Cardiometabolic Health Disparities in Native Hawaiians and Other Pacific Islanders

Abstract

Elimination of health disparities in the United States is a national health priority. Cardiovascular disease, diabetes, and obesity are key features of what is now referred to as the “cardiometabolic syndrome,” which disproportionately affects racial/ethnic minority populations, including Native Hawaiians and other Pacific Islanders (NHOPI). Few studies have adequately characterized the cardiometabolic syndrome in high-risk populations such as NHOPI. The authors systematically assessed the existing literature on cardiometabolic disorders among NHOPI to understand the best approaches to eliminating cardiometabolic health disparities in this population. Articles were identified from database searches performed in PubMed and MEDLINE from January 1998 to December 2008; 43 studies were included in the review. There is growing confirmatory evidence that NHOPI are one of the highest-risk populations for cardiometabolic diseases in the United States. Most studies found increased prevalences of diabetes, obesity, and cardiovascular risk factors among NHOPI. The few experimental intervention studies found positive results. Methodological issues included small sample sizes, sample bias, inappropriate racial/ethnic aggregation of NHOPI with Asians, and a limited number of intervention studies. Significant gaps remain in the understanding of cardiometabolic health disparities among NHOPI in the United States. More experimental intervention studies are needed to examine promising approaches to reversing the rising tide of cardiometabolic health disparities in NHOPI.

INTRODUCTION

The prevalence of cardiometabolic disorders, including cardiovascular disease, diabetes, and obesity, has reached epidemic proportions worldwide. Prevalences of diabetes and cardiovascular disease among ethnic minorities in the United States exceed those seen in the general population (1–7). Because of the excess health burden of cardiovascular disease and diabetes in ethnic minorities, cardiometabolic risk, as the precursor of these diseases, provides a specific target for conducting investigations that aim to reverse and/or eliminate these disparities. Moreover, obesity, as one of the central pathophysiologic mechanisms underlying the syndrome of cardiometabolic risk, might constitute an earlier “upstream” target for treatment and prevention that could be effective in reducing excess morbidity.

In this review, we focused on cardiometabolic health disparities among Native Hawaiians and other Pacific Islanders (NHOPI). As a federal designation, “Native Hawaiian or Other Pacific Islander” refers to persons with origins in any of the original peoples of the islands of Polynesia, Micronesia, and Melanesia (8). The first Polynesian settlers of the Hawaiian Islands are thought to have migrated from the Marquesas Islands as early as 100 B.C.E., some 2,000 years ago (9). Hawaiians lived in isolation until 1778, when European explorers brought with them deadly foreign infectious diseases that decimated the Native population. Colonization and the eventual overthrow of the Hawaiian monarchy resulted in a loss of land and political power, as well as traditional practices and customs, including the near extinction of the Hawaiian language (9).

Events such as the cultural revival referred to as the “Hawaiian Renaissance,” the return of the island of Kahoolawe to the Hawaiian people, the formation of political bodies such as the Office of Hawaiian Affairs, the public recognition of the illegal overthrow of the Hawaiian monarchy, and the ongoing voyages of the Hokule‘a using historical navigation techniques of Native Hawaiians serve to highlight the resiliency and strength of this population. However, despite these positive social and political developments, NHOPI are overrepresented in lower socioeconomic groups, report greater difficulties in obtaining health care, and may be affected by internalized racism as a consequence of their historical experience of disenfranchisement and loss of power within their traditional homeland (10–12). In addition, NHOPI continue to bear a disproportionate burden of disease, including cardiovascular disease and diabetes.

Today, the state of Hawai‘i has the largest population of Native Hawaiians in the United States, followed by California. Among ethnic subgroups in Hawai‘i, Native Hawaiians have the highest prevalence of diagnosed diabetes (11.5%), with reported prevalences ranging from 19% to 22% for type 2 diabetes and from 16% to 35% for impaired glucose tolerance (5, 13, 14). Cardiovascular disease mortality among Native Hawaiians in 2004 was more than twice that in Japanese, who had the lowest rates (372 per 100,000 population vs. 167 per 100,000 population), and diabetes-related mortality was 3 times higher in Native Hawaiians than in Caucasians (39 per 100,000 population vs. 13 per 100,000 population) (6). The Native Hawaiian Health Research (NHHR) Project examined the relation between a clustering of cardiovascular risk factors and biochemical markers of insulin resistance (fasting insulin and C-peptide levels) (7). The investigators found that fasting insulin concentrations were correlated with body mass index, waist-to-hip ratio, blood pressure, and levels of triglyceride, high density lipoprotein cholesterol, and glucose. A significant correlation was also found between increasing insulin resistance and increased clustering of cardiovascular disease risk factors. The NHHR study, in addition to a limited number of other studies on NHOPI, suggests that significant disparities occur between and among these populations. The purpose of this review was to systematically assess the state of the science related to cardiovascular disease, diabetes, and adiposity among NHOPI.

METHODS

Database searches were performed in PubMed and MED-LINE for the time period of January 1998 to December 2008, with keyword combinations of the following racial/ethnic groups in an “OR” search term: Native Hawaiian, Hawaiian, Pacific Islander, Samoan, Tongan, Micronesian, New Zealand, Maori, Melanesia, Chamorro, Guamanian, Fijian, and Polynesian (Figure 1). The racial/ethnic groups were then included in combination with the following terms defined as an “OR” function: minority, minorities, groups, ethnicity, and ethnicities. The above racial/ethnic AND minorities search term was identified as a “Set A” keyword search. Results from the Set A keyword search were then combined with Set B keywords as an “OR” search term which included the following: inequity, inequality, health disparities, health differences, cardiovascular, hypertension, heart, heart failure, heart disease, heart disease risk factors, cardiac, cardiomyopathy, diabetes, syndrome X, metabolic syndrome, insulin resistance, glucose intolerance, prediabetes, cardiometabolic, obesity, adiposity, overweight, physical inactivity, physical activity, nutrition, diet, and smoking in combination with United States.

Figure 1.

