medical community. I don't understand it.

(ProfessorCarlJohanFürst,Interview7)

After expressing their initial reactions, interviewees became more reflective, seeking to communicate their thoughts on the underlying reasons for the problems in the UK and the actions they had taken to avoid these in their own countries. Many of these lessons had already become clear in their own activities over the years in working with the LCP and have already been described here; they concerned the realisation that training, implementation processes and strong governance were essential to a successful process and outcome. In some cases, our interviewees came to the realisation that a close and sustained relationship had to be maintained between specialists and generalists in palliative care.All these prerequisites required funding and sustained co-operation and collaboration. For example, in New Zealand, interviewees reflected that the 'storm' in the UK was surprising but not entirely unexpected. They reflected that as well as exposing the power of the tabloid press intheUK,itrevealedakeyweaknessintheUKversionofLCP, thatitapparentlyprioritised'paperratherthanprinciple': Well, to some extent, unbelievable and yet believable, youknow,tabloidpressbelievable.Thestorm,however, was a bit unbelievable. My reflection is that it exposed the lack of national and regional coordination of this very important tool, i.e. promotion of the paper but not the principles which I think we picked up on very well. (Dr Simon Allan and Bridget Marshall,Interview2,SAspeaking)

A similar emphasis on principle not paper was manifest in the accounts of other interviewees, as a key reason why they had managed to avoid the problems experienced in the UK.

As we have already seen in the DACH collaborative countries, this emphasis, which was in any case necessary because of the lack of a centralised health care system, meant that the LCP was used as a framework of national guidance on palliative and end of life care, but was then subject to local interpretation on the ground. Professor Steffen Eychmüller reflected on the differencesthus: Actually, we had many discussions in our German speaking collaboration, with the Germans and the Austrian people. For us it was not very clear why there was this media led, we would say, hysteric reaction.You alsocanseethatstill inGermanyand alsoin somepartsofSwitzerlandpeopleusethetermLiverpool Care Pathway without any hesitations because it was well established and people thought why change the winning horse? I think the difference, when we looked to your country, we thought that possibly this idea of standardisation and following the rules is very strong in your country and this might really have side effects if you really follow the rules and the guidelinesandpossibly,atleastinourcountry,welikerules and guidelines but we like also to adapt it individually. So there was not this fear that by giving guidance you would possibly exaggerate and put someone on the pathway to hell. I think we thought it's very black and white in your country. It's very much this idea once there is a standard approach and once there is a guideline, you need to follow it. Possibly this might be part of the problem. But it's very difficult for us to judge how strict you follow standards in your country. This is something that is difficult to know. (ProfessorSteffenEychmüller,Interview17)

Both Professor Eychmüller in Switzerand and Professor Voltz in Germany recalled robust discussions and disagreements with Professor Ellershaw about the apparent prioritisation in the UK of the 'document' over its underlying principles. For example, Professsor Eychmüller recalled 'many fights' over the extent to which the document should be taken literally, recalling that once the UK withdrawal occurred, this led to some welcome freedom of interpretation and better enabled adoption of the broad framework provided by the LCP as a basis for nationalpolicyinpalliativecare: Actually I had many discussions with John and others in earlier times [about] how literally we should take the document. I had many fights like this because Ithoughtitcannotbethatweneedtomakethecrosses every four hours and then we guarantee the best care. So I think it's really very much about how strict the document is taken as a guarantee for good quality. … So I think we had a huge discussion about how strict to follow the document and this came also togetherintermsofhowstricttranslationofadocument should be or could be on an international level. Because obviously your document from England represented the style and the attitude and the approach, how you work in your healthcare system. But if we translated it literally in our language, it was very unusual for us. So I think to discuss it on an international level actually opened the door for becoming a bit more, I would say, relaxed and to put it in a place whereitisreallyhelpful,soasaverygoodframework and as a very important area, that you really highlight how important this phase of life is but then leave it to the people to make the best out of it for their use. This is what we actually [do] in Switzerland now.

(ProfessorSteffenEychmüller,Interview17)

In Queensland, Australia, Associate Professor Carol Douglas described her view of the events in the UK as a predictable 'train wreck' relating the unfolding disaster to a lack of emphasis on governance or overarching control of the implementation process, as well as to the key error widely reported at the time of incentivising the use of the LCP in NHS trusts in England:

I just thought … I mean, in a way, I could almost see it was like a train wreck, because LCP was just growing all the time and everyone was taking it up. But, it just seemed like there wasn't an overarchingcontroloftheprocess.Ithinkthefactthat…there were Trusts that were paying per person that went on the pathway, was appalling. I mean I think that was the undoing … It's a bit like our HHSs, the Trusts (in the UK) are a law unto themselves … But that it actually had to be completely withdrawn … I mean it just flies in the face of anything we'd ever heard of here in Australia. (Associate Professor Carol Douglas, Interview8,Queensland)

Dr Mark Boughey in Victoria used the term 'firestorm' to describewhathappenedintheUKbeforegoingontoreflectthat a key reason the same events did not occur in Australia was the presence of quality improvement structures and associated resources,includingeducationalresources.

In almost all cases, the withdrawal of the LCP in the UK was considered to have had negative consequences for the mission of improving end of life care in the interviewees' own countries. For example, in New Zealand it coincided and probably was a causative factor in cessation of funding for the national coordinating office for the LCP programme. Argentina was an exception insofar as there was little negative 'fall out'; instead the Neuberger report was used as an opportunity to try to understand what had happened in the UK. They placed emphasis on the importance of implementation, perceiving that thiswasmissingfromtheNeubergerreport: … it was mainly with us, in terms of discussing. I remember we had a meeting, ourselves, to reconsider, and of course, to read the document, because, not only the important impact, but also, what we are doing, in that sense.We knew about it, but that didn't change ourminds.(DrGustavodeSimone,Interview6) So I was a bit angry about this report because the problem is implementation … The cost I think for the Liverpool team, also for the population, the cost of this situation with the media and this kind of discussion in the media, I think is very bad for people. (DrVilmaTripodoro,Interview3)

A number of interviewees reflected on the 'missing' components of the LCP from their own experience of its use. A key aspect of this was the availability of specialist palliative care advice and help. Dr Costantini in Italy, following his work on the first randomised controlled trial of the LCP, described how a trajectory of research to try and understand and explain the interaction between the specialist palliative care team and the implementation of the LCP was stopped by the international fallout from the UK withdrawal of the LCP, which occurred as the first key publications were emerging from the RCT Others also came to a similar conclusion about the critical role of the specialist palliative care team, albeit from a different form of engagement with the LCP. In Denmark, Dr Otteson reflected on his experience of the work required to provide education for implementation and expressed a lack of surprise that the LCP ran into problems in the UK. He emphasised that it is essential to have specialist palliative care resource availableimmediatelytohelpnon-specialists: I think education and teaching is of course necessary for implementation. But I think our experience is that there has to be a person present at least in the daytime in the department. Not a person you have to call, but a person you can get to, really getting access to information and guidance; that would be one of Interviewees from Norway and New Zealand offered a set of reflections about the infrastructural pre-requisites that they had come to realise were necessary to ensure the safe implementation of the LCP. In Norway, Professor Dagny Faksvåg Haugen described the development of an implementation protocol involving three key people: responsible manager, physician and nurse, who all sign the registration document and are responsible for the implementation process. She expressed the belief that in addition to the safeguards provided by the latter structure, the quality of health care provision in Norway is such that many of the problems experienced in the UK would be unlikely to occur. Norway modelled its implementation protocol on arrangements in New Zealand, as our respondents theremadeclear: Yes, and the strategies that we put in place from a governance level, so we had governance over the work that we were doing and trying to keep those principles in mind. It's probably worth pointing out as well that when you interview in Norway, Norway very much liked the way in which we had set up the national office and the coordinating and the LCP facilitators and very much copied that model very successfully to this day, I believe. (Dr Simon Allan andBridgetMarshall,Interview2,SAspeaking)

In most countries (except for Italy and Japan), the use of the LCP in some form continued after the cessation in the UK, but the terminology used to describe the intervention changed significantly. There was an overarching concern to avoid language used in the UK such as 'putting patients on the pathway'. Moreover, in all cases, the term 'LCP' was abandoned. Which is why it still continues being in this group as well and trying to move things in, as I see, the right direction.'(ProfessorRaymondVoltz).

