Background

Access to appropriate care and support at the end of life is recognised by many as a basic human right [1]. However, ongoing changes in disease and society demand rethinking who should properly receive such care, and how, where and from whom they should receive it. The traditional focus on specialist palliative care (SPC) teams caring for people with cancer in a hospice or community setting has been expanded to include a wide range of life-limiting disease groups and care settings [2]. Changes in living and social circumstances mean that current generations can no longer expect the informal caregiving taken for granted by their forbearers, forcing people to look to formalised healthcare and social services. At the same time, individualistic, consumerist attitudes mean that people demand greater choice in determining and tailoring their healthcare, including the opportunity to be cared for and die in places of preference [3].

Advances in detection and treatment mean that diseases such as cancer and HIV that previously killed people quickly are now chronic conditions that confer an increasing burden of symptoms and functional decline over many years [4, 5]. Medical advances have also contributed to population ageing, facing health systems with an increasing number and proportion of frail people with multiple conditions. Despite the best of intentions, this population is often ‘crisis managed’ within the acute care system rather than being adequately supported in the community for as long as possible [6]. In most countries, access to healthcare varies according to socioeconomic and geographic variables, mandating efforts to decrease health disparities [7, 8]. Given the increasing number and changing profile of people with life-limiting illness, it is neither feasible nor desirable that SPC services provide care for everyone; rather, these services should be reserved for patients with the most complex palliative care needs [9]. A population approach to planning is therefore required that takes a ‘birds-eye view’ across the health system to inform the development of models of palliative care that integrate SPC with other services according to need [10, 11].

The current study set out to inform Australian palliative care policy reform by identifying and synthesising:

  1. 1)

    recommendations for population based palliative care from international policy, and

  2. 2)

    the evidence for improvements on patient, family and health system outcomes available for different population-based models of palliative care from studies of any design comparing one model with another or models to usual care.

Methods

A rapid review of the palliative care literature was undertaken over an eight week period in April 2012. Rapid review methodology utilises similar processes to a full systematic review but generates a more timely synthesis of the evidence by limiting scope (e.g. search terms and inclusion criteria ) and various aspects of synthesis (e.g. data extraction and bias assessment) [12, 13]. In deciding which efficiency measures to introduce, researchers undertaking rapid reviews need to carefully weigh up advantages in time/resource saving against disadvantages in the form of decreased coverage and increased risk of bias; given an appropriate balance, a rapid review can generate adequate advice for the majority of clinical and policy decision when a pre-defined methodology is followed [12]. Decisions made in the current rapid review and associated limitations are summarised in relevant parts of the Methods and Discussion. A protocol was developed and discussed prior to commencement but was not made available publically.

Eligibility criteria

Two kinds of document were deemed informative. First, we were interested to identify how various models of palliative care had been defined in the literature and which of these have been supported by evidence. We also sought international policy documents/reports with most applicability to Australia’s universal health care system and federal/state structure of funding. To be included, documents of both kinds needed to come from an Organisation for Economic Co-operation and Development (OECD) country and have been made publically available in English since 2001. We limited to more recent publications to maximise relevance to contemporary populations and healthcare contexts. Documents had to be concerned with facilitating the delivery of palliative care to people with progressive life-limiting illness in any setting.

Inclusion criteria relating to palliative care were based on the World Health Organisation (WHO) definition on the basis of being the most widely used internationally [14]. Models of care were defined as those providing a framework or system for the organisation of care for people with a progressive life-threatening illness and/or their family, carers or close friends [11]. Eligible care activities included those consistent with the aims of palliative care as defined by WHO (e.g. advance care planning and self-management) as well as meeting the care needs of the patient population (e.g. symptom management and care-giver support). In accordance with the WHO definition, inclusion criteria did not relate to the degree of training and/or experience of those providing care, but rather the nature of care provided. Indeed, as indicated in the Introduction, we were especially interested to identify evidence-based models of care involving generalist or primary palliative care providers as well as specialist services.

Studies were considered eligible for inclusion if they provided data on effectiveness and/or cost-effectiveness based on comparison either between two alternative models of palliative care or between a model of palliative care and usual care. Usual care was assumed to include routine community and hospital services other than SPC models (including private). Comparisons could be concurrent or historical. Studies providing level 1 and 2 evidence (systematic reviews and randomised controlled trials [RCTs]) were prioritised, with studies using other, less rigorous designs (e.g. multiple time series) being treated as secondary sources of information [15].