Procedures used to review the literature on cardiometabolic health disparities among Native Hawaiians and other Pacific Islanders, January 1998–December 2008. Set A: inclusion of Native Hawaiian, Hawaiian, Pacific Islander, Samoan, Tongan, Micronesian, New Zealand, Maori, Melanesia, Chamorro, Guamanian, Fijian, and Polynesian in an “OR” search term along with the following racial/ethnic terms: minority, minorities, groups, ethnicity, and ethnicities, defined as an “OR” search term. Set B: inequity, inequality, health disparities, health differences, cardiovascular, hypertension, heart, heart failure, heart disease, heart disease risk factors, cardiac, cardiomyopathy, diabetes, syndrome X, metabolic syndrome, insulin resistance, glucose intolerance, prediabetes, cardiometabolic, obesity, adiposity, overweight, physical inactivity, physical activity, nutrition, diet, and smoking in combination with United States.

Additional studies (n = 21) were also extracted from the reference lists of the articles identified in the initial search using Set A AND Set B keywords; these studies were reviewed for inclusion/exclusion. The searches were restricted to English-language articles on humans aged ≥19 years that had been published in peer-reviewed scientific journals. Articles were excluded from the review if they were letters, editorials, or literature reviews without new data; if they had been published in a foreign language; or if they were nonempirical.

Using this search strategy, we identified 311 citations, of which 98 were deemed relevant through review of the article title (performed by a single reviewer). All 98 articles underwent abstract review by 2 independent reviewers, using a standard checklist adapted from other reviews of the health-disparities literature (15). Of the 98 abstracts reviewed, 71 articles were selected for a full text review, which was performed by 2 independent reviewers to ensure compliance with all inclusion criteria, as well as ranking on the following study design criteria: 1) use of appropriate indicators for patient characteristics (e.g., race, ethnicity, or ancestry, sex, age, education, income); 2) inclusion of objective measures of the outcomes of interest (i.e., measured height, weight, and systolic and diastolic blood pressure; self-report of or medical chart review to determine obesity, cardiovascular disease, and/or diabetes status, etc.); 3) inclusion of well-defined measures of disease status; and 4) appropriate adjustment for patient comorbid conditions (i.e., age, sex, body mass index for diabetes outcomes, blood pressure for cardiovascular outcomes, etc.).

After full text review, a joint review meeting was convened to determine the final selection of articles to be included in this study. A total of 28 articles were excluded for 1 or more of the following reasons: 1) NHOPI were aggregated with other racial/ethnic groups (i.e., “Asians and Pacific Islanders” was a single category) (18% of articles); 2) the article was a review or editorial (36%); 3) the study included NHOPI but there was no specified outcome related to cardiometabolic diseases (21%); 4) the NHOPI study population lived outside of the United States (14%); 5) the study population was under age 19 years (i.e., children or youths) (7%); and 6) there was another miscellaneous reason for study exclusion (the article had been published in a non-peer-reviewed journal, no NHOPI population was included, etc.) (4%). Thus, the final number of articles included in this literature review of cardiometabolic health and health-care disparities among NHOPI was 43.

RESULTS

Cardiovascular disease

Study characteristics

A total of 12 papers pertaining to cardiovascular disease or its risk factors were reviewed (Table 1). The majority of studies (n = 10) were cross-sectional. In 4 of the 10 cross-sectional studies, investigators had prospectively collected new data, and in 2 they had used retrospective data collected from administrative databases. A single prospective study included a cohort that had been followed for over 4 years. One study included qualitative data collected through focus groups of NHOPI. The only study in which researchers had proposed testing an intervention had had a quasi-experimental, pre-post study design without controls for testing of a Native Hawaiian cultural intervention designed to improve hypertension profiles. Fifty percent of the papers included a study sample of at least 300 participants of NHOPI ancestry.

Table 1.

Literature on Cardiovascular Health Disparities Among Native Hawaiians and Other Pacific Islanders, January 1998–December 2008

