Introduction

Advance care planning for the end of life is a responsibility of a growing number of healthcare professionals and an expanding niche for clinical and consumer IT products. Of the 10 leading causes of death worldwide, only one is a type of cancer, the historical priority of palliative care services.1 Non-communicable diseases accounted for 68% (38 million) of the world's deaths in 2012.2 End of life care is thus falling increasingly within the remit of chronic disease specialists. The specialised and often fragmented nature of healthcare in the developed world and the complexity of end of life needs already demand multiprofessional involvement: 36.1% of US Medicare beneficiaries were treated by 10 or more doctors in the last 6 months of life (2007 data).3 Generalist care, including out of hours, is also key: in a UK study of over 21 500 patients, general practitioners (GPs) were estimated to have recorded data on 7.3 days in the final 90 days of a patient's life.4 Online consumer products, such as My Directives (https://mydirectives.com), Best Endings (http://www.bestendings.com) and My Living Will (https://www.mylivingwill.org.uk), are also becoming popular.

We discuss England's pre-eminent initiative in enabling advance care planning and improved communication and coordination at the end of life—the Electronic Palliative Care Coordination Systems. EPaCCS are also a proof of concept for other conditions requiring multiprofessional and multisetting involvement, communication and coordination around expressed patients' wishes and a range of sources of patient data.

EPaCCS take various forms, which are still insufficiently documented—of web-based electronic registers, systems based on sharing care summaries and plans alongside patients' electronic records, patient portals, real-time extractions from the records of participating organisations, etc. They aim to provide up-to-date key information about patients believed to be in the last year of their life in GP practices, emergency telephone services (111 and 999), GP out of hours services, accident and emergency departments, ambulance services, hospitals, community nursing teams, specialist palliative care services, hospices and care homes. In the UK, each of these settings has their own (electronic or paper) patient record. EPaCCS aim to improve communication and coordination and ensure that all those involved in a patient's care are aware of their wishes, preferences and advance care plan. They are expected, and to an extent have been demonstrated, to enable more patients to die at their preferred place and reduce unnecessary hospital admissions and ambulance journeys, inappropriate interventions, use of unscheduled care and repeated ‘difficult conversations’.5 Provided that they have well-developed reporting functions, EPaCCS also supply detailed outcome metrics and enable continuous quality improvement in local end of life care services.

Striking outcomes have been reported around EPaCCS. For instance, 82.4% of the patients of ‘Coordinate My Care (CMC)’ have died outside of hospital; 77.8% died in their preferred place, with 71.8% of them dying in the place of their first preference (May 2016 data, since inception). Levels of hospital death in patients without cancer on and off the South West EPaCCS were found to be 8.3% vs 49.4%, respectively.6 Yet EPaCCS also face immense challenges, including ones of implementation, sustainability, cost-effectiveness, equality and service capacity.

EPaCCS take the idea of Locality Registers, proposed by the 2008 End of Life Care Strategy,7 into the age of digital communication. Evidence remains sparse. There are two evaluations of early implementers (2011 and 2013);6 ,8 a recent national evaluation offering tentative conclusions, reflecting the substantial challenges of EPaCCS data collection and comparison;9 descriptive studies of CMC incorporating audit data;10–13 internal project reports,14 ,15 only a small part of which are publicly available, and conference abstracts.16–20 The only completed peer-reviewed evaluation has substantial methodological limitations.21 Most evidence on EPaCCS—often anecdotal, at times overinterpreted by virtue of commitment to the initiative, and with limited information on methods and contexts—is solely accessible through policy documents and internal reports.

In terms of their multifunctionality and ambition, EPaCCS appear specific to the UK. They aim to coordinate care across all potential providers of care at the end of life; store a dynamic record of a patient's condition, treatment, wishes and preferences; enable advance care planning and accumulate data for service evaluation, quality improvement and research. Other countries have developed tools addressing some of these aims. Most notably, many maintain registers or databases on end of life care provision and outcomes aimed at quality assessment and improvement and research rather than care coordination. Examples include the Swedish National Quality Register in End of Life Care (The Swedish Register of Palliative Care, SRPC), which, as of 2015, contained data from two-thirds of all deaths in Sweden,22 ,23 the Danish Palliative Database24 ,25 and the PCOC (Palliative Care Outcomes Collaboration) of Australia (http://ahsri.uow.edu.au/pcoc/about/index.html). Transnational initiatives have also been tested, as in the EURO SENTIMELC study.26