We were also interested to see which models of palliative care had been recommended by current international policy. Policy document were defined as any publically available statement of position, standards or recommendations officially put forward by a government. Eligible document types included reports by health services and peak bodies as well as peer-reviewed journal articles and books/book sections.

Information sources

Electronic searches

We searched Medline, AMED, CINAHL, the Cochrane Database of Systematic Reviews, Health Technology Assessment Database and CENTRAL from their earliest records. We also searched the grey literature via internet search engines (Google and Google Scholar), the online Australian palliative care knowledge network, CareSearch, and other relevant online clearinghouses (e.g. Americans for Better Care of the Dying). Deep web searching using Mednar was considered useful for the targeting of scientific material unavailable to search engines like Google [16]. Documents listed in CareSearch’s Review Collection relating to “Service/Systems Issues” (http://www.caresearch.com.au/caresearch/tabid/501/Default.aspx) were also reviewed for inclusion.

Other sources

The reference lists of all included reviews were searched manually for further relevant articles.

Search terms

Searches for literature reporting on palliative care were conducted simultaneously with those aimed at meeting secondary aims of identifying service planning tools and research on the palliative care needs of Australians (not reported in this paper). Database searches used Medical Subject Headings (MeSH) terms or equivalent as well as keywords relating to palliative and end of life care + service models (see Table 1 for an example). Search terms were based on those for PubMed developed by CareSearch.

Table 1 Medline search terms used to identify relevant articles on palliative care models, service planning tools and palliative care needs of Australians in searches conducted on 4 th April 2012
Full size table

Study selection

Articles returned from searches of electronic databases were imported into Endnote (version X4) and coded by a single researcher against inclusion criteria for evaluative studies using a standardised proforma.

Data collection process and data items

Given the rapid nature of our review, we extracted detailed data only from those original studies not contributing to the findings of an included systematic review and data extraction was undertaken by one researcher only. Data on each model of palliative care were extracted using an electronic proforma according to a recently published disease management taxonomy which considered: patient population, intervention recipient; intervention content, delivery personnel, method of communication, intensity and complexity, environment and clinical outcomes [17]. Variables relating to study design, comparator, outcomes and findings were also extracted. Study authors were contacted via email to ask for more information as required.

Assessment of bias

Systematic reviews were quality rated by a single reviewer using the AMSTAR checklist [18]. Any RCTs we identified that were not included in one or more systematic reviews were rated for quality by a single reviewer using criteria set by the US Agency for Healthcare Research and Quality (AHRQ) [19].

Synthesis

Models of care were classified according to definitions provided by a range of sources identified by Medline and Google searches; wherever possible, definitions were taken from Australian sources to ensure relevance to the Australian healthcare system [20–33]. Classification was carried out by one reviewer, seeking input from the team as necessary where classification was not straight-forward. Meta-analysis was planned where two or more studies evaluating models of care met criteria set out in the Cochrane Handbook of Systematic Reviews [19]. Where meta-analysis was not possible, synthesis took a narrative approach using techniques described by Popay and colleagues, namely: tabulation, textual descriptions, grouping and clustering, transformation of data to construct a common rubric, vote counting, and translation of data through thematic and content analysis [34–36]. Initial synthesis was undertaken by one author, with each allocated to consider findings in a particular settings (community, hospital, aged care, paediatric and regional/rural). Iterative discussion was used to distil models and elements thereof. In the absence of studies directly comparing different models of palliative care, inference was made from results comparing models with usual care as to which had most evidence for efficacy and cost-effectiveness. No formal methods were used to examine bias across studies.

Results

Table 2 includes definitions of models of palliative care identified in the literature.

Table 2 Definitions of various models or components thereof for palliative care delivery found in the literature
Full size table

A total of 1,959 articles returned from electronic databases were screened for inclusion as evaluative studies, of which 23 reported systematic reviews, 9 RCTs and 34 non-randomised comparative studies. Systematic reviews included an average of 18 studies (range 2–43) and varied as to whether they defined their focus by setting (day care [41, 42], hospital [43], hospice [44], community [45–49], aged care [50], across settings [51–56]), patient group (transitioning to adult [57], HIV/AIDS [58], dementia [59]) or type of intervention (telehealth [60], caregiver-focused [61], case conferencing [62], UK Gold Standards Framework [63]). Only two of these systematic reviews limited inclusion criteria to RCTs [55, 56], although all but three included RCTs alongside other designs. In total, the reviews included 126 RCTs, 29 of which were included in more than one review. Of the 9 RCTs we identified that had not been included in a review, three tested models using case management [64–67] and five SPC consultation [68–72].

