What is the effectiveness of the support worker role for people with dementia and their carers? A systematic review

Background Dementia is progressive in nature and the associated functional decline inevitably leads to increasing dependence on others in areas of daily living. Models of support have been developed and implemented to assist with adjusting to living with memory loss and functional decline; to navigate the health and aged care system; and to access services. We undertook a systematic review of international literature on key worker type support roles to identify essential components and ascertain how the role can be best utilised to assist community-dwelling people with dementia and their carers. This review of support roles is the first to our knowledge to include both quantitative and qualitative studies and all models of support. Method A systematic review of studies written in English and published between January 2003 and December 2014. Data sources were Medline, PsychInfo and CINAHL, internet, expert consultation and reference lists of included studies. After screening articles to ensure that they reported on a key worker type support role, involved carers and or people with dementia living at home and removing duplicates, eligible papers were appraised and evaluated. Results Thirty six studies were eligible for inclusion in the review. Eligible studies were divided into type of support roles and study type. The heterogeneity of included studies and high risk of bias made a meta-analysis inappropriate and it was therefore difficult to draw overall conclusions. However, essential components shared across support worker models that demonstrated a positive impact on carer burden and improved quality of life included: long term intervention, face to face contact, individualised education and support based on needs, multi-disciplinary teams, collaborative input, health/clinical background of support workers, ongoing follow up and inter professional and inter-sectoral collaborations. There was a lack of studies assessing cost-effectiveness. Conclusions Studies that include a high quality evaluation of holistic, tailored models of support that identify which components of support produce the most valuable outcomes to assist people with dementia and their carers and families to continue to live meaningful lives are needed. There is also a need for a cost effectiveness evaluation of support worker roles. Trial registration PROSPERO international prospective register of systematic reviews: PROSPERO 2014 CRD42014013992. Electronic supplementary material The online version of this article (doi:10.1186/s12913-016-1531-2) contains supplementary material, which is available to authorized users.


Background
Dementia is progressive in nature and leads to a decline in cognitive abilities. In the early stages of the disease, people with dementia may experience difficulties undertaking routine tasks, such as driving, shopping and managing their finances. As the disease progresses there may also be difficulties with self-care, bathing, eating and communication [1]. As a consequence of these difficulties many people with dementia who live at home are supported by informal carers.
The World Health Organisation has estimated that over 46 million people worldwide are living with dementia and that this number will increase to 74 million by 2030 [2]. This rapid increase in prevalence of dementia is expected to pose a substantial challenge to health, aged care and social policy and substantially increase the number of informal care givers [2].
Providing support for a person with dementia can lead to a decline in physical and mental health and can also impact employment and education prospects, finances and participation in social and community life [3]. Disease progression also leads to increasing difficulty in continuing to meet the needs of a person with dementia in the home setting and as a result care has progressively shifted from the private to the public setting through the introduction of home based support services provided in the community [4].
Despite these formal services being available, the use of services by carers (including respite) is quite low [5]. This has been attributed to the services being difficult to navigate, not meeting carer/care recipient needs, and beliefs that service use would result in negative outcomes for the care recipient [5] The non-use of formal services has also been associated with high levels of depression among carers [5].
In order to address the difficulties and the stresses associated with living with cognitive impairment models of support to assist adjusting to living with memory loss; to navigate the health and aged care system and to access services and information have recently been implemented both in Australia and internationally [6]. Support worker type roles include case managers, care workers, counselling support workers and multi-team integrated care. In the United Kingdom there is also the admiral nurse role, that utilises a specialist mental health nurse, and in Australia the role of the key worker, designed to provide support to people with younger onset dementia.
Currently, there is a lack of high level evidence regarding the overall effectiveness of these support roles for people with dementia and their carers [7]. Previous systematic reviews of dementia support worker roles have focused on case management roles [8][9][10][11][12][13][14] with only one extending this to include other support models (integrated care and consumer directed care) [15]. Our comprehensive systematic review of the international literature of models of support for community-dwelling people with dementia and their carers aims to develop an evidence-informed national approach by health and aged care service providers, government and consumers to support people with dementia, their carer's and families. Our extensive systematic review of the international and national academic literature of models of support for community-dwelling people with dementia and their carers is the first to our knowledge to include both quantitative and qualitative studies and all models of support.

Methods
The review questions were: 1. What are the essential components of the key worker type model of support for people living with dementia and their carers 2. How can the role be best utilised to assist communitydwelling people with dementia and their carers?

