We undertook a systematic review, reported here using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [10]. The review was prospectively registered on Open Science Forum in March 2020 (https://​doi.​org/​10.​17605/​OSF.​IO/​X7NR6).

We included published original studies and systematic reviews, and reports of governments / global public health organisations that presented observational data (cross sectional, cohort, case–control) on Pacific subgroup CVD epidemiology or management. We excluded unpublished theses. There were no restrictions on study setting, publication date, or language.

Studies examining atherosclerotic CVD, atrial fibrillation (AF), hypertensive heart disease, or heart failure epidemiology or management for people of at least two Pacific-specific ethnic groups (for differences by Pacific-specific ethnicity) or one Pacific-specific ethnic group (for differences by place (country or region) of birth) were included. We excluded studies of CVD other than atherosclerotic CVD, AF, hypertensive heart disease or heart failure (e.g. rheumatic heart disease, congenital heart disease, cardiomyopathy), and excluded studies which focussed on a broader Pacific group (e.g. Pacific Islanders).

Included Pacific-specific ethnic groups were defined as ethnic groups that are indigenous to the South Pacific sub regions of Niue, Cook Islands, Fiji, Hawaii, Tokelau, Kiribati, Tuvalu, Tahiti, Wallis and Futuna, Nauru, PNG, Solomon Islands, Vanuatu, French Polynesia, Tonga, New Caledonia, Guam, Federated States of Micronesia, Palau, Pitcairn Island, Samoa, Marshall Islands, Easter Island, Northern Mariana Islands, Rotuma (as specified by the Pacific Community (formerly South Pacific Commission) and guidance from a Pacific librarian) [11].

The outcomes sought were differences in the epidemiology and management of CVD among Pacific people according to Pacific-specific ethnicity and place of birth. The epidemiology outcomes were CVD mortality (primary outcome), hospitalisations, and incidence or prevalence (composite, by type). The management outcomes were gaps in the use of the following interventions among people with established CVD: triple therapy (antiplatelet, blood pressure lowering and lipid lowering therapy) (primary outcome), triple therapy components and cardiovascular investigations and interventions (specifically angiography, percutaneous coronary intervention [PCI, with or without stenting] and coronary artery bypass grafts [CABG]). Medication “use” included prescription and dispensing as well as adherence to medication.

We conducted a search on MEDLINE (OVID), EMBASE, Scopus, PubMed and the Cochrane Database of Systematic Reviews up to 24 March 2020. Terms (and synonyms of) ‘atherosclerosis’ or any of the atherosclerotic cardiovascular diseases (ischemic heart diseases, cerebrovascular diseases, peripheral vascular diseases) or cardiovascular diseases related to atherosclerosis (heart failure, hypertensive heart disease, atrial fibrillation), ‘epidemiology’ and ‘treatment, adherence and compliance’ were combined with ‘Pacific people’, any of the Pacific regions or individual Pacific countries. MESH terms were used for each of ‘atherosclerosis’, atherosclerotic cardiovascular diseases, ‘epidemiology’, ‘treatment, adherence and compliance’, ‘Pacific people’, Pacific regions, individual Pacific countries. Full details of our search strategy can be found in Additional file 1: Figs. S1–S5.

Two researchers (JW, VS) also searched for reports of governments and global public health organisations that presented observational data (cross sectional, cohort, case–control) on Pacific subgroup CVD epidemiology or management independently to identify those meeting our inclusion criteria. Title and contents pages were initially searched, before screening the relevant chapters. Disagreements about whether the grey literature met the inclusion criteria were resolved through discussion about the reasoning for their decisions. Full details on government and global public health organisation webpages that were searched are noted in Additional file 1: Fig. S6.

Reference lists of each of the full texts originally assessed for eligibility were also checked for potential citations to include in this systematic review.

References were imported into Covidence (www.​covidence.​org, Veritas Health Innovation Ltd.), which was used to manage the initial title and abstract screening. EndNote was then used to manage full-text screening. Two researchers (JW, VS) screened all titles, abstracts and full texts independently to identify those meeting our inclusion criteria. For disagreements about whether the texts met the inclusion criteria, JW and VS reviewed these titles, abstracts and full texts together, discussed reasoning for their decisions and resolved disagreements. Two other researchers (MH, CG) were available to adjudicate any unresolved disagreements, but this was not needed during the selection process. Neither of the review authors were blind to the journal titles, study authors, or institutions.

For quality assessment, we used the National Heart, Lung, and Blood Institute (NHLBI) Quality Assessment Tool for Observational Cohort and Cross-sectional studies. The authors chose to use the NHLBI tool as we judged it to be able to be applied to both peer-reviewed academic literature and grey literature reporting observational data. Each study was independently assessed by two reviewers (JW, VS) using the NHLBI standardised quality assessment tool. Studies were ranked as ‘yes’ or ‘no’ with regard to selection bias, performance bias, reporting bias and then given an overall judgement of ‘poor’, ‘fair’ or ‘good’. Good quality references were defined as having methodological rigour to enable the evaluation of the internal validity of the broad type of included studies and reports. Where there was disagreement, JW and VS discussed and came to an agreement of the rankings and overall judgement. Two researchers (JW and VS) independently extracted data and discrepancies were resolved by a third reviewer.

