Access to Health Services in Older Minority Ethnic Groups with Dementia: A Systematic Review

BACKGROUND/OBJECTIVES While it is acknowledged that minority ethnic (ME) groups across international settings face barriers to accessing care for dementia, it is not clear whether ME groups access services less frequently as a result. The objective of this review is to examine whether ME groups have longer delays before accessing dementia/memory services, higher use of acute care and crisis services and lower use of routine care services based on existing literature. We also examined whether ME groups had higher dementia severity or lower cognition when presenting to memory services. DESIGN Systematic review with narrative synthesis. SETTING Nonresidential medical, psychiatric, memory, and emergency services. PARTICIPANTS Twenty studies totaling 94,431 older adults with dementia or mild cognitive impairment. MEASUREMENTS We searched Embase, Ovid MEDLINE, Global Health, and PsycINFO from inception to November 2018 for peer‐reviewed observational studies which quantified ethnic minority differences in nonresidential health service use in people with dementia. Narrative synthesis was used to analyze findings. RESULTS Twenty studies were included, mostly from the U.S. (n = 13), as well as the UK (n = 4), Australia (n = 1), Belgium (n = 1), and the Netherlands (n = 1). There was little evidence that ME groups in any country accessed routine care at different rates than comparison groups, although studies may have been underpowered. There was strong evidence that African American/Black groups had higher use of hospital inpatient services versus U.S. comparison groups. Primary care and emergency services were less well studied. Study quality was mixed, and there was a large amount of variability in the way ethnicity and service use outcomes were ascertained and defined. CONCLUSION There is evidence that some ME groups, such as Black/African American groups in the U.S., may use more acute care services than comparison populations, but less evidence for differences in routine care use. Research is sparse, especially outside the U.S.

cognitive assessment / diagnostic delay Items with a / indicate that there were at least two outcomes examined which differed in rating; the total was calculated by taking the average in each box.

Scoring:
• Representativeness: * for representative sample • Sample size: * for justification of sample size (e.g. satisfactory power calculation for the outcome) • Non-respondents: * for comparability between respondents and non-respondents (non-respondents do not differ on ethnicity), *+ where data was collected from routine data, no star if not discussed. • Ascertainment of exposure: * for self-report • Comparability: * controls for age, sex. ** controls for additional sociodemographic variables such as education or socioeconomic status or clinical variables such as severity, comorbidities, functional impairment, etc. • Assessment of the outcome: ** for record linkage or clinician rating for severity/cognition outcomes, * for self-report, no star for no description • Statistical test: * appropriate statistical test, measurement of association presented, including confidence intervals and p-value, no star if no description, missing details, or not appropriate. Scoring:

Supplementary
• Representativeness of the exposed cohort: * for representativeness • Selection of the non-exposed cohort: * if used same population pool for all racial/ethnic groups • Ascertainment of exposure: * for self-report • Demonstration that the outcome of interest was not present at the start of the study: * for first presentation to services or first readmission • Comparability of cohorts on the basis of the design or analysis: * controls for age, sex. ** controls for additional sociodemographic variables such as education or socioeconomic status or clinical variables such as severity, comorbidities, functional impairment, etc. • Assessment of outcome: * for record linkage or independent blind assessment, no star for other or not mentioned • Was follow-up long enough for outcomes to occur: * for >=1 years of follow-up, no star if shorter or not mentioned in article • Adequacy of follow up of cohorts: * for complete follow-up or <20% loss plus description, no star if loss to follow-up was associated with race/ethnicity, *+ if not mentioned but study came from routine health record data. Non-Hispanic White trial participants were more likely to use any outpatient services in the last month than other participants after adjusting for other factors, particularly AD-related outpatient services. However, differences in odds of inpatient hospital use, mental health outpatient use, and medical-surgical outpatient use disappeared after adjustment. The per-month average across the whole sample of service use was: 71.1% using outpatient services, 44% using AD-specific services, and 4.5% using inpatient services. No differences between Native American and White groups in MMSE at evaluation.

Abbreviations: ACE-R-Addenbrooke's Cognitive Examination Revised, AD-Alzheimer's Disease, BDRS-CERAD-Blessed-Roth Dementia Scale Rating (Consortium to Establish a Registry for Alzheimer's Disease version), BME-Black and minority ethnic, CAMCOG-cognitive section of the Cambridge Examination for Mental Disorders of the Elderly, CDR-Clinical Dementia Rating, CI-confidence interval, DSRS-Dementia Severity Rating Scale, ESB-English speaking background, MMSE-Mini-Mental State Examination, NESB-Non-English speaking background, OR-odds ratio, SD-standard deviation
Supplementary

Supplementary Box 1: Interpretation of race/ethnicity in this review
Defining race and ethnicity:

