Breastfeeding Practices among Native Hawaiians and Pacific Islanders

Background. Breastfeeding is associated with a decreased risk of obesity in the early and adult years. Native Hawaiians and Pacific Islanders (NHPI) experience high rates of obesity which is often obfuscated with aggregated data. Using disaggregated data, we examined breastfeeding practices among NHPI. Methods. Seven databases and reference lists were searched. Two independent researchers extracted relevant studies based on predetermined criteria. Nine studies met our inclusion criteria and a meta-analysis was conducted using random-effects, inverse-various weighted models. Results. Few studies disaggregated NHPI populations when examining breastfeeding practices. Most studies were cross-sectional and our search yielded no randomized or quasirandomized control trials. The results of the meta-analysis indicated that 46.5% NHPI women initiated breastfeeding with 40.8% breastfeeding exclusively. These pooled analyses show that NHPI breastfeeding practices are below the recommended national and international goals and guidelines. Conclusion. Breastfeeding practices among NHPI are heterogeneous and critical disparities exist among certain NHPI subgroups and additional research needs to be conducted to determine the reasons for the disparity. Future studies should work to disaggregate data for NHPI and the various subpopulations. Multicomponent, multilevel strategies are needed to support breastfeeding practices among NHPI.

Native Hawaiian and Pacific Islanders (NHPI) are a rapidly growing population and constitute approximately 1.2 million of the Asian American population either alone (44%) or in combination with other races (56%) [24]. Because NHPI are often conflated with Asian Americans and other Pacific Islanders in national and state-level data, disparities among them are often unnoticed. For example, when breastfeeding data are disaggregated by race and ethnicity, studies show that NHPI have lower breastfeeding initiation and EBF rates and have shorter breastfeeding duration than other populations [25][26][27]. This is concerning because NHPI populations experience higher rates of obesity and obesityrelated comorbidities and mortality than other populations, but improving breastfeeding practices can ameliorate obesityrelated issues [28][29][30][31][32][33][34]. Although steps have been taken to separate Asian and NHPI in census data, the problem of aggregating these groups is still prevalent [35].
Currently, there are no systematic reviews on breastfeeding practices that disaggregate data on NHPI. The information from such a systematic review can provide policy makers, researchers, and public health workers a means to support and enhance breastfeeding practices among NHPI while addressing the problem of high obesity. Therefore, the objectives of this study are to provide a systematic review and meta-analysis of breastfeeding practices used by NHPI in terms of initiation, EBF, and duration.

Methods
Searches were conducted in July 2013, October 2014, and updated in January 2016 in the online bibliographic databases PubMed, AGRICOLA, CENTRAL, CINAHL, PsychInfo, Sociological Abstracts, and Web of Knowledge for studies examining breastfeeding practices among NHPI. We used Medical Subject Heading (MeSH) terms in PubMed and other indexing terms for the respective databases, as well as text wording. Our initial searches combined the following terms: breast feeding, breast-feeding, breastfeeding, breastfed, breast fed, breast-fed, child, preschool, infant, Pacific Island, Oceanic ancestry group, Hawaii, breastfeed, bottle feeding, formula fed, formula milk, human milk, infant feeding, weaning, child preschool, infant, and pregnancy. We did not include language or date restrictions in our searches. Two researchers (IA and AK) independently scanned the reference lists of all relevant papers retrieved and extracted relevant studies using predetermined inclusion and exclusion criteria. Discrepancies were resolved through discussion so that consensus could be attained.

Inclusion.
Studies were included if data were disaggregated for NHPI populations and if they were directly related to breastfeeding practices. Excluded articles were those that were conducted outside of the United States or the Pacific Islands (Micronesia, Melanesia, and Polynesia) or qualitative in nature.

Coding.
Two researchers independently coded studies for type, sample size, outcome measures, and effect size. The effect size of interest was the proportion of mothers who adopted certain breastfeeding practices (initiation, exclusive breastfeeding, and duration) or children on whom these practices were used.

Synthesis.
We performed two synthesis techniques: (1) a narrative report of breastfeeding practices among NHPI samples in all studies and (2) a meta-analysis of various breastfeeding practice measures using a random-effects, inversevarious weighted model, with effect sizes grouped according to similar measurement constructs. For example, we synthesized measures of breastfeeding initiation separately from breastfeeding duration and exclusivity. Heterogeneity among the effect sizes was calculated using 2 index [43]. All analyses were conducted using R package metafor [44].

