Palliative care for homeless people: a systematic review of the concerns, care needs and preferences, and the barriers and facilitators for providing palliative care

A systematic review of the research literature was carried out to identify studies that examined the concerns and needs in palliative care for homeless people, and/or provided care to seriously ill homeless people. A review protocol was developed based on the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [18].

Studies eligible for inclusion had to meet the following criteria:

Commentaries, editorials, abstracts, posters for conferences and non-empirical studies were excluded. In addition, studies conducted outside the Western World (outside Northern, Eastern, Southern and Western Europe or Anglo Saxon countries) were excluded. Since Western countries already differ in the way care for homeless people is organized within the health and welfare system, we did however want to ensure comparability in terms of living conditions and welfare levels. There were no restrictions on the setting, year of publication and language of the publication.

The following sources were searched from inception: Embase.​com and Ebsco/PsycInfo (up to 1 April 2016), Ebsco/CINAHL (up to 5 April 2016), Thomson Reuters/Web of Science (up to 3 May 2016) and PubMed (up to 10 May 2016). To identify studies about homelessness and palliative care, we used a pre-defined search strategy. The string for PubMed is shown in Fig. 1, detailed information for all search strings is shown in Additional file 1.

References listed in review articles and references in papers which were excluded in the full text round were also checked. In order to find grey literature, relevant websites of organizations that are involved in palliative care for homeless people or research into it were consulted by searching for relevant keywords using Google (e.g. Simon Communities Ireland – Homeless Charity and St. Mungo Community Housing Association). Duplicate articles were excluded.

All the references obtained by searching databases as mentioned above were independently reviewed by two researchers, using Covidence online software (a primary screening and data extraction tool) [19]. Firstly, titles and abstracts were screened in order to determine whether studies met the eligibility criteria. The exclusion criteria were (1) homeless people could not be distinguished as a separate subgroup (2) study was not about somatically ill homeless adults with a short-life expectancy (3) search outcomes included: comments, editorials, abstracts and posters (4) study was not conducted in N-E-W Europe or Anglo-Saxon countries, and (5) study was not about palliative/end-of-life care. Cohen’s kappa for the first selection of titles and abstracts was 0.92 (unweighted), which is almost perfect according to Landis & Koch [20]. In the second round, the remaining full text papers were independently assessed by two reviewers against inclusion criteria. Cohen’s kappa for the second round was 0.81 (unweighted), thus also reflecting almost perfect agreement according to Landis & Koch [20]. Disagreements about whether or not the criteria were met were solved by discussion and a third researcher was consulted in the event of disagreement. There was disagreement in 8 of the 91 studies (8.8%).

For data extraction and analyses we followed the assumptions for an integrated design of a systematic review, which indicate that qualitative, quantitative and mixed-method studies can be jointly analysed and synthesized [21]. The extraction form was developed by two researchers, discussed by the research group and adjusted in response to comments. Extracted data included information about the country of the research, the research aims and questions, methods and data collection, characteristics of participants, setting, perspective of the publication (homeless people, healthcare providers, relatives/friends, open answer questionnaire), results, strengths and limitations of the study design and key conclusions. The results were extracted with a focus on the research questions; with regard to recommendations for practice we limited ourselves to recommendations given by the authors that were related to the results found in that study. For the first five publications, two researchers extracted the data independently, without any extraction software. When necessary, adjustments were made and conflicts were resolved. For the other papers, data was extracted by one reviewer and checked by a second.

Because our aims were ‘to summarize evidence about the concerns, palliative care needs and preferences of homeless people, as well as barriers and facilitators for delivering high quality palliative care’, we used the findings from the selected studies mainly to describe common themes. Thus, data was analysed using the meta-summary method [21] to identify common themes. The extracted data was classified manually into categories by sorting according to common themes, carried out by one researcher until no new categories came up. These themes were then discussed with a second researcher before discussion in the project team. In the tables the common themes are shown, indicating in which studies they occurred.

The methodological quality of the studies that met the inclusion criteria was assessed the General Appraisal instrument of Hawker et al. [22]. The instrument, which is applicable to quantitative as well as qualitative studies, consists of nine elements (abstract, background, methodology, sampling, data analysis, ethics, results, transferability and implications). Each element is scored on a four-point scale (ranging from very poor to good). Scores for the various are added to give a total score. Total scores range from 9 to 36; scores less than or equal to 18 are rated as ‘poor methodological quality’, scores from 19 to 27 as ‘moderate’ and above 27 as ‘good quality’. All methodological assessments were done by two reviewers independently. If there was a mismatch of more than five points, disagreements were solved by discussion. The scores of assessment can be found in Table 1, more details of the assessments can be found in Additional file 2.

The review process is shown in Fig. 2. We identified 3245 records through database searches, seven additional records were found through websites of organizations. After removing 1656 duplicates, 1596 papers were screened on title and abstract. Of these, the full texts of 91 were checked, resulting in 27 papers meeting our inclusion criteria (Table 1). No additional papers were found by contacting project members.

