Nicole Carlson and Anamarie LeDuc, “The Evaluation of Adolescent Body Size Using Photographs and Participant Recall”
Mentors: Ellen Velie, Emily Guseman, Elizabeth Duthie, and Darek Lucas, Public Health
Adolescence represents a critical period in obesity development, which is associated with negative health outcomes, including breast cancer. However, adolescent body size is difficult to assess retrospectively without historical health records. Data for this project are from a population-based case-control study of breast cancer in women under 50 years of age, the Young Women’s Health History Study (YWHHS). Participants in YWHHS rated their body size at age 12y and 18y on the Stunkard scale consisting of nine body figures ranging from underweight (1) to overweight (9) and also provided historical photographs for those ages. The objectives of this project are to evaluate: 1) the validity of self-reported relative adolescent body size at ages 12y and 18y vs. study rater’s assignment of a body size; 2) agreement between two independent evaluators of participant’s Stunkard body size based on the pictures. Pictures from 19 participants, including both cases and controls, were rated by the two evaluators and compared for assessment of rating reliability. Cohen’s Kappa for the two evaluators’ ratings was low, kappa=0.0, indicating chance agreement. Analysis of the average of the two ratings of the two evaluators vs. participant’s self-reported body size are currently in progress. Though the current sample size is limited, photo evaluation is ongoing so that we may evaluate potential factors associated with recall, such as race, household percent poverty, pubertal status, adult BMI, and childhood food insecurity and physical activity. As such, this study will assess the validity of self-reported adolescent body size that is often used in studies to examine the impact of adolescent body size on adult health.
Click the thumbnail below to open the full sized poster in a new tab.
Very clear presentation. Thank you for sharing your interesting study.
Your third discussion point is interesting, in that it predicts the most variability into the middle range of the Stunkard Figures. As I look at the figures, I have a harder time telling the difference between figures 2-4. There also seems to be a more salient jump between figures 5 and 6 than between the lower numbered figures. It seems that your results also had more variability among the low-mid range of this scale.
Another factor that could influence where a rater would place someone could be breast size or height. For example, someone with a thin body but larger breasts could be placed higher on the scale than someone with the same sized waist who has smaller breasts. The same could happen for shorter vs. taller people. Having raters work together to adjust placement on the scale during training could help with increasing inter-rater agreement.
Very interesting poster!
I find it especially interesting that the two raters had only chance agreement. It would be cool to see how a rater’s own body size (or self-reported body size) effected how they rate images.