Study Authors and Year (Ref. No.)	Sample and No. of Subjects	% of Total Sample Who Were NHOPI	Age, years	Study Design	Main Goal and Outcomes/Findings	Limitations
Novotny et al.,   1998 (16)	66 Native Hawaiian women   out of a total multiethnic   sample of 421	16	25–35	Observational, cross-sectional,   clinical research sample	Examine anthropometric variations   between multiethnic women and their   relation to blood pressure and   cholesterol.	Sample bias; relatively small study.
					Native Hawaiian women had the highest   measures of adiposity. Cholesterol was   not associated with adiposity. Blood   pressure was associated with adiposity   but did not vary by ethnicity.
Verderber et al.,   1999 (20)	23 Pacific Islanders out of   a total multiethnic sample   of 60	38	40–85	Observational, cross-sectional,   hospital surgery sample	Describe risk factors for cardiac disease   and post-CABG outcomes and   complications in multiethnic patients.	Sample bias (single hospital, relatively   small sample). No adjustment for   physician as covariate.
					No ethnic differences in postoperative   complications in first 20 hours. At 48   hours post-CABG, Pacific Islanders   required significantly more care than   Japanese (P < 0.01).
Taira et al.,   2001 (21)	361 Native Hawaiians out of   a total multiethnic sample   of 2,962	11	≥18	Observational, cross-sectional,   administrative database   sample	Examine differences in revascularization   and mortality rates following acute   coronary syndrome in a multiethnic   patient population.	Cross-sectional; claims data   from single health insurer.
					Ethnic differences in types of procedures   received were significant only for men.   Native Hawaiian men were less likely to   receive percutaneous coronary   intervention (adjusted OR = 0.51, 95%   CI: 0.34, 0.75) and more likely to receive   CABG (adjusted OR = 1.8, 95% CI: 1.2,   2.7) than whites. No significant ethnic   differences in women.
Grandinetti et al.,   2002 (18)	572 Native Hawaiians;   nonpregnant	100	≥30	Observational, retrospective   cohort, population-based   sample	Examine degree of Hawaiian ancestry   and blood pressure and relation to other   covariates.	Percentage of Hawaiian ancestry   was self-reported; cross-sectional   design; sample population only from   rural location. Assessment of   adequate blood pressure control   with medications not described.
					Adjusted prevalence of hypertension   increased with degree of Hawaiian   ancestry, except for 100% Hawaiian   ancestry. Hawaiian ancestry was   significantly associated with systolic and   diastolic blood pressure (P < 0.0001).   After adjustment for all significant   covariates, only diastolic blood pressure   was associated with Hawaiian ancestry   (P = 0.008).
Sundaram et al.,   2005 (17)	510 NHOPI women out of   a total multiethnic sample   of 120,035	<1	≥18	Observational, cross-sectional   telephone survey, 2001   BRFSS	Determine prevalence of cardiovascular   disease risk factors among multiethnic   women in 2001 BRFSS.	All data were self-reported. Cross-   sectional; relatively small sample   compared with other ethnic groups.
					NHOPI women had second highest   prevalence of hypertension at 33.7%   (age-adjusted) and highest prevalence of   high cholesterol (23.9%).
Yeo et al.,   2005 (45)	377 Pacific Islanders out of   a total multiethnic sample   of 2,598	14	≥18	Observational, case-control,   hospital administrative   database sample	Examine ethnic differences in   percutaneous coronary intervention   success and/or complications in   hospitalized patients.	Sample bias (recruited from a single   hospital). Cross-sectional data; no   long-term outcomes available.
					Despite higher rates of diabetes,   hypertension, obesity, and renal failure in   NHOPI, there was no difference in   percutaneous coronary intervention   complications compared with whites or   Japanese.
Chiem et al.,   2006 (46)	228 Chamorros	100	≥18	Observational, cross-   sectional telephone survey,   community administrative   database sample	Describe cardiovascular disease risk   factors in Chamorro community to aid in   developing programs.	Sample bias (recruited from community   database). No adjustment for   covariates. Cross-sectional data.   Self-reported health risk factors.
					Crude frequencies of hypertension,   diabetes, hyperlipidemia, and physical   inactivity were higher in Chamorros than   in US whites. Chamorro women were   more likely to have hypertension and   diabetes. Men were more likely to have   elevated body mass index and   cholesterol.
Ezeamama et al.,   2006 (22)	1,289 Samoans (American   and Western Samoan)	100	25–58	Observational, prospective   cohort, population-based   sample	Investigate cross-sectional and   prospective associations between SES   and cardiovascular disease risk factors   and predict the probability of risk factors   by SES level between American Samoa   and Samoa.	Relatively large sample of Samoan   participants; subjects were younger   and thus there were fewer cases of   cardiovascular disease risk factors.   Men lost to follow-up in both   locations were more likely to be   employed in the wage-labor market   and may have introduced attrition   bias.
					High SES was associated with increased   odds of risk factors in less developed   Western Samoa and decreased odds of   risk factors in more developed American   Samoa. Inverse association between   SES and risk factors in Western Samoa   vs. American Samoa is attributable to   heterogeneity across the Samoan   Islands in specific exposures to   economic development and natural   history of individual risk factors.
Grandinetti et al.,   2006 (19)	185 Native Hawaiians out of   a total multiethnic sample   of 588; nonpregnant	31	≥30	Observational, retrospective   cohort; randomly selected   nested study of population-   based sample	Examine angiotensin-converting   enzyme gene polymorphisms and   increased Q-Tc (associated with heart   disease) between ethnic groups.	Cross-sectional prevalence   Cases may lead to bias.
					Increased Q-Tc was highest among   persons with the ACE II   genotype. ACE II genotype was higher in Native Hawaiians   (50.3%) than in whites (21%). After   adjustment, prevalence of increased Q-   Tc was significantly associated with ACE   II genotype independently of ethnicity   (P < 0.01).
Kretzer et al.,   2007 (40)	15 Native Hawaiians out of   a total multiethnic sample   of 23	65	≥30	Quasi-experimental (no   control group), pre-post   intervention, community   sample	Evaluate whether a class on self-identity   via ho‘oponopono would improve high   blood pressure.	No control group, small sample size,   and potentially biased sample. No   adjustment for medication changes.
					Ho‘oponopono intervention (4-hour   group class) reduced mean systolic and   diastolic blood pressure, which was   sustained for 2 months after intervention.   Results for the 15 Native Hawaiians were   aggregated with those for the remaining   8 other non-Hawaiian participants.
Taira et al., 2007   (47)	3,746 Native Hawaiians out   of a total multiethnic   sample of 28,395	13	≥18	Observational, cross-   sectional; health insurance   administrative database   sample	Examine factors associated with   antihypertensive medication adherence   within a multiethnic patient population   using administrative claims data.	Cross-sectional claims data from   a single health insurance plan.   Patient adherence was measured by   filling of prescriptions, not at patient   level. Lack of information on impact   of comorbid conditions on   medication adherence.
					Overall adherence in all ethnic groups   was less than 65%. After adjustment,   Native Hawaiians were less likely to   adhere than whites (OR = 0.84, 95% CI:   0.78, 0.91), and this was consistent   across therapeutic classes. Other patient   factors associated with lower adherence:   younger age, higher morbidity, and   history of heart disease. Seeing   a physician of the same ethnicity did not   improve adherence.
Kaholokula et al.,   2008 (39)	36 NHOPI and family   caregivers	100	≥18	Focus group, community   sample	Identify health beliefs, attitudes,   practices, and social and family relations   important in heart failure treatment   among NHOPI.	Small sample size, and only 30%   were heart failure patients.   Qualitative study design with   potentially subjective responses.
					Native Hawaiians with heart failure   reported coping experiences of denial of   illness, hopelessness, and despair and   relied on spiritual/religious beliefs for   support. Samoans preferred being   treated by physicians, while Native   Hawaiians preferred traditional healers.   Barriers to heart failure care include poor   knowledge, lack of trust of the physician,   a poor patient-physician relationship,   finances, dietary changes, and increased   demands on time.

Abbreviations: ACE II, angiotensin-converting enzyme insertion/insertion; BRFSS, Behavioral Risk Factor Surveillance System; CABG, coronary artery bypass graft; CI, confidence interval; NHOPI, Native Hawaiians and other Pacific Islanders; OR, odds ratio; Q-Tc, corrected Q–T interval; SES, socioeconomic status.

