Diet quality of Marshallese mothers of young children in Northwest Arkansas: an exploratory study

Abstract

Objective:

To characterize the dietary patterns of Marshallese mothers of young children in Northwest Arkansas, informing the cultural adaptation of nutrition education curricula.

Design:

An exploratory cross-sectional study was conducted, in which Marshallese women with children under 12 months completed 3 telephone-administered 24-hour dietary recalls with a trained bilingual Marshallese interviewer. Diet quality was characterized using the Healthy Eating Index (HEI)-2020. A food-level analysis identified top food groupings contributing to total energy and HEI-2020 components.

Setting:

Northwest Arkansas.

Participants:

Marshallese mothers with children < 12 months.

Results:

29 women were recruited, 20 completed 2 or 3 dietary recalls. Median age was 25·5 years. Diet quality by HEI-2020 was 46·4 (max score 100). White rice was the top contributor to total energy; high seafood/plant protein and fatty acid diet quality component scores were influenced by high fish intakes.

Conclusions:

Diet quality was low. Key adaptations include reducing rice portion sizes, while emphasizing lean proteins and fruits/vegetables. Cultural adaptation of nutrition education is essential to improve diet quality among communities with varying dietary practices.

Childhood obesity in the United States (US) has tripled since the 1970’s, reaching 19·7 % in 2017–2020⁽¹⁾. Pacific Islanders face disproportionately higher rates of obesity and obesity-related chronic conditions than other racial and/or ethnic minorities in the US^(2,3). While information is limited, available data suggest Pacific Islander children in the continental US have an obesity prevalence ranging from 17·2 % to 48·6 %⁽⁴⁾. Arkansas has the largest population of Marshallese Pacific Islanders living in the continental US (i.e. migrants from the Republic of the Marshall Islands (RMI))^(5,6). A health screening study of 401 Marshallese adults in Arkansas revealed that 90 % were overweight or obese⁽⁷⁾.

The historical influences of US nuclear testing in the RMI in the 1940–1950s led to a nutrition transition among the Marshallese, contributing to a shift in dietary patterns^(5,8). Whereas Marshallese were once able to consume nutritious produce and seafood produced on the islands, the effects of prolonged radiation contaminated water, plants, and seafood resources^(5,9). Consequently, the US provided imported foods to the RMI; however, foods consisted of highly processed items, including canned meats (e.g. SPAM®), ramen, white rice, and other shelf-stable products^(3,10). Recent qualitative research with Marshallese communities in Arkansas continue to confirm the preferences for and consumption of these foods, with Marshallese infants traditionally receiving early introduction of high simple carbohydrate foods (e.g. rice, starchy fruits like bananas or breadfruit, juice, pureed fruit with evaporated canned milk)^(11–13). Early introduction of solid foods, high maternal pre-pregnancy BMI, and dietary patterns low in fruits and vegetables and high in refined grains, are associated with increased risk of childhood obesity^(14,15). At the same time, nutrition-focused interventions in early childhood (i.e. conception through 24 months) are important in the prevention of childhood obesity⁽¹⁶⁾. The traditional dietary practices among Marshallese mothers must be considered in the development of a culturally appropriate childhood obesity prevention program for this group^(12,13).

Marshallese mothers and caregivers are important decision-makers for feeding practices among children^(12,13). Prior research has identified collectivist values (group over individual) among Marshallese communities, wherein families often eat together from ‘one pot,’ sharing similar dietary patterns⁽¹⁷⁾. Thus, a first step within the development of childhood obesity prevention interventions for Marshallese individuals is to first understand the dietary patterns among Marshallese mothers, which provides insight into food consumed across the entire family. Much of the prior research investigating dietary practices of Marshallese families living in the US has been qualitative and has not included measures of diet quality^(12,13). Further, parental interventions addressing obesity are often not culturally tailored and may not be effective for the Marshallese population where both westernized (i.e. imported rice) and traditional foods and collectivist approaches are more common/valued⁽¹⁸⁾.