The second edition of the LCP handbook, published in 2011, contained a new final chapter, on international development 33 . It gives insight into the principles that underpinned this programme of work and some of the settings where it was taking place. It is the only extended statement, from the proponents themselves, about the process and organisation of the internationalspreadoftheLCP.

The chapter begins by rehearsing the received-wisdom history of the modern hospice movement, its role in fostering the specialty of palliative care and the struggles of that speciality to establish an evidence base to underpin its work. It draws attention to areas where development has been slow and highlights the value of determining best practice for a well-defined group of patients over a well-defined period of time. The LCP is seen as a vehicle for this. Whilst it acknowledges the challenges of LCP adoption in relation to local cultural norms, policies and procedures, and clinical governance and risk frameworks, the chapter highlights an over-arching unifying factor -'the shared dedication and need for change to provide allofuswithadignifieddeath'(p.190).

From the year 2000 this had led the LCP Central Team in Liverpool to work with colleagues from several countries to implement the use of the LCP. The work prompted the creation of an LCP International Programme focussed, as in the UK itself, on four phases of activity: 1) Induction 2) Implementation 3) Dissemination and 4) Sustainability. In each case there are clear requirements and prescriptions spelled out in the 2011 chapter for how the work should proceed. The orientation is generally towards a specific local organisation in which LCP is to be introduced. There is almost no reference to system change,'rollout'orstrategicplansformoreextensivesettings.

Induction requires 'top down and bottom up' approaches and a 'major cultural shift' in the relevant organisation. Registration with LCP Central is important, with attention to branding, intellectual property and copyright. A local steering group is likewise key to taking the project forward and establishing the aims of the programme -to empower 'generic workers', improve care with demonstrable outcomes for the dying patient and relatives and to see care of the dying as part of the core business of the organisation, with its own quality markers. This in turn requires the endorsement of localtranslationsbyLCPCentral,inaccordancewithestablished procedures.Asuccessfulprogrammewillneedarobustapproach to education and training, which must be locally driven, but supported with materials from the Central team. Success will be more than mere numbers of usage, but must lead to changes in knowledge, skills, attitudes and confidence, as well as the physical environment and associated facilities. Research should be part of assessing this. There should be attention to governance and risk, and rigorous use of the core LCP document, with its goals of care unchanged. Careful documentation of 'variance' in achieving the goals is required, including examples where the goals were not deemed to be part of local practice.

The chapter is a remarkably clear statement of purpose, together with detailed modus operandi. It highlights three key elements:implementation,dissemination,sustainability.

Implementation into pilot sites should ideally follow an algorithm of 'plan', 'do', 'study', 'act' to foster continuous learning and some measure of whether improvement has occurred. Eighty percent of local staff should take part in an education programme about LCP before it is first introduced. Periodic statusreportsshouldbesuppliedtoLCPCentral.

Dissemination is key to successful implementation. Data from the first 20 uses of the LCP must be carefully stored, reviewed with the help of the Central team, and then shared. Dissemination should also involve attention to future research, learning and teaching plans, and possible management strategies forextendingLCPbeyondthepilotsite.

Sustainability, in keeping with the local emphasis, is seen to occur when the LCP Continuous Quality Improvement Programme, duly agreed with LCP Central, is embedded across the local institution or the 'local health economy'. But the value of establishing a state, country, or national office for LCP, fully endorsed by LCP Central is also described -perhaps as the pinnacleofsustainability.

Against the background of this 2011 account, we set out in 2017 to dig deeper and to answer a number of questions about the international spread of the LCP. Writing three years later, what lessons can we draw from the sources we have gathered andwhatlighthaveweshedonthequestionsposed?

We have presented here the results of our literature review and qualitative interviews. Numerous overlaps and consistencies can be seen in the two analyses, as well as differing preoccupations. Taken separately we regard the two data sets as complementary and mutually reinforcing. In the interview analysis we found six major themes, which also ran through the body of literature we reviewed. We gained insights into the context and motivation of actors and agencies outside the UK for getting involved with the LCP and its implementation in new settings. Both data sets give accounts of the issues of translation and adaptation to new contexts as well as patterns of LCP deployment and resulting diffusion. The research literature, published commentaries and the interviews all give insights into the perceived benefits, as well as the challenges and drawbacks of using the LCP. The later papers and the interviews in general say something about the consequences for other countries that had become involved with it, of LCP withdrawal in the UK. The six themes identified in the interviews in turn echo and overlap with the original research questions that we set out, drawing from Dolowitz and Marsh (2000) . We now explore the extent to which these questions have been answered, and in the process we weave in some specific and related questions that arose from our analysis, concerning the international spread of the LCP. Before that however, we acknowledgesomelimitationstoourstudy.

Our purposive sampling led us largely to a group of LCP enthusiasts. If some had stepped back to evaluate the pathway in detail, it was always from the perspective that it had prima facie merit. It is quite possible that critical perspectives on the LCP may exist in the countries we studied. However, with one exception, there was no evidence of this in the literature review. It was difficult therefore to identify a process that would lead us to such dissident voices. Additionally, our interviews covered small numbers of individuals, who arguably, madeuptheLCPeliteinthe14countries.Theirviewsmaywell differ from 'rank and file' perspectives within the wider clinical and research workforce. Finally, we have interviews from just 14 countries, whereas the literature revealed interest in six other jurisdictions, from which we have no first-hand accounts.Acknowledging all this, our study provides the biggest literature review yet undertaken on the use of the LCP outside the UK (covering 20 countries) and combines it for the first time, with interviews with key actors in the implementation process(from14ofthesecountries).

Who were the key actors involved? From its point of initiation, the LCP formed part of a discourse concerning ways to improve the generic care of dying people, across multiple care settings. It was about taking the lessons learned in hospices and specialist palliative care settings and 'scaling up', possibly across a whole healthcare system, to include hospitals, care homes and domiciliary services. This was the stimulus for policy transfer to other countries, particularly because in the UK the LCP had gained considerable policy traction and become a key element in the national strategyforpalliativecare.