Study findings

NHOPI women were found to have a high frequency of hypertension and high cholesterol in comparison with whites and other ethnic groups (16, 17). Studies that examined hypertension along with other covariates in NHOPI found that hypertension was significantly related to degree of Hawaiian ancestry and especially diastolic blood pressure after controlling for other covariates (18). The sole genetic study found that increased corrected Q–T interval (Q-Tc), which has been associated with heart disease and sudden death, was associated with the angiotensin-converting enzyme insertion/insertion (ACE II) genotype, which is found with greater frequency among Native Hawaiians than in other ethnic groups (19). Verderber et al. (20) compared post-coronary artery bypass graft (CABG) complications across ethnic groups and found that NHOPI had similar early post-CABG complications (first 20 hours after CABG) but experienced significantly more ventricular arrhythmias requiring medical treatment on postoperative day 2 than Japanese. In another study, NHOPI men with acute coronary syndrome were significantly more likely to receive CABG (odds ratio = 1.8, 95% confidence interval: 1.2, 2.7) and less likely to receive percutaneous coronary intervention following their first hospitalization than were whites (21). No ethnic differences in endovascular treatment for acute coronary syndrome were found in women (21).

In the only longitudinal prospective cohort study, investigators were interested in examining measures of socioeconomic status and cardiovascular disease risk factors in American Samoans versus Western Samoans. Ezeamama et al. (22) found that high socioeconomic status was associated with increased odds of cardiovascular disease risk factors in Western Samoa but decreased odds in more developed American Samoa. The authors attributed this differential effect of socioeconomic status on cardiovascular disease risk factors to the heterogeneity across the Samoan Islands in specific exposures to economic development and the natural history of individual cardiovascular disease risk profiles.

Study limitations

The cardiovascular disease literature reviewed had a number of limitations. First, nearly all of the studies were observational studies with cross-sectional data, which does not permit a clear understanding of cause and effect for significant associations between outcomes and exposures. Half of the studies reviewed had relatively small sample sizes or had serious sample biases that confounded the study’s findings. Finally, several of the studies of sufficient quality were drawn from 2 research groups that have established cohorts in rural communities in Hawai‘i (Grandinetti et al. (19, 23)) and in Western and American Samoa (Ezeamama et al. (22)); those findings may not be generalizable to other NHOPI populations in the United States.

Type 2 diabetes mellitus

Study characteristics

A total of 16 diabetes-related studies were reviewed. Most were cross-sectional investigations (10 studies), although 1 study was descriptive and 2 were retrospective (Table 2). Nine of the cross-sectional studies examined the population-based data of the NHHR Project, including a quasi-experimental study that was a nonrandomized concurrent intervention which included Native Hawaiians with diabetes or at risk for diabetes. Another quasi-experimental study compared “before” and “after” hemoglobin A1c levels in a small sample of Native Hawaiian, Samoan, and Tongan participants undergoing an intervention delivered by community health workers. Sample sizes ranged from 78 participants to more than 3,000. The 2 retrospective studies examined the incidence of macrosomia and gestational hypertension among NHOPI women.

Table 2.

Literature on Diabetes Health Disparities Among Native Hawaiians and Other Pacific Islanders, January 1998–December 2008