In the USA, there is a growing use of electronic registries of Physician Orders for Life-Sustaining Treatment (POLST).27–29 POLST registries appear to focus entirely (Utah, Oregon)27 ,30 or primarily (New York State)31 on patient preferences and wishes concerning treatment options such as resuscitation efforts or use of feeding tubes or intravenous fluids. In addition to providing a record of a patient's wishes and preferences, EPaCCS also contain a summary of their condition and care, medication information and details of carers and services involved, among others. POLST are legally binding, whereas most wishes and preferences recorded on EPaCCS are not. POLST registries thus tend to contain more legal fields, for instance concerning witness requirements, concurrent physician opinion and process of reaching a mental capacity decision.31

The Scottish electronic Palliative Care Summary (ePCS) is a central system updated two times per day from GP records, introduced in 2008 and fully rolled out by 2010.32 ,33 An audit in Lothian found that 75% (year 2012) and 71% (year 2013) of specialist palliative care patients had electronic information available to out of hours services: a significant improvement relative to 49% in 2008.32 Ali et al 34 reported that not having an ePCS was associated with a higher risk of hospital admission (OR=2.43). ePCS have been regarded as a success and led to the development of a new Electronic Key Information Summary for patients with long-term conditions.35 England's EPaCCS and Scotland's ePCS are similar in the information covered and services involved, but also differ in substantial ways. For instance, ePCS is a centralised system, whereas EPaCCS are numerous and varied. A comparison of the two approaches is outstanding.

Two further developments are worth mentioning in contextualising EPaCCS: large-scale disease/treatment registers and health information exchange (HIE) initiatives. Many countries have well-established cancer registers. There is also a growing number of large-scale registers of other life-threatening conditions or treatments of these (eg, Switzerland's AMIS Plus national registry for patients with acute coronary syndrome and the joint Australia and New Zealand Dialysis and Transplant (ANZDATA) Registry).36 ,37 Such registers perform some of the functions of EPaCCS, including patient identification and the generation of data for service evaluation, quality improvement and research. Finally, EPaCCS are a HIE initiative, as much as they have been discussed primarily in an end of life care context rather than healthcare IT. The number of HIE initiatives is rapidly growing. In the USA for instance, a 2012 survey found that 1398 hospitals (30%) and 23 341 ambulatory practices (10%) were participating in 119 operational HIE efforts, relative to 14% of hospitals, 3% of ambulatory practices and 75 operational HIE efforts 2 years earlier.38 A conceptual systematic review is needed to contextualise EPaCCS in the global availability of palliative and end of life registers, registers of life-threatening conditions and treatments for these and the broader HIE literature.

Sources of data and discussion points

We present a critical analysis of EPaCCS against a background of scarce research, limited public awareness and insufficient openness in policy documents and the official discourse. We also offer an extensive structured framework for comparing the features, contexts and key outcomes of existing EPaCCS for use in future evaluations, benchmarking and funding decisions (see online supplementary tables: table S1 includes >60 parameters, table S2 lists the 51 items from the National Information Standard and 12 further fields). Finally, we make suggestions for the way forward, some of which represent a radical shift from what might be a natural trajectory of development.

The paper draws primarily on evidence and lessons from three projects: London's ‘Coordinate My Care (CMC)’, the South West EPaCCS and the Cambridgeshire & Peterborough Project for Data Sharing in End of Life Care (C&P Project). Online supplementary tables S1 and S2 summarise their characteristics. CMC is the largest EPaCCS in England, available to a population of over 9 million people. So far, 29 083 CMC urgent care plans have been created (May 2016 data, since inception, August 2010). It is also the system with the most extensive support infrastructure of all EPaCCS, including user support and training, information materials, clinical and information governance (IG) infrastructure, reporting and analysis work and high profile publicity. The South West EPaCCS is one of the earliest EPaCCS projects, initiated in 2008 before the national pilot (2009–2011)8 and later part of it. It also works in close collaboration with the National End of Life Care Intelligence Network (NEoLCIN) (http://www.endoflifecare-intelligence.org.uk/home). For a proportion of its users, the C&P Project is based on data sharing integrated with routine record keeping—the vision for the ‘ideal EPaCCS’, provided such integration can be achieved for all users.

Sources of data and discussion points include:

• Audit and evaluation data from internal monthly reports and public documents (CMC), existing evaluations (South West) and in-progress evaluations (C&P Project). References are given for published data; none come from peer-reviewed papers.

• Key costs data from the CMC and C&P Project teams.

• Data on features of the geographical areas covered: end of life care statistics from the NEoLCIN;39 data on population sizes and numbers of GP practices from NHS England;40 data on clinical services and teams obtained through Freedom of Information requests (9 for the South West, 1 for London, January–August 2015), key informants and team exercises in local service mapping.