‘CenteringParenting’ is a group-model parenting intervention that takes place from six weeks through 12 months of a newborn’s life, including a series of nine group visits (90–120 min each)⁽¹⁹⁾. The visits follow a structured curriculum, promoting exclusive breastfeeding, appropriate introduction of complementary foods (i.e. foods given to babies in addition to breast milk around 6 months of age), and healthy dietary patterns among infants and mothers⁽¹⁹⁾. Group model parenting interventions have been shown to improve postpartum and well-baby checkups, infant vaccinations, and exclusive breastfeeding in other populations^(20–22). Prior research evaluating the feasibility and acceptability of this group-model parenting intervention has not included Marshallese women^(20,23). Further, culturally adapted intervention approaches using community-based assets and Marshallese cultural values/practices have demonstrated effectiveness in significantly decreasing obesity among adults, but have not been focused on obesity in Marshallese children living in the continental US⁽²⁴⁾. The purpose of this exploratory study was to provide a rich quantitative characterization of a typical diet consumed by Marshallese women with children living in the continental US to inform a culturally appropriate nutrition education curriculum to be used in a future childhood obesity prevention intervention in Arkansas.

Methods

Approach

A Community-Engaged Research (CEnR) approach was used in the design and implementation of this study. CEnR is an approach that may be used to honor and integrate Marshallese cultural values and practices into every aspect of research⁽²⁵⁾. To ensure cultural appropriateness, this study was guided by a Community Advisory Board that included local health care professionals, Marshallese community members, and an interprofessional research team. The interprofessional research team included nutrition/public health and qualitative researchers, as well as Marshallese bilingual study staff to provide feedback on study materials and input on how to modify the CenteringParenting curriculum to be culturally appropriate for Marshallese participants.

Study design, participants, recruitment

The present study was conducted within the context of a larger multi-phase feasibility and acceptability intervention study. Phase one aimed to quantitatively characterize the dietary patterns of Marshallese mothers of young children (described herein), phase two included the cultural adaptation of an abbreviated version of the curriculum, and phase three tested the feasibility and acceptability of the adapted curriculum. A cross-sectional study design was used to describe the diet quality of 20 Marshallese mothers of children under 12 months in Northwest Arkansas. The target sample size for phase one was chosen based on the financial and time constraints of funding and in consultation with the study biostatistician. Marshallese women were recruited from May-August 2023 by trained bilingual female Marshallese community health workers (CHWs) through partnerships with multiple community organizations and programs, including a Healthy Start program, the Marshallese Education Initiative (MEI), Arkansas Coalition of Marshallese (ACOM), and Marshallese pastors. Eligibility criteria included being female, an adult (≥ 18 years), self-reported Marshallese, and having a child/children under 12 months of age. Potential participants who met the eligibility criteria were offered the opportunity to join the study and complete the written informed consent process. The bilingual study staff read the consent aloud to the potential participants in their language of choice (English or Marshallese).

The Community-Engaged Research team used an engaged approach to collaboratively develop a retention plan with Marshallese stakeholders. The retention plan specified that all study staff responsible for retention are bilingual (Marshallese/English). Marshallese bilingual CHWs obtained each participant’s contact information and preferred method of contact. Each participant received a $20 gift card upon completion of each 24-hour dietary recall.

Survey data collection

Upon enrollment into the study, participants completed a 10-item sociodemographic survey in-person. Demographic data collection included age, household size (number of children < 18 years, number of adults), length of time residing in the US, relationship status, education, employment status, health care coverage, birthplace, and participation in the Special Supplemental Nutrition Program for Women, Infants, and Children (WIC). Participation in other federal assistance programs (e.g. SNAP) was not asked of participants, since Marshallese migrants residing in Arkansas during the time of data collection did not qualify for these programs, as Compact of Free Association (COFA) citizens^(26,27). Data were entered directly into the Research Electronic Data Capture (REDCap) tool⁽²⁸⁾.