Despite its underlying policy gaols however, the key actors who became involved in LCP promulgation were typically clinicians, in particular specialists in palliative care, as well as both clinical and non-clinical researchers. The clinicians and researchers were in many cases individuals interested in system change, in finding ways to promulgate palliative care principles in settings where specialist knowledge was often absent. They were not short on enthusiasm about LCP, and can be seen at times evangelising about its benefits, but as we saw in our earlier paper, this could make for vulnerability to hubris and perhaps a disinclination to contemplate the risks or limitations of LCP adoption. Their concerns were mainly with demonstrating the efficacy of the LCP, rather than mitigating the unintended consequences that might result from its deployment. Italy is perhaps an exception to this, where there was a strong view that specialist palliative care teams must be involved in directing and supporting the implementation of LCPinothersettingsifgoodresultsweretobeobtained.

The Netherlands was the first country outside the UK to engage in detail with the LCP. This was no accident. The initiator and leader of the LCP programme in the UK, John Ellershaw, had been a visiting professor at the Erasmus University in Rotterdam from September 1997 and had been active in sharing his LCP work with Dutch colleagues. Ellershaw published a paper on LCP in a Dutch language journal in 2002 45 and co-authored abstracts that were presented at the 8 th Congress of the European Association for Palliative Care (EAPC), held in the Hague in 2003 46, 47 . The work on the use of LCP in the Netherlands is probably the only example where Ellershaw was directly involved in the research process and outputs, which were the second most numerous for any single country, and sustainedoverthelongestperiodoftime.

Initially, in many cases, it was this kind of ad personam link that stimulated the diffusion and transfer of LCP in other places. Many of the key actors were participants at EAPC and other international congresses and built up an informal network of mutual interest. This quickly became bureaucratised and formalised as 'LCP Central' developed a controlling role in transfer and translation, and it was further consolidated through the OPCARE9 collaborative. Only in a few settings, such as Norway and New Zealand, and to some extentAustralia and Sweden, was there a sense of LCP being implemented as a result of a prior and wider policy commitment to end of life improvement. In most instances, the actors involved got to work on trying out the LCP in their local context, and where they saw opportunities they then tried to work LCP into the policy language and architecture of their jurisdiction. We might think of the LCP more as a 'Trojan horse' than a 'step change', seeking to alter a system from within and below, rather than imposing a blueprint from above. LCP Central support seemed more focussed on the clinical organisational setting for implementation than on the wider policy environment. There was a striking absence of policy makers and policy researchers in the authors of the papers we reviewed and the people we interviewed. LCP protagonists could therefore be seen as consistent with Benson and Jordan's sense of those frustrated with current attempts to develop and implement new policies and who were searching for alternative ways to bring about change 22 .

Sometimes actors sought strength for this in bi-and tri-lateral collaborations between countries within the overall group of 20. For example: colleagues in Spain and Argentina worked together and produced some joint publications and presentations; a German language group from three countries formed a collaborative; Norway appeared to draw on experience and approaches from New Zealand; colleagues in Belgium and Italy collaborated in detail on research design and publication.

In several instances, LCP Central in Liverpool used its networks to create wider collaborations with actors in different countries. Likewise, the existence of receptive networks within a country could help to spread LCP awareness and implementation. This was found in strength in Sweden, Norway, New Zealand and Australia, Italy, Belgium and the Netherlands; it was somewhat apparent in Germany, Switzerland and Austria, as well as Japan; but largely absent in Argentina, Spain, India, China,Ireland,Denmark,Singapore,HongKongandSlovenia.

At the same time, we can see evidence of 'push and pull' between the UK actors and those in other countries. There was a 'push' outwards from the LCP Central team through OPCARE 9 and its attempt to exert 'version control' and to standardise reporting processes.Yet there was also a 'pull' from the other countries, that is exposed in both elements of our data.

In the interviews, various motivations and stimulations were revealed, ranging from widely shared aspirations to improve quality of end of life care by applying a structured approach, to research oriented rationales, where the LCP was seen as an opportunity to expand a research programme. In Italy, there was a desire to evaluate the LCP using a rigorous methodology not previously applied, and to establish whether it had benefits in hospital care. In Belgium, a research funding opportunity seemed to be a key motivation. In the literature review, we can see the 'pull' manifested in the enthusiastic marshalling of stakeholders who were eager to try out the LCP in different settingsandorganiseauditsandresearchstudiestocastlightonthe effects.

The key actors involved in the transfer process were therefore to be found in many different roles in varied types of organisation, ranging from hospices, care homes and domiciliary teams to major teaching hospitals. They were primarily clinicians, with some service managers. They also included palliative and end of life care researchers.They in the main did not include policy makers and implementation scientists. As measured by their published outputs, less than half of the countries identified produced research evidence in quantity (81% of the total outputs reviewed) and of significant quality: Japan, Netherlands, Italy, New Zealand, Australia, Sweden, Belgium, Germany,Austria.

What was transferred? Steiner 48 (p 246, cited in Freeman 2009) notes a 'radical tension' in policy transfer between the impulse to implement an exact facsimile of the chosen intervention and the impulse to somehow recreate or modify it. Our study reveals this tension very clearly. As we noted in the review of our interviews, emphases varied from a focus on a precise and exact translation of the LCP to using the LCP as a framework or set of principles, sometimes only referred to in national guidance and not implemented as the pathway itself. Many countries tried to strike a balance between these positions, but few actors seemed to consider the implications of their chosen orientation. We observed in our original paper on LCP, that boundary objects are most effective when subjected to local reinterpretation. In some countries local interpretations were seen as a potential problem (Netherlands, Japan); in others (Switzerland, New Zealand, Australia, Argentina) they were viewed as both necessaryandbeneficial.

Translation is at the heart of this, both linguistically and culturally. Some struggled to achieve a close translation of LCP into another language, and expended significant energy on the forward and back translation approaches as well as in achieving linguistic integrity, for example where particular verbs or nouns did not exist in translation (Japan, Netherlands). Cultural adaptation was visible in several ways, in the creation of new names and acronyms for the LCP (Belgium, Netherlands) and most notably in Argentina, where the chosen acronym had a distinct relevance (PAMPA) to local culture and imagery. In other cases (Australia, New Zealand), there was no significant issue about linguistic translation as the English language was a common factor, but actors did refer to system adaptation, to strengthen the fit with local healthcare practices and procedures. Hong Kong and Singapore however, seemed to combine all three elements of translation (linguistic, cultural, systemic), by translation into the relevant language and alsochangingsomeofthekeycomponentsforlocalrelevanceor significantlycuttingthenumberofgoalsonthepathway.

We noted in our first paper that LCP in the UK was surrounded by ambivalence about its purpose and make up, namely whether it was a document for close attention and implementation, or a broad approach to caring at the end of life, and we also explored whether it encouraged person centred care or standardisation of care processes across a group of patients. This ambivalence was also seen internationally. In some countries (Italy and Belgium, and to a lesser extent Japan), LCP was clearly framed as a complex intervention with several inter-related components and in which its 'support' elements, notably the implementation manual, became crucially important. Actors in other countries focused more narrowly on the introduction of the documentation and its use, without extensive supporting activities, although all were aware that training and education were crucial. These differences were reflected in the types of data gathered for research and evaluation. Some studieditsimpactonpatientorfamilyoutcomesprimarily,while others focused on audit data, especially: 'variance' of recording against the LCP documents or types of document prompted by the introduction of the LCP; rates of use; staff views, experiences, perceptions/ meanings; or process issues such as communicationbetweenclinicians.

