Abstract
BACKGROUND: We tested the validity of the SCOFF, a five-question screening instrument for eating disorders, in a general population sample. METHODS: A random sample of 1863 Finnish young adults was approached with a questionnaire that contained several screens for mental health interview, including the SCOFF. The questionnaire was returned by 1316 persons. All screen positives and a random sample of screen negatives were invited to SCID interview. Altogether 541 subjects participated in the SCID interview and had filled in the SCOFF questionnaire. We investigated the validity of the SCOFF in detecting current eating disorders by calculating sensitivity, specificity, and positive and negative predictive values (PPV and NPV) for different cut-off scores. We also performed a ROC analysis based on these 541 persons, of whom nine had current eating disorder. RESULTS: The threshold of two positive answers presented the best ability to detect eating disorders, with a sensitivity of 77.8%, a specificity of 87.6%, a PPV of 9.7%, and a NPV of 99.6%. None of the subjects with current eating disorder scored zero points in the SCOFF. CONCLUSION: Due to its low PPV, there are limitations in using the SCOFF as a screening instrument in unselected population samples. However, it might be used for ruling out the possibility of eating disorders.