Table 1 shows the characteristics of all the studies included. A number of authors, namely Ko et al. [23, 24], McNeil et al. [25‐27] and Song et al. [28, 29] discussed their own same study in several papers; each paper discussed various aspects of the study. The 27 papers that were included cover 23 different studies. All studies were published in the period 1986–2016 and published in English. Most studies were conducted in the USA (n = 15) or Canada (n = 7).

Fifteen studies had qualitative designs, generally using semi-structured individual interviews and focus groups. Eight studies had quantitative designs using a variety of methods, such as an e-mail survey and a review of medical records. Of these quantitative studies, five studies evaluated an intervention. The methodological quality was assessed as good for fifteen papers, moderate for nine and poor for three (Table 1).

Of all 23 studies, 12 derived data from homeless participants, nine studies from healthcare professionals engaged in caring for homeless people (including review or analysis of medical records) and two studies from both homeless participants and healthcare professionals (Additional file 3). Of the 12 studies that derived their data from homeless people and the two studies with both homeless participants and healthcare professionals, the homeless people were terminally ill in three studies (Table 1) [30‐32].

Homeless people in the studies stayed or lived in a variety of settings. The most frequently mentioned were various types of shelters, e.g. drop-in shelters and homeless shelters. Other settings mentioned were support homes, housing facilities, hospitals and medical centres, healthcare programmes, palliative care services and hospices, hostels, social service agencies and sites or communities for homeless people (Table 1). Additional file 3 shows more information about the characteristics of the study populations. Most studies stated the age, sex and ethnicity of homeless participants. A large proportion of homeless participants were male, with percentages ranging between 60% and 100% of the study population. The average age of homeless participants varied between 43 and 65. In the studies that provided information about ethnicity, homeless people of several ethnic groups participated. The educational level of homeless participants, health status of homeless participants and characteristics of healthcare providers were reported less often.

Table 2 shows the main results we extracted from the publications regarding concerns, care needs and future preferences of seriously ill homeless people about the end of life. The ‘concerns’ considered problems that homeless people had or issues they worried about. Concerns in the physical domain often were about serious illnesses and physical distress [29‐31]. Psychological concerns were mostly related to fear of death and dying [24, 28, 29, 32‐35]. Social concerns were mostly about being a burden to others [24, 28, 31, 35]. Spiritual concerns were hardly mentioned and were regularly described as consisting of fear of the unknown [31, 33]. Frequently mentioned concerns about care included homeless people expecting end-of-life care to be poor [23, 29, 36, 37].

Care needs concerned topics about the care (including palliative care) that homeless people preferred or expected. Attitudes and behaviour of healthcare professionals was a theme that was often mentioned, in which treatment with respect and dignity was stated most often [28, 31, 38]. Needs concerning involvement of the family appeared to be somewhat variable. Some of the homeless want family nearby, others do not want to burden their families [28, 38] and some request some type of social contact with family and friends before dying even if they are estranged [24, 32]. Needs for treatment and care appeared to be an important theme; the most frequently mentioned were spirituality and religion [23, 24, 33, 34]. Although few spiritual concerns were mentioned in included studies, spirituality and religion appear to be important encouraging factors for homeless people when it comes needs for treatment and care. In addition, most mentioned was the possibility of expressing various concerns, such as anonymity, estrangement and maintaining control: advance care planning or documentation can help express these concerns [28, 29, 33, 39]. Only one study looked at the domain ‘after death’, showing an explicit and detailed desire that homeless people’s bodies be laid to rest in a personally and culturally acceptable manner [28]. ‘Preferences for future care and treatment; was where we grouped the preferences homeless people had in advance for the end of life; we grouped them into three domains, namely treatment preferences, the dying process and surrogate-decision making. Regarding the first domain ‘treatment preferences’, a lot of studies mentioned resuscitation and life-sustaining treatments, preferences were found to vary among subgroups of homeless people [31, 40‐42]. In terms of the wishes for the dying process, a natural death was mentioned most often [24, 33, 38], e.g. no prolonged life support without hope of functional recovery. Lastly, surrogate decision-making appeared to be an important theme for homeless people at the end of life, in particular the naming of proxy to make decisions [28, 41, 42].

Tables 3 and 4 show the results in terms of the barriers and facilitators for providing care to seriously ill homeless people. To give an overview of those barriers and facilitators, we identified and described three perspectives. The perspective called ‘barriers and facilitators relating to the homeless person’ revealed a lot of barriers and some facilitators. The most commonly mentioned barriers were themes related to receiving healthcare, such as end-of-life care not being a priority and living on a day-to-day basis [23, 26, 34, 39, 43], themes regarding social relationships such as the absence of support from family members and only having small networks [30, 33, 37], and themes about health-related and other behaviour, such as the limited insights homeless people have into their own health [32, 44]. Although studies reported more barriers than facilitators within this theme, a widely mentioned facilitator for homeless people was the importance of religious beliefs and spiritual experience [24, 28, 39].