The LCP was first developed within the context of the care of terminally ill cancer patients, so in its international spread these patients initially figured prominently, especially in implementation and research that took place with the involvement of specialist palliative care teams. Sometimes, as interest in the LCP gained momentum, the patient groups expanded accordingly, to include patients with non-cancer conditions. Others saw LCP from the outset as a vehicle for broadening the reach of patients who could benefit from its structured approach, in particular older people. When this occurred in Belgium, it required significant adaptation to the documentation of LCP. When this was not undertaken, as in Norway, it brought forth criticism from those who considered LCP inappropriate for patients with dementia in nursing homes.

In some instances (Germany, Sweden), the paper-based format of the LCP looked outdated and did not facilitate integration withahealthcaresystembasedentirelyonelectronicrecords.

The LCP as a document was assiduously transferred elsewhere with relatively minor variations, mostly sufficient only to justify adoption in a new context, perhaps marked by the use of a new name of acronym. This integrity was then leveraged to move LCP into a wide range of settings, supported by a wide variety of agencies and actors across the 20 countries. As this happened, the boundaries around the LCP could be more blurred. LCP Central sought to avoid this and maintain its integrity in various ways. In some countries, where no formal abandonment of the LCP took place (in particular Norway and Sweden) it took on its own momentum, and the relevant actors even suggested their work was made simpler in the absence of the controlling hand of LCP Central. Almost universally agreed however, was that LCP contained a structured approach to end of life care, which was successfully transferred to numerous settings, where it gave a framework for successful care in the final days of life. This could in turn have a wider influence, inflecting policy documents and guidelines with its underlying principles or providing content examples for the training of professionals who may never use the LCPbutcouldlearnfromitscontentandgoals.

From where were lessons drawn?

We have seen that in several countries, where studies were carried out in the early stages of implementation, there was a tendency to interpret the results, however modest in scope, in a very positive light. Limited evidence did not prevent many key actors from being caught up in what might be termed a habitus of optimism that could not countenance poor results or outcomes. The early studies in the Netherlands are good examples of this. Where the research designs were more sophisticated (as in Italy) they could also be prone to over-stating benefit, such as in quasi-experimental before/after studies characterised by two measurement points, one before and one after the intervention, without any external control group, and which weremorelikelytoproduce'benefit'fromtheintervention.

The Italian and Belgian studies (one of LCP, the other of a close derivative) stand out for their systematic approach and a stronger sense of wanting to test the intervention, eventually resulting in a randomised control trial, albeit in both cases with unconvincing results from the LCP perspective.

In Belgium there was also a particular emphasis on learning systematically from the LCP experience within the UK. In Italy and Belgium and to a lesser extent Sweden, the Netherlands and Japan, there seems to be no distinctive body of work on the implementation of the LCP that builds knowledge and experience over time. In other settings (Spain, Argentina, Hong Kong, India, Singapore), the publications were limited in number and seem to set up a discussion and a level of interest based on early enquiries into the use of LCP, that was not followed up with further research or documentation (except in Argentina) to indicate how things developed in the longerrun.

It is perhaps more useful to see LCP as successful in process terms. By highlighting the need for better care at the end of life and the challenges of diagnosing dying, stimulating education and training, helping to leverage new resources and policy change, forging international collaborations (all useful things), it provided a point of focus in an area otherwise rather challenging and daunting for isolated individuals to make any kind of improvements. Yet against the gold standard of RCT evidence, LCP was unable to provide convincing benefit. In the UK there was much anecdotal commentary that LCP could do harm; but this was not shown or even hinted at in the studies of its international use, though in Germany there was an interest in patients on the LCP who did not die and the clinicalimplicationsofthis.

Another 'process' dimension was the way in which the LCP helped to foster international collaboration for clinicians that was perhaps unavailable to them in any other form in a still emerging and evidence-challenged field of specialisation. The enthusiasm with which they engaged with it, and their disappointment and anger when it was attacked in the UK press andgovernment,isthereforeunderstandable.

Most countries were unable to engage with the 'plan, study, do, act' cycle of implementation and several did not conduct any extensive level of evaluation or research. Rather, they seemed to rely on the wave of international interest in the LCP, seeing this as somehow sufficient as a recommendation for use. They placed a high value on the UK/Liverpool origins of the LCP and its creation by a prominent professor in the field, who had indeed trained with Dame Cicely Saunders, and who in turn had endorsed the LCP in her foreword to the handbook, first published in 2003. Later in the cycle, some, through their own observations or attempts to introduce LCP beyond their own setting, came to realise its limitations, in time reading the 'Neuberger' report for ideas that would preempt similar issues in their own jurisdiction. Some remained puzzled by what had occurred in the UK and perceived no deleterious consequences for the use of the LCP. But others, like Italy, stopped short of introducing it more widely or actively encouraging its growth, even though its effects seemed useful,ifnotasmuchashoped.

The lessons were drawn from several sources. Initially these focussed on specific experiences in the participating countries, which were shared via LCP Central, OPCARE9, or in wider fora such as conferences and publications. In time a body of publications appeared that individually and severally offered numerous lessons, mainly at the micro-level of implementation. Macro-level lessons were drawn by inference from the UK experience, leading reluctantly to the conclusion that if LCP could not work effectively there, then its long-term chances ofsuccesselsewherewereprobablyslim.

The form of policy transfer we have been describing here is 'voluntary' rather than 'compulsory' in character. The various sets of actors collaborated through their own motivation, volition and desire to improve palliative care, rather than through any form of requirement, constraint or target to be met. Indeed, as we have seen, their goal was in part to foster a policy mandate for LCP, that if successful would have led to morecompulsionaboutitsuse.

We need to distinguish between two elements within the 'transfer' described here: 1) the diffusion of ideas and practices around the LCP and 2) the 'spread' of LCP implementation. Both were mainly confined to the Global North. Successful sharing of ideas and practices however did not lead inevitably tospread.

It was in clinical settings that evidence around ideas and practices was most visible. Clinicians in all the countries we studied can be seen engaging with the LCP in ways that were fostered through international meetings, networks, journal publications and conference presentations. These undoubtedly facilitated aspects of transfer, and the formation of LCP Central in Liverpool and then of OPCARE9 provided a secure framework in which to foster these forms of knowledge exchange,albeitmainlywithinthosedefinedasmembers.

It seems evident that a lack of material resources to access these networks served to restrict transfer and those who most easily gained entry to them were in some cases already prominent within the international palliative care field, established as clinicians and academics and members of groupings of collaboration that even preceded the advent of LCP. Only two sets of actors from low and middle income countries (LMICs) -Argentina and India -were able to benefit from such resources or alternatively, to see the LCP as relevant to their work. Additionally, where palliative care specialist knowledge was weakly developed or not widely available, then this generated problem for the successful use of the LCP in hospital wards, care homesanddomiciliarysettings.

