Assessing food selection in a health promotion program: Validation of a brief instrument for American Indian children in the Southwest United States.

ABSTRACT

Objective Brief dietary assessment instruments are needed to evaluate behavior changes of participants in dietary intervention programs. The purpose of this project was to design and validate an instrument for children participating in Pathways to Health, a culturally appropriate, cancer prevention curriculum.

Design Validation of a brief food selection instrument, Yesterday's Food Choices (YFC), which contained 33 questions about foods eaten the previous day with response choices of yes, no, or not sure. Reference data for validation were 24-hour dietary recalls administered individually to 120 students selected randomly.

Subjects The YFC and 24-hour dietary recalls were administered to American Indian children in fifth- and seventh-grade classes in the Southwest United States.

Results Percentage agreement for all items was greater than 60%; for most items it was greater than 70%, and for several items it was greater than 80%. The amount of agreement beyond that explained by chance (K statistic) was generally small. Three items showed substantial agreement beyond chance (K [greater than or equal to] 0.6); 2 items showed moderate agreement (K=0.40 to 0.59); most items showed fair agreement (K=0.20 to 0.39). The food items showing substantial agreement were hot or cold cereal, low-fat milk, and mutton or chile stew. Fried or scrambled eggs and deep-fried foods showed moderate agreement beyond chance.

Statistical analyses performed Dietary recalls were coded for food items in the YFC and results were compared for each item using percentage agreement and the K statistic.

Conclusions Previous development and validation of brief food selection instruments for children participating in health promotion programs has had limited success. In this study, instrument-related factors that apparently contributed to poor agreement between data from the YFC and 24-hour dietary recall were inclusion of categories of foods vs specific foods; food knowledge, preparation, and vocabulary; item length; and overreporting of attractive foods. Collecting and scoring the 24-hour recall data may also have contributed to poor agreement. Further development of brief instruments for evaluating changes in children's behavior in dietary intervention programs is necessary. Factors related to the YFC that need further development may be issues that are also important in the development of effective, brief dietary assessments for children as individual clients or patients. J Am Diet Assoc. 2000;100:205-211.

Among American Indian peoples in the Southwest United States, mortality rates for coronary heart disease and cancer have been lower than rates among the general population and other American Indian groups [1-4]. However, increasing prevalence of obesity, type 2 diabetes, and cardiovascular disease, as well as evidence of changing dietary patterns and lifestyles, have suggested the need for health promotion interventions [1,3-8].

Diet and nutrition are important factors in programs to promote change in health behavior for the prevention of coronary heart disease and cancer. Such programs draw on scientific understanding of dietary risk factors to translate nutrient-based information into recommended food choices [9-11]. Nutrition education programs for schoolchildren have the potential to influence food practices throughout life and possibly modify disease risk factors at a young age [12]. Moreover, school-based programs are a practical approach to serving widely scattered communities in the rural Southwest [13-15].

In school-based nutrition education, as in other community-level interventions, brief instruments are needed to evaluate behavior change. Although considerable research has been conducted on dietary assessment methods for general use, there is relatively little research on validation of methods for measuring dietary changes resulting from a nutrition education or behavior change intervention [16,17]. Successful instruments should be efficient to administer and measure group behavior change with respect to targeted food choices [16-18]. Instruments designed for school-based programs must also be appropriate for developmental and grade level and consistent with local and regional food patterns [17,19,20].

Successfully validated instruments for children include a modified food record, a detailed food frequency questionnaire, and a 1-day frequency of consumption form [21-24]. In these validations, children's literacy level seemed to be an important factor [21-23] or the children were trained intensively on how to complete the form [24]. Brief food selection instruments that have been used with children either had poor validity [25-27] or lacked formal validation [28,29].

We developed an instrument for fifth- and seventh-grade American Indian children in the Southwest who were participating in a culturally appropriate, 16-lesson, cancer prevention curriculum, Pathways to Health [14,30]. The Pathways to Health curriculum, designed in cooperation with advisers from these American Indian communities, builds on traditional customs and values related to healthful living and foodways. Yesterday's Food Choices (YFC), adapted from an instrument for adults named the Food Behavior Checklist, is a 1-day food selection assessment tool to describe mean behavior of a group participating in a health promotion program [18]. The purpose of this article is to describe the design and validation of the YFC. Because the instrument is a questionnaire measuring self-reported food selections the previous day, 24-hour dietary recalls were the reference method for validation.

METHODS

Subjects

In spring 1992, 120 randomly selected students from 9 schools completed the YFC and 24-hour dietary recall. Subjects were from elementary and middle schools that had agreed to participate in Pathways to Health, a classroom cancer prevention program for American Indian students in the Southwest [14,30]. Subjects were representative of the students that were to receive the full curriculum in fall 1992. In each of the 20 participating classrooms, the YFC was administered to all students in class on the day of testing (n=285) according to a standard protocol developed for Pathways to Health. After the YFC instruments were completed, 6 students from each class were randomly selected to participate in the validation. Project staff generated random numbers before going to the school and students were selected from class lists using these numbers.

All schools are located in rural New Mexico and serve American Indian communities in which English is a second language for some children. A few classes included students who were not American Indian but whose families lived within these isolated communities. Table 1 shows the gender, grade, age, and ethnicity of the participating students. Occasionally, classes contained mixed grade levels such as a fifth- and sixth-grade class or a seventh- and eighth-grade class, which accounts for the small number of sixth- and eighth-grade students.

Instrument

The YFC was designed to measure self-reported eating behavior related to the Pathways to Health behavioral goals in nutrition: eating a diet low in fat, eating a diet high in fiber, and eating a diet high in fruits and vegetables [30-32]. The YFC consists of a list of foods and response choices indicating whether the items were eaten the previous day. It was developed based on an adult instrument, the Eating Behavior Checklist of Kristal and coworkers [18]. The Eating Behavior Checklist contains 18 high-fat or high-fiber foods that meet the goals of the California Nutrition and Cancer Prevention Program [18]. Generation of the food list for the YFC was based on the behavioral goals for …