Development and Implementation of a Food System Intervention to Prevent Childhood Obesity in Rural Hawai‘i

Abstract

This paper presents details the Healthy Foods Hawai‘i (HFH) intervention trial, aimed to improve children's dietary behavior to prevent child obesity, by modifying the food environment with community-selected foods. Four communities were selected by ethnic composition, income level, two on O‘ahu and one neighbor island. On each island one community was randomly assigned to intervention and one to control. The intervention was implemented through food stores in the intervention communities. HFH was designed to strengthen the network between local food producers, food distributors, storeowners and consumers, to increase the availability of healthier less energy dense foods for children in underserved rural communities of Hawai‘i. The intervention includes phases: healthier beverages, snacks, condiments, and family meals. Moderate to high fidelity was achieved for educational materials (shelf labels, posters and educational displays). The number of educational displays varied by intervention phase and community. Posters were found in place 100% of the time. Shelf labels were found intact in the correct location. Low to moderate fidelity was achieved for distributors, with some products not stocked. In the intervention communities, 6–8 week phases focused on target foods with 40 food demonstrations. A total of 1582 food related samples were distributed. A high to moderate dose and reach of the overall intervention was achieved in delivery of the cooking demonstrations. A high to moderate dose and reach of the intervention was achieved overall; fidelity to the intervention protocol was moderate. To improve healthy local food availability in stores in rural communities in Hawai‘i, agricultural producers reported needing additional support to sell and transport product to local stores, rather than to centralized distributors.

Introduction

This paper describes the development and implementation of the Healthy Foods Hawai‘i (HFH) project, which aimed to modify the food environment of rural underserved communities to shift food availability and consumption to healthier local foods, to ultimately prevent and reduce child obesity. HFH built on strategies, goals and methodologies developed and tested in previous Healthy Store Projects. Previous Healthy Stores intervention trials showed success through focus on retailers and consumers (which was also a component in HFH). For HFH we additionally aimed to identify and create linkage opportunities between local food producers, food distributors, and food store owners, as a strategy to give consumers improved access to healthier food choices.

The authors conceptualize a multilevel model of influences on obesity¹³ while focusing on the food environment to influence food behavior. We also draw on social-cognitive theory, recognizing that dietary behaviors are influenced by individual/personal factors. We measured the impact of the HFH intervention on parent and child diet and cognitive factors relating to diet. The HFH intervention program had significant dietary and psychosocial impacts, improving diets of children and self-efficacy of parents.⁷ Thus, the HFH intervention was designed to modify several aspects of the food environment in order to incur healthier food consumption among minority children and their families in rural underserved communities in Hawai‘i. Here we present details of the HFH program implementation and food system that was identified through formative and summative evaluation of the food environment of multiethnic rural Hawai‘i. The programmatic approaches finally selected and insights gained into the Hawai‘i food environment may be useful when designing other programs intended to improve the environment to change behaviors that can prevent child obesity.

Methods

Intervention Development and Design

Healthy Foods Hawai‘i (HFH) intervention trial was a project of the Healthy Pacific Child Program (HPCP), which was developed after participatory strategic planning with the Healthy Living in the Pacific Islands Initiative (HLPI).⁸ HFH was conducted in two of Hawai‘i's communities, on two of its islands. Four communities were included in the study, two control/comparison communities and two intervention, with one control/comparison and one intervention on each island. Matching of intervention with comparison/control community was based on similarity of ethnic and income distribution, which was 10–27% Native Hawaiian and Pacific Islander, and >75% below the poverty level.⁹ Five stores in the two intervention communities were randomly selected to implement the intervention.

HFH was a child-focused intervention program with a long term goal of reducing child obesity through increased healthy eating in Hawai‘i's rural multiethnic communities. The HFH project was unique in its focus on children, and in its efforts to integrate additional components of the food distribution system (Figure 1); we use the term food “getting” since food was obtained by a variety of both monetary and non-monetary transactions. Figure 1 illustrates the food system as identified for the HFH project, based on community workshops and in-depth interviews with local informants. Stores were an important source of food, though food was also obtained at farmers markets, from gardens, from friends and family, and from restaurants. Food in stores was obtained from national distributors and local producers.