The spread of LCP within a jurisdiction was also dependent on the support of other stakeholders beyond the adopters and champions of LCP. For example, in the Netherlands, critical to the scaling up was endorsement of LCP by the Comprehensive Cancer Centre of the Netherlands and then the 'roll out' in 66 regional palliative care networks. Key actors in Bergen gained support from the Norwegian Medical Association and then the local health authorities. In New Zealand and Sweden there were national LCP centres that took a lead. In Switzerland the process involved inserting elements of LCP into the principles and competencies contained in the national strategy for palliative care, rather than a concerted process of diffusion. In many countries there appear to have been examples of organic and un-regulated use of LCP ideas, unconnected to Liverpool Central, and that could be found in bespoke local and usually re-named variants that seemed to be well received and perhaps quite resilient, even when LCP was withdrawn intheUK.

Educational and workforce limitations restricted spread, as did the absence of a national framework for palliative care.

Research or its lack did not seem to have a bearing on spread, although when research was conducted in its most rigorous form (in Italy) it eventually led to the conclusion that the LCP was neither sustainable nor safe unless implemented with ongoing and intensive specialist palliative care support, which was unavailable at scale.Actors in other countries reached similar conclusions, albeit through different routes (Denmark). In only one country (Norway) was there evidence of conflict betweenactorsaroundtheuseoftheLCP.

Transfer of the LCP was restricted by several factors, including high level policy commitment in the most of the 'recipient' countries, limited resources and still under-developed systems of palliative care, and the absence of research evidence that could be used to make a convincing argument for its adoption. It was facilitated by the work of LCP Central and the resources for collaboration that it marshalled, as well as the individual qualities and enthusiasm of actors who were eager to bring about system level improvements in end of life care.

How did transfer relate to 'success' or 'failure'? On the eve of the publication of the Neuberger report, John Ellershaw and his colleagues posed the question: '… as the debate continues in England, the LCP's country of origin, could an international perspective provide the next steps in improvingcareofthedying?' 14 .Thecurioussyntaxbeliestheintention. If LCP was discredited in the UK, it might yet survive, modify, eventhrive,elsewhere.

Early adopters outside the UK (like the Netherlands, New Zealand, Japan and Hong Kong) did not have the shadow of growing criticism of the LCP to contend with, at least initially. Italy was in the paradoxical position of publishing the most important RCT of the LCP several months after the Neuberger report had recommended its withdrawal. Others (Australia, Austria, India, Germany) were somewhere in between and were proceeding with LCP plans even as a 'gathering storm' of criticism was beginning in the UK from September 2009. Late adopters (Sweden, Norway, Belgium) did so even as a torrent of professional and public criticism was being heaped upon the LCP in the UK, from 2012 onwards, yet found success initsuse.

It does not appear that the media storm which circulated around the LCP in the UK was replicated elsewhere. It is difficult to determinewhythisshouldbethecase.Asoneintervieweenoted, the 'Murdoch press', seen to be key to the UK attack on the LCP, is also present inAustralia, where no such storm occurred.

In Norway there was robust opposition to LCP from one quarter, but this did not escalate in the manner seen in the UK. Rather the reactions elsewhere to LCP withdrawal in its home base were disbelief, annoyance, puzzlement and regret. In Italy, LCP seemed poised for wide adoption, but the equivocal results of the RCT, coming hard on the heels of the Neuberger report, led to a considered and regretful decision to abandon it. The main consolation here was that the studies conducted in Italy had demonstrated that research can be done among people in the last stages of life and can produce insightsrelevanttopolicyandpractice.

There can be no definitive answer to this question of how transferrelatesto'success'or'failure'.

Across the 20 countries, many successes can be seen to result from engagement with LCP. It fostered new collaborations and partnerships which proved enduring even when the LCP work tailed off. It created significant awareness of the need for a structured approach to end of life care that could be operationalised across skill levels, professional groups, and care settings. It was reported as beneficial in some way by many studies, none of which suggested it did harm. It was widely accepted and liked by caregivers in many capacities, and whom founditsupportive.

If it failed it did so on two fronts. First it rarely gained the policy recognition and investment it needed to become established and integrated at the level of the care system. It remained mainly in the domain of those who were enthusiastic about it, but with a few exceptions, it did not diffuse far beyond the pilot sites. Second, it lacked a cumulative evidence base that fully recognised its complexity as an intervention. The highest quality studies were conducted in only three countries, and in no instance was there an overwhelming case for the success of the intervention, but it was only in one of these (Italy)thatthepathwaywasabandoned.

The transfer and translation of the LCP to 20 countries beyond the UK had several underlying properties. It contained elements of reciprocity in the giving and receiving of an idea, in the sharing of its subsequent modification and development, and in the actions needed to evaluate the outcomes.

These reciprocal actions were voluntary and not mandated. The actors' own agency drove the quest for new knowledge, skills and improvement in end of life care. The spread of the LCP took place in defined spaces, mostly in prosperous countries, and was sustained over around 15 years. It took in differing geographies and cultures, and a variety of linguistic, policy and practice contexts. If it did not succeed in a wider transformational goal, it appears to have been well received and perceived as beneficial in many contexts. Those who promoted the international spread of the LCP created a set of values around it akin to a 'culture' or 'movement' which went beyond its technical dimensions. It was not reported to have done harm and it did not generate major public concern or scandal.YetLCPwasnotfullyestablishedinanyofthe20countries and in some only achieved the most tentative foothold. But in several settings its influence could be seen in policy documents, inspired 'spin offs', or frameworks for education. This also begs the question 'what is the LCP?', which nobody seemed to agree upon. Its protagonists, led by the Liverpool team conducted damage limitation after the withdrawal of the LCP in the UK, maintained the interest of the international group members, and quietly re-labelled the work, in particular to avoidthecontestedideaofanendoflife'pathway'.

If 20 countries outside the UK experimented with LCP, it still left more than 170 that did not. This was not just about resources. It is difficult to know why there was no move to adopt the LCP in the USA, given that much 'pathway' thinking originated there, though some American clinicians were quick to comment on where LCP had gone wrong in the UK.

Nor did countries such as Canada, South Africa, or France (all with long histories of palliative care development) emerge as nodes for LCP implementation. There was however a synergy between interest and development relating to LCP and the overall level of palliative care in a country. A study based on data from 2017 showed 30 countries, mainly in the global north, to be in the highest level of palliative care development. Of these, 15 were countries that had shown interest in LCP, per our literature review 49 . We might conclude from this that LCP had appeal in rich countries where palliative care was already well developed, but seemed of little attraction where the reverse was the case. Even so, our interviews show that diffusion of LCP within the countries studied was also extremely variable. If the LCP in Denmark and Austria was confined to just one specific local service, in Norway, New Zealand, Australia, Sweden and the Netherlands, and sometimes with the support of a national co-ordinating centre, a degree of national spread took place, sometimes within a comprehensive range of patient groups and care settings. Many other countries were somewhere in the middle with a degree of take-up regionally, orlocally.

Among the 20 countries included in our review, less than half made significant traction with published outputs and higher level research studies. Yet in these global north settings, with pre-existing palliative care infrastructures, it did prove possible to conduct experiments with LCP and where favourable initial results emerged, to build alliances that would support its implementation within local or even regional jurisdictions. Even where this advanced to a high level, such as in the Netherlands, unlike in the UK, it did not generate wider criticisms or concerns. The LCP never obtained in other countries the level of national policy endorsement that it gained in the UK -and this, paradoxically, may have protected it from criticism. Only in Norway, where LCP came to be widely used in nursing homes, did LCP come under critical scrutiny from specialists in the care of older people, who saw it as unsuitableforuseforpeoplewithdementia.

