Background

Eating disorders are considered an important public health problem, due to the chronicity and mortality to which they are associated. They are time-dependent diseases, in which early detection and treatment can improve the prognosis.

Objectives

The objective of our work was to assess the psychometric properties of the SCOFF questionnaire as an instrument to detect the presence of eating disorders quickly and reliably in a primary care centre.

Material and methods

Observational, cross-sectional and multicentre study carried out on 291 people in all the health centres of the capital of Guadalajara. Stratified, multistage and proportional random sampling according to age, gender and health centres. The study was carried out in 2 phases: completion of the SCOFF questionnaire and clinical interview. Sensitivity, specificity, positive and negative predictive value, global value, positive probability ratio, ROC curve, Cronbach’s alpha, and mean time to perform the test were observed.

Results

Sensitivity: 91.6% (95% CI: 71.8–100%); Specificity: 91% (95% CI: 87.5–94.5%); Positive predictive value: 30.5% (95% CI: 14.1–46.9%); Negative predictive value: 99.6% (95% CI: 98.6–100%); Positive probability ratio: 10.2 (95% CI: 6.8–15.4%); Efficiency: 91.1% (95% CI: 87.6–94.5%); Area under the ROC curve: 0.9099 (95% CI: 0.8082–1.0117); Cronbach’s α: 0.725; Mean performance time: 3.98 (± 0.12) minutes.

Conclusions

The SCOFF is a short questionnaire, easy and brief to fill out, reliable and with indications of validity and sufficient diagnostic sensitivity to be applied with guarantee in primary care centres.