Our interpretation of ethnicity in this review is informed by Bhopal's definition that
"ethnicity is a multi-faceted quality that refers to the group to which people belong, and/or are perceived to belong, as a result of certain shared characteristics, including geographical and ancestral origins, but particularly cultural traditions and languages." 1 While "race" conceptually differs from "ethnicity", particularly in its historical implications, both are socially constructed and often used interchangeably such that for the purposes of our review we used either the hybrid "race/ethnicity" or "ethnicity" alone. 1

Race/ethnicity terms in our search strategy:
In our search terms, we included race/ethnicity categorisations from national censuses in addition to "race" and "ethnicity". We expected many studies to rely on these as standard (albeit imperfect) groupings, particularly where ethnicity was gathered from routine data sources. 1  Consistency of terminology to describe race/ethnicity categories in our review: While we maintained the authors' chosen terms when discussing individual studies in order to reflect the specific category used (which could affect how participants self-ascribed), many of the broader terms used may hide important disparities within groups. For example, the use of terms such as "White", "European", and "Caucasian" has been criticised because they are imprecise, can imply geographic or even genetic (rather than social) divisions, and can refer to heterogeneous populations with different health burdens. 5

Supplementary Box 2: Further context: Social and long-term care services
Long-term residential care services, such as nursing homes, comprise an important source of care as well for people living with dementia. While they are distinct from the other routine and acute medical services described in this review due to the additional social services included, disparities in use of these services have also been reported and may also impact use of services included in this review.
Previous reviews and meta-analyses have found that ME groups such as African American and Hispanic American groups were less likely to use long-term or nursing home facilities versus comparison groups. [6][7][8] In qualitative studies from the UK, South Asian families and carers have also expressed negative views about using residential care services and a strong preference for being cared for at home. 9 These patterns of nursing and residential care services may then impact use of the services included in this review, for example, if nursing homes or residential care services are better able to reduce preventable hospitalisations and manage dementia outside the hospital. 10 Within nursing home facilities, however, there have also been reports of ethnic disparities in dementia care management. 11 Future research might examine the disparities in use of social services, including updating existing reviews on the topic, as well as investigating how use of social services influences the use of the hospital, memory clinic, and primary care services studied in our review.

PROSPERO International prospective register of systematic reviews
Systematic review 1. * Review title.
Give the working title of the review, for example the one used for obtaining funding. Ideally the title should state succinctly the interventions or exposures being reviewed and the associated health or social problems.
Where appropriate, the title should use the PI(E)COS structure to contain information on the Participants, Intervention (or Exposure) and Comparison groups, the Outcomes to be measured and Study designs to be included.
Differences in rate of health care service utilization in ethnic minorities with dementia: a systematic review

Original language title.
For reviews in languages other than English, this field should be used to enter the title in the language of the review. This will be displayed together with the English language title.
3. * Anticipated or actual start date.
Give the date when the systematic review commenced, or is expected to commence. Indicate the stage of progress of the review by ticking the relevant Started and Completed boxes. Additional information may be added in the free text box provided. Please note: Reviews that have progressed beyond the point of completing data extraction at the time of initial registration are not eligible for inclusion in PROSPERO. Should evidence of incorrect status and/or completion date being supplied at the time of submission come to light, the content of the PROSPERO record will be removed leaving only the title and named contact details and a statement that inaccuracies in the stage of the review date had been identified. This field should be updated when any amendments are made to a published record and on completion and publication of the review. If this field was pre-populated from the initial screening questions then you are not able to edit it until the record is published.
The review has not yet started: Search was performed on 7 November 2018. Title and abstract screening is ongoing at the time of this submission.
The named contact acts as the guarantor for the accuracy of the information presented in the register record.

Melissa Co
Email salutation (e.g. "Dr Smith" or "Joanne") for correspondence: Give the name and affiliation of any individuals or organisations who are working on the review but who are not listed as review team members. NOTE: email and country are now mandatory fields for each person.

* Review question.
State the question(s) to be addressed by the review, clearly and precisely. Review questions may be specific or broad. It may be appropriate to break very broad questions down into a series of related more specific questions. Questions may be framed or refined using PI(E)COS where relevant.
To determine if ethnic group differences in rates of contact with health services (e.g. inpatient admission or outpatient visit with general medical care, psychiatric services, memory clinics, or emergency services) have been quantified in the literature for individuals with dementia.
State the sources that will be searched. Give the search dates, and any restrictions (e.g. language or publication period). Do NOT enter the full search strategy (it may be provided as a link or attachment.) The search will be conducted in Ovid and will include English-language, peer reviewed articles from: Embase Give a link to a published pdf/word document detailing either the search strategy or an example of a search strategy for a specific database if available (including the keywords that will be used in the search strategies), or upload your search strategy.Do NOT provide links to your search results. https://www.crd.york.ac.uk/PROSPEROFILES/118132_STRATEGY_20181128.pdf Alternatively, upload your search strategy to CRD in pdf format. Please note that by doing so you are consenting to the file being made publicly accessible.
Do not make this file publicly available until the review is complete 18. * Condition or domain being studied.
Give a short description of the disease, condition or healthcare domain being studied. This could include health and wellbeing outcomes.