Data sources and search strategy
Literature indexed in the scientific databases MEDLINE, CINAHL and PsychINFO, was searched. Google Scholar was used to identify studies that did not appear in the scientific databases. The search conducted in EBSCO MEDLINE, CINAHL and PSYCHOINFO used key words and subject headings limited to English language published between 2003 and December 2014. Subject headings included: ("Alzheimer disease" OR "Dementia" OR "Dementia, Multi-infarct" OR "Fronto-temporal Dementia" OR "Dementia, Vascular" OR "Lewy Body Disease") OR (keywords "dementia" OR "Alzheimer's"). Key words included: "key worker", "link worker", "support worker", ("case management" as key word OR "Case management" as subject heading), "case manager", ("nurse clinician" as key word OR"Nurse clinicians" as subject heading), "clinical nurse consultant", "admiral nurse", ("patient navigation" as key word OR "Patient Navigation" as subject heading), "navigator", "nurse specialist" then all of these searches were combined with OR. Finally, the combined search of roles (i.e. key worker etc.) was added together with the combined search of dementia using AND to produce the final search.

Inclusion and exclusion criteria
To ensure that our review was relevant to current practice we included research papers written in English language and published between January 2003 and December 2014. All study designs of articles that examined key worker type support roles for people with dementia living at home and carers of people with dementia living at home were appraised.
The key worker type support roles included were: case managers, care managers, support worker, admiral nurse, link worker, key workers, counselling roles and team based/multi-agency/integrated care roles.

Outcomes
Evaluation of key worker type roles; reduced carer burden; improved quality of life, improved symptom severity for people with dementia and reduced institutionalisation rates.
We excluded articles published prior to 2003 and not written in English. We also excluded articles that were case reports, editorials and opinion pieces rather than reports of an intervention or description of a support worker model.

Study selection process
All evaluations, descriptive and comparative studies of the utilisation or role of key worker type support models assisting community-dwelling people with dementia and their carers were screened independently by two authors. Initially, the title and abstract of the all indentified studies were screened for eligibility. An eligibility instrument was used to guide the decisions (see Additional file 1).

Potential effect modifiers and reasons for heterogeneity
As the types of studies included in the systematic review were heterogeneous, and after consideration of the risk of bias, they were not suitable for inclusion in a meta-analysis. Therefore we undertook a comparison of the studies and their outcomes synthesising the data into tables according to types of support worker roles and study types.

Quality assessment
One of the purposes of conducting research is to provide evidence of efficacy, however, not all evidence is considered equal [16]. Consequently, we considered the levels of evidence before summarising the information. Two authors independently appraised the quality of all included quantitative and qualitative articles. Where there were discrepancies in appraisal, papers were re-read by both assessing authors and consensus reached through discussion.
Acceptable levels of information were decided using the NHMRC Grade levels (see Table 1) and the Cochrane and CASP Risk of Bias Tools to guide decisions.

Risk of bias
To assess the presence/risk of bias of the studies we identified we used the Cochrane Risk of Bias assessment tool for RCTs and non-randomised or quasi-experimental studies. For observational studies, we adapted the Critical Appraisal Skills Programme (CASP) checklist for cohort studies and the CASP checklist for case control studies to determine risk of bias.

Data extraction
All identified studies were screened for eligibility based on titles and abstracts using an eligibility assessment tool to determine if the study utilised or discussed a support worker type role, if the participants had dementia or cognitive decline or were carers of people with dementia or cognitive decline and whether the participants were community dwelling/living at home or were carers of community dwelling people with dementia or cognitive impairment (see Additional file 1).

Data synthesis and presentation
The selected studies/papers were categorised into type of role and study type. In most cases, but not all the studies/ papers were mutually exclusive to their categories. Country of origin, year of study and whether the study was registered was recorded for each study/paper. Electronic PDF versions of all eligible studies were retrieved prior to undergoing a critical appraisal. No attempts were made to contact authors for additional information.
Duplicates were removed, and titles identified in the electronic search were read, to identify those that were relevant. Abstracts were reviewed, and where they were identified to meet the inclusion criteria, the full publication was obtained and assessed for eligibility.
Two researchers screened records for inclusion in the review using the GATE framework tool to undertake a critical appraisal of the quantitative studies [17] (see Additional file 1). The schedule derived from this framework considered: population, exposure and comparison groups, outcomes, time, results and applicability (generalisability, feasibility and relevance, applicability and potential harms [17].

Qualitative data
Qualitative studies exploring the experience of those involved in interventions and evaluating factors that shape the implementation of interventions have an important role in ensuring that systematic reviews are of maximum value to policy, practice and consumer decision making [18][19][20]. Therefore, we also included a synthesis of the qualitative evidence in this review. As many of the studies/papers were qualitative Greenhalgh & Taylor's paper [21] and Britten & Pope's work on synthesising qualitative studies were drawn upon to appraise these qualitative studies [22]. This appraisal considered: if the paper described an important problem and if the authors answered the question they set out to, methodological approach (were methods appropriate, setting, participants, recruitment, aims, recruitment bias, researcher perspective, interview schedule design, data collection, recording and transcription, data analysis, validity and reliability, if the results were credible, conclusions and if these were justified and whether the findings were transferable to other settings (see Additional file 1).
The appraisal was undertaken by two independent members of the research team (DG and ER). Results of appraisals by the two independent researchers were compared and differences resolved through discussion and revisiting the criteria associated with each of the critical appraisal tools. Consensus was achieved in all cases.