Study Authors and Year (Ref. No.)	Sample and No. of Subjects	% of Total Sample Who Were NHOPI	Age, years	Study Design	Main Goal and Outcomes/Findings	Limitations
Grandinetti et al.,   1998 (5)	574 Native Hawaiians from 2   rural communities in   Hawai‘i; nonpregnant	100	≥30	Observational, retrospective   cohort, population-based   sample	To estimate prevalences of type 2   diabetes and impaired glucose   tolerance.	Cross-sectional; self-reported   ancestry, contributing to possible   misclassification.
					Prevalence of impaired glucose   tolerance was 16%; diabetes, 20%.   Prevalence of impaired glucose   tolerance was significantly higher in   women and significantly associated   with body mass index, waist   circumference, and waist-to-hip ratio.   Age-adjusted prevalence of diabetes   was 4 times higher than in the   Second National Health and   Nutrition Examination Survey   population.
Grandinetti et al.,   2000 (26)	581 Native Hawaiians from 2   rural communities in   Hawai‘i; nonpregnant	100	≥30	Observational, retrospective   cohort, population-based   ample	To examine the association between   diabetes and CES-D depressive   symptoms.	Cross-sectional design; thus, the   temporal relation between   glycemic control and CES-D   depressive symptoms could   not be determined.
					Among participants reporting a prior   history of diabetes, both mean CES-D   score and depressive symptom   prevalence were significantly higher   than in participants with no prior   history of chronic illness, after   adjustment for age and social support.   High hemoglobin A1c level (≥7%) was   also associated with higher   prevalence of CES-D-assessed   depressive symptoms. Results   suggest that hyperglycemia may   explain the high prevalence of   depressive symptoms among   participants with known and newly   identified diabetes.
Mau et al.,   2001 (29)	147 Native Hawaiians with   diabetes or metabolic   syndrome and their ‘ohana   (family) support person	100	≥30	Quasi-experimental,   nonrandomized, controlled   trial; subjects recruited from   population-based sample	To examine association of stage of   change with diet and exercise in   response to lifestyle intervention.	Nonrandomized intervention;   lack of true control group.
					Stage of change was significantly   associated with positive dietary   and exercise behaviors. Participants   in the family support intervention   group were more likely to advance   from pre-action to action/maintenance   regarding fat intake and physical   activity than the standard intervention   group.
Grandinetti et al.,   2002 (18)	578 Native Hawaiians from 2   rural communities in   Hawai‘i; nonpregnant	100	≥30	Observational, retrospective   cohort, population-based   sample	To investigate the relation between   glucose and percentage of Hawaiian   blood quantum.	Cross-sectional survey; self-   reported ancestry, contributing to   possible misclassification; self-   reported lifestyle behaviours
					Increased Hawaiian blood quantum was   significantly associated with increased   fasting glucose level, body mass   index, waist-to-hip ratio, and age. Full   Hawaiians had higher glucose   concentrations than part-Hawaiians   after adjustment for age, sex, body   mass index, and waist-to-hip ratio.
Kaholokula et al.,   2003 (27)	59 Native Hawaiians out of   a total multiethnic sample of   141; nonpregnant	41	≥30	Observational, nested case   study from a population-   based sample	To examine correlations between   glycemic status and health-related   quality of life and depressive   symptoms.	Limited generalizability; little   variability in body mass index and   waist-to-hip ratio.
					No correlation between depressive   symptoms and glycemic status was   observed. Health-related quality of life   had the greatest magnitude of effect   on depressive symptoms in people   with diabetes compared with glycemic   status and knowledge of diabetes   diagnosis. Relation between   depressive symptoms and health-   related quality of life was influenced by   glycemic status, sex, education,   marital status, and social support.
Grandinetti et al.,   2005 (48)	510 Native Hawaiians out of   a total multiethnic sample of   1,447; nonpregnant	35	≥30	Observational, retrospective   cohort, population-based   sample	To estimate the prevalence of metabolic   syndrome.	Cross-sectional
					Prevalence of metabolic syndrome was   significantly higher among Native   Hawaiians and all other minority ethnic   groups than among Caucasians.   Prevalences were similar in all non-   Caucasian groups. Prevalence of   abdominal obesity and low high   density lipoprotein cholesterol was   highest in Native Hawaiians.
Wu et al.,   2005 (49)	228 Chamorros in San   Diego, California	100	≥18	Observational, cross-   sectional telephone survey,   administrative database	To assess diabetes risk   status, incidence, and morbidity.	Data were population- and   geography-specific and may not   be generalizable. Self-reported   data.
					Diabetes prevalence was 16.2%; 60% of   respondents with diabetes were   obese as compared with 21% of those   without diabetes. Respondents   without diabetes reported more days   of moderate physical activity than   those with diabetes. More than half of   all men and women reported   consuming less than the   recommended 5 or more fruits and   vegetables per day. Prevalence of   high blood pressure was 42.5%,   higher than the nationwide 2003   Behavioral Risk Factor Surveillance   System prevalence of 24.8%.
Kaholokula et al.,   2006 (25)	80 Native Hawaiians out of   a multiethnic total sample of   190; nonpregnant, with   diabetes	50	≥30	Observational, nested case   study from population-   based sample	To examine relation between   depressive symptoms and aspects   of health-related quality of life   in type 2 diabetes.	Cross-sectional survey,   small sample sizes.
					Ethnicity moderated the   relation between depressive   symptoms and quality-of-life aspects   of physical and role-emotional   functioning, bodily pain, vitality, and   general health.
Silva et al.,   2006 (31)	614 NHOPI women out of   a multiethnic total sample of   2,155 women	28	≥18	Observational, retrospective   cohort, medical-right-   based sample	To examine ethnic differences in   perinatal outcomes among women   with gestational diabetes.	Retrospective study; ethnicity was   self-reported.
					Being NHOPI was a significant predictor   of fetal macrosomia. Higher   percentage of NHOPI women required   insulin during pregnancy and before   20 weeks’ gestation, suggesting that   there may be a larger subset of NHOPI   women with preexisting undiagnosed   diabetes.
Grandinetti et al.,   2007 (13)	510 Native Hawaiians out of   a total multiethnic sample of   1,452; nonpregnant	35	≥30	Observational, retrospective   cohort, population-based   sample	To assess prevalences of diabetes and   glucose intolerance.	Cross-sectional; self-reported   lifestyle behaviors.
					Threefold higher prevalence of diabetes   among Asian and Native Hawaiian   groups than among Caucasians;   diabetes prevalences were similar   across non-Caucasian ethnic groups   despite differences in body mass   index.
Mau et al., 2007   (32)	196 Native Hawaiians out   of a total multiethnic sample   of 793	25	≥18	Observational, cross-   sectional, community   clustered sample	To examine associations between   factors associated with chronic kidney   disease.	Cross-sectional; community sample   may have been biased.
					Chronic kidney disease was highest   among Native Hawaiians. Diabetes,   hypertension, and lower education   were significantly associated with   increased chronic kidney disease in   Native Hawaiians.
Beckham et al.,   2008 (50)	78 Native Hawaiians,   Samoans, and Tongans out   of a total sample of 116   diabetes clinic patients	67	≥18	Quasi-experimental, refusal   control group, pre-post   intervention with lack of   control for number of visits,   clinic sample	To examine the effectiveness of a   culturally tailored diabetes   management program delivered by   community health workers, using   intervention refusers as the   comparison group.	Small sample size, pre-post study   design, biased sample without true   control group. 42% of the control   group vs. 10% of the intervention   group was lost to follow-up.   Intervention group received more   visits than controls.
					72 of 80 participants in the community-   health-worker-delivered intervention   had a postintervention decrease in   hemoglobin A1c level of 2.2% (SD,   1.8), as compared with 21 of 36   participants who declined community   health worker intervention, who had   a 0.2% (SD, 1.5) decrease in   hemoglobin A1c.
Elstad et al.,   2008 (38)	64 Samoans (35 with   diabetes and 29 caregivers)	100	≥18	Focus groups, community   sample	To study perceptions of diabetes to   design a culturally appropriate   program.	Small sample size; focus groups   were mixed with persons with   diabetes and caregivers, which   may have biased responses.
					American Samoans with type 2 diabetes   experienced multiple types of stress.   Environmental and familial stress   worsened their glucose levels.   Despite the effects of family stress on   diabetes, family members were often   the primary caregivers.
Kaholokula et al.,   2008 (51)	495 Native Hawaiians from   2 rural communities in   Hawai‘i; nonpregnant	100	≥30	Observational, retrospective   cohort, population-based   sample	To examine associations between   modes of acculturation and diabetes   prevalence.	Cross-sectional survey. Modes of   acculturation included only Native   Hawaiian vs. American; other   ethnic cultures were not included.
					Native Hawaiians with a traditional mode   of acculturation were more likely to   have diabetes (27.9%) than persons   with integrated (15.4%), assimilated   (12.5%), or marginalized (10.5%)   modes.
Kim et al.,   2008 (28)	434 Native Hawaiians out of   a multiethnic total sample of   1,257; nonpregnant	35	≥30	Observational, retrospective   cohort, population-based   sample	To examine associations of   diabetes with dietary pattern   and ethnicity.	Cross-sectional; possible recall bias   on food frequency questionnaire;   measurement error in estimation of   food portions.
					Consumption of local ethnic foods was   correlated with body mass index,   smoking, waist-to-hip ratio, and   glucose. Consumption of these foods   was higher for Native Hawaiians than   for other ethnic groups. Native   Hawaiians had the highest total   energy intake.