* Participants/population.
Give summary criteria for the participants or populations being studied by the review. The preferred format includes details of both inclusion and exclusion criteria.
Individuals with a diagnosis of dementia, of different ethnic groups, of any age. This can include vascular dementia (whose incidence is associated with ethnicity) as well as mild cognitive impairments (MCI). Studies focusing on services for caregivers of people with dementia will be excluded. Studies defining their population based on other disorders will be excluded (e.g. studies whose population is defined to be people with cancer, some of whom may/may not have dementia). Studies which also include participants with other diseases, entire hospital registers, or more general populations may be included only if they separately report results for a dementia-specific group (e.g. studies which include both participants with cancer and participants with dementia but separately report the results for those with dementia). Studies of dementia diagnosis in a general population (since the population does not have a dementia diagnosis at the start of the study) will be excluded.
Give full and clear descriptions or definitions of the nature of the interventions or the exposures to be reviewed.
Where relevant, give details of the alternatives against which the main subject/topic of the review will be compared (e.g. another intervention or a non-exposed control group). The preferred format includes details of both inclusion and exclusion criteria.
Ethnic majority individuals (e.g. White British in the UK) with dementia. Studies which do not compare multiple ethnic groups will be excluded. Studies comparing multiple ethnic minorities will also be included.
Give the pre-specified main (most important) outcomes of the review, including details of how the outcome is defined and measured and when these measurement are made, if these are part of the review inclusion criteria.
Use of healthcare services: This is defined to include inpatient admission or outpatient visit with general medical care, psychiatric services, memory clinics, or emergency services. This may include studies on referrals, use of services, presentation to services at different rates, etc. These can be dementia-specific services or general services, but not services specific to other unrelated conditions (e.g. studies looking at how dementia patients access cancer services would be excluded). This does not include nursing and residential homes, which are considered social services, or diagnostic services with general practitioners/primary care providers. Studies about willingness to use services or advanced care directives will not be included, as they represent attitudes towards health services rather than use of services (as defined above).

* Measures of effect
Please specify the effect measure(s) for you main outcome(s) e.g. relative risks, odds ratios, risk difference, and/or 'number needed to treat.
Service use must occur for a population which already has dementia (i.e. not diagnostic).

* Additional outcome(s).
List the pre-specified additional outcomes of the review, with a similar level of detail to that required for main outcomes. Where there are no additional outcomes please state 'None' or 'Not applicable' as appropriate to the review None * Measures of effect Please specify the effect measure(s) for you additional outcome(s) e.g. relative risks, odds ratios, risk difference, and/or 'number needed to treat.
Not applicable.

PROSPERO International prospective register of systematic reviews
Describe how studies will be selected for inclusion. State what data will be extracted or obtained. State how this will be done and recorded.
All hits for the search will be de-duplicated in Ovid and exported to EndNote, where further duplicates will be removed. Then, the remaining papers will be imported to Rayyan (rayyan.qcri.org, Ouzzani 2016), a web application which facilitates screening and collaboration for systematic reviews. The review author will first screen all records for inclusion, and a secondary reviewer will screen a percentage of the records. Relevant records after title and abstract screening will then be assessed for eligibility after reading the full-text article.
The secondary reviewer will also assess a percentage of the relevant records for eligibility. Disagreements between the two reviewers will be discussed and brought to a third reviewer for a final decision, and the review author will contact authors of articles in question if more information on the study design or measures is necessary. As the search strategy is being shared with another review (protocol 118129), screening for inclusion in either will be done concurrently. Additional articles may also be included in the results after manually reviewing reference lists of included articles from the search. Results from the screen and reasons for exclusion after full-text screening will be presented in a PRISMA flow diagram. Data will be extracted by the review author into a standard form in Excel which will include information about authors, funding source, publication date, setting/service, study design, recruitment method, inclusion/exclusion criteria, sample size, type of dementia diagnosis, ethnic groupings, other variables included in analyses, and health service use findings.
Describe the method of assessing risk of bias or quality assessment. State which characteristics of the studies will be assessed and any formal risk of bias tools that will be used.
Quality of studies and risk of bias will be assessed by the primary researcher using the Newcastle-Ottawa Scale (http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp). The checklist will also be modified to include an item identifying whether ethnicity was the primary exposure being investigated or a secondary exposure variable. Studies will be given an overall quality rating of high, moderate, low, or unclear (not enough information reported).
Provide details of the planned synthesis including a rationale for the methods selected. This must not be generic text but should be specific to your review and describe how the proposed analysis will be applied to your data.
It is expected that there will be multiple health service types and multiple study designs included in the review, so narrative synthesis will be used to summarize findings from these articles. Number of studies discussing each health service type as well as how often disparities were found between ethnic groups and in what direction will be reported. Quality of the studies will also be taken into consideration.
State any planned investigation of 'subgroups'. Be clear and specific about which type of study or participant will be included in each group or covariate investigated. State the planned analytic approach.