Risk of bias
Qualitative studies were critiqued according to Greenhalgh & Taylor's [21] and Britten & Pope's [22] frameworks for assessing/synthesising qualitative studies. This process examined the context, theoretical approach, categories, concepts and interpretation of each study.

Results
The search strategy identified 513 references (Fig. 1). After removal of duplicates 434 abstracts were examined for relevance and full text for 71 references were obtained for full screening. Hand-searching of reference lists of included articles yielded an additional 41 articles. In total 112 articles were assessed for eligibility, of which 36 articles were selected for data extraction and analysis.

Study characteristics
Of the 36 included studies, 24 were Randomised Controlled Trials (RCTs), eight were qualitative, two were mixed method, one was a case report and one was a cohort study. The studies were conducted in the United States of America (n = 16), Europe (n = 5), United Kingdom (n = 8), Hong Kong (n = 3), Australia (n = 2), Canada (n = 1) and one was conducted across the United Kingdom, United States of America and Australia (n = 1).
The majority of studies evaluated counselling support roles. The rest evaluated support worker (including key worker, link worker, Admiral Nurses), case manager, team-based/multi-agency/integrated support roles, and care manager roles.

and 5).
Level II evidence -randomised controlled trials Outcomes of study and effective/non effective components of the model The four RCTs utilising case management models evaluated the roles impact on outcomes for the carer and person with dementia (Table 2 below).
The two RCTs conducted by Chien and Lee [23,24] with intervention periods of 6 months produced significant outcomes for people with dementia and carers. These outcomes included: reduction in carer burden and improvement in quality of life]; and reduced institutionalisation rates at 12- [23] and 18-months [24] post intervention. Additionally people with dementia showed improved symptom severity at 6-months [23] and 18-months [24].
The remaining two RCTs with intervention periods of four [26] and 12-months [25] found no significant differences in carer and person with dementia health or social outcomes measured in any of the follow-up assessments (4, 6, 12-months) [25,26]. However, Lam et al. did show a significant increase in family carers of people with dementia seeking external support at both 4 and 12-months. [26]. Jansen et al. indicated that the lack of significant results may have been attributed to either the intervention being offered too early or it lacking the intensity or duration to achieve a change in outcomes [25].
The interventions showed variance in length and mode of support as well as the role and qualification of the case manager. The components in the case manager roles in the two RCTs producing significant results included: a 6-month intensive intervention; input from a multi-disciplinary committee; training of the case manager; clinical backgrounds (nurses as case managers); collaborative care; continuity of care (same case manager); structured needs assessments and individualised education and support programs for each participant.

Risk of bias
Overall the methodological quality in three of the four RCTs investigating case manager roles was high [24][25][26] (see Table 6).

Mixed methods evidence
Outcomes of study and effective/non effective components of models Mixed method evaluations of a case management model consisted of quantitative client data and in-depth interviews [27] and a four phase Delphi Survey and focus group [28] (Table 4). Iliffe et al., [27] showed that case management offered potential benefit to people with dementia, their carers and community based professionals through continuity of care by a named trusted individual that could act proactively to prevent a crisis. However, it was also shown that needs may be overlooked. Verkade et al., [28] found that the essential components of dementia case management were: information of the patients and their systems; support to the patients and their systems; coordination and monitoring of the care provided by others and to a lesser extent practical help. It is suggested that the appropriate way of offering case management is through a patient-centred approach and that successful case management requires that case managers be able to rely on a shared case management vision to give direction to day-to-day care provided in practice.

Risk of bias
The methodological quality of both studies was good in terms of design, recruitment and data analysis. Limitations of the studies included a small sample size [27] and an inability to generalise results to other countries [28].

Qualitative evidence
Outcomes of study and effective/non effective components of models Minkman et al., [29] undertook a qualitative case study analysis in order to describe and analyse a new approach in extensive case management programs concerned with long-term dementia care in the Netherlands (Table 5) [29]. It was found that the success for case management in long-term dementia care concern the expert knowledge of case managers; investment in a strong provider network and coherent conditions for effective inter-organisational cooperation to deliver integrated care. The failure factors were: distrust of the programme by local providers and competition for delivering care; inadequate or no structural funding; little or no involvement of primary care specialists; doubt about the added value of case managers; and not including patients without a confirmed diagnosis of dementia.

Risk of bias
The methodology was sound however only one case manager from each program was included in the interviews and no consumers were interviewed to determine their views.