If our analysis tells us something about the transfer of palliative care interventions internationally, it demonstrates that such transfer can be achieved with discrete measureable successes, rather than in ways that fully endorse the intervention in toto. Our analysis confirms Freeman's point that in the translation process, a boundary object may allow a wide variety of actors to come together to communicate about a problem, even when -as in end of life care -there is still a lack of consensus about how a solution should be delivered, to who and with what expectations. In this way the LCP provided an international point of focus and action to try to improve end of life care, even though the actors involved often had different ideas about it and used it in a varietyofways.

Central to this was the almost universally recognised point in the literature and among our interviewees that LCP provided a structured approach that could give direction and confidence to practitioners as they entered into the often obscured and clinically charged areas of care in the very last hours and days of life. This manifested itself in new approaches that were given further encouragement by involvement with the LCP, to improve: advance care planning, interdisciplinary communication, and aspects of nursing practice. But there were also drawbacks and costs. Effective implementation of LCP required investment in education and training for those who would use it. There were also funding challenges, as well as issues relating to staff turnover, which could hamper progress. Likewise, difficult decisions had to be made about how far a standardised approach should be taken to LCP, and to what extent its successful implementation required local variation. Sometimes new care 'plans' emerged that were inspired by LCP but were often formulated quite differently. Many actors adapted their work on the LCP into the healthcare system in which they were located. This could mean less rigidity about the terms of reference, a more sophisticated approach to implementation, or simply taking the best ideas from the LCP and building them into clinical mentorship, education and training.

Yet ultimately, the structured approach that was so much valued seemed to bring very little measureable improvement when tested out in controlled studies. The entire LCP international initiative saw such studies in only three countries. In each case the measureable benefits were very small, restricted in Sweden to two symptoms (shortness of breath and nausea), in Italy to improvements in respect, dignity and kindness and in the control of breathlessness. In Belgium a negative effect emerged from the RCT, relating to reduced satisfaction with care among family members. Such stark results should perhaps not obscure the evident acceptability of and enthusiasm for LCP on the part of many of the professionals and service users who encountered it, but they do seem a meagre outcome from so much effort.Withdrawal in the UK did not prevent continued use of LCP ideas in many places, but it did kill off the fledgling research agenda, which was about to be taken to the next level, for example in Italy, by focussing on the elements withintheLCPwhichwerebeneficialornot.

Some of those involved with it internationally were shocked and confused by the withdrawal of LCP in the UK and expressed this in strong terms. Yet in several instances, through the knowledge transfer of OPCARE9 and the wider LCP network, some countries had been able to mitigate against the problems seen in the UK -building in a greater emphasis on training, focussing more on implementation principles, and ensuring good governance. For some the post-Neuberger era even meant a removal from the constraints imposed by LCP Central, and greater freedom for self-determination locally. Only Italy andJapanabandonedtheLCPcompletelyafteritsUKwithdrawal.

What then does this analysis tell us that we did not know before about the transfer of palliative care interventions internationally?

Conceptually we can identify five dimensions that shaped the transfer of the LCP from its country of origin to 20 jurisdictions beyond the UK and which need further attention in futurework.

1) We need to understand policy transfer in a period of historical time, in this case when the discipline of palliative care was seeking to gain traction and promote its knowledge and methods beyond the specialistservicesinwhichithadoriginated.

2) The transfer is in turn shaped by particular actions and processesthatinourworkedexamplesoughttopromote, implement and validate the LCP as a transformative carepathwayfortheendoflife.

3) To do this LCP had to be lodged within systems in which it could operate effectively and deliver measureablechange.

4) How this was achieved depended significantly on the cultures, geographies and settings in which it was placed,andthesecouldtakemanyforms.

These five dimensions may well form a useful checklist for others to consider as they contemplate embarking on the extensive transfer of an end of life care intervention across multiple jurisdictions. They serve as a useful antidote to the moreformulaicalgorithmof'plan','do','study','act'.

If we judge the 20 countries included in this study against the yardstick of the UK, then collectively their encounters with LCP can be regarded as moderately successful. First and foremost, no scandal resulted, as had been the case in the UK. There was no almost evidence for or claims about patient harm. Beyond that helpful alliances and understandings were fostered, as disparate practitioners in varied settings made common cause in a bid to improve the quality of dying and saw the LCP as helpful in this process. Researchers worked diligently to produce robust study designs and useable results. Almost 100 published outputs described aspects of these processes and what had emerged from them. By 2017, the key actors were still working together on improving care in the last days of life, but LCP as a complex intervention, with all its associated systems and machinery, had faded from view. Its legacy remained, some even continued to mourn its passage or to want further discussion of what had occurred, but most, not least its creator, wanted to close that particular chapterinthedevelopmentofpalliativecare.

We have noted elsewhere that understanding an end of life intervention should include some account of the motivations of its instigators, the processes of its implementation, the field of discourse in which it is located and the presence or absence of unintended consequences relating to it 50 . We have tried here to capture these elements in relation to the LCP, through our literature review and our interviews. We remain concerned that the field of palliative care continues to be unduly committed to finding workable models that can be quickly scaled up and transferred across jurisdictions, as seen in the approach taken by a recent Lancet Commission with its 'essential package of palliative care medicines, basic equipment, and human resources that could alleviate much of avoidable suffering in LMICs' 51 . We contend that alongside this approach something more nuanced is also required, which acknowledges actor motivations, takes account of transfer and translation, can make balanced judgements about equivocal research findings, and at the same time gain the ear of policy makersandfunders.

Our literature review and our interviews surfaced many statements about the importance of 'implementation' when it came to understanding the story of the LCP. Four key elements in a theory of implementation have been mapped out by May 52 . He describes 1) the involvement of agents in the intentional modification of social systems that occupy a field of action, 2) expressions of agency that shape and are shaped by their context, 3) these in turn interact with endogenous and exogenous contingencies and confounders, and 4) they must then be negotiated by the agents involved who seek to shape them in ways that can govern the process and its outcomes. May's approach is 'founded on the notion that implementation expresses "agency" and should be understood and evaluated against the problem of how human agents take action in conditions of complexity and constraint'(p2). To paraphrase this approach, rather than seeing the LCP as a 'thing' to be implemented, it is better understood as a set of practices, with varying degrees of workability and multiple opportunities for deployment, and with the strong likelihood that the actors involved will shape expectations surrounding andconditionsofitsuse.

The international initiative around the LCP demonstrates the saliency of all these interacting elements. Overall, it led to changes in policy and practice in certain jurisdictions and it fostered a sustained international collaboration that continued after the abandonment of its use in the UK. LCP was maintained in modified form in certain settings, and it largely avoided accusations of misuse and harm that had occurred in its country of origin. If ultimately it did not succeed in promoting transformational change in the end of life care systems of the countries that engaged with it, nevertheless its influence remained, and as one of our interviewees reported when writing to us just as this paper was completed -in 2020 it even 'rose out of the ashes' to influence hastily formulated palliative care strategies for generalist settings, in the face of COVID-19.

The full underlying transcripts contain many passages that are confusing, in poor English and are difficult to understand. In fairness to our interviewees we therefore opted to include the full analysis of the interviews as underlying data 28 , rather than the full transcripts. We gave the interviewees an opportunity to check their statements to make sure that the analysis reflected the full interviews performed. Therefore, although the full transcripts are not provided, the analysis contains all relevant information required to reproduce the analysis performed in thestudy. 

