Article Text
Abstract
Introduction Cardiovascular disease (CVD) is the leading cause of death worldwide, with significantly worse CVD outcomes in ethnic minorities in developed countries, especially South Asians, compared with the prevailing white ethnic group. This protocol outlines the process for conducting a systematic literature review to investigate the CVD outcome inequalities between South Asian and white Caucasian ethnic groups.
Methods Studies that compared the South Asian ethnic minority with the predominant white ethnicity in developed countries with CVD will be included from inception to 22 April 2021. We will search MEDLINE, Embase, Web of Science and grey literature to find all relevant peer-reviewed articles, reports and online theses. Articles will be screened using inclusion/exclusion criteria applied first at the title and abstract level, and then full texts, both by two independent reviewers. Articles kept in the review will undergo a risk of bias assessment using the Quality In Prognosis Studies tool and data will be extracted. Random-effects meta-analysis and heterogeneity tests will be undertaken, and tests for publication bias, outlying highly-influential observations. If insufficient data is founded or studies are highly heterogeneous, a narrative synthesis will be conducted.
Ethics Formal ethical approval is not required for this review.
Dissemination The results and findings of this systematic literature review will be disseminated through peer-reviewed publications and reports.
PROSPERO registration number CRD42021240865.
- Cardiac Epidemiology
- PUBLIC HEALTH
- EPIDEMIOLOGY
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
Strengths and limitations of this study
This systematic review protocol aims to assess how cardiovascular disease-related mortality differs in growing ethnic minorities in developed countries.
This review follows the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols guidelines.
Comprehensive investigation of bias, quality and meta-analysis assumptions.
Potential for confounding variables when comparing results from different countries, introducing bias.
Introduction
Cardiovascular diseases (CVD; a full list of abbreviations are provided in online supplemental appendix 1) are a group of disorders of the heart and blood vessels. They include coronary heart diseases such as angina, myocardial infarction and heart failure, strokes, transient ischaemic attacks, peripheral arterial disease and aortic disease. The WHO estimated that 17.9 million people died from CVD in 2016,1 representing 31% of all global deaths. Additionally, over 75% of these deaths occur in low-income and middle-income countries. However, they still pose a substantial mortality risk in developed countries. In the UK alone, heart and circulatory diseases cause a quarter of all deaths each year.2
Supplemental material
In the 2011 census, Asian British people amounted to 7.5% of the UK population. This was split into around 2.5% Indian, 2.0% Pakistani, 0.8% Bangladeshi, 0.7% Chinese and 1.5% Other Asian. In particular, the UK’s South Asian population was the largest minority ethnic group.3 This was an increase for all the Asian ethnicities from the 2001 census of England and Wales (figure 1). In the USA, Asian Americans (5.9%) made up the third largest ethnic minority group, after Hispanic and Latino, and Black or African American4 and, of these, 1.9% are South Asian. In Canada, South Asian Canadians make up about 5.6% of the total Canadian population as of 2016,5 and in Australia, Asian Australians make up about 16.3% of the population, amounting to about 4% from the South Asian countries.6
Current understanding of CVD is derived largely from studies of Caucasians of European origin.7 However, certain ethnic groups are susceptible to different types of CVD due to the high prevalence of these diseases in certain populations.
In the UK, CVD is more common in people of South Asian, African or Caribbean background,8 as people of these ethnicities are more likely to have other risk factors for CVD, such as hypertension or type 2 diabetes mellitus.9–11 In most cases, the risk of first heart attack is thought to be related to modifiable risk factors, for example, smoking, high cholesterol, inactivity and excess alcohol consumption.12
A 2017 study13 investigating the ethnic differences in the initial lifetime presentation of clinical CVD in over one million people from the CALIBER platform found that age of CVD onset was the lowest in South Asians, and significantly lower in South Asian women compared with South Asian men. However, an older study14 found CVD death rates were significantly lower in all Asian ethnic groups compared with the other groups from the REACH registry.
A systematic literature review (SLR) will help to quantify and provide clarity on CVD-related mortality inequalities between a major migrant group in some developed countries and the prevailing white ethnicity, and provide guidance for policies promoting health equality. To the best of the authors’ knowledge, there have been no SLR which compares the South Asian ethnic population against the prevailing white ethnic population in the UK and other Western, developed countries in patients with any type of CVD.
Research question
What is the magnitude of difference in CVD-related mortality between South Asian ethnic group and white population in developed countries?