Counselling support roles
Fourteen RCTs that evaluated counselling support type roles were identified (Tables 7 and 8).
Outcomes of study and effective/non effective components of models The studies which implemented counselling support type roles focused on a range of outcomes for carer and people with dementia (see Table 7).
Three RCTs [30][31][32] report on different outcomes from an intervention that provided enhanced counselling and support to carers over a 4 month period. Mittelman and colleagues [31] found that at the 5 year follow up after baseline differences were controlled for cares in the intervention group had significantly fewer depressive symptoms compared to controls. These effects were sustained for 3.1 years after baseline and after nursing home placement or death of the patient. A further report by Mittelman and colleagues [32] found that the intervention group had significant delays in nursing home placement when compared to controls. Gaugler et al., [30] then aimed to determine whether the intervention reduced the burden and depressive symptoms of carers during the transition to nursing home placement. It was found that nursing home placement itself reduced burden and depressive symptoms in carers for both groups but that the intervention resulted in significantly lower burden and depressive symptoms at the time of and after nursing home placement.
Two RCTs [33,34] reported on different outcomes from the Cleveland Alzheimer's Managed Care Demonstration. The aim of the demonstration was to evaluate the effect of a 12-month care counselling consultation (a multi-component telephone intervention) delivered within a partnership between a managed health care system and Alzheimer's Association during the 12-month study period. The intervention was shown to significantly decrease depression symptoms in carers and reduce strain in non-spousal carers [33], significantly reduce feelings of embarrassment and isolation and decrease 'difficulty in coping' due to memory problems in people experiencing memory problems or with a diagnosis of dementia [34]. Additional intervention effects were shown for people with more severe impairment. There was less direct impact of the intervention on health service utilisation (hospital, emergency department, physician) with significantly lower utilisation only occurring in services that provided that same types of assistance as the intervention.
A further three RCTs [35][36][37] were a part of the Resources for Enhancing Alzheimer's Caregiver Health (REACH) multisite research program. Each RCT implemented a different social and behavioural intervention targeting carers of people with dementia. Mahoney and colleague's [37] 12 month computer mediated interactive voice response system intervention involving counselling showed no significant effect for the intervention in reducing bother scores, depression or state anxiety scores. However, people who were wives or had low mastery scores at baseline did show a significant decline in bother scores, depressive symptoms and anxious complaints as a result of the intervention. Burns et al., [35] reported on parallel simultaneous interventions (behaviour care versus enhanced care) over a 24-month period and found that carers who received either intervention showed significant improvements for bother associated care recipient behaviours. However, those who received the behaviour care component only, compared with those who also received the stress-coping component    information, support and counselling, coordination of the care provided, and to a lesser extent practical help. A patient centred approach was found to be one of the key aspects. • Essential preconditions were: vision, care relationship, structured methodology, integration of case management into the health care chain, and the case manager's level of training/expertise.
It is recommended that the essential components and preconditions be used as a basis for developing minimum quality criteria for case management in people with dementia to enhance quality of care and reduce undesirable differences.
Note: Assessment of bias was not relevant for the mixed method studies as their study design did not meet the criteria for the risk of bias tools; instead the methodology was critiqued according to Greenhalgh & Taylor's [21] paper and Britten & Pope's [22] (enhanced care), had significantly worse outcomes for general wellbeing and a trend toward increased risk of depression. Eisdorfer and colleague's [36] combined family therapy and technology intervention which ran for 18-months reduced depressive symptoms in carers at 6-months. However at the 18-month follow-up this result was only sustained for Cuban American husband and daughter carers indicating that the intervention has differing effects according to ethnic group and carer-care recipient relationships [36]. The remaining six RCTs were all independent and involved counselling support roles singularly [38] or combined with: psycho-educative programs [39]; structured education [40]; exercise training and behavioural management techniques [41]; telephone-based education [42]; and care consultation [43]. The counselling roles in all six RCTs were focused on supporting the carer.
One study investigated the effects of a 10-week carer telephone support group intervention on cost of care for the care recipient and found a significant short term cost saving benefit at 6-months compared to usual care [42]. However this result was not maintained at 1 year.
Two of the RCTs with counselling roles implemented for 12-months [43] and 2 years [38] looked at nursing home admission as a primary outcome measure. Brodaty and colleagues [38] found, over an average of 5.4 years, no differences in nursing home placement or mortality between groups. Similarly Fortinsky and colleagues' [43], intervention did not lead to a statistically significantly lower rate of nursing home admission, although there was a trend in favour of the intervention group during the 12-month study period. Fortinsky and colleagues [43] also found no significant intervention effects on the secondary outcomes of carer self-efficacy, depressive symptoms or burden at 12-months.
The remaining three RCTs showed that: a 4 month psycho-educative program significantly reduced carer reactions to behaviour problems at the 4 month posttest [39], a structured intervention (on two occasions) reduced frequency of problem behaviours particularly delusion and psychic agitation in people with dementia at 12-months [40] and a 6-month exercise training program combined with teaching carers behavioural management techniques improved physical health in    [41]. There were no significant effects shown of any of the other outcomes for these three RCTs as listed in Table 7.
The heterogeneity in interventions, variance in outcomes measured and conflicting results meant that the effective and non-effective components of each intervention were unable to be quantified.

