With regard to diet, health-promoting environments are those which facilitate healthy food choices. Put simply, in a healthy nutrition environment, the healthy choice is the easy choice. The number of reports of various dimensions of nutrition environments is increasing, however, there is no guidance in the literature on how best to measure these environments in a comprehensive manner. Research on school food environments; neighborhood food environments (stores, restaurants); and state policies are illustrative of well-developed measurement tools and important needs in this area. While there are a few items on perceptions of the nutrition environment included within larger surveys of neighborhood environments (e.g., Echeverria et al., 2004), no comprehensive survey instruments have been reported.
This section provides examples of accomplishments and needs in the area of measurement of nutrition environments in schools, stores, and restaurant settings. The tools addressed here emphasize observational measures, as this is the focus of substantial emerging research.
Schools. A number of measures of school food environments have been carefully developed, most often for use in intervention research. Large-scale studies of school food policies and environments have been conducted using surveys of school administrators and food service managers (Delva et al., 2007; Wechsler et al., 2001). These data are limited by the usual concerns with self-report (bias, forgetting, etc.) and may also suffer from non-response bias. Recently, a state level nutrition-environment policy classification system has been developed to track developments in eleven policy areas, including school meal environments, reimbursable school meals, BMI screening, and competitive foods. This system is based on a social-ecological model and should enhance the surveillance opportunities for all 50 states and the District of Columbia (Masse et al., 2007).
Local and regional studies typically use a combination of data collection methods, including surveys of food service managers, observations and data-based inventories of foods available, observations/analysis of students’ bag lunches, and food service sales data. Often, the food availability and/or sales data are combined with nutritional information and subjected to nutrient analyses (Sallis, McKenzie, et al., 2003; French et al., 2003). These measures are carefully designed and subjected to quality assurance, but few psychometric data are available. A key limitation of on-site measures is that the sales data are usually recorded manually rather than obtained from automated cash register systems. Details of the instruments and protocols used in peer-reviewed research have not been widely disseminated, most likely because the tools were developed in specific settings as part of larger intervention studies.
Neighborhood food environments: the community nutrition environment. Key categories of food sources in neighborhoods include stores and restaurants. It is useful to distinguish where people get food and what type of food they can get within those establishments. The community nutrition environment is comprised of the number, type, location and accessibility of food outlets such as grocery stores, fast-food restaurants, and full service restaurants. The consumer nutrition environment is what consumers encounter in and around places where they buy food, such as the availability, cost, and quality of healthful food choices (Glanz, Sallis, Saelens & Frank, 2005). Community nutrition environment data are available from various commercial sources such as Dun & Bradstreet business lists (Powell et al., 2007), as well as from county health or agriculture department food license lists, telephone books, and the internet. While national studies may rely on business lists, local and regional studies suggest that more complete and accurate enumeration of food sale locations can be achieved using a combination of sources (Glanz, Sallis et al., 2007; Saelens, Glanz et al., 2007) and supplemented with ‘ground truthing’ by systematically walking or driving each street in a neighborhood.
Consumer Nutrition Environments in Stores. Some of the earliest published measures of availability of healthy foods in stores were reported nearly two decades ago by Cheadle and others (Cheadle et al., 1991), who calculated the percentage of shelf space used for healthy food options, such as low fat milk, whole wheat bread, cheese and lean meats. They found high inter-rater reliability (0.73 to 0.78) and test-retest reliability ranging from 0.44 to 1.00. These measures are theoretically robust but may be difficult to apply in contemporary grocery stores that are larger and more varied in layout than they were two decades ago. Horowitz and others (2004) measured availability of 5 diabetic-recommended foods in grocery stores and reported excellent inter-rater reliability ranging from 0.94 to 1.00. Other published reports have been less clear about the rigor of their methods or did not report reliability of the measures.
Recently, the Nutrition Environment Measures Study developed observational measures of the nutrition environment within retail food stores (NEMS-S) to assess availability of healthy options, price, and quality for ten indicator food categories, aligned with the U.S. Dietary Guidelines (Glanz, Sallis et al., 2007). Using an iterative process involving field work, research team deliberation, and expert consultation, 10 indicator food categories were developed: fruit, vegetables, milk, ground beef, hot dogs, frozen dinners, baked goods, beverages (soda/juice), whole grain bread, and baked chips. Inter-rater reliability and test-retest reliability of availability were high: inter-rater reliability kappas were 0.84 to 1.00, and test-retest reliabilities were .73 to 1.00. These measures are being disseminated through training workshops at Emory University and, as of mid-2007, raters and trainers in 28 states have learned to use these tools and the NEMS-R restaurant measures.
Consumer Nutrition Environments in Restaurants. Research on the environment within restaurants is limited. There have been some recent advances in the measurement of food environments within restaurants, including good inter-observer reliability for availability of fruits and vegetables (Edmonds et al., 2001). Cassady and colleagues (2004) developed a reliable restaurant menu checklist for use by community members that assesses food preparation, number of healthful choices, and fruit/vegetable availability. However, this checklist did not assess the whole restaurant environment and was tested in only 14 family-style restaurants.
The NEMS-R observational measure for restaurants (see Appendix B) was recently developed to assess factors believed to contribute to food choices in restaurants, including availability of more healthy foods, facilitators and barriers to healthful eating, pricing, and signage/promotion of healthy and unhealthy foods. Inter-rater and test-retest reliability were assessed in 217 sit-down and fast food restaurants in four neighborhoods, and inter-rater reliability was generally high, with most kappa values >.80 (range .27 - .97) and all percent agreement values >75% (range 77.6% - 99.5%). Test-retest reliability was high, with most kappa values >.80 (.46 – 1.0) and all percent agreement values >80% (range 80.4% - 100%); (Saelens, Glanz et al., 2007). Like the NEMS-S store measure, it has been widely disseminated and continues to be adopted for research and community program use.