In 2004, the Massachusetts Department of Public Health in partnership with Blue Cross Blue Shield of Massachusetts launched the Healthy Choices (HC) Initiative, a three-year school-based intervention to promote healthy eating and physical activity and reduce TV viewing and overweight in middle school youth. [8
] Administrators at all public middle schools in the state were notified about HC; all of the 51 schools entering the study in 2005 were invited to participate in an evaluation of the intervention that involved collecting survey data from students at several time points. The sociodemographic composition of these schools was similar to that of middle schools statewide. In fall 2005, before intervention activities began, 47 (92.2%) schools administered a baseline, self-report survey assessing sociodemographics, height, weight, and weight-related behaviors was administered to all 6th
, and 8th
graders. Self-reported height and weight, which have moderate validity in adolescents [9
], were used to classify overweight (body mass index [BMI]≥85th
percentile for age and sex) following U.S. Centers for Disease Control and Prevention (CDC) guidelines (http://www.cdc.gov/growthcharts/
). To assess DWCB, students were asked, “In the last 30 days
, have you done any of the following to lose or maintain your weight?” and they chose as many as applied from the options: “Vomit or throw up on purpose after eating”; “Take laxatives”; “Take diet pills without a doctor’s permission.” This item was modified from validated DWCB items used by the CDC’s Youth Risk Behavioral Surveillance System [10
]. Information on school characteristics (percent of students eligible for free or reduced-price lunch, percent of residents in school census tract living at or below poverty) were gathered through publicly available data sources (http://profiles.doe.mass.edu/
). Passive parental consent and adolescent assent procedures were used. Harvard School of Public Health institutional review board approved this study.
Generalized estimating equation (GEE) methods were used to estimate adjusted odds ratios (ORs) of reporting DWCB (defined as any vomiting or laxative or diet pill use for weight control in the past month) in youth of color compared to their same-gender, white peers. Multivariate models were used to cross-sectionally examine potential individual- and school-level confounders and explanatory factors and to account for school clustering. Subgroup analyses used GEE models controlling for gender and grade, stratified by race/ethnicity group, to examine gender differences in prevalence of DWCB. A total of 20,367 surveys were collected; 3,389 were excluded due to missing or implausible values, resulting in an analytic sample of 16,978 (83.4%).