The purpose of this study is to find out an better screening protocol to identify female adolescents at risk for eating disturbances.
We administered a newly-developed eating disturbance onset prediction test and the Eating Attitude Test-26 (the EAT-26) to 1061 Japanese female high school students at Time 1 (T1). EAT-26 was re-administered about 10 months later, designated as Time 2 (T2). Based on EAT-26 findings, students were classified into four eating behavior groups : onset-free group, recovered group, persistent group, and onset group. The normal groups at T1 (onset-free and onset groups) were extracted in Step 1, and the onset group was predicted from the extracted normal groups in Step 2. Useful items for Step 2 were investigated through the eating disturbance onset prediction test, and four items were selected. As a result, the correct discrimination rate in the onset group was obtained as 73.3%, and the negative predictive value was obtained as 98.3%.
These results show that the protocol using the four items is useful in predicting the onset of eating disturbances among female high school students. False negative cases accounted for 1.2% of all students, thus, reduction of the false-negative rate by modification of measurement items may be needed.