Abbreviations: CES-D, Right for Epidemiologic Studies Depression Scale; NHOPI, Native Hawaiians and other Pacific Islanders; SD, standard deviation.

Study findings

Among studies using the NHHR data, Grandinetti et al. (5) found prevalences of type 2 diabetes and impaired glucose tolerance to be higher among NHOPI than among Caucasian participants. The overall prevalence of diabetes was 4 times higher in the NHHR participants than in the Second National Health and Nutrition Examination Survey population, and the prevalence of diabetes was also significantly higher among full Hawaiians than among part-Hawaiians. In comparison with global estimates of standardized prevalence rates (24), 1 study revealed that the prevalences of diabetes and impaired glucose tolerance among Hawaiians in the NHHR study were among the highest reported, except for Pima and Nauruan populations (5). Grandinetti et al. (5) also found that the age-adjusted prevalence of impaired glucose tolerance was higher in Hawaiian women than in men and was significantly associated with measures of adiposity (i.e., body mass index, waist circumference, and waist-to-hip ratio). Similarly, Kaholokula et al. (18) reported that increased Hawaiian blood quantum was significantly associated with higher fasting glucose concentration, body mass index, and waist-to-hip ratio.

Three studies examined the relation between ethnicity, depressive symptoms, and diabetes among NHHR participants. Among Native Hawaiians with diabetes, depressive symptoms were associated with poorer physical functioning, poorer perception of general health, more severe and limiting bodily pains, less energy, and more emotional problems (25). NHHR participants with elevated hemoglobin A1c levels reported more depressive symptoms and a lower quality of life than participants with normal hemoglobin A1c levels (26, 27). Another cross-sectional study of NHHR participants examined dietary patterns, ethnicity, and the prevalence of diabetes and found that consumption of local ethnic foods was positively correlated with body mass index, smoking, waist-to-hip ratio, fasting glucose, and 2-hour glucose (28). Native Hawaiians were found to have significantly higher consumption of these foods and the highest total energy intake in comparison with all other ethnic groups. These results suggest that total energy intake may be a more significant risk factor for diabetes than a specific dietary pattern among Native Hawaiians (28). In the nonrandomized concurrent intervention study that enrolled Native Hawaiians with diabetes or at risk for diabetes, participants in a family support intervention were more likely than a standard intervention group to advance from the pre-action stage of change to the action/maintenance stage with regard to fat intake and physical activity (29).

Three additional studies examined diabetes-related conditions. Mau et al. (30) found that the prevalence of chronic kidney disease was higher among Native Hawaiians than among Asian and Pacific Islander participants in the National Kidney Foundation’s Kidney Early Evaluation Program community screening. In a retrospective study of perinatal outcomes in NHOPI women by Silva et al. (31), a higher percentage of NHOPI women required insulin during pregnancy and before 20 weeks’ gestation, suggesting that there may be a larger subset of NHOPI women with preexisting undiagnosed diabetes.

Study limitations

A major limitation of the diabetes-related research with NHOPI populations is the lack of studies that have tested the efficacy of interventions. A large number of studies were observational, cross-sectional studies that precluded causal inferences. Several of the studies also had small sample sizes, resulting in limited generalizability.

Obesity

Study characteristics

There were 15 obesity-related studies reviewed (Table 3). Four studies examined data from the Multiethnic Cohort Study, a population-based cohort study designed to examine risk factors for cancer (i.e., obesity) that included Asian, black, Hawaiian, Latino, and white adults from Hawai‘i and California. One additional study was a population-based prospective cohort study of ethnic groups residing in Hawai‘i. Another study pooled data from 18 population-based epidemiologic studies conducted in Hawai‘i over a period of 25 years to examine trends in body mass index among different ethnic groups in Hawai‘i and to explore associations between food intake and excess weight. Seven studies were cross-sectional. Two studies, 1 cross-sectional and 1 longitudinal, examined genetic associations with body mass index among Samoans residing in American Samoa. One study was qualitative; the researchers conducted focus groups with 32 Native Hawaiian community college students to explore facilitators and barriers to living a healthy lifestyle.

Table 3.

Literature on Obesity Health Disparities Among Native Hawaiians and Other Pacific Islanders, January 1998–December 2008