Risk of bias
None of the identified RCTs that analysed counselling support roles met all the criteria for low risk of bias. In all 14 of the identified RCTs evaluating counselling support roles the risk of bias was unclear or high in the majority of categories. Therefore the results cannot be considered as the higher level-two high quality evidence (Table 6).

Care manager roles
Five RCTs that evaluate interventions trialling the care manager role were identified (Tables 9 and 10). Three of the studies report on different findings from the same RCT [44][45][46].

Level II evidence -randomised controlled trials
Outcomes of study and effective/non effective components of models Four of the five study's care manager interventions ran for a 12-month period and were specifically designed for people with dementia and their carers in a primary practice setting [44][45][46][47]. The further RCT care management intervention was implemented for 15-months and targeted people with dementia and carers already integrated within traditional case management systems in eight different countries [48]. Outcomes measured are shown below in Table 9.
The three RCTs reporting on the same care manager intervention revealed higher quality of care in regard to dementia guideline recommendations [46], that healthcare organisation care managers were essential for dementia care quality improvement and that additional coordinated interactions with primary care and community agency staff yielded even higher quality of care for people with dementia [44]. However, no significant cost offsets for the intervention were found [45].
The remaining two care manager RCTs produced some significant health outcomes, mainly for carers during the intervention period. This included: significant improvements in stress at 12-months (but not 18months) [47] and significantly improved carer stress, well-being and endurance potential at 15-months [48]. Callahan and colleagues [47] also showed their trial of care management to significantly reduce behavioural and psychological symptoms of dementia during and post-intervention (18-months). However neither intervention showed significant improvements in depressive symptoms, activities of daily living, cognitive status or kind and frequencies of behaviours for people with dementia during or post-intervention [47,48].
Components of care manager roles that were successful and consistent across interventions included: a 12-month intervention; collaborative care planning; education for people with dementia/their carers; structured assessments; and multi-disciplinary input/collaboration.

Risk of bias
The RCTs undertaken by Callahan et al., [47] and Vickrey et al., [46] showed low risk of bias for all assessment indicators indicating high methodological quality (Table 6). Chodosh et al., [44], Duru et al., [45] and Vickrey et al., [46] did not state in any of their three papers whether there was any blinding of personnel or outcome measures. The intervention by Specht et al., [48] had the lowest methodological quality with only category being rated as low risk of bias (random selection generation, selection bias) ( Table 6).

Level II evidence -randomised controlled trials
Outcomes of study and effective/non effective components of models      • Five overall categories emerged which summarised the major differences between the services, encompassed the views of all groups and provided a rationale for why the specialist service was better than the standard service. These categories were structure and function; responsiveness; control and autonomy; building relationships; and reducing carer burden. • The specialist service demonstrated greater flexibility and responsiveness to the particular needs and circumstances of service users and family carers, who were encouraged to take part in routine decision making and activities. • By sharing responsibilities, the specialist service helped reduce carer stress and prevent crises. • These outcomes depended on the configuration of the service, including multidisciplinary health and social services input, care worker autonomy and independence, continuous reassessment of clients' circumstances and preferences and the capacity to develop long term relationships, through care worker continuity. The standard service, which used a task oriented approach, lacked these characteristics.
This qualitative study provides evidence of the benefits of a specialist multi-agency home support service over standard home care in the opinions of service users, carers and care workers.
Note: Assessment of bias was not relevant for the qualitative studies as their study design did not meet the criteria for the risk of bias tools; instead the methodology was critiqued according to Greenhalgh  Eloniemi-Sulkava and colleagues, [49] aimed to determine whether community care of people with dementia could be prolonged with a 2-year multi-component intervention program that included a family care coordinator, a geriatrician, support groups for care givers, and individualised services [49]. The effects of the intervention on total usage and expenses of social and healthcare services were also analysed. At 1.6 years, a larger proportion of people in the control group were in long-term institutional care when compared to the intervention group however, at 2 years, the difference was no longer statistically significant. The intervention did lead to a reduction in use of community services and expenditure however when the intervention costs were included this result was no longer significant.

Risk of bias
There was a high risk of bias with the majority of bias categories being rated as high or uncertain (Table 6).
Level IV evidence -observational descriptive (case series/case report) Outcomes of study and effective/non effective components of models Stevenson et al., [50] used a naturalistic study design to evaluate the impact of a multi-agency enhanced community assessment and support team that provided coordinated care management services to older people with a confirmed diagnosis of dementia [50]. The multiagency community team comprehensively assessed and supported at home patients with dementia who previously would have been referred to the local psycho geriatric admission ward and day hospital, and consequently reduced utilisation of these hospital facilities. Health workers, voluntary agencies and carers were positive about the service.