Dietary assessment

The dietary recall data collection protocol included collecting three 24-hour dietary recalls on non-consecutive days within a one-month time frame, including two weekdays and one weekend day for each participant. Collecting and averaging data from at least two non-consecutive dietary recalls per participant is recommended to provide a cross-sectional estimate of an individual’s usual dietary intake, accounting for day-to-day variation⁽²⁹⁾. A trained bilingual Marshallese CHW familiar with traditional Marshallese foods and recipes collected all dietary recalls. The Nutrient Data System for Research (NDSR) software version 2022 (Minneapolis, MN) was used to facilitate a multiple-pass approach to dietary data collection to obtain complete intake data. NDSR provides intake estimates for 178 nutrients, nutrient ratios, and other food components and has been used to assess the dietary intake of racially/ethnically diverse populations⁽³⁰⁾. Following data collection, a Registered Dietitian Nutritionist (RDN)/Nutrition researcher (ES) met with the data collector to discuss culturally-specific foods and recipes reported in dietary recalls, and ensure accurate capture within NDSR. For example, dietary recalls contained multiple grain products, including pancakes and rolls. Upon discussing the preparation of these foods, a closer match in NDSR was substituted to more accurately capture ingredients used (e.g. switching a plain dinner roll to a sweeter Hawaiian roll).

Data analysis

Descriptive statistics were used to provide a rich characterization of the diet quality of Marshallese mothers of children under 12 months. Survey and dietary intake data were summarized using frequencies and percentages (categorical variables) or median and interquartile ranges (IQR) (continuous variables).

Dietary intake data was applied to the Healthy Eating Index (HEI)-2020, a tool used to evaluate the extent to which dietary intake meets the Dietary Guidelines for Americans (DGAs) 2020–2025⁽³¹⁾. The HEI is regularly updated to reflect the current release of the DGAs; thus, the HEI-2020 was chosen as the appropriate version to evaluate data collected during the 2020–2025 release of the DGAs⁽³¹⁾. The HEI has been used widely to describe diet quality among diverse populations, including pregnant and breastfeeding individuals^(32–35). The HEI-2020 consists of 13 separate nutrient components summed to create a total score (0–100), with higher scores representing higher diet quality. The HEI-2020 consists of nine adequacy components to emphasize in the diet (e.g. total fruit, whole fruit, total vegetables, greens and beans, whole grains, dairy, total protein, seafood and plant protein, fatty acids) and four dietary components to consume in moderation (e.g. refined grains, sodium, added sugars, saturated fat). Some HEI-2020 nutrient components include groupings of foods (e.g. greens and beans, seafood and plant proteins) to align with Dietary Guidelines for Americans healthy eating pattern recommendations to consume a variety of protein foods, including seafood, lean meats and poultry, eggs, legumes, nuts, seeds, and soy products^(31,36). The National Cancer Institute’s Simple HEI scoring algorithm was used to derive ratios from the dietary data for each of the 13 nutrient components⁽³⁷⁾. For example, the ratio for total fruit was calculated using the following formula: total cup-equivalents of fruit consumed/(total energy consumed/1000). Nutrient component ratios were averaged across the three dietary recalls and then applied to scoring standards. The 13 component scores were then summed to create an overall HEI-2020 score for each participant. The four moderation components were scored so that higher scores represent a lower (and recommended) intake.

To provide an in-depth understanding of foods contributing to the HEI-2020 component score results and inform specific nutrition curriculum adaptations, a food-level analysis was conducted using methods described by Taylor et al. ⁽³⁸⁾ This analysis looked across all dietary recalls to identify the foods/beverages that contributed the highest percentage to total energy consumption and to each HEI-2020 nutrient component. Individual foods/beverages reported in dietary recalls were first grouped into categories comparable to the What We Eat in America (WWEIA) member-level food categories using an NDSR supplemental data file^(39,40). To calculate the percentage (%) contribution to total energy and HEI-2020 components by each food category, the following formula was used:

The top food categories contributing to total energy and HEI-2020 nutrient components were identified. Analyses were conducted using Stata 18.0 (StataCorp LLC, 2023).

This study was reviewed and received expedited approval by the University of Arkansas for Medical Sciences Institutional Review Board (#274 752).

Results

Sample characteristics

Twenty-nine participants were recruited, 20 participants completed two (n 2) or three (n 18) 24-hour dietary recall interviews, and 18 participants completed the demographic survey. Two participants did not complete the demographic survey due to human error within data collection. The two or three dietary recall interviews were completed across a median (IQR) of 18 (8·5–36) days. Participants were a median (IQR) age of 25·5 (22·3–32·8) years, with an average of 4·0 (range: 3·0–6·0) children and an average of 4·5 (range: 4·0–6·0) adults living in their household (including themselves) (Table 1). Participants reported living in the US between two to 33 years, and almost all participants were born in the Marshall Islands (n 16, 89 %). Half of the sample reported high school graduate as their highest level of education (n 9, 50 %). Seven participants (39 %) reported being currently employed, 8 (44 %) reported currently taking care of their family/home, and 3 (17 %) reported being out of work. Almost all women were enrolled in WIC (n 16, 89 %).