Study Author(s) and Year (Ref. No.)	Sample and No. of Subjects	% of Total Sample Who Were NHOPI	Age, years	Study Design	Main Goal and Outcomes/Findings	Limitations
Maskarinec et al.,   1998 (52)	4,321 Native Hawaiians out of   a total multiethnic sample of   27,678	16	>30	Observational, prospective   cohort, population-based   sample of 2% of the Hawai‘i   State population	To investigate effects of alcohol intake and   body weight on mortality from all causes,   cancer, cerebrovascular disease, and   coronary heart disease.	No information on preexisting   cardiovascular disease, serum   cholesterol level, hypertension,   diabetes, or family history.
					Native Hawaiians had the highest mortality   rate and a higher prevalence of obesity.   BMIa >29.3 was associated with 50%   higher risk of death. Coronary heart   disease mortality was higher in Native   Hawaiians with BMIs >29.3.
Galanis et al.,   1999 (53)	946 Samoans in Western   Samoa and American   Samoa	100	25–55	Observational, cross-   sectional, retrospective   cohort, community and   workplace sample	To describe dietary intake as measured   by 24-hour recall of American Samoans   and Western Samoans.	No biochemical measurement of   cardiovascular disease risk or   correlations with anthropometric   measurements; potential sample   bias; recall bias on dietary   assessment.
					American Samoans consumed a   greater proportion of carbohydrates and   protein but less fat or saturated fat than   Western Samoans. The mean BMI of   American Samoans was 35.2 as   compared with 30.3 for Western   Samoans.
Grandinetti et al.,   1999 (32)	567 Native Hawaiians in 2   rural communities in Hawai‘i;   nonpregnant	100	≥30	Observational, retrospective   cohort, population-based   sample	To assess the relation of degree of Native   Hawaiian ancestry with BMI and waist-   to-hip ratio.	Cross-sectional; self-reported   ancestry and dietary recall.
					Combined prevalence of overweight and/   or obesity was 81.5% in Native   Hawaiians as compared with the US   prevalence of 52.6%. 49% of Native   Hawaiians were obese as compared   with the US prevalence of 21%    Increased waist circumference was   found in 51% of Native Hawaiians. More   women (59%) than men (39%) had   increased waist circumference. Age,   percentage of Native Hawaiian ancestry,   and total dietary energy intake were   significantly associated with increased   BMI and waist-to-hip ratio. Adiposity   increased with percentage of Native   Hawaiian ancestry.
McGarvey et al.,   2002 (54)	181 Samoans and   American Samoans	100	25–55	Observational, nested study   in a prospective cohort;   population-based sample	To test the association of 6 genetic   microsatellite markers related to the   human leptin (LEP) locus and the pro-   opiomelanocortin gene region in adult   Samoans and American Samoans.	Functional significance of present   finding remains unclear.
					Significantly greater frequency of allele   226 at the LEP locus in the nonobese   Samoans than in the obese subjects.
Mampilly et al.,   2005 (36)	585 Native Hawaiians out of   a total multiethnic sample of   3,732	16	≥18	Observational, cross-   sectional telephone survey;   Behavioral Risk Factor   Surveillance System   population sample	To assess the physical activity levels of   multiethnic groups in Hawai‘i.	Telephone survey; self-report of   physical activity. Self-reported   weight and height.
					Native Hawaiians were more   active (39% moderate, 24% vigorous)   than other Asians and Pacific Islanders   but less active than whites. 48% of   Native Hawaiians reported being   overweight as compared with 58% of   whites, 42% of Filipinos, and 41% of   Japanese.
Henderson et al.,   2006 (55)	159 Native Hawaiians out of   a random subsample of 811   persons from Hawai‘i and   California	20	45–74	Observational, nested study   in a prospective cohort;   randomly selected sample	To examine the relation between   circulating levels of 2 primary proteins   (IGF-1 and IGFBP-3) in the insulin-like   growth factor pathway and obesity in 5   racial/ethnic groups using BMI as an   indicator of adiposity.	Unmeasured confounding factors;   racial/ethnic differences in BMI   cutpoints may have confounded   study. Plasma IGF-I and IGFBP-3   were measured at a single time   point.
					No significant interaction was   found between IGF-1 and BMI in Native   Hawaiians as compared with Japanese   and Latinos, in whom decreased IGF-1   was associated with increasing BMI.
Howarth et al.,   2006 (56)	433 Native Hawaiians out of   a total multiethnic sample of   2,326 persons from Hawai‘i   and California	19	45–74	Observational, prospective   cohort; population-based   sample	To determine whether dietary   energy density was related to current   BMI and risk of overweight/obesity in   a multiethnic population.	Food frequency assessment of dietary   energy density based on recall.   Dietary recall may vary by body   weight.
					Native Hawaiian men had the highest BMI;   weight and dietary energy density were   significantly related to BMI. Native   Hawaiian women were second-heaviest   after African Americans. Higher energy   density was significantly associated with   greater likelihood of being overweight in   all ethnic and sex groups. Native   Hawaiians were unusual in that low   energy density was associated with high   BMI.
Maskarinec et al.,   2006 (34)	9,994 Native Hawaiians out of   a total multiethnic sample of   76,163	13	≥18	Observational, prospective   cohort; pooled data from 18   population-based studies   carried out over 25 years in   Hawai‘i	To describe trends in BMI and the relations   of nutrient and food intake with excess   weight.	Limited in comparability of nutritional   measures across studies over time,   mainly because dietary assessment   methods have improved over the   years; recall bias.
					Native Hawaiians had the highest   prevalence of excess weight at all times.   Carbohydrates were positively   associated with excess weight among   Native Hawaiians. Nutritional   determinants of excess weight were   similar among whites, Japanese, and   Native Hawaiians, despite marked   differences in BMI.
Albright et al.,   2007 (33)	12,306 Native Hawaiians out   of a multiethnic total sample   of 200,003	6	45–74	Observational, prospective   cohort; population-based   sample	To examine BMI in persons with ethnic   admixture as compared with persons   who were monoracial but shared   a common ethnicity/race.	Self-reported height, weight, and   ethnicity; unable to quantify the   genetic contribution of each ethnicity   reported.
					Native Hawaiians had the highest BMI   compared with other ethnic groups.   Ethnic admixtures that included Native   Hawaiian heritage had higher BMIs than   most other ethnic combinations.
Boyd, 2007 (37)	32 Native Hawaiian   community college students	100	18–25	Focus group, university   convenience sample	To assess perceptions of healthy lifestyles   and supports for and barriers to healthy   living.	Small sample size; biased sample.
					Demanding lifestyle and laziness were   cited as barriers to being physically   active. Preferences for group-oriented   and college-course-based opportunities   to learn about healthy living and how to   become more physically active.
Dai et al., 2007 (57)	583 American Samoans	100	≥18	Observational, nested study   in a prospective cohort;   random selection sample	To detect trait loci influencing adiposity-   related phenotypes using a whole   genome linkage scan approach in   families from American Samoa.	No adjustment for other important   genetic and environmental factors   that contribute to adiposity, such as   diet and physical activity.
					Strong evidence for a major locus on   chromosome 6q23.2 influencing serum   leptin levels. Another genetic region,   16q21, appears to be a susceptibility   locus that affects phenotypes for BMI,   percentage of body fat, leptin levels, and   waist circumference.
Maskarinec et al.,   2007 (35)	16,079 Native Hawaiians out   of a multiethnic total sample   of 117,065	14	≥40	Observational, prospective   cohort; population-based   sample	To investigate changes in risk factors in   Hawai‘i over 20 years and compare   health behaviors among ethnic groups.	Differences in data collection across   2 studies; improvements in   nutritional assessment over time;   validity of ethnicity assignment (self-   reported); little information on   socioeconomic status.
					Native Hawaiians had the highest chronic   disease risk scores in comparison with   other groups. BMI was highest for Native   Hawaiians.
Novotny et al.,   2007 (58)	55 Samoan women	100	18–28	Observational, cross-   sectional, convenience   sample	To examine anthropometric cutoff points   as indicators of chronic disease risk.	Cross-sectional, small sample of   Samoan women; biased sample.
					80% of sample was overweight or obese.   BMI and dual-energy X-ray   absorptiometry lean mass were   significantly and positively associated   with glucose levels.
Maskarinec et al.,   2008 (59)	254 Native Hawaiian women   out of a multiethnic total   sample of 1,418	18	≥21	Observational, cross-   sectional, breast-cancer-   related participant sample	To examine relation of soy intake with   body weight over the life span of women.	Self-reported height, weight, and diet   data; lifetime recall of soy intake was   difficult.
					Meat and vegetable intake   and total energy intake (kcal/day) were   higher for Native Hawaiians. Eating more   soy foods in adulthood did not predict   lower BMI or lower annual weight gain for   Native Hawaiians.