Figure 1.

Food getting system* in rural Hawai‘i

Formative work involved community workshops to develop HFH intervention messages, with an emphasis on local foods and agricultural products consumed by children.¹⁰ The intervention aimed to increase the availability of healthy foods in stores in target communities through work with store owners and managers, and food producers and distributors; and to promote healthier food choices and food preparation methods through intervention messages in stores and local media. Messages were designed to encourage and foster gradual change in specific eating and food getting habits by adult caregivers and children. Promoted food items, themes, mass media material, and giveaways were implemented to resonate with the children.

Implementation of the intervention in the two intervention communities differed slightly. In one community, the intervention was delivered primarily by HFH project staff; in the other community, a local not-for-profit was contracted to deliver the intervention. Process data were only collected on the intervention in the intervention communities. Dietary outcome data was collected in all four communities (two intervention and two comparison/control) and is reported elsewhere⁷.

The intervention was comprised of four phases, each running for 6–8 weeks. The phases targeted: i) healthier beverages (water, diet soda, lite nectars and 100% juices: ii) healthier snacks for children (whole grain, lower sugar cereals (WIC), low fat milk, fruit and vegetables with low fat dips, pretzels and baked chips): iii) healthier condiments (lite mayonnaise, low fat salad dressings and homemade dressings with herbs): and iv) healthier meals (drain and rinse ground meat, lite/low sodium Spam™, tuna in water, fresh herbs, locally produced “chop suey” (greens) mix and watercress). These four phases were applied in both intervention communities in partnership with store owners and managers, food distributors, and local food distributors. Phase-specific educational materials were posted in various food, health, and community locations, and culturally-relevant cartoons were published in local neighborhood newspapers. Popular local recipes were modified, creating healthier versions.

In-Store Components

Selected stores were the primary sources for food purchasing in each community. Cooking demonstrations and taste tests were conducted in the five intervention stores (three in one community, and two in the other). Cooking demonstrations/ taste tests planned for 4–6 times per phase at each intervention store, with brochures and recipe cards distributed during the demonstrations/ taste tests. In-store posters, educational displays and shelf labels (Lower in Fat, Lower in Sugar, Healthy Food Choice, Healthy for Keiki (Child), Local Produce) were used as educational tools, with one set of materials per phase.

Agricultural Producer and Food Distributor Components

Two local producers and four local distributors collaborated with the project by providing promoted products and/or promotional items for taste tests and cooking demonstrations, and responding to key informant interviews. One producer had already worked with one of the local stores and had a small area in the produce section. The other producer had already supplied his product to the participating local stores but had only a limited amount of product available and, therefore, did not want to increase demand. This agricultural producer also provided a fixed amount of produce to one of the participating food distributors on a weekly basis, to be combined as part of the “chop suey” (greens) mix that was promoted during the intervention. Both of these local agricultural producers and their products were highlighted in store through the use of a “producer biography,” which was a 4″×6″ laminated card with a photo and brief biography of the farmer and farm which was hung above their product in the produce section.

As distributors already had products in most of the stores, collaboration centered on increasing the availability of HFH targeted products. Of the four food distributors involved in the project, one distributed canned nectar juice drinks, one snack foods, one milk and milk products, and the other acted as both an agricultural producer and a food distributor, growing produce, processing the produce, and distributing their own and others products. One or more of the food distributors provided product and/or promotional materials in each phase, and the agricultural producers supplied product during the specific phases that their produce was promoted. During phases 1–3, the food distributors (for milk, chips, nectar drinks and local produce) provided their products for use in the taste test/cooking demonstrations. Some also provided gift certificates and giveaways (e.g. pens, visors, fresh produce).