Table 1.

Sociodemographic characteristics of Marshallese mothers (n 18) ^*

Characteristic	Median	IQR
Age (years)	25·5	22·3–32·8
Number of children living in household	4·0	3·0–6·0
Number of adults living in household	4·5	4·0–6·0
Years living in the US	11·5	7·0–18·3
	Frequency	%
Relationship status
Single	7	39 %
Married	7	39 %
Unmarried couple	4	22 %
Education
Some high school	5	28 %
High school graduate	9	50 %
Some college or technical school	1	6 %
College graduate	3	17 %
Employment status
Out of work for less than 1 year	1	6 %
Out of work for 1 year or more	2	11 %
Taking care of your family/home	8	44 %
Employed	7	39 %
Health care coverage †
Yes	9	50 %
No	7	39 %
Don’t know/not sure	2	11 %
Birthplace
Marshall Islands	16	89 %
US born	1	6 %
Other	1	6 %
Enrolled in WIC
Yes	16	89 %
No	2	11 %

WIC: Special Supplemental Nutrition Program for Women, Infants, and Children.

^* Two participants did not complete the sociodemographic survey but completed the dietary recalls, n 18.

^†Do you have any kind of health care coverage, including health insurance, prepaid plans such as HMOs, or government plans such as Medicare?

Diet quality

The median (IQR) HEI-2020 score across the sample was 46·4 (43·0–48·5) out of a possible 100 (Table 2). HEI-2020 nutrient component scores were high within total protein (5·0 (5·0–5·0)), seafood/plant protein (5·0 (5·0–5·0)), fatty acids (7·3 (5·3–8·7)), added sugars (10·0 (10·0–10·0)), and saturated fats (9·9 (7·7–10·0)), indicating more favorable intake. Notably, the adequacy components total fruit, whole fruit, greens and beans, and whole grains, had a score of 0·0 (i.e. median participant intake of no fruit, whole fruit, greens and beans, or whole grains). Scores for total vegetables (0·9 (0·1–1·8)), dairy (0·4 (0·0–1·6)), and refined grains (0·0 (0·0–0·0)) were also low. The score for sodium was moderate (6·0 (3·7–9·7)).

Table 2.

Median diet quality of Marshallese mothers of young children (n 20)

Diet quality component (Minimum to maximum possible score)	Median	IQR	Standard for maximum score	Standard for minimum score
Total HEI-2020 Score (0–100)	46·4	43·0–48·5	–	–
Adequacy components
Total fruit (0–5) *	0·0	0·0–1·1	≥ 0·8 c eq/1000 kcal	No fruit
Whole fruit (0–5) †	0·0	0·0–0·0	≥ 0·4 c eq/ 1000 kcal	No whole fruit
Total vegetables (0–5) ‡	0·9	0·1–1·8	≥ 1·1 c eq/1000 kcal	No vegetables
Greens and beans (0–5) ‡	0·0	0·0–0·3	≥ 0·2 c eq/1000 kcal	No dark green vegetables or beans or peas
Whole grains (0–10)	0·0	0·0–0·0	≥ 1·5 oz eq/1000 kcal	No whole grains
Dairy (0–10) §	0·4	0·0–1·6	≥ 1·3 c eq/1000 kcal	No dairy
Total protein (0–5) ‡	5·0	5·0–5·0	≥ 2·5 oz eq/1000 kcal	No protein foods
Seafood and plant protein (0–5) ||	5·0	5·0–5·0	≥ 0·8 oz eq/1000 kcal	No seafood or plant proteins
Fatty acids (0–10) ¶	7·3	5·3–8·7	(PUFA’s + MUFA’s)/SFA’s ≥ 2·5	(PUFA’s + MUFA’s)/SFA’s ≤ 1·2
Moderation components
Refined grains (0–10)	0·0	0·0–0·0	≤ 1·8 oz eq/1000 kcal	≥ 4·3 oz eq/1000 kcal
Sodium (0–10)	6·0	3·7–9·7	≤ 1·1 g/1000 kcal	≥ 2·0 g/1000 kcal
Added sugars (0–10)	10·0	10·0–10·0	≤ 6·5 % of energy	≥ 26 % of energy
Saturated fat (0–10)	9·9	7·7–10·0	≤ 8 % of energy	≥ 16 % of energy

