EAT-26 Self-Test :: Screening

SCREENING

SCREENING AND CASE FINDING FOR THE GENERAL PRACTITIONER

David M. Garner, Ph.D.

Introduction.

It is widely acknowledged that eating disorders are prevalent and constitute significant health problems among young women. The aim of this chapter is to provide the rationale for general practitioners screening for patients suffering from eating disorders as well as suggest practical methods for case identification, early intervention and referral. The two main eating disorders anorexia and bulimia nervosa have serious physical and psychiatric morbidity c.f. Garner, Vitousek & Pike, 1997; Garner & Dalle Grave, 1999; Note 1), and in the case of anorexia nervosa, mortality is higher than any other psychiatric disorder). There is evidence that effective treatment exists for both bulimia and anorexia nervosa; however, findings from community studies indicate that only a minority of cases are in treatment. These are some of the factors that have led researchers to employ various screening strategies in order estimate the prevalence of eating disorders as well as to detect cases for the purpose of earlier interventions.

Incidence and Prevalence.

Incidence rates are defined as the number of new cases in the population per year whereas prevalence rates refers to the actual number of cases in the population at a certain point in time. Eating disorders have been studied most commonly using prevalence studies and, in most cases, these have been conducted on high risk populations like college students and athletes (Garner, Rosen & Berry, 1998). There are serious limitations to estimates of the incidence and prevalence of eating disorders, because most have been derived exclusively from self-report instruments and on samples that may not reflect important demographic differences in base rates. In general, estimates based exclusively on questionnaires yield much higher rates of eating disorders.

Estimates of incidence based on detected cases in a primary care practices yielded rates of 8.1 per 100,000 persons per year for anorexia nervosa and 11.5 for bulimia nervosa (Hoek et al., 1995). The most sophisticated prevalence studies using strict diagnostic criteria report rates of about 0.3% for anorexia nervosa and about 1% for bulimia nervosa among young females in the community. This compares to surveys using questionnaires that find that as many as 19% of female students report bulimic symptoms. Prevalence studies of higher risk samples indicate that serious eating disorders occur in as many as 4% of female high school and college students (Hoek et al., 1995). Suspected cases of clinical eating disorders or subclinical variants are even more common among “high risk” groups of athletes competing in sports that emphasize leanness to enhance performance or appearance (Garner et al., 1998).

Prevalence estimates for eating disorders vary widely depending on the methods used and definitions of the disorder. Self-report methods have generally produced higher prevalence estimates than interview-based methods. In contrast, data from epidemiological studies have led to the conclusion that current diagnostic criteria may be overly restrictive in that many of those who fail to meet certain diagnostic thresholds have similar psychiatric morbidity to those with the full syndrome.

Screening Versus Case Finding.

Screening and case finding are based on the assumption that early identification of a disorder can lead to earlier treatment thereby reducing morbidity and mortality. Screening for a variety of medical disorders has become routine in a range of different settings and involves testing presumably healthy volunteers from the general population for the purpose of separating them into groups that have either a high or a low probability of having a particular disorder. An example of screening would be a national program aimed at identifying those who are HIV positive, or who have breast cancer in a particular population. In screening, the initiative is taken by the health professional rather than volunteered by the patient.

In contrast, case finding involves testing patients who have voluntarily sought health care or information as part of a comprehensive assessment of health. Health care workers may screen for certain disorders during routine physical examinations in patients who are at risk or in community-based voluntary settings. An example of case finding would be blood pressure assessment in a shopping mall or mammographies offered at community centers. The routine practice of a primary care physician closely resembles this definition of case finding since it is common for a complaint-related or non-complaint-related illness to be identified in the course of the patient seeking care.