Institute of Nursing and Health Research and School of Nursing, Ulster University, Newtownabbey, UK

Institute of Nursing and Health Research and School of Nursing, Ulster University, Newtownabbey, UK Thank you for the opportunity to review this important report that seeks to ascertain how and why the Liverpool Care Pathway (LCP) was transferred to 20 countries beyond the UK and with what consequences.

This report seeks to address a significant question and gap in terms of the literature. As noted by the authors the discontinuation of the LCP in the UK had a significant impact on modern palliative care at the time and yet a gap existed in terms of international context and questions around transferability and transference of such tools and initiatives in practice. The authors suggest this should assist in offering wider lessons about policy transfer in end-of-life care. Given the increasing emphasis on global working and World Health Assembly Resolution in 2014 and recent developments and lessons from COVID-19 pandemic, this would seem to be a timely report. Furthermore, it is important that as a community we do engage in critical reflection about the risks and benefits of transferring such interventions across international boundaries.

Overall, the report offers a critical and thorough review of the transference of the LCP internationally, and is well supported by appropriate and current literature. The aim was to construct a detailed case study of how the LCP was adopted internationally. To do this the authors drew on a theoretical framework derived from policy transfer literature. This posed 6 key questions: 1) who the key actors were involved; 2) what was transferred; 3) from where lessons were drawn; 4) the different degrees of transfer that took place; 5) what restricted or facilitated transfer; and 6) how transfer was related to 'success' or 'failure'. The authors contend that the use of such an approach helps to make sense about the international reach of LCP and assists in deeper understanding and helps to inform the research questions.

Using this, the methods include two aspects: 1) a historical narrative review drawing on published and grey literature relating to LCP in international contexts (n=95) and 2) qualitative interviews (n=18) with key actors in involved in LCP implementation, research or discussion in countries outside the UK. The design and methods flow logically from the research question and are methodologically sound. Sufficient detail is provided, inclusive of practical methodological detail, to allow for replication and assessment of rigour. For example, in Phase 1: narrative review sufficient detail is provided in terms of inclusion and exclusion criteria, clearly outlined that the focus was published articles, conference abstracts and presentations and wider commentaries that focus on the use of the LCP in countries outside of the UK. Whilst phase 2 qualitative interviews employed both purposive and snowball sampling involving a breadth of experts from clinicians with experience of LCP implementation, researchers outside of the UK, policy makers, and global experts in palliative care. A strength of the study was the use of initial sampling frame derived from LCP framework, building to include those who were identified from the narrative review of the literature and from other informants. Conversely, however, the authors acknowledge the limitation with purposive sampling including those who could be viewed as LCP enthusiasts, which may not reflect the wider views of both clinical and research population. Other limitations include only having 14 countries included in the interview process, whereas the literature would indicate interest in 6 other areas.

Phase 1 Narrative Review: The reporting of results for the narrative review is both comprehensive and well structured. The use of a clear structure categorising studies as audit, implementation analysis; controlled studies and commentaries enabled the reader to clearly follow the evidence base and note the conclusions from an international cross-country perspective over time. The overall conclusions from the narrative review are clearly outlined and well justified from the extensive review and evidence base. One limitation of the narrative review to be acknowledged is the challenge of comparing many studies with varying deviances from the original LCP model. Whilst some papers followed the LCP in detail, others only used LCP guiding principles and developed additional documentation, such as handbooks for example, to support the implementation in practice. Therefore, this raises questions around the evidence and whether the successes of each study be directly linked back to the original LCP model or at least in part to the iterations subsequently made.

Clinical specialists and researchers in palliative care across multiple organisations were identified as the key actors involved in promoting and transferring the LCP. The authors clearly noted those key informants that either declined or were unable to be contacted, again enhancing the transparency and rigour of the approach. The interview questions were informed by the 6 areas related to policy transfer noted previously. It was noteworthy that the nature of "what" was transferred ranged from a precise and exact translation of the LCP to a guiding set of principles underpinning end of life care. In terms of clinical representation from a multidisciplinary perspective it was noted that all but one of the participants was from a medical background, yet statements highlight the importance of ACP for nursing staff and yet their participation in the interview phase was lacking. This could be included within the acknowledged limitations of including those who were identified as 'LCP enthusiasts'. Questions also remain around why further views from policy makers were not included within this phase of the report.

The discussion and conclusion of the report draws on the key points identified in the two phases and data sets, again using policy transfer as a structure. The use of the two complementary and yet distinct data sets help to clearly reinforce and provide evidence for the final conclusions. The key elements of failure were clearly noted and evidenced from the 2 data sets: (a) the lack of policy recognition and investment it needed to become established and integrated at the level of the care system and (b) a lack of cumulative evidence base that fully recognised its complexity as an intervention.

Lessons drawn from the report in the conclusion section point to the need for caution with interpretation of results, due to a tendency for modest results to be interpreted in a positive light. Despite evidence from studies using rigorous designs that the effect of the LCP in comparison to normal care was mostly not statistically significant, smaller studies reported inflated benefits. Stakeholder support, educational and workforce limitations, and resource availability impacted on transference. Whilst new partnerships and collaborations in clinical practice, clinician awareness of the need for a structured approach to end of life care, and acceptability of the model by caregivers were seen as successes, failures included limited policy recognition and investment, and insufficient gathering of an evidence base to support the LCP. The results and conclusions presented reflect the aims and research questions driving the research and can be clearly supported by the data available.

The overall report concludes by providing a useful synthesis of the learning gained from the 2 data sets by identifying 5 dimensions that shared the transfer of the LCP beyond the UK. This provides a useful synthesis of the results and a key aspect for future considerations. We need to understand policy transfer in a period of historical time, in this case when the discipline of palliative care was seeking to gain traction and promote its knowledge and methods beyond the specialist services in which it had originated.

1.

The transfer is in turn shaped by particular actions and processes that in our worked example sought to promote, implement and validate the LCP as a transformative care pathway for the end of life.

To do this LCP had to be lodged within systems in which it could operate effectively and deliver measurable change.

How this was achieved depended significantly on the cultures, geographies and settings in which it was placed, and these could take many forms.

Success in all this required clarity about goals, outcomes and related consequences.

This is an interesting and highly relevant report that adds to the knowledge base and provides a useful platform for future discussions for the discipline of palliative care.