Of the six included texts (four reports, one webpage, and one article), four were of Pacific populations in Pacific Island countries and two were of Pacific diaspora in other countries (Table 1) [1, 2]. All included studies reported data on CVD epidemiology [1‐3, 21‐23], but only 1 reported data on CVD management [23].

The results are reported according to comparisons by Pacific country using the international studies, and by Pacific ethnicity using the NZ studies.

There were no included studies that reported on the following: CVD hospitalization rates, the use of individual components of triple therapy, CVD investigations or interventions. Full details of hospitalization data extracted from all records assessed for quality can be found in Additional file 2: Table S4.

There were a number of limitations related to the evidence included in the review. Firstly, for the management of CVD it was unclear which medications were considered for managing CVD risk [23]. Secondly, there may be differences in the recording and reporting of health data and classification of cause of death that may explain some of these differences which contribute to interoperability concerns in the Pacific region [27], and may over- or underestimate the differences in prevalence and mortality attributed to CVD. For example, there was a difference in mortality rate due to hypertensive heart disease being 20 times higher in the Cook Islands than in Palau, yet, over the ten year period of 2009 to 2019, IHD remained the leading cause of death in Palau (preceding diabetes) and the second leading cause of death in the Cook Islands (preceded by diabetes) [28, 29]. Hypertensive heart disease was recorded as the fourth leading cause of death in the Cook Islands but 75th in Palau [28, 29], despite its clear relationship to IHD.

It is important to note that the GBD study makes many adjustments and imputations when deriving the country estimates, often based on very limited primary data. This is particularly common among LMICs, of which 10 of the 17 Pacific Island countries whose data were included in this review are [30‐33]. In contrast, other studies are more likely to have used primary data. So what at first glance may look like reasonably precisely enumerated estimates in the GBD study could be considerably influenced by imputations for missing data [31].

Ethnicity was re-classified differently between the two texts that reported on the burden of CVD in Pacific-specific groups in NZ [1, 2]. In the metro-Auckland population of people who had identified as Pacific in 2011, Pacific-specific ethnic groups were assigned in reverse population size order in order to attain the largest subgroup sizes [2]. Niuean ethnicity was prioritised (if Niue had been selected as their ethnicity on their health contact record for 2011), followed by Cook Islands Māori (if present in 2011 health contact record), then Tongan (if present in 2011 health contact record), then Samoan (if present in 2011 health contact record) [2]. Other Pacific was assigned if none of the previous Pacific-specific ethnic groups were present [2]. This differed from the total response (overlapping) count used in the other text, which allowed for each person to be assigned to each of the ethnic groups they reported [1]. This is considered beneficial for representing people in each ethnic group they identify with, but presents complexities with making comparisons between groups, as people can be assigned to multiple groups [34]. The lack of consistency in NZ analyses of defining Pacific-specific ethnicity is problematic for tracking the burden of CVD over time.

There are several limitations of this review which need to be considered when interpreting findings. In terms of the nature of the review, we assumed that data reported by Pacific Island country (excluding Fiji) represented homogenous indigenous populations. It is possible that Pacific countries, particularly those which have been or are currently territories of USA or France do not have a homogenous indigenous Pacific population anymore. Combining data from indigenous and non-indigenous populations may have masked the true size of the effect for indigenous Pacific populations in these countries.

In terms of the studies identified, there are a number of notable observations. First, the literature on this topic was sparse. The limited evidence are worrying [1] for the South-Pacific region that is seeing an increase in the burden of non-communicable diseases such as CVD [35], and [2] because the Pacific diaspora now outnumber the population of many Pacific countries but experience inequities in exposure to CVD risk factors in their new home countries [36]. Second, none of the named first authors of the 23 texts for which quality was assessed were (to our knowledge) of indigenous Pacific ethnicity (see Table 2). A likely implication of this is that the data collection, interpretation and/or reporting is likely not representative of indigenous Pacific worldviews and perspectives. In contrast, the first and last author of this study are each of indigenous Pacific ethnicity. Third, there were important variations between studies in the age of participants, methods and data collected (which is why we were unable to undertake a meta-analysis).

Despite the strategies that were used to ensure its completeness, this review could have missed relevant data.

Health and mortality data reporting is limited across the Pacific [37]. Of the 25 Pacific countries listed, data were available for only 17 of those countries in this review. This may reflect delays in implementation of critical infrastructure [38], such as recording cause of death, which has only been recorded in French Polynesia since 1984 [39]. In addition to this, a notable observation was the reporting of diabetes as a leading cause of mortality in Pacific countries [12, 13, 18, 20, 28, 29]. Differences in diabetes mortality between countries were reported as being over eight times higher in Kiribati (203.99 (158.28, 252.68)) than in Guam (23.70 (19.89, 28.63)) [3]. We did not specifically search for diabetes as a cause of mortality as we instead considered it to be a risk factor for developing CVD as a complication [40]. There may be inconsistencies in the way mortality associated with diabetes is recorded. This, along with the limited reporting of health and mortality data in the Pacific, will need further exploration, which is outside the scope of this paper.

Given the differences in burden of CVD evident in the literature, the homogenised approach to understanding the epidemiology of CVD among Pacific peoples, that is evident in the USA, NZ, and Australia where most Pacific people migrate to, may not be appropriate. Exploration of Pacific cardiovascular epidemiology and management by Pacific-specific ethnicity, location, and place of birth is urgently needed to better understand the diversity of CVD health needs and hence health service requirements of Pacific peoples.