See Figure 1 for a flowchart of inclusion/exclusion of peer-reviewed articles and Tables 3 and 4 for details of systematic reviews and RCTs respectively.

Figure 1
figure 1

Flowchart for inclusion of articles reporting evaluative studies identified through searches of electronic databases.

Full size image
Table 3 Systematic reviews evaluating the efficacy of palliative models of care
Full size table
Table 4 Randomised controlled trials (RCTs) comparing models of care to ‘usual care’ and reported in the peer-reviewed literature
Full size table

In keeping with international policy, this review found a high level of interest in addressing the palliative care needs of populations beyond people with cancer to those with illnesses such as chronic heart failure [74], end-stage kidney disease [75, 76], chronic obstructive pulmonary disease [77] and dementia [78]. Research has highlighted the importance of better identifying the palliative phase of these conditions in order to appropriately time advance care planning, access to symptom management and provision of support to patients and their families. Many studies included patients with a range of diagnoses and did not distinguish care or effectiveness by disease group.

Variation in the content of models, contexts in which these were implemented and lack of detailed reporting meant that no two studies met the requirements for meta-analysis that had not previously been reported in a published review. Heterogeneity in the ways models were configured and described led to a focus on the attributes of effective palliative care and service elements effective at delivering these as the most meaningful unit of analysis, rather than models of care per se.

Attributes of effective palliative care

Table 5 contains a summary of the attributes of palliative care provision recommended by English-language national policies from OECD countries.

Table 5 Attributes of models for palliative care recommended by national policy documents from OECD countries available in English
Full size table

Our review of research evidence found that few studies have been conducted across care settings, with most focusing on the provision of palliative care either in the community, acute care or aged care settings.

Attributes of home-based models of palliative care

Most commonly, models of palliative care have been aimed at supporting home-based end of life care, optimising use of SPC expertise, avoiding futile treatments and providing support for family-care givers and community health professionals [44–49, 51–53, 55, 56, 58, 61]. The most important characteristics of home-based models of care have been documented as those that support communication and coordination, engage and enable skill enhancement both for the primary palliative care team (including general practitioners [GPs]) and informal caregivers/patients, and clarify goals of care through advance care planning.

Attributes of acute care models of palliative care

Models of palliative care adopted in the acute care sector largely consist of specialist consultative services, in-patient palliative units/beds or nurse practitioner models [82, 83]. In a landmark study from the US, SPC consultation was found to improve not only quality of life but also surival for patients with advanced lung cancer [72]. Specialist consultative service models have tended to focus on: 1) discussions about prognosis and goals of care; 2) pursuing documentation of advance directives; 3) discussion about foregoing specific treatments and/or diagnostic interventions; 4) family and patient support; 5) discharge planning; and 6) symptom management [84]. Consultative services provided by hospital palliative care teams have been shown to improve symptom control and quality of life, alleviate emotional burden and improve caregiver and patient satisfaction [85, 86]. In addition they have resulted in hospital cost saving [87, 88]. Currently, SPC in the US acute care setting is more likely to be available in larger hospitals, academic medical centres, not-for-profit hospitals, and Veterans Affairs (VA) hospitals compared to others [89]. Dedicated palliative care units have been established but struggle to meet increasing demands.

The increasing pressure on emergency departments and recognition of their role in end-of-life care highlight the dearth of community based services and failure of advance care planning [90–92]. Commonly, emergency presentations result from inadequate symptom control in the community and/or absence of adequate care givers [93–95]. In some countries, financial issues also act as an incentive for patients to access treatment through the emergency department in preference to other services [96]. A particular issue is the uncertainty that emergency department health professionals face when forced to make decisions in the absence of a detailed case history and advance care plans [92]. Studies have identified the capacity of coordinated models of care to decrease unnecessary emergency department usage and inappropriate admission, especially to intensive care [97, 98].