Abbreviations: BMI, body mass index; IGF-1, insulin-like growth factor 1; IGFBP-3, insulin-like growth factor binding protein 3; NHOPI, Native Hawaiians and other Pacific Islanders.

^aWeight (kg)/height (m)².

Study findings

Prevalences of overweight and obesity were consistently higher among Native Hawaiians than in other ethnic groups (whites, blacks, Latinos, Asians, and Filipinos) across studies. Grandinetti et al. (32) reported a combined prevalence of 82% for overweight and obesity in NHHR study participants, as compared with a national prevalence of 53%; 49% were obese as compared with 21% nationally. Body mass index was also higher in persons with an ethnic admixture that included Native Hawaiian ancestry, as compared with most other ethnic combinations (32, 33). In pooled data from 18 population-based studies carried out over 25 years, Native Hawaiians had the highest prevalence of excess weight at all times (34).

Energy intake was consistently higher among NHOPI than in other ethnic groups in Hawai‘i. Both the NHHR and Multiethnic Cohort studies found that total dietary energy intake was significantly associated with Native Hawaiian ancestry and increased body mass index (32, 34). In 2 large population-based prospective studies, Native Hawaiians had the highest chronic disease risk scores in comparison with other ethnic groups, primarily because of high prevalences of overweight and obesity, higher rates of smoking, and chronic alcohol use (35). In the NHHR study, increased body weight was strongly associated with glucose intolerance (5). Despite the high prevalence of overweight and obesity, NHOPI reported a higher prevalence of physical activity in the 2001 Hawai‘i Behavioral Risk Factor Surveillance System survey than did other ethnic groups (36).

Study limitations

The obesity-related studies reviewed had several limitations. First, the cross-sectional design of many of the studies did not allow for determination of causal relations. Second, many of the studies were questionnaire-based and may have been vulnerable to recall bias or a propensity towards giving socially desirable answers. Finally, investigators in several studies were unable to measure confounding variables, which limited the potential for understanding the true association between exposure and disease.

DISCUSSION

Studies of cardiometabolic disparities among NHOPI are sparse. The 43 studies in this review were published in the last 10 years and provide growing evidence that NHOPI are one of the highest-risk US populations affected by cardio-metabolic diseases. Some progress has been made in addressing these disparities, as evidenced by the handful of studies that have shifted from observational research towards program development and then to experimental and clinical trial-type studies that include NHOPI. However, there were a number of methodological issues apparent during the course of this literature review. For example, there were several studies that were limited by sample bias (convenience samples, etc.) and relatively small sample sizes (i.e., <50 subjects). NHOPI comprise less than 1% of the US population, and thus recruitment of NHOPI into research studies remains a challenge. Despite these challenges, a number of research teams have been successful in enrolling sufficient-sized samples or have taken advantage of existing data or administrative databases to better understand cardiometabolic diseases in this population. Moreover, recent developments in the use of community engagement approaches have served to increase the participation of this population in research activities and ensure that studies are relevant and translatable to NHOPI communities.

Aggregation of NHOPI with Asian Americans in several publications limited the number of available studies for this review. There was also a paucity of experimental studies that were adequately designed to reduce treatment bias (i.e., randomization) and longitudinal prospective cohort studies that would allow elucidation of cause-and-effect relations in cardiometabolic diseases. However, a few focus groups and quasi-experimental studies provided preliminary data that offer potential for designing intervention studies in the future.

Several studies (29, 37–40) provided initial insights on promising approaches in NHOPI populations, such as social and/or family support and the inclusion of cultural and/or traditional healing methods as alternatives or supplements to conventional medical regimens. Other studies provided empirical evidence with which to develop scientifically informed and culturally specific diet-based interventions for prevention and treatment of cardiometabolic disparities. Health care differences in cardiovascular disease treatment suggest that more study is needed in order to determine the best medical treatments for high-risk ethnic groups such as NHOPI (20, 21). Further investigation is needed to examine both provider factors and patient factors that may underlie the treatment differentials between patients who may receive different treatments and hence have different outcomes.

There remain significant gaps in our understanding as to why cardiometabolic diseases occur more frequently in the NHOPI population in the United States (Figure 2). Any number of factors, alone or in combination, may contribute to the creation of disparities in health within this population. Compared with most other US ethnic groups, NHOPI are overrepresented in the lower socioeconomic strata, under-represented in higher education, and more likely to be marginalized from the larger society (41). Behavioral risk factors for diabetes and cardiovascular disease, such as tobacco use and psychological distress, are highly prevalent in NHOPI (42). In the case of Native Hawaiians, many health professionals have suggested that the health disparities experienced by Native Hawaiians are associated with their lower social status and adverse historical relations with Western governments (43, 44). Thus, it would seem appropriate in future studies to explore psychosocial stressors that may contribute to health disparities in NHOPI.

Figure 2.

Conceptual model of health and health-care disparities.

Future research aimed at eliminating cardiometabolic disparities in health and health care among NHOPI needs to move beyond observational studies into intervention studies that will engage NHOPI communities in the process while maintaining scientific rigor. Researchers should consider the whole spectrum of types of scientific studies—ranging from genetic, bench studies to clinical studies to effectiveness studies that test interventions in real-world settings. NHOPI can participate in this research not only as study subjects but also as investigators. In this way, they can both obtain health equity and, more importantly, help to promote health and wellness for all.

Abbreviations

CABG

coronary artery bypass graft

NHHR

Native Hawaiian Health Research

NHOPI

Native Hawaiians and other Pacific Islanders