Methods and analysis
Protocol design and registration
This systematic literature review protocol has been prepared according to the Preferred Reporting Items for Systematic Reviews and Meta-Analysis Protocols (PRISMA-P) statement and checklist (online supplemental appendix 2). The review has been registered in the PROSPERO (International prospective register of systematic reviews; https://www.crd.york.ac.uk/prospero). Any changes will be updated on PROSPERO accordingly.
Supplemental material
Patient and public involvement
No members of the public have been involved in the design process of this SLR.
Eligibility criteria
Population
The population will be restricted to that of the UK and other western, more economically developed countries where the prevailing ethnicity is Caucasian or other white ethnicities, and a comparator group includes South Asians. Studies will be limited to population-based samples and include populations with CVD of any form. Although age is an independent risk factor for CVD,15 especially in older patients, we will consider all patients aged 18 or older.
An initial comparison of Asian immigration to the European Union and North America between 2000 and 201016 suggests that, from European countries, the UK was the main country of destination for immigrants from India and Pakistan, and found a 296.5% net migration for Indian immigrants to the UK in the 10-year period to 2010, and both Indian and Pakistani immigrants migrated to the USA and Canada.
Therefore, we anticipate relevant countries for this review to be the UK, the USA, Canada and Australia only, and will be searched accordingly.
Ethnicity
Ethnicity can be self-reported or defined by proxy, such as country of birth, country of birth of parents or ancestry. Box 1 shows how ethnic groups were categorised in the 2011 Census of England and Wales. The 2021 census asked about ethnicity in a similar way: asking respondents ‘What is your ethnic group?’, where the response tick-boxes are grouped under the headings shown in box 1. The 2021 census also included the option for the ‘Roma’ ethnicity under the White category,17 whose numbers are estimated to exceed 100 000 in the UK alone.18
Ethnic groups categories included in the 2011 Census of England and Wales
White
English, Welsh, Scottish, Northern Irish or British.
Irish.
Gypsy or Irish Traveller.
Roma.*
Any other white background
Mixed or multiple ethnic groups.
White and black Caribbean.
White and black African.
White and Asian.
Any other mixed of multiple ethnic background
Asian or Asian British.
Indian.
Pakistani.
Bangladeshi.
Chinese.
Any other Asian background.
Black, African or Caribbean background
African.
Caribbean.
Any other black, African or Caribbean background.
Other ethnic group
Arab.
Any other ethnic group.
*Roma was included as an option under the White ethnicity group as part of the 2021 census.
Where ethnicity is reported as Indian, Pakistani or Bangladeshi, they will be combined to create the South Asian ethnicity. Moreover, South Asia generally also constitutes Afghanistan, Bhutan, Maldives, Nepal and Sri Lanka. Where this is reported by specific country, this will also be included as South Asian ethnicity.
The UK census groups the East Asian ethnicity, consisting of countries such as China, Japan and South Korea, together with the South Asian countries. However, due to observed differences in mortality between the two,7 19–21 data from studies that combine these two ethnicities will be excluded. We will attempt to contact the authors of such studies to request data for South Asians and East Asians separately if possible.
The UK census also reports the Caribbean and African ethnicities under one larger group, as seen in box 1 under the Black, African or Caribbean background group. Again, due to differences between the two ethnicities in terms of mortality, all-cause22 and cause-specific,23 these ethnicities will be reported separately in any subgroup analyses, if data is available.
The corresponding census documentation will be consulted for ethnicity categorisation when considering studies not from the UK, such as the 2020 US census for any relevant studies in the USA.
Comparators
The comparator group is the ethnic majority population which includes:
White British for the UK only comparison.
White Caucasian ethnicity from any country.
All other ethnic groups apart from white and South Asian.
Outcomes
The outcome will be CVD-related mortality between the South Asian ethnicity and the prevailing white ethnicity. This can be reported as HR, relative risk or mortality ratio. Where absolute risk of mortality is reported, studies will be included if the estimation of relative risk is possible or by contacting the author for the pertinent information. Outcomes stratified by the confounders will be included when adjusted for age and sex, and all other confounders. For completeness, we will also extract the mortality estimate between other ethnicities. All-cause mortality will be included as a secondary outcome.
We will present a summary of findings table reporting the outcome and key characteristic variables listed in the following section.
Confounders relevant to all or most of the studies
Both age and gender are important risk factors in CVD.24 The prevalence of CVD has been shown to increase with age, and the American Heart Association reports that the incidence of CVD in US men and women is around 38% from 40 to 59 years, increasing to 79% for men and 86% for women aged 80 years or over.25
Results from the PURE (Prospective Urban Rural Epidemiology) study26 found that the incidence of CVD in women (4.1/1000 person-years) was statistically significantly less than in men (6.4/1000 person-years), as well as better outcomes being consistently observed in women than in men.