If applicable, is the statistical analysis and its interpretation appropriate?

and Olsen, 2016) 1 on the use and experience of the LCP in Norway. This publication is authored by nurses and describes nurses' experiences of the LCP in nursing homes. Descriptions on how the LCP was implemented and experienced in one municipality ( https://www.utviklingssenter.no/prosjekter/lindrende-behandling/implementering-av-liverpoolcare-pathway-i-vestfold), and the closing report from the same municipality ( https://www.sandefjord.kommune.no/globalassets/helse-sosial-og-omsorg/hsodokumenter/utviklingssenter/administrative-dokumenter/lcp-sluttrapport-sept.-2013-2.pdf), may also shed light on the transfer and translation of the LCP to the Norwegian context. Under the heading 'Commentaries and other publications', the authors of an article (Husebø et al., 2017) 2 are referred to, stating that no open and critical debate has taken place about the LCP in Norway (or in other Scandinavian countries). This may be the case among researchers and healthcare professionals; however, in my experience, critique has been raised, but mostly in local newspapers. The point I want to make for the case of Norway, and which I guess also applies to other countries, is that it is possible to find further outputs, which may inform the conclusions of the historical narrative review. To include newspaper or social media debates on the LCP may be worth considering, perhaps in another paper. When it comes to the interviews, I want to make the same point. By including only clinicians, policymakers and global experts in the paper referred to as LCP enthusiasts and the LCP elite, the views and experiences from particular nurses and other healthcare professionals working close to dying patients are missing. The authors describe this as a possible limitation, and my claim is that this is a weakness in the study. Particularly from a Norwegian perspective where, in a short time, the LCP spread to many of the municipalities, nurses played a very important role in transferring and translating the LCP to local, often small, contexts. They were the ones who were practising the LCP on a daily basis. When critique was raised about the LCP in Norway, nurses may have had a hard time explaining and defending the pathway. Despite my critical comments, there is no doubt that the paper contributes to a richer understanding of the complexity involved in transferring and translating ideas and practices from one country to another. I hope for further analysis on the same issue.

Is the work clearly and accurately presented and does it cite the current literature? Yes

Diffusion of innovations in service organizations: systematic review and recommendations

PubMed Abstract | Publisher Full Text | Free Full Text 2

The Liverpool Care Pathway for the dying: what went wrong?

PubMed Abstract | Publisher Full Text | Free Full Text

What's in a name? From pathways to plans in end of life care

Ethics and end of life care: the Liverpool Care Pathway and the Neuberger Review

Jones AR: The Liverpool Care Pathway for the dying patient: Euthanasia through the back door, or the sign of poor death education?

The Demise of the Liverpool Care Pathway? A Cautionary Tale for Palliative Care

Lessons from the Liverpool Care Pathway--evidence is key

Assessing the uptake of the Liverpool Care Pathway for dying patients: a systematic review

� Life after the Liverpool Care Pathway (LCP): a qualitative study of critical care practitioners delivering end-of-life care

OPCARE9 -A European perspective on the last days of life

Care of the dying and the LCP in England: An international perspective

Switzerland Dagny Faksvåg Haugen Lead Consultant, Head of Centre Regional Centre of Excellence for Palliative Care, Western Norway Haukeland University Hospital Bergen, Norway Urska Lunder -MD Palliative Care Unit Lead University Clinic for Respiratory and Allergic Diseases Golnik, Slovenia Marisa Martin-Rosello CEO and Medical Director Cudeca Hospice Foundation Centro

The Liverpool Care Pathway for the Dying Patient: a critical analysis of its rise, demise and legacy in England

PubMed Abstract | Publisher Full Text | Free Full Text 20

Who Learns What from Whom? A Review of the Policy Transfer Literature

What Have We Learned from Policy Transfer Research? Dolowitz and Marsh Revisited

Common or multiple futures for end of life care around the world? Ideas from the 'waiting room of history'

PubMed Abstract | Publisher Full Text | Free Full Text

What is 'Translation'? Evidence and Policy

Institutional ecology, 'translations' and boundary objects: amateurs and professionals in Berkeley's Museum of Vertebrate Zoology

Learning from abroad: the role of policy transfer in contemporary policy making

International transfer and translation of an end of life care intervention: the case of the Liverpool Care Pathway for the Dying Patient

Slovenia was the only country for which we could find no published outputs, beyond the description in the LCP handbook chapter

Effectiveness and efficiency of search methods in systematic reviews of complex evidence: audit of primary sources

PubMed Abstract | Publisher Full Text | Free Full Text

Observation and interviewing: Options and choices in qualitative research

Using the framework method for the analysis of qualitative data in multi-disciplinary health research

International development of the Liverpool Care Pathway for the Dying Patient (LCP)

Dublin: Irish Hospice Foundation

Análisis de los resultados de un programa de calidad en cuidados paliativos para los últimos días de vida. Diez años de experiencia

Framework for the design and evaluation of complex interventions to improve health

PubMed Abstract | Publisher Full Text | Free Full Text

Validation of toolkit after-death bereaved family member interview

VOICES redesign and testing to inform a national end of life care survey

End-of-life care pathways for improving outcomes in caring for the dying

Developing and evaluating complex interventions: the new medical research council guidance

Being CAREFuL about improving end-of-life care in hospitals

An integrated care pathway for the dying

Moving to setting standards of care for the dying

Outcomes of care in the dying phase: comparison between the United Kingdom and the Netherlands. 8th Congress of the European Association for Palliative Care

Reference Source

Mapping levels of palliative care development in 198 countries: the situation in 2017

Interventions at the end of life -a taxonomy for 'overlapping consensus

The Lancet Commission on Palliative Care and Pain Relief-findings, recommendations, and future directions. The Lancet Global Health

Towards a general theory of implementation

Omsorg ved livets slutt -En kvalitativ studie av sykepleieres erfaringer med bruk av Liverpool Care Pathway i sykehjem. Tidsskrift for omsorgsforskning

The Liverpool Care Pathway: discarded in cancer patients but good enough for dying nursing home patients? A systematic review

We are grateful to several colleagues who assisted us in various ways with our study and the preparation of this article: Amy McCreadie, Anthony Bell, Erin Craighead, Dr Alexander Whitelaw, Dr Chao Fang, Miho Tanaka, Dr Anne Grinyer and Dr Mila Gerson, as well as others who must remain anonymous. In particular, we thank our interviewees, who have all agreed tobeingnamedinthearticleandinthedata.

Are all the source data underlying the results available to ensure full reproducibility? Yes Are the conclusions drawn adequately supported by the results? Yes © 2020 Thoresen L. This is an open access peer review report distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Department of Interdisciplinary Health Sciences, Institute of Health and Society, University of Oslo, Oslo, Norway Reviewer Report: This is indeed a very interesting and important paper. Congratulations to the authors. For all of us who have followed the rise and fall of the UK-based Liverpool Care Pathway (LCP) from a distance, this paper brings new insights into why and how the pathway could actually spread internationally as it did, and how it impacted end-of-life care practices in different countries. Together with a critical paper on the LCP by Jane Seymour and David Clark in 2018 (also published by Wellcome Open Research), the present paper may work as a key reference not only to the story of the LCP but also to knowledge transfer and translation in general. The paper is based on a historical narrative literature review on the international implementation of the LCP, together with data from 18 qualitative interviews with 19 people. Those interviewed are described as key actors involved in transferring and translating the LCP into 14 countries. The concept of 'policy transfer' is part of the theoretical framework, and implies that knowledge translation is also about transferring ideas and practices internationally, from one country to another. Another theoretical concept is 'the boundary object', which was also used earlier in the aforementioned 2018 paper. When it comes to the number of published articles included in the literature review, I was particularly interested in contributions from my home country, Norway. The literature search resulted in only three Norwegian publications. I suggest adding an article by (Brattgjerd

Is the study design appropriate and is the work technically sound? Yes

If applicable, is the statistical analysis and its interpretation appropriate? Not applicable Are all the source data underlying the results available to ensure full reproducibility? Partly

Competing Interests: No competing interests were disclosed.Reviewer Expertise: My area of research is interdisciplinary studies and perspectives on dying and death I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard.