Use of routine and cohort data globally in exploring dementia care pathways and inequalities: A systematic review

The aim of this systematic review was to evaluate studies which employed routine and cohort data sets to understand inequalities in dementia care pathways.


| INTRODUCTION
The number of people diagnosed with dementia in the United Kingdom is set to increase in the next 20 years, which will exacerbate already strained models of health and social care provision. By 2040, Wittenberg et al. project the number of people with dementia (PwD) to double and the reflected cost of care to increase threefold. 1 Crucially, the greatest proportional increase is expected to occur among people who will be diagnosed with severe dementia. Yet staff and services are already struggling to meet current demand. With further increases in the number of people with pressing needs, and their reflected costs, this will likely result in a greater levels of unmet care needs. 2 Currently, there are marked socio-economic and geographic inequalities in the prevalence of dementia, availability of informal care, access to and use of formal care. Socio-economic status [3][4][5] and place of residence 6,7 influence access to a diagnosis, and aspects of care, treatment and the support a PwD may receive. Those most socioeconomically disadvantaged are more likely to bear the brunt of these inequalities.
Existing socio-economic inequalities will exacerbate many current unmet needs in those affected by dementia, with post-diagnostic care often being underfunded and understaffed. 8 People who are unable to afford their own care will endure worse outcomes relative to more affluent PwD. 4,9 In a time of restricted finances, both individually and in the public provision and staffing for dementia care, it is imperative we better understand inequalities within dementia care pathways. We must provide evidence to inform policy and applied change in health and social care for PwD.
Routine data sets in dementia-such as the National Alzheimer's Coordinating Center in the United States and SveDem in Swedenare large databases containing standardised clinical and service interaction records for PwD. Such registries are developed with the aim of harnessing data to identify issues in access to, and quality of care, as a means to improve health and social outcomes. Other routine data sets, such as those focussing on hospital admissions, accident and emergency attendances, primary care records and social care interactions can also be used to uncover a wider picture of care pathways for PwD.
The aim of this systematic review was to evaluate studies which employed routine and cohort data sets to understand inequalities in dementia care pathways. Existing systematic reviews have explored specific socio-economic inequalities in dementia care pathways, including age, ethnicity, gender, deprivation and country of residence. [10][11][12][13][14] However, this is the first to explore how routine data sets have been used internationally to understand dementia care pathways, variance in care and inequalities in dementia care pathways. With continued restricted government funding for dementia care and an increase in the number of PwD reliant on state-funded care, the use and application of routine data is crucial to understand where and how inequalities emerge. This knowledge can enable improved person-centred care that generates a better quality of life for PwD and their carers, and attempt to reduce related inequalities in care.

| METHODS
This systematic review was registered on the PROSPERO International prospective register of systematic reviews (ID: CRD42020162934).

| Search strategy
We searched the Cochrane Central Register of Controlled Trials, CINAHL, Google Scholar, MEDLINE, PsycINFO, PubMed and Web of Science databases up to and including 2020. The search terms 'dementia' AND 'care' AND 'routine data' AND 'cohort' were used in combination. Initially, titles and abstracts of papers were read by JW and CG, retaining those which were relevant. Discrepancies over inclusion of papers were discussed between reviewers and an agreement reached. JW and CG read through the remaining papers in their entirety, to define against inclusion and exclusion criteria. Any ambiguity over inclusion of papers was discussed between reviewers until an agreement was reached.
A search of grey literature and snowballing of references from the papers already included were used to find further papers meeting the inclusion criteria.

| Inclusion criteria
The population used in the studies had to be dementia-specific and PwD need to have received a formal diagnosis of dementia. Only English language papers, published since 01 January 1990, looking at the use of routine or cohort data to quantitatively explore issues of

| Exclusion criteria
Papers published in any language other than English, published before 01 January 1990, using qualitative research methods, or quantitative papers not using routine or cohort data were excluded; reviews of any kind were also discounted. Papers with populations with no formal dementia diagnosis, non-dementia study populations or those with mixed diagnoses, for example, dementia and mild cognitive impairment were also excluded.

| Assessment of quality
The National Institutes of Health (NIH) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies was employed in this review; a 14-point checklist (see Table 3 footnotes). This tool has been used as it is a practical and pragmatic method of identifying potential flaws in the methodology of cohort studies which may increase bias. Such biases can undermine research strength and quality, casting potential doubt over subsequent results.

| Data extraction
To process the literature to conduct the data synthesis, the relevant findings were taken from the final research based on the aims and objectives of the systematic review. The following information was taken from each paper: Author name(s), research title, year of publication, country/countries of study, dementia population/subtype, socio-economic analysis in study (if yes, which socio-economic categories are included), outcomes and stages(s) in pathway.

| Data synthesis
A narrative summary of the evidence taken from the final included papers was conducted. This process involves drawing out specific findings from quantitative papers when a meta-analysis is not feasible. 15 From the included papers we looked at: what their findings tell us and what part(s) of the dementia care pathway they related to-as defined in the four sub-headings of the results section: transitions in nursing care (in the United Kingdom, this includes both residential care homes-which provide support for people with personal care and provide accommodation-and nursing homeswhich provide personal care, but also have a greater number of qualified nurses providing specific condition-related care; e.g., dementia or neurological conditions), anti-dementia medications, health and social care interactions, disease progression, mortality and survival.

| Search outcomes
We identified 1506 studies via database (n ¼ 1486) and grey literature searches (n ¼ 20), 796 of which remained once duplicate records were removed (see Figure 1 for PRISMA flowchart). Screening of these records based on titles and abstracts led to the exclusion of 583 records; the remaining 113 records were reviewed as full-text papers. Sixty-seven papers were removed leaving a total of 46 studies which met the defined inclusion criteria.  Table 1 shows the descriptive characteristics of the 46 included studies. The majority of these studies focused their analysis on one country, using one set of routine data. However, some analysed data from multiple routine or cohort data sets, and in some cases, different registries across a variety of countries. The vast majority of studies were conducted in the US (13), England/UK (12), or mainland Europe (29), with one study conducted in Australia, Hong Kong and Puerto Rico, respectively.

| Characteristics of included studies
Twenty-seven studies have highlighted differences or inequalities in care and, or outcomes as a result of a protected characteristic-such as age, gender or ethnicity-income or deprivation, or personal circumstance-such as availability of informal care. These 27 studies go beyond merely stating, for example, the number of people of each age group, or who were male and female.
Of these 27 studies, seven specifically investigated socio-economic deprivation or geography as factors in care use and quality, and or health and social outcomes for PwD (see Table 2).

| Study outcomes/stage in care pathway
Literature investigated various areas of dementia care and support (transitions to nursing care, anti-dementia medication, health and social care interaction, and disease progression, mortality and survival). Several studies looked at different stages of the care pathway (i.e., disease progression, medication initiation and mortality), so that some studies were discussed across different areas.

| Transitions to nursing care
Nine studies investigated care transitions for PwD, primarily analysing the significance and degree to which socio-economic factors 254can influence the probability of moving into nursing care. Being older was associated with a significantly greater likelihood of admission to nursing care. 7,35,36 Stevnsborg et al. 48 used data from three Danish health registries to demonstrate that the oldest PwD (those aged 70-79 and ≥80 years vs. to 60-69 years) were more likely to live in nursing care when first diagnosed with dementia. Korhonen et al. 36 reported the same association between age and institutionalisation into long-term care, but discovered it was starker for women than men.
Gender was as a factor in care transitions in several studies, revealing differences in the likelihood of transitions to nursing care. Stevnsborg et al. 35 found women were significantly more likely to be diagnosed with dementia whilst living in nursing care, and Smith et al. 7 discovered that men with dementia were significantly less likely to be admitted to nursing care generally. However, in the final 2 years of life, a greater proportion of men with dementia had care transitions, with a higher median number of transitions than women. 16 The timing of transitions in disease progression was also found to differ by gender. Neumann et al. 40 found Alzheimer's disease severity to progress at a faster rate for men, with men being more likely to transition to nursing care if they had severe Alzheimer's disease. Women, however, were more likely to be admitted to nursing care with mild Alzheimer's, indicating they would enter institutional care earlier in their disease progression.
Knapp et al. 35 investigated ethnicity as a factor in care transitions, finding people of Caribbean/African and East/South Asian ethnicities to be significantly less likely to be admitted to institutional care (hospital or nursing home). In Denmark, Stevnsborg et al. 48 compared native-born, Western and non-Western immigrant dementia populations, discovering non-Western immigrants were significantly less likely to be living in a nursing home when receiving their dementia diagnosis.

| Anti-dementia medication
Eighteen studies investigated diagnostics and use of anti-dementia medications, with nine analysing socio-economic variables as factors in likelihood of anti-dementia drug initiation.   medications. Stevnsborg et al. 48 found that Western and non-Western immigrants to Denmark were significantly less likely to receive anti-dementia medications than people born in Denmark.

| Health and social care interaction
Of 17 studies exploring health and social care interactions, eight involved socio-economic variables. The greater number of socioeconomic risk factors somebody has, the greater their likelihood of hospitalisation, 44 for example. It was found that PwD living in the most socio-economically disadvantaged areas had a significantly higher risk of emergency hospital admission, and significantly lower probability of elective admissions. 47 Evidence varies on the exact impact that ethnicity can have on admissions to hospital for PwD. 25,35,47 However, the variance in findings may be reflective of the country in which the studies were carried out, and their relevant health and social care systems.
Older PwD were at greater risk of hospitalisation. 25,35,44 However, the type of admissions demonstrates that they may be at greater risk of emergency hospital admissions, but a significantly reduced risk of both mental health inpatient 35 and elective admissions. 47 However, among PwD, those who are older (aged ≥80 years) and younger have a reduced probability of coming into contact with health and social care services in a more elective capacity. Cooper et al. 12 identified people aged under 70 years as less likely to come into contact with healthcare-even for annual dementia reviews-and those aged 80 and over were less likely to receive surgical consultations or weight monitoring checks. Furthermore, Wattmo et al. 55 discovered that for PwD, younger people had a longer delay in accessing home help services.
Care quality can also be impacted by socio-economic factors.
Connolly et al. 20 found people living in the community had greater quality annual dementia reviews and overall care than those living in care homes, and Scalmana et al. 45 discovered that people with less education were less likely to access health and social care services.