Other important risk factors as identified by NHS England include hypertension, smoking, hypercholesterolaemia, diabetes, inactivity, overweight or obesity, a family history of CVD, ethnic background and excessive alcohol consumption.
These will be tabulated for each eligible study.
Study types
All observational studies that meet the PICO (population, intervention, comparison, outcome) criteria will be considered for inclusion, such as:
Case–control studies.
Cross-sectional studies.
Longitudinal studies.
Cohort studies.
A systematic review27 of cardiovascular cohort studies in the US and Europe found a shortage of information on racial or ethnic minority populations. Moreover, only a few studies gave details on the ethnic composition of the study setting, therefore inclusion will be considered for any cohort studies which included a small amount of data on ethnic minority populations only as a narrative assessment.
Search strategy
Searches will be conducted according to PRISMA guidelines in MEDLINE, Embase and Web of Science. Additionally, searches will be conducted through the Cochrane Library and PROSPERO databases to find pertinent systematic reviews. We will conduct searches of grey literature through OpenGrey and EThOS (e-theses online service). Finally, searches will be conducted in Google Scholar and using the Google search engine to find any unpublished works, such as reports. If we detect additional relevant key words during any of the electronic or other searches, we will modify the electronic search strategies to incorporate these terms and document the changes. We will place no restrictions on the language of publication when searching the electronic databases or reviewing reference lists in identified studies. Searches will be carried out from inception. The search strategy will be repeated prior to publication to find any new articles that have been published since the original search. The Ovid MEDLINE search strategy is provided in online supplemental appendix 3.
Supplemental material
Data management
All search results will be exported to EndNote X9.3.3 for screening. A Microsoft Excel file will be used to document the full selection process, including the number of studies identified by each database, the number of studies removed plus reasons for exclusion, additional studies included via pre-prints or grey literature, number of abstracts and full-texts screened and the number of studies included in the final analyses. These numbers will be entered into a PRISMA flow diagram.
Selection process
Two authors (MP and SA) will screen titles and abstracts identified by the search independently for selection into the next step of the review. The next stage involves independent review of the full-text articles, by MP and SA, to confirm their inclusion into the study. Disagreements will be resolved by consensus or, where necessary, by a third reviewer OAU. If multiple studies are identified that analysed the same data set, the study with the longer-term data will be used. If this is the same, then the most recent study will be used.
Data extraction
Data will be entered into data collection forms independently by two authors (MP and SA), who will test the data extraction form prior to data extraction for this review. This form will be based on the Cochrane data extraction forms and past data extraction forms so that all relevant information is extracted for each study included in this SLR.
These data extraction forms will include the following information: study details (study ID, design, duration, funding, conflicts of interest and type), study eligibility (study arms, groups), participant characteristics, study flow, baseline characteristics, outcomes, adverse events, risk of bias assessment and author’s conclusions.
The authors will review both sets of data extraction forms to check for disagreements, which will be resolved either by consensus or with the help of an additional author, if required. Once agreement is reached, data will be collated into a Microsoft Excel spreadsheet. Where important data is missing, we will contact the lead authors requesting this data, or the raw data if possible. Where SD is missing, we will impute these values by assuming the SD of the missing outcome to be the average of the SD from those studies where this information was reported.
Risk of bias assessment
Two authors (MP and SA) will assess the risk of bias of each included study independently. Disagreements will be resolved by consensus, or by consultation with a third author (OAU) if required.
In observational studies, as with other study types, the threats to validly are confounding bias, selection bias, performance bias, detection bias and reporting bias, and the threats to precision are inadequate study size and lack of study efficiency.28 29
Risk of bias will be assessed using the Quality In Prognosis Studies (QUIPS) tool.30 This tool assesses study participation, study attrition, prognostic factor measurement, outcome measurement, study confounding and statistical analysis and reporting. Each domain will be rated as having either ‘low’, ‘moderate’ or ‘high’ risk of bias. A study with ‘low’ risk in all six domains will be rated as having a low risk of bias. A study that has a ‘high’ risk of bias for any domain will be rated as having a high risk of bias. All other studies will be rated as having a moderate risk of bias. The QUIPS tool is provided in online supplemental appendix 4.