Attributes of residential aged care models of palliative care

A setting that has proven especially challenging to improvements in quality of end of life care is residential aged care [99]. Older people in aged care are less likely to be referred to SPC services for consultation or ongoing management and more likely to have poor symptom control, unnecessary hospitalisations, sub-optimal communication, inadequate advance care planning and families who are dissatisfied with end of life care [50]. A recent Cochrane Review [50] examining multi-component palliative care interventions for older people in nursing homes identified three studies [100–102] graded as ‘poor quality’ that provided weak evidence for the following model of palliative care elements: i) communication - identifying residents who would benefit from an SPC referral and negotiating this with their doctor and family [100]; ii) development of palliative care leadership teams, technical assistance meetings for team members, education in palliative care for all staff, plus feedback on performance [102]; and iii) targeted symptom control strategies to improve discomfort [101]. Systematic reviews on the efficacy of palliative care in dementia have identified a very limited evidence-base with which to develop appropriate interventions or services [54, 59].

Attributes of care required during transitions

Models of care are faced with special challenges during transitions between care settings (community, aged care and hospital) where support is needed to avoid patients ‘falling through the cracks’ [103] and/or when a rapid response is required in the context of quickly changing clinical status or patient preferences for place of care (e.g. wishing to return home while still possible) [104]. As patients and caregivers may lack knowledge of what services are available and how to access them [105], navigating the transition from inpatient to community based care requires intensive effort and coordination to put management plans and caregiver support in place. The importance of supporting transitions is especially underscored in advanced dementia where, unless a care plan is in place, health professionals in acute care may lack awareness that a palliative approach is appropriate and initiate treatments inappropriately aimed at prolonging life with negative effects on quality of life [106, 107]. Transitional care between paediatric and adult palliative care services is also a focal point requiring intensive support [57].

Elements of effective models of palliative care

This review identified a number of dynamic elements that have been integrated into palliative care models in a range of care settings to enable access to appropriate services, improve communication and coordination between providers, enhance palliative care skills of non-specialist and informal carers, and inrease capacity to respond rapidly to individual patient needs and preferences as these change over time.

Case management

Case management is a recurring feature of many successful models [43–45, 48, 49, 51, 53–56, 58, 62, 64–69] that seeks to assess and meet the full range of each individual’s palliative care and other needs, including those relating to activities of daily living (e.g. house-work) and social wellbeing. As a result, case management frequently requires coordination of services beyond the healthcare sector, including social services and pastoral care. Case management is informed by the principles of patient-centred care [108]; as such, patients and families themselves often play an active role in determining which services they receive.

Shared care

Whilst defintions of shared care have varied (Table 2), it has been frequently reported as an element of effective palliative care delivery, utilised by a number of different models [109]. Characteristics of shared care seem to have commonly included: an identifiable lead clinician working together with health professionals from other disciplines, a focus on communication and coordination, and a rapid needs-based response and navigational strategies.

A model of care that incorporates case management and shared care and has been recommended by policy in Australia in the absence of evaluation data is the ‘pop up’ model. This model was originally developed to extend palliative care to rural/remote adult services and has since been recommended for paediatric palliative care [110]. The model develops a rapid-response team around the patient and their family drawn from primary, community-based and SPC services as required to address each client’s care plan. The model relies on excellent coordination, established networks and a system of triggers for referrals, re-assessments and re-referrals to provide intensive support over brief periods. In the UK, a coordinating role for a similar model has been assigned to paediatric oncology outreach nurse specialists to support children dying from cancer [111, 112]. The outreach nurse role is described as ‘empowering the primary healthcare team through advice and direct patient care; providing an interface between primary, secondary, and tertiary care services; and coordinating services’ [111] (p.4474).

Specialist outreach services

Internationally, specialist outreach services have been widely adopted to improve care outcomes for underserved populations through the establishment of: i) specialist clinics in urban primary care practices; ii) specialist clinics in rural hospitals where no specialist services exist; and iii) sub-specialist clinics in regional centres [113]. A Cochrane review examined efficacy of specialist outreach services in primary care and rural hospital settings implemented as one element of complex multifaceted interventions involving collaboration with primary care, education or other health services [113]. This review concluded that specialist outreach services can improve health outcomes, ensure delivery of more efficient and consistent evidence-based care, and reduce the use of inpatient services. The additional costs associated with the provision of specialist outreach appear to be balanced by improved health outcomes. None of the studies in the review included comparisons of palliative care specialist outreach services; their widespread use raises a need for evaluation [114].