PUFA: polyunsaturated fatty acids; MUFA: monounsaturated fatty acids; SFA: saturated fatty acids; oz: ounce; c: cup; g: gram; intakes between minimum and maximum standards are scored proportionately; all components are scored per 1000 kcal or percentage of energy, except for fatty acids; energy from alcohol is included in total energy intake.

^* Includes 100 % fruit juice.

^†Includes all forms except juice.

^‡Includes legumes (beans and peas).

^§ Includes all milk products, such as fluid milk, yogurt, and cheese, and fortified soy beverages.

^|| Includes seafood, nuts, seeds, soy products (other than beverages), and legumes (beans and peas).

^¶ Ratio of PUFA and MUFA to SFA.

Food-level analysis

Food-level analyses provided deeper insight into the top three food categories contributing to total energy intake and each HEI-2020 nutrient component (Table 3). Across the 20 participants, 58 total dietary recalls were collected, and 568 individual food/beverages were reported. Individual food/beverages were grouped into 73 different member-level categories.

Table 3.

Top three food category contributors to total calories and HEI-2020 nutrient components across dietary recalls collected with Marshallese women (n 20)

Nutrient	Top food category contributor *	Second highest food category contributor *	Third highest food category contributor *
Total calories	Cooked cereals, rice	Chicken	Finfish
	36·4 %	9·8 %	9·5 %
Total fruit (c eq)	Fruits, not berries	Citrus fruit juices	Fruit juices, not citrus
	49·2 %	38·4 %	7·2 %
Whole fruit (c eq)	Fruits, not berries	Berries	Grain/pasta/bread mixtures
	91·3 %	6·8 %	1·9 %
Total vegetables (c eq)	Raw vegetables	Potato salad	Cooked vegetables
	31·2 %	24·2 %	8·7 %
Greens and beans (c eq)	Raw vegetables	Green/leafy vegetables	Grain/pasta/bread mixtures
	43·6 %	31·4 %	12·1 %
Whole grains (oz eq)†	Ready-to-eat cereals	Salty snacks	–
	63 %	37 %
Dairy (c eq)	Milk, fluid	Cream cheeses	Grain/pasta/bread mixtures
	62·3 %	5·9 %	5·8 %
Total protein (oz eq)	Finfish	Chicken	Frankfurters, sausages
	36·4 %	30·1 %	11·9 %
Seafood and plant protein (oz eq)	Finfish	Shellfish	Grain/pasta/bread mixtures
	88·1 %	3·1 %	2·9 %
Fatty acids (% of energy)‡	Chicken	Finfish	Frankfurters, sausages
	19·5 %	11·1 %	10·3 %
Refined grains (oz eq)	Cooked cereals, rice	White breads, rolls	Pancakes
	68·5 %	9·5 %	6·2 %
Sodium (mg)	Chicken	Frankfurters, sausages	Soups with grains
	19·3 %	13·0 %	11·9 %
Added sugars (% of energy)	Teas	Soft drinks, carbonated	White breads, rolls
	26·7 %	25·8 %	11·2 %
Saturated fat (% of energy)	Chicken	Frankfurters, sausages	Chicken eggs
	15·3 %	10·6 %	8·0 %

oz eq: ounce-equivalents; c eq: cup-equivalents; mg: milligrams.

^* The percentage contribution of a food category to the nutrient is shown below each category. Top 3 food groupings are shown; total percentage across each nutrient may not equal 100 %.

^†The only reported foods containing whole grains across dietary recalls fell into two categories.

^‡Percent of energy from unsaturated fatty acids was calculated rather than fatty acid ratio (PUFA + MUFA/SFA), to understand which food groupings contributed to a higher fatty acid ratio value.

