The third School Nutrition Dietary Assessment Study: background and study design.

OBJECTIVE: This article describes the background and design of the third School Nutrition Dietary Assessment Study (SNDA-III).

DESIGN: SNDA-III is a nationally representative cross-sectional study of the National School Lunch Program and the School Breakfast Program in 2005. The three-stage sample design allowed description of district and school food environments and policies, analysis of foods and nutrients in school lunches and breakfasts, and assessment of the role of school meals and competitive foods in students' diets. Surveys of district and school staff were by telephone or in person; school menu data were collected in a mail survey with telephone assistance; and student and parent interviews were conducted in person and in school, except that parents of secondary-school students were interviewed by telephone. Student interviews included a 24-hour dietary recall, as well as measurement of height and weight. Response rates were 83% for districts, 95% for schools, and 63% for students, whose participation was constrained by consent issues and school schedules.

SUBJECTS/SETTING: Data were collected from 130 public school food authorities (districts that offer federally subsidized school meals), 398 schools within those districts, and 2,314 public-school students in grades 1 through 12 in these schools. Of the 2,314 students, a random subset of 666 (29%) completed a second recall to permit estimation of usual nutrient intake distributions.

STATISTICAL ANALYSES PERFORMED: Descriptive tabulations were used to summarize the background characteristics of schools and students and most study outcomes. Multivariate regression models and propensity score matching were used to compare the nutrient intakes of school meal participants and nonparticipants.

CONCLUSIONS: SNDA-III data provide a rich resource for examining interactions among the school meal programs, the school food environment, students' diets, and child obesity. Subsequent articles in this Supplement present analyses in all these areas.