Improving palliative care in selected settings in England using quality indicators: a realist evaluation

The science of clinical decision-making is well developed but the science of the delivery of care is relatively new and unfamiliar to most health professionals [1]. As a consequence there are gaps between available evidence of best practice and its application in everyday care, across all disciplines within health care [2]. This includes palliative care [3, 4], which aims to “improve the quality of life of patients and families who face life-threatening illness, by providing pain and symptom relief, spiritual and psychosocial support from diagnosis to the end of life and bereavement” [5].

The ageing of the population and the longer survival of people with life-threatening conditions have resulted in rising demand for palliative care from patients with multiple and complex problems [6, 7]. For example, in Europe within the next decade the incidence and prevalence of cancers will increase by about 20 % whilst the prevalence of dementia is expected to double before 2050 [8, 9]. Meeting the palliative care needs of this growing population is a big task, since the aim is to optimise the quality of life of people who have complex, incurable and life-threatening health problems by addressing their physical, emotional, psychosocial and spiritual needs.

Palliative care for people with cancer is relatively well developed, in terms of its conceptual framework and evidence base [10]. The evidence base to guide practice with those dying with complex non-cancer conditions (for example, dementia) is less well developed, although now evolving [11]. The organisation of palliative care has been described extensively [12‐14], but it is not clear how best to improve its quality, especially for non-cancer conditions.

Quality Indicators (QIs) can be used to assess where care can be improved. Campbell et al, [15], following Danabedian [16], describe QIs as evidence-based, explicitly defined and measurable items that evaluate and describe the structure, processes and outcomes of health care, and that can indicate problems in achievement of good quality care. They can be used to assess and feed back to professionals their actual performance against benchmarks, as a starting point for quality improvement [17]. QIs have been used effectively to assess and improve hospital care [18], primary care [19], and dementia care [20]. Several international studies have also developed QIs to improve the structures and process needed for the delivery of good quality palliative cancer and dementia care [21‐23]. However, these indicator sets have not yet been widely applied in everyday clinical practice. This study developed and implemented a QI package to improve the organisation of palliative care in five European countries [24]. This paper describes the experiences of using a QI set to improve palliative care in different settings in one of the five countries, England.

Table 2 shows recruitment and retention of sites in the English arm of the study. General practices could not be recruited to the study despite considerable effort over six months. A variety of approaches was used to recruit including the engagement of palliative care ‘champions’ and focusing upon trainee practitioners and practitioners recommended by palliative care teams and hospices for their interest in the topic.

Care homes were recruited but not retained in the study. Three care homes were recruited from one commercial chain, with the support of their senior management, and two agreed to take part in the study. In one care home the manager spoke frankly about her difficulties in engaging the staff in palliative care and welcomed the opportunity to have staff workshops run by IMPACT consultants, but had moved to a different care home before these workshops could begin. In the second care home the manager explained at the launch meeting that it was the home’s policy to avoid providing palliative care to residents, preferring to transfer dying residents to other homes or to hospital. This manager agreed that wider staff use of pain scales would be an appropriate change for the care home to explore, but thereafter did not respond to further contacts with the research team.

Overall the participant sites performed relatively well against the Quality Indicator set. All had access to palliative care expertise outside normal office hours and at weekends, although not necessarily to all disciplines involved in multi-disciplinary palliative care teams. Regular meetings of multi-disciplinary teams occurred at all sites, as did training activities for team members. Medical records were not accessible across organisational boundaries, so those providing end of life care in a hospital setting would not be able to access records held in the nearby but organisationally separate hospice or community palliative care team. All sites provided bereavement support (albeit in different forms) to relatives, and most also offered bereavement support for staff. Opioids and co-analgesics could be obtained when needed. Communication about transfer of patients to other settings was systematised in all sites, using mixtures of written, electronic (email or fax) and verbal (telephone, face-to-face) methods. All sites had access to necessary equipment (drug administration pumps, anti-decubitus mattresses, stoma care supplies etc.) when needed. Single rooms for dying patients were available in all sites, except of course in community services supporting patients in their own homes. Use of standardised instruments (for example, for pain assessment, or assessment of emotional needs) varied considerably between sites. Similar variation was found in the systematic capture of relatives’ experiences of the services, and in use of palliative care guidelines.