Process Evaluation Methods

The process evaluation measures evaluated reach (number of participants), dose (amount and frequency of exposure to intervention elements), and fidelity of the intervention (how closely the intervention was implemented as compared to the planned implementation and were categorized as: high=75–100%, medium 50–75%, low <50%). Evaluation instruments examined the amount of customer exposure to promotional materials and provided information on the nature and amount of interaction between customers and interventionists in the two intervention communities. A Store Visit Process Evaluation form (SVPE) and the Cooking Demo/Taste Test Process Evaluation form (TTPE) evaluated the process in the stores and in the community. Each form was completed twice a month by site visit in Community 1, and weekly in Community 2. The SVPE form evaluated success at keeping the promoted food items on the shelf, proper and intact labeling for promoted items, and phase-specific posters visible in the store settings. The TTPE form evaluated the process of bimonthly cooking demonstrations and/or taste tests conducted in the stores. This included the number of people who fully participated in the activity, the number of people that partially participated, and the number of people who passed by without engaging with the HFH staff. The Cooking Demo/Taste Test Participant Evaluation (CDPE) form evaluated community response to promoted items and/or promoted behavioral changes and rated the likelihood to purchase the promoted items or cook using the promoted methods.

The study was approved by the University of Hawai‘i Committee on Human Studies. Informed consent was obtained for individual level data that was obtained. Quantitative data used in the present analyses of store data were entered using Microsoft Excel 2003, which was used to calculate means of intervention frequency, reach, and dose.

Results

Moderate to high fidelity was achieved for educational materials (shelf labels, posters, and educational displays), which were readily available in intervention store locations (Table 1). The number of educational displays varied by phase and community, from two to nine; in every case, community two used more educational displays. Posters were found in place 100% of the time. Shelf labels were most often found both in the correct location and intact (63 to one hundred percent of the time, depending on phase), though some foods had higher rates of missing/damaged labels or labels incorrectly placed under non-promoted foods. However a low to moderate fidelity was achieved for distributors/producers stocked items (75 to 100 percent of the time); some products were not stocked. Incorrectly placed labels occurred most often for foods that had high turn over, where items have to be frequently restocked (though not necessarily the promoted foods); these included chips, canned nectars, and luncheon meat. Incorrectly placed labels also occurred often in the produce section where produce was rearranged often, based on the season and quantity delivered to the store, or was damaged by the water sprayed on the shelves.

Table 1.

Fidelity of intervention (availability of promoted foods and print materials) in intervention communities by HFH intervention phase

HFH Intervention		Community 1					Community 2
Phase	Foods promoted	No. Store visits	Times stocked during phase (%)	Times shelf label correctly placed (%)	Times poster posted (%)	No. educational displays	No. Store visits	Times stocked during phase (%)	Times shelf label correctly placed (%)	Times poster posted (%)	No. educational displays
1	Water	6	100	100	100	2	8	100	100	100	3
	Lite Drink		100	75				100	97
2	Fresh fruit/vegetable	12	100	100	100	4	12	100	100	100	6
	Low-sugar cereals		100	100				100	100
	Low-fat milk		100	100				83	100
	Baked chips		83	58				75	75
3	Lite mayonnaise	8	100	100	100	4	36	94	83	100	6
	Low/fat free dressing		100	63				94	94
	Herbs		100	63				94	81
4	Fresh local vegetable	6	83	83	100	4	16	100	100	100	9
	Lite spam		100	83				100	88
	Low sodium spam		100	33				94	100
	Tuna in water		100	100				100	94
	Cooking spray		100	100				100	94

In interviews, food producers expressed concern about having adequate product, and cost of delivery to stores as compared to working with one central distributor, who will often pick up their product. Food distributors expressed need to sell as much product as possible in the shortest period of time, often resulting in removing newer (healthier) products on the shelf when existing (often less healthful) products with higher turnover could be stocked instead, even when the store owner was prepared to take the risk of lower turnover for a period of time, in order to support local farmers and the provision of healthy products. Although the price per unit of produce sold to food distributors tended to be lower than produce sold directly to local stores, the producers believed the cost benefit was not sufficient to cover the additional labor and transportation costs for them to work directly with local stores. Large produce processors/distributors pick up produce directly from farms, and provide transport of the produce the 30 to 40 miles to central processing facilities for cleaning, sorting, and packaging for subsequent re-distribution of the produce island-wide. Agricultural producers were also, on occasion, unable to produce sufficient quantity of produce at the required/requested time, due to production constraints such as weather, pests, and diseases.