Managed clinical networks and/or health networks (clinical networks)

Across the globe, clinical networks have been integrated into many healthcare systems as part of a wider reform agenda to ensure that underserved populations and those with poorer outcomes have better access to quality, clinically-effective health services [115, 116]. Clinical networks facilitate the formal linking of groups of health professionals and organisations from primary, secondary and tertiary care to work in a coordinated manner, unconstrained by existing professional and organisational boundaries [117]. Many of these boundaries are driven by funding models and geographical boundaries. Although conceptually appealing, few empirical studies have been undertaken to evaluate the effectiveness of clinical networks. A literature review identified eight empirical studies, including comparative and observational designs [117]. The review concluded that clinical networks - when formally established, with governance and guidelines in place - facilitated access to care for people in underserved communities.

Integrated care

Numerous studies identified the crucial role of integrated care [51–56]. Integration refers to coordination of disparate services centred on the needs of each individual patient and family with the aim of ensuring continuity of care. Integrated care requires that patients and families are involved in informed decision-making and goal setting. It is based on principles of advocacy and respect that provide seamless, continuous care from referral through to bereavement and across organizational boundaries. Positive effects of integrated care in paediatrics have been demonstrated not only for patient and family outcomes, but also on organisational efficiencies and staff satisfaction [57].

Integrated care is especially important when supporting adults or children in the community, the enablement of which is increasingly prioritised by policy in many countries [118, 119]. While the role of primary care at the end of life is important everywhere, palliative care support for primary healthcare is most essential in rural and regional areas, where the burden for coordinating and providing medical care falls predominantly on general practitioners (GPs) and nursing care to community nurses [120, 121]. Data suggest that in some jurisdictions, including Australia, many GPs want to be involved in palliative care delivery but have decreasing capacity to undertake visits to homes or aged care facilities due to workload, time constraints and inadequate remuneration [7, 122–126]. Whilst there are no evidence-based models for palliative care in the primary healthcare setting [127, 128], there is emerging evidence that the UK’s Gold Standards Framework (GST) has improved communication, collaboration, assessment and planning since its introduction in 2001 [63]. It should be noted, however, that the UK’s National Health Service has unique drivers not readily transferrable to countries such as Australia with different healthcare funding models and multiple jurisdictions.

Volunteers

Use of volunteers may have potential where informal caregivers are lacking; however, appropriate governance models are needed. Volunteer models have been used across a range of palliative care settings but evidence of implementation and evaluation is limited [129–132].

Cost-effectiveness

Most studies that have examined cost-effectiveness of palliative care services versus usual care have found either no significant difference or palliative care to compare favourably [44, 45, 48, 51, 53, 56]. However, there remains controversy as to appropriate methods of measuring cost-effectiveness in care for the dying. The limited survival of this patient population proves a challenge for cost-utility methods; most analyses to date have focused on costs alone, with little integration of data on efficacy. Furthermore, relatively little attention has been given to costs incurred by family caregivers who may absorb costs shed by the healthcare system via community care interventions aimed at avoiding hospital admissions. No data were found comparing cost-effectiveness of different models of palliative care beyond usual care.

Discussion

Like previous systematic reviews in palliative care [133], we found few well-designed RCTs comparing models of palliative care with each another, or even with usual care. Systematic reviews have tended to include service-level interventions defined by setting (e.g. day care [42]) and/or the population served (e.g. people with dementia [59]) rather than by model of care. This consideration led us to redirect our synthesis away from whole models to focus on service elements consistently featured in models found to be effective. Of these elements, case management has been perhaps the most commonly supported [43–45, 48, 49, 51, 53–56, 58, 62, 64–69], albeit usually contributing to a complex intervention alongside a number of interacting components, different in each study. These considerations limit our ability to state with confidence that positive outcomes have resulted from case management per se.

Our review also identified the role required of political drivers in leveraging health system reform. Case management demands care across jurisdictions and care settings, which is not easy to achieve in a state/federal funding structure of the kind seen in Australia. The influence of local drivers also means that some models of care have been based on geo-political boundaries that may not be readily transferrable to other regions [63, 134–136].