The food category with the highest energy contribution was rice (i.e. white and jasmine rice), which comprised 36·4 % of total energy reported across all dietary recalls. Finfish was the top food category contributing ounce-equivalents to both total protein (36·4 %) and seafood/plant proteins (88·1 %) and was among the top contributors to energy from unsaturated fatty acids (11·1 %). Individual foods reported in this category included salmon, canned tuna, tilapia, sashimi, bass, pompano, bonito, and eel. Chicken was the top food category contributing energy from saturated fats (15·3 %) and milligrams of sodium (19·3 %). Upon evaluation of individual foods reported in this category, chicken was often prepared with soy sauce marinades and consumed with skin. Frankfurters/sausages/lunchmeats contributed substantially to total proteins (11·9 %), energy from saturated fats (10·6 %), and sodium (13·0 %).

Teas (26·7 %) and soft drinks (25·8 %) contributed the highest percentage of energy from added sugars and were most often purchased ready-to-drink. Rice contributed 68·5 % of the ounce-equivalents to the refined grains component (i.e. 346·0 ounce-equivalents rice/505·1 total ounce-equivalents of refined grains). On the other hand, only two food categories included whole grain ounce-equivalents (i.e. ready-to-eat cereals, salty snacks from grain products), aligning with the HEI whole grain component median (IQR) of 0·0 (0·0–0·0). The only foods reported within these categories included Honey Bunches of Oats® granola, Doritos® (made with whole-grain corn, although high in saturated fat in sodium), and pretzels (made with just 0·18 ounce-equivalents of whole grain). Milk was the top contributor to dairy (62·3 %).

Consistent with low HEI-2020 component scores for total and whole fruit, very few fruits were reported across the 58 dietary recalls (a total of 7·4 cup-equivalents of whole fruit, 13·8 cup equivalents of total fruit). Of the few cup-equivalents of fruit reported, most were grouped within ‘fruits, excluding berries’ for both the total (49·2 %) and whole fruit (91·3 %) components (e.g. honeydew, banana). Similarly, few vegetables were reported across dietary recalls (24·7 cup-equivalents of total vegetables, 3·1 cup-equivalents of greens and beans). The top food category contributor to total vegetables (31·2 %) and greens/beans cup-equivalents (43·6 %) was ‘other vegetables, raw,’ which included tossed salad, green/string beans, green cabbage, onion, kelp, and cucumber.

Discussion

To our knowledge, this study was the first to provide a rich characterization of dietary patterns in Marshallese mothers of children under 12 months living in the continental US. The median overall HEI-2020 score (46·4) was low compared to a nationally representative sample of breastfeeding women (age 20–44) in the US (mean HEI-2015 score 62)⁽³⁵⁾. Diet quality was also lower than two prior studies including children living in the RMI—Dela Cruz et al. (2023) found a mean diet quality score (HEI-2005) of 50·1 in a sample of 191 RMI children, while Hingle et al. (2023) found a mean diet quality score (HEI-2005) of 54·7 among children (n 829) living in the Freely Associated States, which includes the RMI^(41,42). The comparison of diet quality scores across versions of the HEI (i.e. HEI-2005, HEI-2015, HEI-2020) should be interpreted within the context of changes to nutrient component categories and scoring across time. The HEI-2015 and HEI-2020 are fully aligned across all nutrient components and scoring standards⁽³¹⁾. The HEI-2005 differs in some components; for example, the HEI-2005 includes whole grains and total grains, whereas HEI-2015 was updated to include both whole grains and refined grains to address high levels of refined grain consumption across US adults⁽⁴³⁾. However, all versions reflect similar aspects of the diet, use a density approach for scoring standards, and may be used to understand compliance of a set of foods to the version of the Dietary Guidelines for Americans that aligns with the time period data was collected⁽⁴³⁾.