Risk of bias
Overall the study methodology was poor with no clear information provided on how the data was collected or analysed (Table 6). In addition, no comparative statistical analysis was performed and there was no randomisation of participants meaning that any changes observed cannot be solely attributed to this model of care under evaluation.

Qualitative evidence
Outcomes of study and effective/non effective components of models Rothera et al., [51] used a qualitative approach and soft systems methodology to compare a specialist multiagency home care service for older people with dementia to a standard service in a demographically similar area. The specialist multi-agency home support service demonstrated greater flexibility and responsiveness to the particular needs and circumstances of service users and family carers, who were encouraged to participate in routine decision-making and activities. By sharing responsibilities, the specialist service helped reduce carer stress and prevent crises. These outcomes depended on the configuration of the service, including multidisciplinary health and social services input, care worker autonomy and independence, continuous reassessment of clients' circumstances and preferences and the capacity to develop long-term relationships, through care worker continuity. The standard service, which used a taskorientated approach, lacked these characteristics.

Risk of bias
The methodology used was sound but given the design the results are susceptible to researcher bias, acquiescence bias, inconsistency in the identification of outcomes and false attribution of causation.

Key worker/Link worker/Admiral Nurse/Clinical nurse consultant roles
The seven studies whose interventions involved support worker roles included one observational analytic cohort and six qualitative studies (Tables 14 and 15). Four of the seven studies evaluated the Admiral Nurse role [52][53][54][55] and the remaining three job roles included key workers [56], link workers [57] and a clinical nurse consultant [58].

Level IV -observational study designs: analytic studies (Cohort Studies)
Outcomes of study and effective/non effective components of models Woods et al., [52] aimed to provide an evaluation of the outcomes association with the Admiral Nurse Service for both the family carer and the person with dementia in comparison to conventional multi-disciplinary community mental health teams for older people in similar areas [52]. Both services resulted in lower distress scores for carers of people with dementia over an 8-month period however carers receiving the specialist Admiral Nurse Service showed a greater reduction in anxiety and insomnia. Outcomes for people with dementia (in terms of institutional placement) were no worse in the Admiral Nurse group, despite the carer focus.

Risk of bias
The methodological quality of the trial was good, however given the trial was not randomised and therefore unidentified differences between interventions would have existed and may have influenced the findings.

Qualitative
Outcomes of study and effective/non effective components of models  Complexity of carer's situation; Admiral Nurses' perception of their specialist role; mode of referral and information received and cross-functional working/ trust-wide provision. • The most significant factor that came out of the data was the perceived complexity of the presenting situation, one in which both the circumstances of the carer and the person with dementia were considered. It was also shown that decisionmaking was influenced where significant risk was identified to either party.
This qualitative study found that the decision to offer the Admiral Nursing service to carers was influenced not only by perceived need but also by the nurses feeling professionally responsible for perceived gaps in service provision. It is suggested that Admiral nurses may need to limit their involvement with carers in line with the service aspirations and become more confident in promoting on-referrals and discharging individuals from the service. It is concluded that it does not appear practical for Admiral nurses to provide a specialist service that meets the needs of all the carers who require support. • Main outcome was development of a specialist nursing competency framework. • The competency framework was made up of a set of eight core competencies: therapeutic work/interventions; sharing info about dementia and carer issues; advanced assessment skills; prioritising work load; preventative and health promotion; ethical and This qualitative study developed a competency framework that reflected the needs of the service, was owned by the majority of practitioners and project commissioners which had a positive impact on implementation. It is suggested that the competency framework will enable Admiral Nurses to demonstrate their level of specialist practice as individuals and as a service while also promoting

