Impact of home healthcare on end-of-life outcomes for people with dementia: a systematic review

Background Home healthcare (HHC) comprises clinical services provided by medical professionals for people living at home with various levels of care needs and health conditions. HHC may reduce care transitions from home to acute hospitals, but its long-term impact on homebound people living with dementia (PLWD) towards end-of-life remains unclear. We aim to describe the impact of HHC on acute healthcare utilization and end-of-life outcomes in PLWD. Methods Design: Systematic review of quantitative and qualitative original studies which examine the association between HHC and targeted outcomes. Interventions: HHC. Participants: At least 80% of study participants had dementia and lived at home. Measurements: Primary outcome was acute healthcare utilization in the last year of life. Secondary outcomes included hospice palliative care, advance care planning, continuity of care, and place of death. We briefly reviewed selected national policy to provide contextual information regarding these outcomes. Results From 6831 articles initially identified, we included five studies comprising data on 4493 participants from USA, Japan, and Italy. No included studies received a “high” quality rating. We synthesised core properties related to HHC at three implementational levels. Micro-level: HHC may be associated with a lower risk of acute healthcare utilization in the early period (e.g., last 90 days before death) and a higher risk in the late period (e.g. last 15 days) of the disease trajectory toward end-of-life in PLWD. HHC may increase palliative care referrals. Advance care planning was an important factor influencing end-of-life outcomes. Meso-level: challenges for HHC providers in medical decision-making and initiating palliative care for PLWD at the end-of-life may require further training and external support. Coordination between HHC and social care is highlighted but not well examined. Macro-level: reforms of national policy or financial schemes are found in some countries but the effects are not clearly understood. Conclusions This review highlights the dearth of dementia-specific research regarding the impact of HHC on end-of-life outcomes. Effects of advance care planning during HHC, the integration between health and social care, and coordination between primary HHC and specialist geriatric/ palliative care services require further investigation. Supplementary Information The online version contains supplementary material available at 10.1186/s12877-022-02768-3.


Introduction
Dementia is a life-limiting, progressive neurodegenerative syndrome affecting multiple cognitive and physical functions [1]. It is currently one of the most common causes of death in high-income countries, and globally, leads to an escalating need for end-of-life care [2][3][4]. People living with dementia (PLWD) are at a high risk of experiencing care transitions (i.e. transfer from home or care homes to acute hospital admission or emergency department visit), particularly towards the end-of-life [5][6][7][8]. Research in the USA and Taiwan has shown that potentially non-beneficial life-sustaining treatments such as tube feeding or mechanical ventilation are associated with care transitions [9][10][11] which may not improve the quality or length of life and are burdensome for the PLWD and their carers [7,8,10,12]. Strategies to support this population with complex care needs and high care cost living and dying well in the community are vital [13,14].
Palliative care has been considered an important service that contributes to the quality of care and fulfills the care needs of PLWD near the end-of-life [15]. According to World Health Organization's definition, palliative care is an approach that involves the identification and management of problems associated with a life-threatening illness for patients and their families, prevents or relieves their suffering and improves their quality of life [16]. Its impact on reducing transitions for PLWD in care homes has been investigated [17][18][19]; however, the evidence on effective palliative care for PLWD living at home remains scarce and inconclusive [20,21]. Furthermore, the provision of palliative care for PLWD is at a low coverage level across countries and the referral of the service is usually late in the potentially long, slow decline of disease trajectory when PLWD are approaching their end-of-life [11,22,23]. It is important to better understand the longterm impact on interventions such as home healthcare (HHC), which is usually provided earlier and more common for PLWD living at home, on reducing their high risk of acute healthcare utilisation or other health outcomes towards the end-of-life [15,24,25].
HHC comprises a spectrum of clinical care provided by healthcare professionals for people living at home with various levels of care needs and health conditions at different stages throughout the life course [26]. HHC types vary in terms of acuity, type of care provided, and degree of physician involvement, including patient-centered medical home, hospital at home, home-based primary care, physician or nurses house calls, skilled home healthcare, rehabilitation, and medication managements [26,27]. HHC does not include case management, exercise coaching, social care (such as hygiene care or nutrition support) or self-management.
HHC is increasingly recognised as an integrated and value-based service for the ageing population including PLWD [14]. The demand for promoting better integrated and continuous care at home has been advocated by groups of PLWD, caregivers, and health professionals [27][28][29], and some reforms of policy to integrate the fragmented services and quality-based payment schemes to enhance multidisciplinary approach at a national level for improving quality of HHC is established in high-income countries [27,[29][30][31][32][33]. However, in the existing literature of HHC, only a few studies have focused on the PLWD and none of them reviewed long-term effects of HHC on acute healthcare utilisation and outcomes at the end-oflife [34,35].