Screening and case finding are not appropriate for every condition or disorder. Key indications for employing a screening for a disorder or condition are that it constitutes an important health problem, it is treatable, and that early identification leads to a favorable outcome. The decision to screen always depends on weighing factors including potential beneficial and harmful effects of testing for the individual as well as the population surveyed. In addition, the screening test employed should have desirable psychometric characteristics (including sensitivity, specificity and positive predictive value) as well as be relatively simple, economical, and acceptable to those asked to complete it. The efficiency of screening is depends on the validity or accuracy of the testing as well as on the prevalence of the disorder. Again, the main difference between screening and case finding is who initiates the initial contact. In both case finding and screening, the health professional must carefully evaluate the risks and the benefits of the procedures used as well as the practicability, effectiveness and the efficiency of the measures employed.

Screening and case finding carry different ethical obligations. If the practitioner initiates the screening, there needs to be conclusive evidence that the procedure can positively affect the natural history of the disorder. Moreover, the risks of screening on those who are unaffected must be carefully considered since the target individual who has not asked the health professional for assistance. This situation is somewhat different from case finding where the patient has asked for some level of assistance. While the patient should be assured of the highest standard of care available at all times, case identification occurs in the context of a patient asking for assistance. In this situation, there is no guarantee of benefit and, it could be argued, that there is at least some implied exposure to risk. In other words, the implications of an uninvited intrusion on a person with the suggestion that they could have a health concern or that they indeed manifest an illness requires a higher burden of proof that the potential benefits far outweigh the risks.

In the eating disorder field, both screening and case finding studies have been conducted with little discussion of the relative risks of these procedures. It is not uncommon for eating disorder patients to reveal in an initial consultation that they learned about self-induced vomiting or laxative abuse from a well-meaning research survey “educational” program. Thus, case finding must be conducted in a general practice setting with caution and careful consideration of the potential benefits and risks to the individual.

The Eating Attitudes Test (EAT) for Screening and Case Finding.

The EAT is a standardized, self-report measure of symptoms and concerns characteristic of eating disorders. It is designed to be economical both in administration and scoring time. The EAT has been used as a screening and case finding instrument in non-clinical populations. A factor analysis of the original 40-item EAT (Garner & Garfinkel, 1979) produced a 26-item abbreviated measure, the EAT-26 (Garner, Olmsted, Bohr, & Garfinkel, 1982). The total score of the EAT-26 is the sum of the scores of the individual scores on the test.

The EAT does not yield a specific diagnosis of an eating disorder but studies have shown that it can be an efficient case finding or screening instrument to identify those who are at increased risk for serious eating disorders. Neither the EAT, nor any other screening instrument, has been established as highly efficient as the sole means for identifying eating disorders. This is attributable in large part to the relatively low prevalence of eating disorders in most populations of interest. A disorder must have a prevalence approaching 20% in order for the test to be efficient in detection. Thus, even with a highly valid test, it is very difficult to achieve high efficiency in detecting eating disorders that have a prevalence between 2% and 4% in populations of adolescent or young women. The relatively low incidence of eating disorders has led to recommendations that screening be confined to “high risk” groups and that a “two stage” method be used in which a screening questionnaire is administered to a sample or to the entire population and only with high scores are interviewed.

The two-stage survey process is illustrated in a general practice setting by King (1989; 1991) who studied the men and women between the ages of 16 and 35 years of age were asked to compete the EAT-26 in the waiting rooms of several general practices. Of the 748 people contacted, 96% completed the EAT-26. Of the 76 high scorers, 7 refused an interview and of the remaining 69 individuals, 7 cases of bulimia nervosa were found (6 female and 1 male). King (1989, 1991) found that very few of those who scored at or below the cut-off on the EAT had eating disorders or serious eating concerns on interview (few false negatives). Of those who scored above the cut-off on the EAT, a third had clinically significant eating concerns or weight preoccupations. In a follow-up of high scorers 12-18 months later, 20% of those who initially had a “partial syndrome” now met diagnostic criteria for an eating disorder. Moreover, more than 30% of the initial “normal dieters” became “obsessive dieters” (King, 1991).