• Descriptive EPaCCS data from the project leads and coauthors, organised around a framework developed for this paper (see online supplementary tables S1 and S2b) and the National Information Standard for EPaCCS (see online supplementary table S2a).

• Anecdotal evidence and discussion points concerning the drivers and challenges of the projects and the way forward from the project leads and coauthors. These were obtained through structured and unstructured email exchanges, interviews, meetings and telephone calls, the majority of which conducted between June and September 2014.

• The national EPaCCS Conferences in July 2015 and March 201641 ,42 and a critical review of policy documents were additional sources of data and considerations.

These data and discussion points were brought together by the first author in an extended internal report (available from the corresponding author), later distilled for this paper.

Stage of development, outcomes and costs of EPaCCS

A 2013 survey suggested that work was underway on at least 82 EPaCCS—33 developed in partnerships involving 139 of the 211 Clinical Commissioning Groups (CCGs) in England and 49 independently.43 In November 2014, 91 (43%) CCGs reported that they had a functioning EPaCCS (or similar system), 53 (25%) had ‘plans’ and 2 (1%) had no plans for an EPaCCS. Progress was unknown in 65 (31%) CCGs.44

It is unclear how commonly used descriptors of ‘functioning’ and ‘operational’ relate to the usability and use of EPaCCS. A total of 26 249 patients were on the 18 systems (49 CCGs) for which data were provided in the 2013 survey.43 More up-to-date national data are not yet available although data collection for an EPaCCS Baseline Review 2015/2016 has been completed (to be published by NHS England in 2016/2017).9 Even established systems achieve far from optimal coverage: CMC, for instance, covers 16.6% of its estimated end of life population (December 2014).

Striking outcomes have been observed for EPaCCS patients, yet high quality evidence is lacking. Data from the South West on 3012 EPaCCS patients and over 67 000 total deaths demonstrated differences in hospital deaths of 9.8% vs 33.9% for patients with cancer and 8.3% vs 49.4% for patients without cancer, respectively, on and off EPaCCS.6 Of the CMC patients who died (12 362) between August 2010 and May 2016, 7614 had a preferred and actual place of death documented and 77.8% of these died in their preferred place (first preference achieved for 71.8%). An independent evaluation suggested average savings of £2100 per person who dies with a CMC care plan, through reductions in hospital attendance and length of stay aligned with patient wishes.45 My Care Choices, the Essex EPaCCS, reported that 43% of all deaths in their 38 participating practices were preidentified by the register and 86% of those patients were able to die in their preferred place.20 Estimates from the Nottinghamshire EPaCCS, concerning a caseload averaging 3345 patients, suggested a 45% lower rate of hospital admissions for EPaCCS patients (2.5% vs 4.5%).14 Data from the Bedfordshire EPaCCS show that 69.7% (235/337) of patients have not been conveyed to hospital after a contact between the ambulance service and the service hosting the EPaCCS.46 Those are impressive outcomes, but ones concerning potentially biased patient samples, affected by limited opportunities to control confounders reliably, representing evidence of association, not causation, and coming from a small range of successful projects with good reporting capabilities. The latest national evaluation9 was not able to identify a statistically significant difference between EPaCCS and non-EPaCCS sites in terms of improvements in death in usual place of residence, hospital admissions and resource use. An economic assessment could not be completed, largely due to limitations of the data that could be collected. Rigorous evaluation and research is urgently needed.

There is a concern that EPaCCS projects may be under-resourced. The 2013 Economic Evaluation uses as its default assumption costs of £21 104 for set-up and £8235 for annual maintenance, per 200 000 population per annum, with expectations of wider end of life care investment.6 While IT costs may be low with simpler solutions, those projects demand resources for securing stakeholder support, providing training and information, negotiating workflow and system changes, upgrading the EPaCCS, auditing and reporting, etc. Between August 2010 and April 2014, CMC cost ∼£1.5 m. Using the Economic Evaluation reference points, adjusting for a population size of 9.27 m and assuming 1.75 years have been dedicated to set-up and 2 years to annual maintenance, the projected cost of CMC is over £1.7 m for this time period. The C&P Project cost £245 000 between July 2012 and March 2016, whereas its estimated cost is £165 109 (adjusting for a population size of 0.86 m, and accounting for 20 months for set-up and 2.1 years for maintenance). While substantial progress has been made in local data sharing-in end of life care and more broadly-much further work remains to be undertaken.

Key challenges

Key drivers

Where next?