Methods
We registered the protocol on PROSPERO (CRD42019151250)and adhered to PRISMA statement in reporting the review [36].

Eligibility criteria
This review included peer-reviewed original articles of quantitative and qualitative studies. The inclusion and exclusion criteria on the study design, definition of PLWD and HHC and details of the study outcomes are summarised in Table 1. The comparison groups include any type of usual care, routine care or no intervention. In this study, home-based palliative care is not included in the HHC interventions because the effects of the services for PLWD have been reviewed [21], and palliative care is identified as an outcome of interest.

Search strategy and study selection
We applied a three-step search strategy: An initial limited search of Medline was performed, followed by the analysis of the terms used in titles and abstracts, and of the index terms used to describe articles. A second search using all identified keywords and index terms for 'dementia' , 'home healthcare' , and a series of outcomes such as 'acute healthcare utilisation' , 'continuity of care' , 'palliative care' or 'place of death' were then undertaken across five electronic databases, including OVID Medline, EMBASE, PsycINFO, Cochrane Library and CINAHL, from inception to September 2020. In the third search, the reference list of all identified articles was searched for additional studies. Search terms were used in combination with MESH headings, controlled vocabulary and free-text terms to cover the topics and detail (see Supplementary file). Two authors (PJC and LS) read the abstracts of half of the retrieved records to identify potentially relevant publications. These publications were marked as 'include' or 'uncertain' after the exclusion of irrelevant studies. A random 15% of selected records were independently checked by a second reviewer (JYL). The two authors then retrieved the full texts of identified studies and screened them according to the eligibility criteria. The final list of articles was checked by the three authors and any disagreements were discussed with the third reviewer (ELS) to reach consensus. We constructed a PRISMA flowchart to describe the selection process and a table containing excluded studies with the rationale for exclusion. References were managed and deduplicated by citation management software.

Data extraction
We extracted relevant data into a standardised table using Microsoft Excel. The table format was pilot-tested on three articles to ensure consistency and was approved by the research team. Extracted data included country, time, study design, data source and collection, research questions (aims), participants, content of interventions, comparison, and outcomes. Information from included studies was extracted by PJC and JYL independently and checked for accuracy by RM. Discrepancies were discussed with ELS to reach consensus.

Quality assessment
The Critical Appraisal Skills Programme (CASP) toolkit was used by PJC and LS to appraise the quality of the included studies [37]. Studies were rated as strong, moderate or weak based on the following components: study design, data collection method, bias of selection and outcome measurements, intervention integrity, confounding factors, appropriate analysis and implication for practice. Discrepancies were discussed with RM to reach a consensus.

Data synthesis
We narratively described the effectiveness of HHCrelated outcomes. We used an adapted multilevel framework of Ferlie and Shortell [38] to synthesise core properties of HHC. The empirically derived model was used for summarising and classifying the various characteristics related to end-of-life care provision in care homes across countries and focus on three levels of the implementation: macro-such as national policy, legislation, or financial provision; meso-such as training or service model/framework; micro-effects or components in an individual programme [39]. We were unable to conduct a meta-analysis because of clinical and statistical heterogeneity across studies. Results were set out in a table with data reported from the included study (e.g. p-values).