Contrasting with the previous theme, a lot of studies in the theme ‘relating to the interaction between homeless people and healthcare professionals’ described facilitators and substantially fewer studies described barriers between homeless people and professionals. The attitudes of healthcare providers towards homeless persons proved to be a major theme, e.g. building and establishing relationships of trust [25, 32, 35, 37]. The treatment of homeless people was also reported to be an important theme as facilitator, e.g. a pragmatic and flexible approach from staff [25, 37, 45]. Furthermore, providing activities and therapies was also often mentioned as facilitator for the interaction between homeless people and healthcare professionals, e.g. counsellor-guided advance directive completion [41, 42, 46]. Feelings of being ignored, discriminated against and disrespected by healthcare providers and a lack of trust were often mentioned as barriers [26, 33, 35‐38, 45]. For barriers and facilitators in the third theme, ‘relating to healthcare professionals’, substantially more barriers than facilitators were mentioned. The most frequently mentioned barriers were lack of knowledge and skills of professionals, e.g. the difficulty for staff in determining when a patient is nearing the dying phase and meeting the palliative care needs [32, 33, 35, 37]. Another barrier mentioned relating to healthcare professionals was the organization of care, e.g. minimal access to palliative care [26‐28, 32, 36, 38, 47]. On the other hand, facilitators relating to the knowledge and skills of professionals such as optimizing management of pain, symptoms and functional decline were often mentioned [48, 49]. Facilitators regarding the overall organization of palliative care for homeless people were not found in many papers; one facilitator mentioned in a Canadian study by Podymow et al. was in-shelter hospice care, which also substantially lowers the costs [48].

A significant number of studies made evidence-based recommendations for practice regarding (palliative) care for seriously ill homeless people, themes are shown in Table 5 and more detailed information on the themes is shown in Additional file 4. Training, education and knowledge; delivering care, and overall organization appeared to be the comprehensive themes. A very often mentioned recommendation relating to training, education and knowledge was training for staff working with homeless people to provide palliative care as health deteriorates and death approaches [26, 27, 34, 36, 45]. Related to recommendations on delivering care, addressing themes related to a patient-centered approach concerning dignity and asking questions about death and dying in advance directive formats were most often mentioned [24, 31, 35‐37, 42, 46]. Trustful and respectful relationships were also mentioned as a recommendation for delivering care; as well as attention for different domain of concerns of homeless people compared to healthcare providers, flexible programs and availability and support after death. Recommendations concerning overall organization of palliative care to homeless people concerned mostly the availability of accommodation, involved persons and coordination, policies and guidelines and partnering and exchange of knowledge between organizations. This included partnering social communities with the end-of-life care system, such as accessibility and availability of palliative care beds [34, 45].

One strength of this systematic review is that it looks at the concerns, care needs and preferences, barriers and facilitators and recommendations for practice, thereby providing a broad overview of topics that are relevant to palliative care for homeless people. In addition, the broad inclusion criteria resulted in a large number of studies being included (given the limited size of the field being researched). Moreover, this review combines both qualitative and quantitative studies. Finally, another strength of this systematic review is that doing a grey literature search meant that we also included studies by organizations working in the field, such as Simon Communities and St. Mungo’s.

An initial limitation of this study is that the definition and terminology of palliative and/or end-of-life care differ according to the study. Studies may therefore include other aspects of palliative or end-of-life care while using the same definition and terminology. A second limitation is that both studies of seriously ill homeless people and studies of homeless people who expressed their expectations about being seriously ill have been included. Expectations about the end of life in advance may differ from the reality later. Another limitation was that, although we aimed only to summarize the recommendations from the studies’ results, it is was not always certain that the recommendations were not also reflections of the author’s opinions. As a fourth limitation, this systematic review lacks studies in published in languages other than English. Finally, a methodological limitation was that in some studies it was difficult to assess the methodological quality because some information was missing. It is possible that in those cases the actual study was conducted in a more thorough way than reported on in the article.

Homeless people at the end of life experience a range of problems and barriers concerning access to palliative care. A tailored, flexible and low-threshold approach consisting of awareness about the fear of death among homeless people (as well as priorities and needs of homeless people other than those assessed by healthcare professionals) can be used to help provide appropriate care in good time. This tailored, flexible and low-threshold approach should at least involve awareness of the concerns of homeless people (fear of death and negative experiences with healthcare providers). This requires sensitivity and patience of healthcare professionals. In addition, awareness about the meaning of dignity and respect to the homeless patient is important when it comes to understanding the needs of homeless people, as well as recognizing important components such as religiosity and documentation of future preferences. Finally, healthcare professionals need to be aware that future preferences may be different for homeless patients compared to a non-homeless patient, and therefore ask specific questions about it. Training, education and experience of healthcare providers can accomplish this.