Admiral nurse role
Three studies undertook qualitative analyses of the Admiral Nurse role [53][54][55]. The studies focused on different aspects of the role and service and therefore it was not possible to synthesise their findings. Data collection techniques included structured interviews [53]; emancipatory action research [54] and semi-structured interviews [55]. The studies showed that: The desire of Admiral Nurses to fulfil a case management role while attempting to provide a service that is of a specialist nature and of limited capacity generated tension in the role. It was determined that it is not practical for Admiral Nurses in the UK to provide a specialist service that would meet the needs of all those carers who require support and that in order to maximise potential there is a need to further define the services' remit and enhance its level of specialism [53] The development of a specialist nursing competency framework for the Admiral Nurse role in the UK to demonstrate the level of Admiral Nurses specialist practice and core competencies of the role. These were: therapeutic work (interventions); sharing information about dementia and carer issues; advanced assessment skills; prioritising work load; preventative and health promotion; ethical and person centred care; balancing the needs of the carer and the person with dementia; and promoting best practice [54] The triadic relationship between the carer, carerecipient and the Admiral Nurse was encompassed under 'negotiating the balance' as an overarching process. The findings emphasised the importance of exploring the perspective of all three members in order to improve the quality of support that is provided [55] The remaining three qualitative studies evaluated three different support worker roles: a key worker [56], link worker [57] and clinical nurse consultant [58]. Data collection methods included grounded theory [56], an empirical investigation [57] and participatory action research [58]. There was qualitative evidence for: Positive outcomes in the carer/key worker relationship to be linked to the quality of the relationship and involve the carer and professional care worker actively including and working with the person with dementia [56] Link workers to perceive and experience many different influences on decisions made about family caregiving. A shared approach to care was found to be vital in decreasing burden among family members and that due to their close relationship and knowledge of families, multicultural workers can offer an important perspective that is invaluable in informing the provision of carer education and support within CALD communities [57] A dementia Clinical Nurse Consultant to show benefit to those living with cognitive impairment and or/their carers and families. The importance of the relational aspect of the role including face-toface contact and opportunities to explain their needs and concerns in a time and manner of their preference were found to be integral to the person with dementia and carer's ability to adjust to change [58]

Risk of bias
The six qualitative studies discussed above all had limitations with their methodological design. Some of this was due to a lack of clarity around how the data was collected and analysed and some was related to more serious issues such as researcher bias, recruitment bias, limited data analysis methods, low sample sizes or the utilisation of the wrong methodological approach. An overall limitation of these qualitative studies is that the results are quite specific to the population and setting under investigation and thus cannot be generalised to other settings or communities.

Discussion
Our systematic review of the international literature on models of support for people with dementia and their carers revealed 36 papers which were evaluated in this review.
Systematic reviews of dementia support worker roles have been undertaken previously. These reviews have primarily focused on case management roles and not any of the other support models of care identified in this review. These reviews have investigated case management's impact on: health care costs and resource utilisation [8]; general wellbeing [9]; consumer and client outcomes [10]; risk of long-term care placement [11]; clinical outcomes and utilisation of resources [12] and its potential for people with dementia [13] and barriers to implementation [14]. One systematic review by Low et al. [15] looked at outcomes for older adults including those with dementia from three different models of care: case management, integrated care and consumerdirected care. This review builds on previous reviews, as it is the first of its kind to analyse the essential components of multiple key worker type support roles for people with dementia. The results of this review have the potential to inform future research and practice through the incorporation of these essential features into future trials or current support worker roles operating in the community. The results from our comprehensive systematic review of support models for people with dementia and their carers provide level 1 evidence in regard to evaluations of current models of support for community-dwelling people with dementia and their carers nationally and internationally.
The findings from the systematic review that positively changed characteristics of programs compared to those that did not lead to change suggest that the essential components for support worker roles/interventions were: Having an intervention duration of at least 6-12 months in order to significantly impact on measures such as carer burden, general health or wellbeing measures or the person with dementia's symptom severity Having a multi-disciplinary/inter-disciplinary team Having collaborative input to determine what support is needed/provided (e.g. with the person with dementia, their carer and family) Inter-professional collaborations and a shared approach to care Providing individualised support for each person based on a needs assessment Ensuring the support worker has a skilled background (e.g. a nurse, occupational therapist, social worker, trained in dementia) Providing ongoing follow-up (home visits, telephone contact) that is based on needs Providing individualised education based on needs Investment in a strong provider network including linking with and having close contact with the physicians/GPs of the person with dementia and coordination and monitoring of care Capacity to develop relationships While the models we examined were categorised according to the definition of the type of support worker there were similarities in the support provided by the workers. It could be inferred that the chosen terminology was just used by the authors as a way to define various multi-component interventions that were under investigation. In fact on closer inspection, many of the models identified: case management/support workers/ key workers/link workers/Admiral Nurses were performing very similar roles e.g. information provision and education, referrals to services, and support and advice yet none of the roles were uniform across the studies. Many of the studies identified and previous systematic reviews have only reviewed roles classified as case management. Case management has been defined as "a process encompassing a culmination of consecutive collaborative phases that assist clients to access available and relevant resources necessary to the client to attain their goals" [59]. Taking this definition into account all of the roles identified above in some way performed according to this definition. It is therefore important for future research to further concentrate on determining which aspects, of all support worker type roles, provide the most benefit for people with dementia, their carers/ families so that these key features can be incorporated into roles being implemented in practice.
The inconsistencies in results between the studies identified in this systematic review were notable. The heterogeneity in inclusion criteria, design, study populations, recruitment strategies, methods of delivery, role implemented, outcomes measured and the health and social care systems in which they are conducted made it difficult to synthesis results and draw conclusions. It should also be noted that the methodological quality of the majority of the studies included in this review was quite low. Only four studies were rated as having high quality according to the quality criteria. The majority of the trials lacked blinding and allocation concealment (or didn't clearly state their method) which compromised their quality. Some studies were also underpowered to detect statistically significant differences in effect size between the intervention and control groups. Very few of the RCTs were registered with a clinical trials register and therefore it was not possible to determine if selective reporting occurred. Many of the studies did not describe their data analysis techniques in enough detail which limited the validity and reliability of their results. Mention of confounding factors and the methods used to control for these confounders was also low.