Study design
· Any type of trials · Uncontrolled before and after studies · Interrupted time series · Observational studies · Qualitative study · Exclude Reviews, case reports, commentaries, conference abstracts, qualification theses, and non-English articles

Results
Of all retrieved studies, five met the inclusion criteria. The selection process is illustrated in the PRISMA flow chart (Fig. 1).

Studies' characteristics
We identified five studies; one prospective cohort study [40], two retrospective cohort studies [41,42], and two case-control studies (Table 2) [43,44]. No clinical trials or qualitative studies were included. In total, 4493 participants with dementia were included in these quantitative papers. Three studies (3936 participants, 87.6%) were conducted in the USA [41,42,44], and one each from Italy [40] and Japan [43]. All studies included both males and females and most participants were over 80 years old. The HHC provided in these five studies were all forms of home-based primary care.

Micro-level
Four studies mentioned that the HHC was provided by the multidisciplinary team [40,[42][43][44]. HHC in the other study emphasised advance care planning in a nurse-led programme [41]. Only one study mentioned the duration of the HHC intervention [42], and no study reported the time between dementia diagnosis and the first HHC.
Two studies compared outcomes in people receiving HHC with those in nursing home care [40,42], and another study compared outcomes of treatments for acute events between the HHC group and the hospitalised group [43]. Four studies investigated outcomes related to end-of-life issues or palliative care [40][41][42]44], whereas survival and mortality rate was the main outcome outcomes in the other study [43].

Meso-level
Three studies mentioned coordination with social workers, non-clinical social care support and external specialists such as palliative care consultants or geriatricians in HHC [40,42,44].

Macro-level
We did not find information on national policy or financial schemes that influence HHC for PLWD in the selected articles.

Quality of the evidence
No study received a high-quality rating ( Table 2). The main reasons for low scores were authors not taking account of confounding factors appropriately in the analysis and the absence or insufficiency of follow-up period because of study design.

Impact of HHC on end-of-life outcomes for PLWD
Results identified from included papers were summarised in Table 3. All the observed impacts were at the micro-level.

Acute healthcare utilisation in the last year of life
Three studies have reported results regarding our primary outcomes of interest [40][41][42].  [42]. In terms of specific procedures, fewer people in the HHC group were given life-supporting therapies such as oxygenation or feeding tube than those in the nursing home group at the end-of-life. Jennings et al. described the effects of an HHC programme in California, which specifically focused on advance care planning including Physician Orders for Life-Sustaining Treatment (POLST), on end-of-life care in PLWD [41]. A higher proportion of PLWD who received HHC with the completion of POLST experienced hospitalisations in the last 6 months of life compared with those receiving HHC but without a POLST. No significant difference was found between the length of hospital stay, intensive care unit admission, and frequency of emergency department visits between the two groups.

Hospice palliative care use
In Toscani's study, physicians in the nursing home group were more likely to consider/make decisions that focused on reducing suffering or on improving quality of death for PLWD than physicians providing HHC [40]. Two studies in the USA reported a higher percentage of HHC recipients who used hospice or were referred to hospice care before their death compared with nursing home residents or the control group [42,44]. In California, Jennings et al. demonstrated that PLWD who received HHC with a completed POLST were more likely to have hospice care discussion or consultations, use hospice care when they died and died at home than HHC recipients who did not have a completed POLST [41].

Advance care planning
Only Mitchell's study indicated that fewer HHC recipients had advance directives before death than did nursing home residents, despite a higher proportion of HHC recipients having a life expectancy of less than 6 months [42].

Place of death
Each study in the USA and Japan reported this outcome [41,43]. A higher proportion of PLWD in HHC with POLST died at home than those in HHC without POLST. Information regarding PLWD's preferences of place of care/ death is not found.

