Health & Medical Public Health

Setting Policy to Address Eating Disorders and Weight Stigma

´╗┐Setting Policy to Address Eating Disorders and Weight Stigma

Methods

Sample


Study participants were drawn from two distinct samples. First, to assess support for policy actions among the general public, we conducted a survey of a diverse, national sample of U.S. adults, who were recruited through a survey panel administered by Survey Sampling International (SSI; http://www.surveysampling.com). SSI recruits participants through thousands of websites with data aggregators that reach millions of users. Panelists were aged 18 years and older, actively indicated their intention to join an SSI panel, and provided validated geographic and demographic information. SSI set quotas to approximate U.S. Census demographics. Of the 1157 participants who entered the survey, 223 (19%) were excluded due to listwise deletion of item non-response missing data, resulting in a final sample size of 934 participants from the U.S. general public.

Second, to obtain a sample of individuals from the eating disorders field, the survey was advertised on websites, electronic newsletters, and/or list-servs of existing professional organizations in the U.S. that specialize in eating disorders. These organizations included the Academy for Eating Disorders, Binge Eating Disorder Association, and National Eating Disorders Association. Announcements about the study contained a weblink to the online survey. Participants who clicked on the weblink were transferred to the survey website (hosted by Qualtrics.com) and were provided with information explaining the survey and inviting them to participate. Of the 1977 participants who began the survey, 320 (16%) withdrew at the beginning of the survey prior to responding to any questions. Of the remaining 1657 participants, 253 (15%) were excluded due to listwise deletion of item non-response missing data, yielding a final sample of 1404 individuals from the eating disorders field. Table 1 presents sample characteristics.

All participation was voluntary and anonymous, and participants in both samples completed identical surveys. The survey software (Qualtrics) enabled features to prevent the same user from completing the survey more than once. Data collection occurred during May through July of 2013. All participants provided informed consent, and the study was approved by the Yale University IRB.

Survey Questionnaire


We developed an online self-report survey instrument to assess level of support for potential policy actions to address eating disorders and weight stigmatization. A list of 37 potential policy actions were generated through reviews of the literature, identification of related policies being implemented in other countries, ideas proposed at scientific meetings, and discussions with researchers and advocates in the fields of eating disorders, psychology, and public health. After carefully reviewing this list, items were excluded that were too vague or were redundant with other items (n = 10), resulting in 27 items that formed an initial version of the questionnaire. We fielded this questionnaire to 10 international experts in the eating disorders field, who pre-tested survey questions and provided feedback on wording and content for each item. Based on their feedback, item wording was revised for increased clarity and four items were removed, yielding a total of 23 questions. The survey asked participants to indicate the extent of their support (on a 5-point Likert rating scale, ranging from 1 = definitely oppose to 5 = definitely support) for each of 23 potential policy actions related to eating disorders and weight stigmatization. Scale items were later recoded into binary items to assess the percentage of participants who either "somewhat" or "definitely" supported each policy action (reflecting a "4" or "5" on the 5-point Likert rating scale).

Policy actions were focused in five different content areas (see Table 2 ): 1) schools (e.g., "Schools should conduct screening for eating disorders"), 2) weight stigma and discrimination (e.g., "Existing civil rights laws should include body weight to protect people from weight discrimination"), 3) healthcare (e.g., "Insurance companies should be required to reimburse for eating disorder treatment"), 4) weight control products (e.g., "Selling over-the-counter diet pills and laxatives to minors should be restricted by the government"), and 5) the media (e.g., "The use of very underweight fashion models should be restricted by the government").

After indicating their level of support for each policy action, participants were asked to choose the five policy actions from the list of all 23 policies that they believed would have the most positive impact on efforts to address weight stigma and eating disorders. Participants were then asked to select the five policy actions from the full list of 23 policies that they believed would be the most feasible to implement.

Participants also responded to questions assessing demographic characteristics. Participants in the general public sample were asked to indicate their age, gender, race/ethnicity, height and weight, level of education, household income, and political orientation. Participants recruited from the eating disorders field were asked their gender, age, ethnicity, and their profession. Both samples were additionally asked whether they, or anyone in their family, have had an eating disorder.

Analysis


Descriptive statistics and regression models to assess demographic predictors of policy support were used for analyzing the data. Since the study outcome variables (mean scale scores derived by averaging across respective items within each of the five content areas described above) were negatively skewed with high probability mass around their theoretical maximum score, tobit models for censored data were used. Separate models were fit for each of the five outcome variables and for each of the two samples. For the general public sample, missing values for BMI were multiply imputed (20 datasets), utilizing information from all outcome and predictor variables that were used in the regression analyses ( Table 3 ). All analyses were performed using the statistical analysis software Stata, version 13.



You might also like on "Health & Medical"

Leave a reply