Implications for research
It is vital that any future research in this area has sound methodology and that the interventions and trials are rigorous in design and delivery. The outcome measures need to be valid and reliable and the methodology clearly defined and well-documented to enable critical appraisal and interpretation of results. With clearly defined sound-methodology there is less risk that the results and outcomes will be subject to bias. In addition to quantitative outcome measures, a qualitative component included in the evaluation would add richness to the data collection by providing direct information from the people with dementia, their carers and family about the real implications and effectiveness of the role which are often not captured in clinical tools.
While it was not clear which aspects of the support worker roles produced the most effective outcomes for people with dementia, their carers and families however some key areas of importance where identified. These areas of importance were drawn from studies that demonstrated significant outcomes and low risk of bias and identify the essential components for an ideal model. The essential components for key worker type support roles/interventions identified in this review provide guidance on how the key worker type support role can be best utilised to assist people with dementia living in the community and their carers. It is essential that a full description of the type of support model and the support provided in both the intervention and control groups is provided in any further research. These clear descriptions will also be useful for others looking to replicate the trial or implement the support model in other settings.
High quality randomised controlled trials of multidisciplinary/collaborative holistic models of support are urgently needed. High-quality trials will also provide robust evidence in regard to cost-effectiveness and potential for cost savings of the support model as well as the emotional, physical and social benefits (quality of life, wellbeing, social support, reduction in symptoms and carer stress) for people with dementia, their carers and families.

Implications for policy and practice
This review identified how dementia support workers are able to respond to the needs of people with dementia and their families throughout the course of the disease. Despite a paucity of high level evidence for the role the findings highlighted that dementia support workers have a unique potential to achieve person centred care and continuity of service through offering a single point of long term contact to the consumer. The needs of a person with dementia and their families vary over time and with these changes the need for assistance from health services also varies accordingly. The personalised nature of the support worker service mitigates the risk of this population reaching 'crisis' point which is when many have been observed to access services [6]. The inherent nature of the support worker service means the model/ role can overcome issues such as fragmentation of services, poor service co-ordination and poor collaboration between providers by providing a 'real person' to assist with dementia related needs [6].
Despite limitations in the current evidence base for the support worker roles revealed by this systematic review there is enough evidence to warrant further exploration so that the essential components of the role can be incorporated into the design and funding of current and future community support services. The culmination of these findings has led us to recommend that the role be further examined so that greater evidence for the support worker models ability to contribute to the delivery of dementia care and the cost effectiveness of this role can be gathered.

Strengths and limitations of the review Strengths
Our extensive systematic mixed studies review of the international and national academic literature of models of support for community-dwelling people with dementia and their carers is the first to our knowledge to include both quantitative and qualitative studies and all models of support. Previous systematic reviews have focused mainly on the case management role, our review looks at all models of support for people with dementia, their families and carers. The investigation of both international and national models of support is also a key strength.
Limitations A limitation to this review was that it was not possible to conduct a meta-analysis of results due to the heterogeneous nature of the articles and the interventions implemented. Furthermore, it is also possible that some studies were not identified as a result of the search terms that were used in each database.

Conclusion
The strength of our synthesis of evidence is that it identifies the essential components of how key worker type support models could enhance current support models and how they can best be utilised to assist communitydwelling people with dementia and their carers. This review also reveals the poor evaluation design of many studies published to date: in the majority of cases, studies did not allow sufficient follow up time, many were not randomised and there was insufficient reporting in regard to blinding of outcome measures. Also as most studies were not registered there was an inability to determine if selective reporting occurred.
Studies that include a high quality evaluation of holistic, tailored models of support that identify which components of support produce the most valuable outcomes to assist people with dementia and their carers and families to continue to live meaningful lives are needed. There is also a need for a cost effectiveness evaluation of support worker roles.

Additional file
Additional file 1: Appendices -Eligibility and critical appraisal instruments, studies assessed for eligibility, eligibility appraisal and included studies. Availability of data and material All data generated or analysed during this study are included in this published article and its Additional file 1.
Authors' contributions SK and DG conceived the study and initiated the systematic review. ER and DG appraised all articles identified from the academic and grey literature searches. DG and ER undertook the appraisal of all studies for methodological quality, CS undertook an appraisal of some of the academic studies. All authors undertook the final drafting of the article, revised it critically for important intellectual content, read and approved the final version of the report and accept accountability for all aspects of the work.