Discussion
To the best of our knowledge, this is the first review that explored the association between primary HHC and endof-life outcomes among homebound PLWD who are at a high risk of mortality [24,42]. The comparison groups and outcomes measured in the five included studies vary, and we found the results were heterogeneous and very limited to conclusively examine the effects of HHC on end-of-life outcomes. The existing literature suggests that HHC may be associated with an inverse risk of acute healthcare utilisation in the early and late periods (e.g. 90 vs 15 days before death) of the disease trajectory towards the end-of-life in PLWD. HHC seems to increase referrals to hospice palliative care, whilst advance care planning may influence the effects of HHC on end-of-life outcomes. HHC providers' difficulty in making treatment decisions for PLWD at the end-of-life may require further training and external support. The coordination between HHC and social care is important but not well implemented and investigated.

Micro-level
The differential effects of HHC on acute healthcare utilisation among PLWD in the early or late period imply different care needs at various stages in the disease trajectory among PLWD, for which distinct components and models of HHC service may meet their needs better. A systematic review showed that home-based primary care mostly reduces the events and length of hospitalisation [35]; however, this effect was observed within 1 year after HHC but not followed up to the recipients' death. In addition, we have not been able to clarify the influence of the duration, continuity, or intensity of HHC from the current literature. Among homebound PLWD approaching the end-of-life, identifying care needs and treatment decisions are complicated because the person may be unable to express their care preferences [40]. Multidisciplinary approaches in HHC may be beneficial to PLWD towards the end-of-life [40,[42][43][44], but none of the selected studies have examined the effectiveness of skill mix across various professions or quantified the contribution of each discipline. Given that people with dementia may lose the capacity to make decisions, advance care planning in HHC may substantially influence end-of-life outcomes. A recent review showed that advance care planning for PLWD was associated with decreased hospitalisations and increased concordance between prior preferences and actual care received [45]. In primary care settings, key barriers of professionals to conduct advance care planning for PLWD includes the time restraints of medical staffs; the insufficiently trusted relationship between PLWD and medical staffs due to infrequent contact; staffs' attitude, knowledge, skills, and moral considerations towards talking end-of-life issues and death; and inadequate reimbursement [46,47]. However, the context of home visits is preferred by PLWD and family caregivers because it is a trusted environment to have care plan conversations addressing not only medical but also non-medical preferences, such as valued abilities and activities, family support and relationship, and place of care/ death [46,48]. Both interactive training and clinical practice of conducting advance care planning during HHC can be facilitated by involving interdisciplinary professionals, such as nurses, social workers, or care managers [46,48].

Meso-level
Providing training programmes and seamless palliative care support from external specialists to HHC practitioners may improve the capability of primary end-of-life care in the home setting [49,50]. HHC physicians were less likely to initiate palliative care for PLWD [40], even though eventually the specialist palliative care or hospice referral is higher for people in the HHC group than those in the nursing home or control group [42,44]. Primary healthcare workers may more likely to consider that palliative care is not meaningful in PLWD than in people with other life-limiting illnesses [51] or only acknowledged its benefit in terminal care, so they refer the people to the service late [49]. The resistance of timely palliative care approaches provided by HHC professionals, such as symptom management and initiating advance care planning discussion, can be further overcome through education, skills training, and discussion of moral dilemmas [47,49,52]. Further service commissioning and integration between HHC teams and external specialists such as geriatricians or palliative care may contribute to PLWD living and dying well at home [23].
Good-quality HHC requires strong coordination between health and social care services to achieve better end-of-life outcomes. A UK cohort study found that the need for social care services increased among PLWD towards the end-of-life [23], and the lack of social care support at home may lead to a higher risk of acute healthcare use [53]. In the selected studies, only one US study assessed the use of social services in the HHC setting and found such services were not used to its full potential, showing that the coordination between health and social care may be an area for improvement [44]. The barriers of care integration for PLWD towards the end-of-life may include the conflicting relationships and communication between disciplines and settings, lengthy referral processes, minimal care planning, diffuse responsibility, and fragmented reimbursement system [50,54]. At the local services, organising the cross disciplinary network between healthcare and social care sectors, increasing the communication and establishing the practice guidelines with shared goals, peer supporting to formal and family caregivers, and highlighting good practices can improve the quality of integrated care [50].