| Disease progression, mortality and survival
Nine studies investigated direct outcomes and illness progression for PwD, with seven studies focussing on socio-economic factors.
Both age and gender were found to correlate with disease progression and mortality risk. Older people and men with Alzheimer's disease transitioned at a faster rate. 40 Women and those who were younger when receiving a dementia diagnosis were shown to have better survival, 6,32 with older people and men at greater risk of dying. 51,52 Greater support is indicative of better survival, long-term independence and condition maintenance among PwD. As well as noting that less education is associated with poorer survival for PwD, Huang et al. 32 identified living in the community with a caregiver or in institutional care acted as a protective factor for survival among people with Alzheimer's disease. Survival was worse for people living in a nursing home compared to those living at home, 6 living in the community with a caregiver or in institutional care acted as a protective factor for people with Alzheimer's disease and living with others and having multiple carers led to a greater likelihood of a PwD maintaining their dependence at the level at seen at study start. 34 Pujades-Rodriguez et al. 41 compared people with and without dementia in retrospective analyses of mortality records. They illustrated PwD were more likely to live in deprived areas. Van de Vorst et al. 51,52 identified mortality risk as significantly higher for PwD living in the most socio-economically disadvantaged areas.

| Study quality
Using the NIH Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies, some of the 14-points used to assess quality were not applicable (Table 3) and so had a maximum potential score of 10 or 11. One paper received a quality rating of 5 (out of 10), but the remainder received a rating of between 7 and 10. Thus, the studies included in this review are of moderate to high quality. As a whole, they were clear in their objectives and findings and tended to describe and deliver robust research methods.

| DISCUSSION
This is the first systematic review evaluating the use of routine and cohort data sets to investigate inequalities in dementia care pathways. Findings from this review highlight the advantage of using national longitudinal databases to explore inequalities in dementia care across the globe, highlighting numerous gaps and current inequalities in care which need to be addressed.
The most socio-economically deprived areas have higher rates of undiagnosed dementia. 56 Deprivation can be reflective of wider, social and structural factors, such as income, employment, housing and transport. GP practices from such areas are more likely to only have one GP and therefore have less time to identify, diagnose and manage dementia. 57 With greater unmet needs, 4 people from deprived areas are more likely to present to emergency healthcare services later in their disease trajectory when their condition is less manageable, resulting in poorer management and treatment.
Inequality in primary care provision needs to be addressed, and promoting earlier identification and diagnosis of dementia needs to be prioritised as a means to enabling support for PwD in more disadvantaged areas.
We found evidence of inequalities in care transitions and medication use between ethnic groups. However, it is important to note that the inclusion criteria for papers in this review was for papers that included only people with a formal diagnosis of dementia.
Although people from BAME groups are more likely to have dementia, they are less likely to be diagnosed 14,58 and are therefore underrepresented in services, 59 and may therefore be WATSON ET AL.  72 and heightened dementia severity, lead to older PwD being more likely to spend longer in hospital and encounter poorer outcomes. 73 We find that greater caregiver burden and less support increase emergency healthcare use and ILTC. PwD in ILTC have poorer primary healthcare experiences and worse survival. Carers feel a sense of duty 74 and take on great responsibility. 75

| Strengths and limitations
To our knowledge, this is the first review to identify inequalities in dementia care pathways globally through the use of routine data. We

| CONCLUSIONS AND IMPLICATIONS
This systematic review has identified clear inequalities in current dementia care pathways across the globe, and the advantages of using existing routine and cohort data sets to explore and highlight these. Whilst there is a burgeoning literature on inequalities due to some socio-economic factors, there is a dearth of research identifying the impact of such factors in combination and the specific pathways WATSON ET AL.
-267 through which they operate. Our findings however are important to guide the production of improved care plans to ensure that everyone living with dementia and affected by the condition receives the right care at the right time. Maintaining care at home is mutually beneficial and can narrow inequalities but requires informal carers to be supported-we need to identify ways to reduce carer burden, aid care at home and improve outcomes as a result. Moreover, there is a need to provide a more equitable service to PwD, improve the availability of culturally appropriate services and to provide services to PwD who are not in the position of being able to call on informal care from family or friends.