In the two intervention communities, during nine months of intervention in four six to eight week phases that focused on target foods, there were 40 food demonstrations (22 in community one and 18 in community 2), that lasted a total of 84 hours (55.5 in community one and 28.5 hours in community two), Table 2. 1150 individuals participated in the food demonstrations (646 in community one and 508 in community two). A total of 1582 food related samples were distributed (868 in community one and 713 in community two). A high to moderate dose and reach of the overall intervention was achieved in delivery of the cooking demonstrations/taste tests (Table 2). The majority of the customers liked the promoted products (data not shown). For example, a blind taste provided participants with three types of luncheon meat (regular, low sodium and lite) and more than 90% of respondents (n=36) stated they would like to buy low sodium or lite luncheon meat. There was a reported increase in sales of those items. Healthier beverages were best liked among promoted and tested products.

Table 2.

Dose and reach of cooking demonstrations and taste tests in stores of intervention communities by HFH intervention phase, mean (SD)

HFH Intervention		Community 1				Community 2
Phase	Foods promoted	Number of Demos	Hours per Demo	Number of Participants Per Demo	Number of Food Samples Per Phase	Number of Demos	Hours per Demo	Number of Participants Per Demo	Number of Food Samples Per Phase
1	Water, Lite drink	4	10 (2.5)	245 (61.3)	349 (87.3)	5	1.5 (7.5)	125 (25)	314 (62.8)
2	Fresh fruit/vegetable, Low-sugar cereals, Low-fat milk, Baked chips	6	15.5 (2.6)	137 (22.8)	229 (38.2)	5	1.8 (9)	164 (32.8)	182 (36.4)
3	Lite mayonnaise, Low/fat free dressing, Herbs	6	17.5 (2.9)	135 (22.5)	160 (26.7)	5	1.5 (9)	138 (23)	136 (22.7)
4	Fresh local vegetable, Low sodium spam, Tuna in water, Cooking spray	6	12.5 (2.1)	132 (22)	131 (21.8)	5	1.5 (3)	81 (40.5)	81 (40.5)
Total All Phases		22 (5.5)	55.5 (13.9)	646 (29.5)	869 (39.5)	18 (4.5)	28.5 (7.1)	508 (28.2)	713 (44.6)

Discussion

To our knowledge, this is the first store intervention trial to actively incorporate food distributors and producers. We found that the local food distribution network needed more support to increase availability of local produce in stores.

This study showed high fidelity, dose and reach of store intevention components, comparable with other studies.^11,12 Availability was a challenge. Stocking decisions are not always controlled by store owners/managers. Greater support to agricultural producers would be an important future approach to improve healthy food availability in stores. Unique elements of this intervention that demonstrated good reach, dose and fidelity included the identification and promotion of “local” products in stores, and the provision of products by distributors for taste tests and cooking demonstrations. The taste tests and cooking demonstrations provided a unique opportunity for interactions between four intervention elements: producer/distributor (macro) and store/consumer (micro) environments. Interventionists provided educational and promotional activities to consumers using the store as a venue, while producers and distributors were offered the opportunity to showcase their products at minimal cost to themselves.

Working with multiple stakeholders to enhance use, the food distribution system (stores, food producers, food distributors) for healthy foods proved challenging, presenting institutional barriers to successfully integrate food system components needed to sustain the intervention. Nonetheless, the integration of community-based organizations into intervention delivery enhanced implementation and likelihood of sustainability. Researchers, clinicians and other health professionals may find application of formative processes to be useful for identifying and modifying other behaviors that influence health. Further, familiarity with community driven targets for strengthening healthy food behaviors, will help align efforts to shift the food environment toward healthier food, a critical component to prevent child obesity.

Disclosure

Funding was provided by the US Dept of Agriculture, grant 2004-35215-14252 (Novotny PI), and Hawai‘i Dept. Health, contract 436851 (Novotny).