Macro-level
Reform of national policy and payment schemes would be vital in promoting better care for individuals [55,56], or in building up interdisciplinary collaborations and delivery systems between health and social care services [50]. However, contextual information, including descriptions of the related policy and payment schemes in each country, was not mentioned in the selected papers.
To understand this context better, we summarise some international examples through a brief policy review and discussion in our research network (Table 4), including those from where the five papers of this review originated (USA, Italy and Japan) and the top-ranked countries in related regions in the Quality of Death Index Report [56], such as Australia, Canada, UK, and Taiwan. The key lessons from the policy comparison are building up person-centred continuous care at home, with a seamless connection between primary care and palliative care throughout the disease trajectory, quality-and valuebased payment for interdisciplinary collaboration, as well as comprehensive networking with coordination between health and social care sectors [26][27][28]31].

Strengths and limitations
We comprehensively and systematically searched the literature by applying a wide range of search terms including synonyms of HHC and types of HHC programmes. The identified studies were rigorously checked by quality assessment tools. The international members of our research team provided insights and interpretation. This study has several limitations. Firstly, the majority of studies evaluating the effects of HHC had short followup periods, often less than 12 months after the initiation of HHC [34,63,64]. Data about end-of-life outcomes that occurred within the final years before death were neither investigated nor analysed separately, leading to fewer papers meeting our selection criteria [64,65]. Secondly, none of the included studies were rated as high quality in critical appraisal. Outcomes were heterogeneous, and it was not possible to pool data and perform a meta-analysis of the findings. In addition, the lack of information about the duration, intensity or components of HHC interventions meant that we could not explore the 'dose-response' relationship between characteristics of HHC and outcomes [66]. Finally, the small sample size in the HHC group and lack of random sample selection may lead to poor external validity [67].

Implications for research and practice
The sparse evidence in our review suggests that the role of primary HHC may have been overlooked as a key service that could deliver better quality end-of-life care for PLWD. HHC services may vary widely across countries, and details of the components and contextual factors of HHC and how they are implemented are important to evaluate the effects of complex interventions, which were not reported in the included studies [68]. For future studies, it is essential to better understand the effective components of HHC, such as advance care planning, and the mechanism of how they influence end-of-life outcomes for PLWD. Conducting a randomised trial or a prospective cohort study with a longer follow-up period of HHC would be challenging in practice [24,35]. A more pragmatic, hybrid paradigm incorporating quality improvement or service evaluation may be more useful and realistic to conduct [69]. Large real-world datasets containing whole population samples, with complete follow-up are also good sources to evaluate HHC programmes throughout the disease trajectory, though appropriate and robust methodologies should be applied [67,[70][71][72]. This would reduce selection bias and prevent missing data due to the attenuation of study cohorts. Current metrics of care quality, which were developed for individual diseases, are not holistic and do not capture more value-based dimensions such as continuity of care or level of care integration [73].
Regarding the clinical practice, advance care planning, comprehensive geriatric assessment, and a palliative approach which focuses on patients' care preferences and improving quality of life should be emphasised in HHC for PLWD [15,74,75]. Stakeholders should enhance education for HHC users and providers, strengthen the training of the interdisciplinary workforce, and promote a service model supported by external professionals including social care or even telemedicine during the pandemic to meet complex end-of-life care needs in PLWD [23,26,49,74]. Policymakers are encouraged by the experience of national policy and payment scheme reform in some countries to build up the continuously integrated care framework that improves the synergy of various services [26,27,74].

Conclusion
This review added the new knowledge that different care needs at various stages in the disease trajectory towards the end-of-life among PLWD urge more integrated services with effective components to respond to their demand better. Effects of advance care planning, multidisciplinary approach, integration between health and social care, and coordination between primary HHC and specialists' support in local healthcare networks for better continuity of care at home should be emphasised in clinical practice and policy-making. Population-based large databases may provide opportunities to examine more clearly the long-term impact of HHC and its synergy with other clinical services on end-of-life outcomes in a longitudinal study design.
Additional file 1: Supplementary file. Search strategy and terms in electronic databases.