Cuzzolaro & Petrilli (1988) translated the EAT-40 into Italian and validated in Italy. More recently, the EAT-26 has been validated by Dotti and Lazzari (1998) in a sample of 1,277 roman high school students. Several research teams have used the EAT to estimate prevalence of eating disorders in Italy (Santonastaso, Zanetti, Sala, Favaretto, Vidotto, & Favaro, 1996; Dalle Grave, DeLuca & Oliosi, 1997; Vetrone, Cuzzolaro & Antonozzi, 1997). Using a two-stage design, Dalle Grave et al. (1997) surveyed 795 students in southern Italy using the EAT-40 and found that 17.3% of the females and 1% of the males scored above the EAT cut-off score with a point prevalence for eating disorders of 5.4% among the female students. In the Vetrone et al. (1997) study of 297 Italian schoolgirls, 24 cases (8.1%) were identified as cases of eating disorders (5 bulimia nervosa and 19 eating disorder NOS; however, only 16 of these (all 5 bulimia nervosa and 11 eating disorder NOS) scored above the cut-off score on the EAT. The 8 participants who scored below the EAT cut-off score were identified using other criteria that proved useful. Dotti and Lazzari (1998) surveyed Roman high-school students and interviewed 95 students with a score of 20 or more. They found that 13.1% of the girls and only 1.3% of the boys scored over this cut-off score on the EAT-26. They also interviewed and diagnosed 40 randomly selected students with low scores. Results indicated that the EAT-26 was more sensitive to the presence of an eating disorder than to a specific clinical entity and it was concluded that the EAT-26 be used to isolate cases at risk of clinical spectrum eating disorders.

National Eating Disorders Screening (case finding) Program (NEDSP)

The EAT-26 was used in the 1998 National Eating Disorder Screening Program (NEDSP) in the United States (Garner et al., 1999) and behavioral questions were added to the EAT-26 to improve the instrument’s ability in “case finding”. The aim of the NEDSP was to determine the effectiveness of a national effort to identify those suffering from eating disorders and to encourage them to seek professional help. In February 1998, the National Eating Disorders Screening Program conducted screening for eating disorders at 1083 sites in the United States. A total of 69,374 individuals attended the screening and 35,897 individuals were screened for eating disorders. More than half of those screened were college students. Follow-up by telephone interview was conducted two months after the initial case finding on a representative sample of 937 participants. Of those screened and then interviewed at follow-up, 34.5% scored positively (20 or more) on the EAT, and 89% of these individuals were not in treatment at the time of screening. Of those interviewed, 15% reported vomiting in the preceding 6 months to control their weight, 15% reported abusing laxatives, 33% used diet pills, and 11% took diuretics. Results from the follow-up indicated that 38% of the sample was referred for further treatment. Of those who scored positively on the EAT, and were referred to a clinician, 42% actually followed through and saw a clinician and 76% of this group continued in further treatment. Of those interviewed, 82% felt that the screening program was helpful in at least some way and 32% noted an improvement in their eating attitudes or behaviors following the NEDSP. The EAT proved to be a psychometrically sound and useful screening instrument, particularly when supplemented by behavioral questions asking about eating disorder symptom frequencies. It is concluded that voluntary screening for eating disorders is an effective way to bring certain untreated individuals to treatment.

It is important for the health professional to understand that the EAT is not a substitute for an interview for those with eating disorders. Although self-report measures like the EAT should never be used as the sole basis for diagnostic and treatment decisions, they do provide an efficient means of obtaining information for clinical and research purposes. The EAT has the advantages of economy in administration and scoring. It also may minimize interviewer bias and other potential threats to validity that stem from responses being derived from the interaction between the interviewer and the subject. When used as the initial step in a two-stage process, the EAT can alert the practitioner to patients who have potentially serious eating disorder symptoms.

Screening for Eating Disorders in Athletes.