In the present study, almost all Marshallese women were born in the RMI, although have been living in the US a median 11·5 years. The historical nutrition transition among the Marshallese continues to have a lasting impact on Marshallese women dietary patterns living in the US, evidenced by low diet quality scores within the refined grain HEI component, although moderate scores within sodium (often high in processed foods). This study and other qualitative research have identified white rice as a cultural staple food consumed with most meals, due to its affordability, but also the belief that rice is necessary as a means for survival, a perception that may have stemmed from the historical import of rice provided to Marshallese after nuclear testing^(10,12,13). Thus, recommendations set by the Dietary Guidelines for Americans (e.g. switching to whole grains) may not be culturally acceptable to this group, and was a sentiment shared in discussions with Marshallese Community Advisory Board members throughout this study. Rather, nutrition recommendations may focus on smaller portions of rice, especially given that rice made up the largest percentage of energy consumed across dietary recalls in this study. Additional recommendations to emphasize more lean proteins, fruits, and vegetables at meals may be more acceptable in combination with reduction in portions. The recommendation to reduce portion size of rice was found to be acceptable in focus groups with Marshallese women of young children⁽⁴⁴⁾. Prior research among other racial/ethnic groups that include white rice as a predominant food have tested the acceptability of substituting white rice with brown rice^(45,46). The main barriers to brown rice consumption included rough texture, unpalatable taste, cost, and longer cook times, although perceptions improved after discussing the nutritional benefits of brown rice^(45,46). Given the strong suggestion shared by Marshallese Community Advisory Board members that the substitution from white to brown rice would not be acceptable in this group, future research is needed to assess the impact of the culturally acceptable recommendation to decrease portion sizes of white rice in combination with an increase healthy proteins, fruits, and vegetables on measures of nutrition and health indicators (e.g. diet quality, blood sugar, BMI).

This study provided a deeper insight into the food groupings that drove overall HEI-2020 and component score results. High diet quality component scores in seafood/plant protein and fatty acids were consistent with the food-level analysis results that found fish as a substantial contributor to both ounce-equivalents of seafood/plant proteins and energy from fatty acids. These positive results may be used in the cultural adaptation of the Centering Parenting nutrition education curriculum. The Dietary Guidelines for Americans 2020–2025 recommends the inclusion of seafood within the introduction of complementary foods at 6 months as an important source of iron, zinc, Vitamin D, and PUFA critical for growth and development⁽⁴⁷⁾. However, it is also important to incorporate safe fish consumption guidelines for mothers and children as outlined by the Food and Drug Administration (e.g. choosing fish varieties lowest in mercury, serving size guidance)⁽⁴⁸⁾. Given the strong relationship between maternal and child dietary intake among Marshallese individuals^(12,13), emphasizing incorporation of fish within safe consumption guidelines is one strategy to promote healthy food intake for both mothers and their children.

Fruit (total/whole fruit) and vegetable (total/greens and beans) diet quality component scores were notably low in this study, consistent with findings of low HEI-2005 fruit and vegetable diet quality scores among children living in the RMI reported in prior research^(41,42). Additional cross-sectional studies including children living in the RMI (n 892) have reported similar findings, with less than half of children meeting fruit and vegetable consumption guidelines (e.g. daily consumption of fruits and daily consumption of vegetables) in one study (n 892)⁽⁴⁹⁾ and a mean of 0·22 daily cups of vegetables and 0·34 daily cups of fruits reported in a second study (n 191)⁽⁵⁰⁾. Although very few fruits and vegetables were reported in dietary recalls, food-level analyses provided insight into the different fruits and vegetables chosen by study participants (e.g. honeydew, banana, tossed salads, greens/beans) that may be emphasized in nutrition education. Qualitative research has found that Marshallese individuals have a desire to consume fresh fruits and vegetables, yet economic constraints and larger household sizes limit their ability to do so⁽¹⁰⁾. Further, larger household size (median 4·0 children and 4·5 adults in the current study) may contribute to available produce being consumed quickly after purchase and, therefore, more challenging to capture within dietary recalls. The finding of low fruit intake was also surprising to the Marshallese research team members in this study, given the perception of a higher fruit intake among their community. Research team discussions about how to incorporate more fruits and vegetables into meals led to Marshallese team members sharing the Marshallese word ‘Leen Wijket’ that refers to any produce that comes from a tree or the ground. This word was recommended to use when encouraging a higher intake of produce for both infants and mothers. Although the encouragement to increase produce consumption is important to include within nutrition education, it is also important to consider barriers to consumption, such as cost. In 2024, COFA citizens in Arkansas became eligible for the Supplemental Nutrition Assistance Program (SNAP), which could be one avenue in making it easier to purchase and consume fruits and vegetables^(26,27). Further, financial incentive programs have been shown to significantly increase fruit and vegetable purchasing and consumption among low-income families (e.g. SNAP Double Up Food Bucks)^(51,52). Future research should consider evaluating the impact of SNAP program participation on dietary intake among Marshallese individuals.