Two new systematic reviews published since our search was conducted have provided important data on cost-effectiveness of palliative care. The value of home based palliative care has been demonstrated in a recent meta-analysis which found that receiving home palliative care doubles the odds of dying at home and reduces symptom burden, especially for patients with cancer, without having an adverse impact on caregiver grief [137]. A systematic review using narrative synthesis concluded that palliative care of all kinds was generally found to be cost-effective compared with usual care, usually statistically so [138].

Limitations and areas for future research

The rapid nature of the current review is associated with a number of methodological limitations [12, 13]. Limiting the scope of our search and associated terms is likely to have resulted in relevant references having been missed and increased the risk of publication and country/language biases [139]. Our inclusion criteria and approach to synthesis favoured reviews over original research and relied on a somewhat ‘blunt’ comparison that did not control for overlap between reviews. Limitations in time and resources also required us to forego the level of documentation commonly associated with full systematic reviews. These limitations were moderated somewhat by the use of the online resource ‘CareSearch’ which was designed by experts specifically to identify palliative care evidence [140] and quality assessment involving experts, including the authors of key research [141]. However, the emphasis we placed on models of care relevant to the Australian healthcare system will inevitably limit applicability of findings to some other countries.

As mentioned, the current review was also limited by variations in reporting of service models that precluded comparison and accumulation of evidence for any given model. The term ‘model of care’ was itself used inconsistently and relatively infrequently in the literature; a Medline search using terms for ‘palliative care’ combined with ‘model(s) of care’ returned only 1% of articles returned by searching for palliative care alone. Inconsistency and incompleteness in reporting impairs not only synthesis of research but also replication of successful models in future evaluations and implementation into practice. Researchers are encouraged to follow guidance on key variables to report that would enable greater comparability and support replication and refinement of models in research and practice [142].

The literature’s focus on elements rather than models raises important questions about how these elements might interact to the betterment or detriment of care quality and outcomes. The pop up model is one example of a model of care that has been recommended by policy without evidence for its effectiveness as a whole but rather an assumption that effective elements can be combined to optimise benefit [110]. Future evaluations should use factorial designs and process measures to clarify causal mechanisms between elements and identify influential contextual factors to inform ongoing development and tailoring to local needs and resources [143].

Finally, our review was limited by the problem we encountered in mapping between evidence at the outcome levels of patient (e.g. symptoms), caregiver (e.g. satisfaction), provider (e.g. knowledge of palliative care needs) and service (e.g. hospital days). A recent systematic review identified 15 patient-level domains alone, including quality of life, quality of care, symptoms and problems, performance status, psychological symptoms, decision-making and communication, place of death, stage of disease, mortality and survival, distress and wish to die, spirituality and personality, disease-specific outcomes, clinical features, meaning in life and needs [144]. The plethora of outcomes and associated measures is a recognised barrier to comparability between studies [145–150]. Whilst the WHO palliative care definition provides a framework for evaluating palliative care at the levels of the patient, provider and system, this has not yet been undertaken for any known model of palliative care. There is also a need for comprehensive economic evaluations that include descriptions of patient preferences as well as consideration of costs incurred by family caregivers and sub-group analyses examining the influence of disease and socio-demographic factors [138, 151].

Conclusion

Heterogeneity in definitions and reporting mechanisms limit the focus of conclusions from this rapid review to attributes and elements of successful palliative care services rather than whole models. Best practice palliative care should be accessible to all who need it, tailored to individual patient and family’s palliative care needs in a timely manner, and extend beyond organisational and disciplinary boundaries as required via strategies that support communication and coordination. Population-based models of palliative care should therefore include elements that support case management via integration of SPC with primary and community care services, and enable transitions across settings, including residential aged care.

While palliative care models may have once been relatively homogenous, dynamic models are increasingly required to accommodate rapidly changing population demands and health system structure and drivers. Access to specialist services for rural and regional patients and carers has been identified as especially in need of targeted intervention. The current focus on medical and nursing service delivery should also be broadened to incorporate services addressing social and environmental determinants of health as required.

Increasing complexity in service configuration warrants consideration by future research of the roles played by contextual factors such as funding and policy in order to inform planning at the population level. Research should ideally test the impact of changes over time both within and between regions using standard measures of process and outcomes.