In recent years, there has been a growing interest in eating disorders among athletes. Most studies of athletes competing in sports that emphasize leanness to enhance performance or appearance (e.g. dance, gymnastics, distance running, wrestling) have shown that they are at increased risk for the development of eating disorders. The reasons for the increased prevalence of eating disorders among certain subgroup of athletes has been open to considerable speculation. The most common view has been that the pressures to diet and become thin may actually trigger eating disorders in those with specific vulnerabilities (Garner, Rosen & Berry, 1998; Garner & Dalle Grave, 1999). Alternatively, it may be that certain sports may attract athletes with pre-existing eating disorders. There have also been suggestions that certain personality and family factors common in athletes may predispose them to develop eating disorders. It is important for general practitioners to be aware of the connection between eating disorders and athletes in order to improve case identification.

The failure of some studies to find higher rates of eating disorders among certain groups of athletes has raised questions regarding their actual prevalence as well as the utility of conventional assessment methods. One criticism of self-report instruments such as the EAT is that they are highly vulnerable to denial of symptoms. However, in arguably the most methodologically sophisticated and comprehensive study of disordered eating among athletes, Sundgot-Borgen (1994) examined risk factors for eating disorders among elite athletes representing six different groups of sports. The Eating Disorder Inventory (EDI), a measure with many similarities to the EAT, was administered to all participants and careful steps were taken to address the issue of respondent truthfulness. Of the 522 elite female athletes participating in one phase of the study, 117 (22.4%) were classified “at risk” for an eating disorder based on scores on the Drive for Thinness and Body Dissatisfaction subscales of the EDI (Sundgot-Borgen, 1994). Of the at “at risk” athletes who participated in a clinical interview (N=103), 48% met criteria for anorexia or bulimia nervosa and 41% had clinically significant eating disorder symptoms but failed to meet all diagnostic criteria. The prevalence of eating disorders was greatest in sports where athletes are encouraged to be thin to meet performance or appearance standards. Although negative findings in some studies have been used to cast doubt on the utility of self-report instruments, the fact that these same measures have been used successfully to screen for eating disorders in other studies suggests that the false negatives may be due more to poor study design than to inherent flaws in the assessment instruments. Valid findings require procedures that assure respondents that results will be kept strictly confidential and that identification of eating problems will not lead to some negative outcome.

There also has been controversy regarding the meaning of eating disorders among athletes. There have been those who have raised alarm about high prevalence rates and others who have suggested that eating disorders in athletes are a benign form of the clinical syndrome. This is certainly a controversial position that we do not share. The topic of eating disorders among athletes addressed in detail in a review by Garner et al. (1999) and we will discuss it here further.

Final Comments.

There potentially tremendous benefit from a general practitioner screening for an eating disorder. Eating disorders have devastating physical, psychological and social consequences. This is reflected in the high level of mortality, morbidity and the poor quality of life that they confer. Screening has the potential for allowing early identification of eating disorders and this can lead to earlier treatment which has been shown to lead to a more favorable prognosis

One of the criteria for determining if screening should be conducted for a particular disorder relates to whether or not a effective treatment is available. Over the past two decades, there has been a rapid expansion in controlled psychological and pharmacologic research which has resulted in a number of very effective treatments (see Garner & Dalle Grave, 1999). Morever, there is evidence that simple education regarding the effects of starvation on behavior can lead to the reduction of serious eating disorder symptoms (Garner, 1997; Garner & Dalle Grave, in press). Nevertheless, there are risks of screening for eating disorders. It is possible that well-meaning efforts at screening may actually cause certain people to develop eating disorder symptoms. The intention to change the course of a participant in a screening program must exercise extreme caution to ensure that the change is for the better rather than for the worse. These risks are somewhat less in case finding methods since the target of the survey has sought assistance from the health care professional.

Another important principle in screening and case finding is the availability of an economical and valid instrument for detecting the particular disorder of interest. It has been more than two decades since the Eating Attitudes Test (EAT) was described as a screening instrument in high-risk populations. The EAT-26 has a number of limitations including its reliance on the respondent’s honesty during testing. Nevertheless, a significant number of research studies have shown that the eating is an economical, reliable and valid instrument that can assist in identifying in different cultures (Nasser, 1997). The recent addition of specific behavioral questions to supplement the eating may improve the EAT’s ability to detect clinical cases of eating disorders; however, the EAT remains a face-valid test of the level of concern the patient is expressing surrounding eating symptoms. If used with specific interview probes suggested in this chapter, the EAT can be a useful tool in the arsenal of assessment procedures used by the general practitioner.