Another surprising finding in this study was the high added sugar HEI component score (i.e. a lower/more desirable intake), given prior research identifying frequent consumption of sugar-sweetened beverages among Marshallese living in Arkansas (e.g. iced tea, vegetables juices, fruit juices, sweet tea)⁽¹⁰⁾. One reason for this finding could include that the added sugar HEI component is scored as a percentage of total energy, with participants receiving the maximum score if ≤ 6·5 % of total energy come from added sugars (Table 2). With the insight that white rice was the top contributor to energy across dietary recalls, added sugars may have proportionately contributed a lower percentage of total energy consumed, and thus, a more desirable HEI score. Food-level analyses provided deeper insight that much of the added sugars within dietary recalls came from sugar-sweetened beverages, and nutrition education tailoring may include a consideration of beverage alternatives (e.g. changing to unsweetened tea varieties, mixing small amounts of sweetened tea with unsweetened, encouraging water consumption).

This study had several limitations. Although this population was hard to reach to conduct dietary recalls over the phone (e.g. disconnected phone numbers, sharing phones across family members, no answer), integrating Marshallese CHWs into the research team allowed for trust and understanding to be built between the study team and community. A reason for the challenge in reaching Marshallese participants may be related to frequent travel between the US and the Republic of Marshall Islands (RMI) as COFA citizens, making follow-up data collection difficult⁽²⁶⁾. Due to the focus of this study on Marshallese women with young children, limited generalizability may be made regarding the dietary patterns among Marshallese men, older adults, and other Pacific Islander groups. Although the sample size of this study was small, which may limit the precision of quantitative findings, the sample was appropriate for the aim of the study to provide a rich understanding of diet quality to inform the cultural adaption of nutrition education. A strength of this study included the use of a food-level analysis to provide a deeper understanding of which food groupings influenced HEI-2020 score results.

Results of this study will be used to inform tailoring of the nutrition education within a parenting intervention, CenteringParenting, aimed to reduce obesity among Marshallese children. Future research should consider the cultural adaptation of dietary assessment tools to better capture cultural foods consumed in diverse racial/ethnic groups. Future researchers may also build on the findings within the present study by exploring the foods and food groupings that drive varying HEI scores across different racial/ethnic groups. This study highlighted the importance of adapting interventions for diverse groups and capturing cultural nuances that may be incorporated into nutrition education. For example, incorporating Marshallese language, such as ‘Leen Wijket,’ to encourage intake of fruits and vegetables, may help improve the understanding and acceptability of healthy eating interventions for communities afflicted with a high prevalence of diet-related diseases across the US.

Financial support

Research reported in this publication was supported by the National Institute of General Medical Sciences of the National Institutes of Health (NIH) (5P20GM109096), University of Arkansas for Medical Sciences Translational Research Institute funding awarded through the National Center for Advancing Translational Sciences of the NIH (1U54TR001629-01A1, KL2 TR003108, and UL1 TR003107), and the National Institute of Nursing Research of the NIH (1R21NR020677 – 01). This work is also supported by A1344 Diet, Nutrition and the Prevention of Chronic Diseases (grant no. 2020-68015-30734/project accession no. 1021697) from the U.S. Department of Agriculture (USDA) National Institute of Food and Agriculture. The content of this paper is solely the responsibility of the authors and does not necessarily represent the official views of the funders. Funders had no role in the design, analysis, or writing of this article.

Competing interests

The authors declared no conflicts of interest.

Authorship

B.L.A.: Conceptualization, funding acquisition, supervision, writing – reviewing & editing. E.S.: Data curation, formal analysis, methodology, writing – original draft, reviewing & editing. S.K.C: Project administration, Writing – reviewing & editing. A.A., R.N., C.C.: Writing – reviewing & editing.

Ethics of human subject participation

This study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving research study participants were approved by the by the University of Arkansas for Medical Sciences Institutional Review Board (#274752). Written informed consent was obtained from all subjects/patients.