Supplemental material
A subgroup analysis is planned based on a study’s risk of bias rating. Furthermore, to measure the extent to which highly biased studies influence the overall results, a sensitivity analysis is planned where the high risk of bias studies will be removed.
Data synthesis
Quantitative syntheses will be conducted provided that at least two studies for the comparison between the South Asian and white ethnicities for CVD-related mortality are found; this will also include other ethnicities where data is provided, and if there is sufficient homogeneity. This will be tested alongside the main evidence synthesis, and the details of which are written in subsequent sections.
The main meta-analyses will be conducted using a Bayesian random-effects model with a 100 000 burn-in sample and 100 000 subsequent iterations, and non-informative priors for the true pooled effect size and between-study heterogeneity. We will check for model convergence by checking in the output; =1 signifies model convergence.
We will favour measures that stratify for the important confounders, like age or gender, over measures that are adjusted for them.
We anticipate that studies will report mortality differently, for example as event rates or estimates of effect size. For all estimates. We will extract SEs or, where only CIs are reported, we will use these to calculate SEs. The definitions of each CVD diagnosis and outcomes will be extracted to facilitate subgroup analyses, both by CVD type and by cause-specific outcomes.
As age remains a fundamental predictor of CVD risk and, according to NHS England, CVD is most common in people over 50 years of age, the majority of people included in this SLR are likely to be over 50 years old, therefore we will conduct subgroup analyses by age (over 50 years vs 50 years or younger). Further planned subgroup analyses will consist of assessing mortality based on the type of CVD, as there are various types, and cause-specific mortality.
The following tests will be performed to test the assumptions of the meta-analysis: (a) heterogeneity, see next section; (b) 95% prediction interval to see if, in some studies, the true outcome may favour one group over the overall estimate; (c) an examination of studentised residuals for outliers; (d) an examination of Cook’s distance to check for influential studies; (e) funnel plot (SE vs log estimate) to check for publication bias. If any highly influential or outlying studies are identified, they will be removed for sensitivity analyses. Results and plots from these tests will be provided in the appendices.
All analyses will be conducted by MP using RStudio.31
Statistical heterogeneity
Statistical heterogeneity will be tested using the and Cochran’s Q () statistics. A high signifies high heterogeneity. However, the low does not signify no heterogeneity. As the test for heterogeneity is not very powerful in detecting significant results, and that a non-statistically significant result does not indicate the absence of heterogeneity, the significance level will be set at 10%.
If one or both tests concludes the possibility of heterogeneity, p>0.10 for the Cochrane’s Q test or , representing substantial to considerable heterogeneity, the feasibility of a random effects meta-regression model will be explored to try to explain statistical heterogeneity, provided a large enough sample size. This model will include the aforementioned confounders. Furthermore, subgroup analyses, detailed in the subgroup analysis section, will be explored to explain heterogeneity.
Subgroup analyses
Cause-specific mortality (other than CVD-related).
Type of CVD.
Age groups (below 50 vs 50+ years).
Geo-political regions (Americas vs European studies).
Risk of bias rating.
The following subgroup analyses will be undertaken to explore heterogeneity if it is sufficiently high:
Subgroups of ethnicities included as part of a larger ethnic-minority group (where sample size is adequate). For example, in the case of South Asians, a subgroup analysis of Afghanistan, Bangladesh, Bhutan, India, Maldives, Nepal, Pakistan and Sri Lanka individually.
Method of reporting used to determine ethnicity.
Sensitivity analyses
The following sensitivity analyses are planned:
Removal of studies identified with a high risk of bias.
Removal of non-peer reviewed articles (such as reports or conference articles).
By study design.
Method of imputation, if applicable.
By effect measure.
Removal of outlier studies or studies with high influence.
Multiple testing
Due to the high number of hypotheses being tested, the Bonferroni-Holm method32 will be used to correct for multiple testing.
Ethics and dissemination
As this review will not collect any individual patient data and will only include published data, no ethnical approval is required. Findings will be published in an open-access peer-reviewed journal and plain language summaries will be created to disseminate to members of the public. To the best of the authors’ knowledge, this will be the first SLR to investigate differences in CVD-related mortality between South Asians and white ethnicities in developed countries, and will be of interest to those involved in public health.
Ethics statements
Patient consent for publication
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Contributors MP conceived the original idea for the study and planned and designed the protocol with respect to the PICO criteria, search strategy, quality assessment and methods of data synthesis, with the assistance of SA and OAU. All authors have contributed and approved the final version of this protocol.
Funding Open access fees were covered by financial support from the University of Warwick Research Development Fund.
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.