Systematic Review
We identified 718 articles from which we removed 336 duplicates; and, of 382 articles screened for potential relevance, 348 were excluded. Thirty-four articles were assessed for eligibility. Further studies were excluded because they (1) were outside of the United States or Pacific Islands ( = 8), (2) addressed complementary feeding or did not disaggregate NHPI from other ethnic groups ( = 16), (3) described interventions (without reporting effects) ( = 1), and were secondary data from other included studies. Hence, nine studies were included in our final results. See Figure 1

Breastfeeding Initiation
Four studies examined breastfeeding initiation, three among Hawaiians and one in the Pacific Islands. In the first study, an initiation rate of 29.0% for Hawaiian mothers was observed as compared to 26.0% Caucasians and 50.0% Japanese [36]. In the second study, a 52.6% breastfeeding rate at hospital discharge was observed for Native Hawaiian women as compared to 70.4% Caucasian, 55.4% Japanese, and 33.3% Filipino [39]. In the third study, a 1-month initiation rate of 91.0% was observed for Native Hawaiian mothers attending WIC clinics [27]. Finally, among Pacific Islanders, all mothers on Honiara and Nggela-Sandfly breastfed their children [37]. The results of the meta-analysis indicated that less than half of mothers initiated breastfeeding (M = 46.5%, 95% CI (26.9-66.1), 2 = 95.22).

Breastfeeding Duration
Four studies in our review measured breastfeeding duration, three among Pacific Islanders and one among Native Hawaiians. The first study observed that 45.7% of all infants in Kiribati were breastfed at 12 months [38]. The second study among mothers in the Commonwealth of the Northern Marianas Island reported that 22.0% received breastmilk for 12 or more months [42]. The third study [37] mentioned that most mothers in both Honiara and Nggela-Sandfly were breastfeeding at 6 and 12 months. The fourth study in Oahu [27] reported a 6-month breastfeeding duration rate of 52.0% for all mothers. Due to the variability in measures of breastfeeding duration a meta-analysis could not be appropriately performed.

Discussion
The high incidence of obesity, low breastfeeding rates, and other chronic diseases among NHPI makes it imperative to study differences in breastfeeding practices in these populations. Our review found that breastfeeding practices among NHPI are heterogeneous, highlighting disparities in breastfeeding practices among certain NHPI subgroups [26]. Examining breastfeeding practices among NHPI can expose hidden health issues and trends, making it possible to unearth disparity [45]. Our systematic review yielded only nine studies spanning a 44-year period. Moreover, we found limited disaggregated data in national surveys and surveillance systems for these groups. We found no studies examining breastfeeding patterns and disaggregating data for NHPI in the US where approximately two-thirds of NHPI reside [34]. The NHPI population is a young population that is expected to increase   [46,47], and it is also critical that mandates related to disaggregating NHPI data by distinct ethnic and racial groups separate from Asian Americans are implemented [34,46]. There is a need to enhance the monitoring of breastfeeding practices in all US NHPI states, territories, and commonwealths to better identify risk factors for poor breastfeeding outcomes.
In addition, we found that varying terms and time periods were used for breastfeeding initiation, exclusivity, and duration among studies in our review. Such variations have also been observed among federally funded datasets, indicating a need for standardization of breastfeeding outcomes [45]. Future studies should implement consistent terminology for the purpose of comparing effects and replicating studies. In addition, it is important that studies use time frames (breastfeeding at 6 and 12 months and exclusive breastfeeding at 3 and 6 months) that are aligned with national objectives and national data collection and monitoring systems.
The result of our meta-analysis indicated that less than half of NHPI women initiated breastfeeding or breastfed exclusively. These pooled analyses show that NHPI breastfeeding practices are below the Healthy People 2020 goals. Qualitative studies have highlighted several factors that may affect breastfeeding practices among NHPI, namely, lack of knowledge, lactation problems, poor family and social support, social norms, embarrassment related to breastfeeding in public, employment and child care issues, and the education of health care providers, so that they can fully support the breastfeeding efforts of NHPI mothers [35,48,49]. Further research needs to be conducted to determine the reasons for the disparity in breastfeeding practices. We suggest, at a minimum, disaggregating survey data by the race/ethnicity in order that future analyst can use such data. Our search produced neither randomized nor quasirandomized controlled trials related to breastfeeding among NHPI. Studies are also needed to systematically explore the types of interventions that can enhance exclusive breastfeeding rates and duration among NHPI.
There are few important considerations that need to be noted in interpreting the findings of this review. The unclear definitions of breastfeeding terms, variability in the measurement periods, and the small number of studies are limitations. However, despite these concerns, the need for this study cannot be overemphasized as it illuminates disparities within the NHPI populations and brings to light the needs of a group that could potentially remain on the outskirts of policy decisions and focus.

Conclusion
Future studies should work to disaggregate NHPI data and identify barriers to breastfeeding among NHPI. Moreover, multicomponent, multilevel strategies directed toward NHPI communities are needed to support breastfeeding practices. Such studies can support existing culturally appropriate practices, advance health promotion activities, and provide directions for policies to minimize health-related morbidity, mortality, and disparities among the NHPI population.