Table 3 summarises changes planned in the implementation study in each site: ticks represent achievement of changes, as agreed by the research team and the participants at each site. Using the QI set hospital wards were recruited and retained, and achieved some of their desired changes. Hospices and community palliative care teams were able to achieve almost all their desired changes, and develop plans for future quality improvements. Changes that were not achieved were implementation of validated pain assessment scales and documenting conversations with patients. In all hospital ward, hospices and related community teams the changes pursued after applying the QI package took longer than the study’s six month follow-up period (by up to three months) because of intercurrent problems such as staff changes, funding difficulties and competing organisational demands, delaying data transfer on outcomes to the IMPACT research team.

Table 4 shows the main headings of the realist evaluation, drawing together the research team’s conclusions on the impact of the contexts and the mechanisms on achievement of outcomes. Important contextual factors with a negative impact were the implementation of the controversial Health & Social Care Act 2012, which redefined the roles of general practitioners (in England) in particular, and the reorganisation of the care home chain involved in the study. The most evident mechanism with a negative impact on outcomes was ‘top down’ decision-making about joining the study; where senior management (in care homes or hospitals) decided to take part in the study, less senior staff resisted involvement.

The novel feature of the IMPACT project was the use of a QI set as part of a complex intervention to facilitate improvement in palliative care across a range of settings. The IMPACT QI set was comparable to the EUROPALL Quality Indicators [45] but more organisational and less clinical than those reviewed by Lorenz and colleagues [46].

The stepwise implementation used in this study enabled us to tailor the intervention at a setting and country level and so to translate the intervention into everyday clinical practice [12]. It followed the Batalden and Davidoff model of change, beginning with the generalisable scientific evidence that was applied in known contexts to give measureable performance improvements [22]. The realist evaluation focussed on developing explanatory models, using multiple data sources and methods, and sought to investigate contexts, mechanisms and outcomes in specific configurations [35].

Uptake and implementation of changes in organisation and practice identified using the QI set varied across settings, with maximal impact on in-patient hospices and community palliative care teams, and no impact in general practice or care homes. Although claims have been made that palliative care is the “territory” [47] of generalists the failure to engage general practices in the study is not surprising. The context (major recent changes in the British National Health Service across healthcare delviery and in general practice in particular, lack of funding for GPs to carry out palliative care) added to the known difficulties that GPs have in caring for dying patients [48], the challenge of maintaining palliative care skills [49], and the persistent taboo about discussing death [50].

Two objectives were not reached by any of the hospices and community paliative care teams: documentation of discussions with family carers; and routine use of validated pain assessment tools. Our impression is that the former had low priority compared with other communication tasks, such as transfering information if the patient moved settings. Pain management, of which formalised assessment tools are an important part, remains a problem in palliative care [51], and our participants’ inability to achieve their intended goal reflects this.

Realist evaluation suggested that four factors operated in the English sites:

These conclusions are consistent with Berwick’s view that any complex multi-component intervention is essentially a process of social change that is in turn influenced by leadership, changing environments, details of the implementation and organisational history [53]. It is also consistent with the argument that mechanisms are usually hidden and sensitive to variations in context [54].

The findings of this study should be treated cautiously. Althought case studies yield important lessons about service changes in complex organisational environments, and are therefore very appropriate to palliative care [55], their generalisability may by limited by their close attention to contexts. Before-and-after studies tend to over-estimate effect sizes [56], although this probably applies more to quantitative outcomes than to organisational processes, and realist evaluation has methodological problems that may limit its potential [57]. More involvement of generalists in the development of the QI set, with public and patient involvement too, might have resulted in a QI set that was more applicable to general practice and care homes.

Nevertheless, the findings of this study can contribute to the further evolution of palliative care, as well as indicating where further research and development efforts are needed. For example, findings from the IMPACT project informed the European Declaration on Palliative Care in 2014 (www.​palliativecare20​20.​eu/​declaration/​). In our view a single, generic QI set is unlikely to serve the needs of different settings, and further work is needed to find levers for change in general practice and care homes. Even in hospices and community palliative care teams the implementation of a QI set seemed to depend on the pre-existing research and development culture of the organisation. A range of methods for engaging services with less developed cultures is needed [58], especially if palliative care expertise is to be made available in England to those with non-cancer conditions, as will be recommended by the forthcoming NICE guidance (http://​www.​nice.​org.​uk/​guidance/​indevelopment/​gid-cgwave0694)

EUROPALL, European Palliative care study; IMPACT, IMplementation of Quality Indicators in PAlliative Care sTudy; INTERDEM, interventions in dementia research group; MRC, Medical Research Council; NRES, National Research Ethics Service; QI, quality indicator; RCT, Randomised controlled trial; SMART, specific, measurable, acceptable, realistic and timely.