Eating disorders are among the most common psychiatric disorders in young women. Early detection and treatment improve the prognosis, but the presentation of eating disorders is often cryptic—for example, via physical symptoms in primary care. The ability to diagnose the condition varies and can be inadequate, 1 and existing questionnaires for detection 2 , 3 are lengthy and may require specialist interpretation. No simple, memorable screening instruments are available for nonspecialists. In alcohol misuse, the CAGE questionnaire (questions about Cutting down, Annoyance with criticism, Guilty feelings, and Eye-openers) 4 has proved popular with clinicians because of its simplicity. We developed and tested a similar tool for eating disorders, with questions designed to raise the suspicion that an eating disorder might exist before rigorous clinical assessment.
We developed five questions addressing core features of anorexia nervosa and bulimia nervosa, using focus groups of patients with eating disorders and specialists in eating disorders. We tested the questions in a feasibility study of patients and staff at an eating disorders unit. None of these participants was involved in the subsequent study. We created the acronym SCOFF from the questions (see box).
We recruited patients sequentially from referrals to a specialist clinic: 116 women aged 18 to 40 years who were confirmed as having either anorexia nervosa (n=68 [35 binge eaters, and 33 restricted their food intake]) or bulimia (n=48), according to the criteria specified in the Diagnostic and Statistical Manual of Mental Disorders, fourth edition. 5 For controls, we recruited 96 women (through advertising at local colleges), aged 18 to 39 years, who were confirmed not to have an eating disorder. Patients and controls were asked the SCOFF questions orally; they also completed the eating disorder inventory 3 and the Bulimic Investigatory Test, Edinburgh (BITE), a self-rating scale for bulimia. 2
No significant differences existed between patients and controls for age or ethnicity. As expected, more patients than controls were in the highest socioeconomic groups (P3 2 =47.4), and patients were more likely to be single, separated, or divorced (P1 2 =13.0). The mean length of illness for patients was 8 years (SD, 4.8; range, 1-25 years). The mean (SD) body mass index (weight [kg]/[height (m)] 2 ) for controls, patients with bulima, and patients with anorexia was 22.3 (1.9), 24.4 (1.8), and 15.1 (0.8), respectively. All scores on the eating disorder inventory and the BITE scale were consistent with published data for women with or without eating disorders. 2 , 3
The SCOFF questions *
Do you make yourself Sick because you feel uncomfortably full? Do you worry that you have lost Control over how much you eat? Have you recently lost more than One stone (14 lb) in a 3-month period? Do you believe yourself to be Fat when others say you are too thin? Would you say that Food dominates your life?All participants found the questions and the term “SCOFF” acceptable. Setting the threshold at two or more yes answers to all five questions provided 100% sensitivity for anorexia and bulimia, separately and combined (all patients: 95% confidence interval, 96.9%-100.0%; patients with bulimia: 92.6%-100.0%; and patients with anorexia: 94.7%-100.0%), with a specificity of 87.5% (79.2%-93.4%) for controls ( table )
Numbers of cases (true positives) and controls (true negatives) identified by SCOFF questionnaire as likely to have an eating disorder *
| Cases | Total no. of subjects | No of participants identified by SCOFF as likely to have eating disorder |
|---|---|---|
| All cases | 116 | 116 |
| Bulimic cases | 48 | 48 |
| Anorectic cases: | 68 | 68 |
| Bingeing | 35 | 35 |
| Restricting | 33 | 33 |
| Controls | 96 | 12 |
The SCOFF questionnaire seems highly effective as a screening instrument for detecting eating disorders. It is simple, memorable, easily applied and scored, and has been designed to suggest a likely case rather than to diagnose.
We consider that the SCOFF questionnaire performed well against the 10 questions suggested by Greenhalgh to assess screening tests. 6 The false-positive rate of 12.5% is an acceptable trade-off for high sensitivity.
Further work is needed to establish validity and reliability in a wider population, particularly in those in the general population who are at risk for eating disorders. Nonetheless, the evidence of validity is sufficient for it to be used routinely in all patients considered at risk for eating disorders.
We thank Martin Bland, professor of medical statistics, for guidance in planning the research methods.
Competing interests: None declared.
This paper was originally published in the BMJ 1999;319:1467-1468
* Each “yes” equals 1 point; a score of 2 indicates a likely diagnosis of anorexia nervosa or bulimia
1. King MB. Eating disorders in a general practice population: prevalence, characteristics and follow-up at 12 to 18 months. Psychol Med Monogr Suppl 1989; 16 : 191-194. [PubMed] [Google Scholar]
2. Henderson M, Freeman CPL. A self-rating scale for bulimia: the “BITE.” Br J Psychiatry 1987; 150 : 18-24. [PubMed] [Google Scholar]
3. Garner DM, Olmstead MA, Polivy J. Development and validation of a multidimensional eating disorder inventory for anorexia nervosa and bulimia. Int J Eating Disorders 1983; 2 : 15-34. [Google Scholar]
4. Ewing JA. Detecting alcoholism: the CAGE questionnaire. JAMA 1984; 252 : 1905-1907. [PubMed] [Google Scholar]
5. American Psychiatric Association. Diagnostic and statistical manual of mental disorders . 4th ed. Washington (DC): American Psychiatric Press; 1990.
6. Greenhalgh T. Papers that report diagnostic or screening tests. BMJ 1997; 315 : 540-543. [PMC free article] [PubMed] [Google Scholar]