References

Note 1: Adapted from: In: Anoressia, Bulimia Binge Eating Disorder (edited by: Fabio Piccini, M.D.) Il ruolo del medico nello screening dei DCA.

Becker, A., Grinspoon, S., Klibanski, A., & Herzog, D. (1999). Eating Disorders. The New England Journal of Medicine, 340:14, 1092-1098.

Cuzzolaro & Petrilli (1988). Validazione della versione italiana dell’EAT-40 di D.M. Garner e P.E. Garfinkel. Psichiatria dell’infanzia e dell’adolescenza, 55, 209-217.

Dalle Grave, R., De Luca, L., & Oliosi, M., (1997). Eating attitudes and prevalence of eating disorders: A survey in secondary schools in Lecce, southern Italy. Eating and Weight Disorders, 1, 34-37.

Dottti, A., & Lazzari, R. (1998). Validation and reliability of the Italian EAT-26. Eating and weight disorders. 3, 188-194.

Garner, D. M. (1997). Psychoeducational principles in treatment. In D. M. Garner & P. E. Garfinkel (Eds.), Handbook of Treatment for Eating Disorders (pp. 145-177), New York: Guilford Press.

Garner, D.M. & Dalle Grave, R. (1999). Terapia cognitivo comportamentale dei disturbi dell’alimentazione.

Garner, D.M. & Dalle Grave, R. (In press). Review of the Classic Key's "Starvation" Study and its Relevance to the Understanding and Treatment of Eating Disorders. Eating & Weight Disorders.

Garner, D.M. and Garfinkel, P.E. (1979). The eating attitudes test: An index of the symptoms of anorexia nervosa. Psychological Medicine, 9, 273279.

Garner, D.M. Garner, M., Powers, P., Meszler Reizes, J., Deutsch, N., & Jacobs, D. Results from the 1998 national eating disorder screening program. Paper presented at the annual eating disorder research society meeting, San Diego, November, 1999.

Garner, D.M., Olmsted, M.P., Bohr, Y., & Garfinkel, P.E. (1982). The Eating Attitudes Test: Psychometric features and clinical correlates. Psychological Medicine, 12, 871-878.

Garner, D.M., Rosen, L. and Barry, D. (1998). Eating Disorders in Athletes (839-857). In: Child and Adolescent Psychiatric Clinics of North America., 7, New York: W.B. Saunders.

Garner, D. M., Vitousek, K., & Pike K. (1997). Cognitive-behavioral therapy for anorexia nervosa. In D. M. Garner & P. E. Garfinkel (Eds.), Handbook of Treatment for Eating Disorders (pp. 94-144), New York: Guilford Press.

Garner, D.M. and Wooley, S.C. (1991). Confronting the failure of behavioral and dietary treatments for obesity. Clinical Psychology Review, 11, 729-780.

Hoek, H. W. (1995). The distribution of eating disorders. In K. D. Brownell & C. G. Fairburn (Eds.), Eating Disorders and Obesity (pp. 207-211), New York: Guilford Press.

King, M. B. (1989). Eating disorders in a general practice population. Prevalence, characteristics and follow-up at 12 to 18 months, Psychological Medicine, Monograph Supplement 14, 1-34.

King, M. B. (1991). The natural history of eating pathology in attenders to primary medical care. International J. of Eating Disorders, 10, 379-387.

Nasser, M. (1997). The EAT speaks many languages: Review of the use of the EAT in eating disorders research. Eating & Weight Disorders, 2, 174-181.

Olmsted, M.P., Davis, R., Garner, D.M. and Rockert, W. & Eagle, M. (1991). Efficacy of brief psychoeducational intervention for bulimia nervosa, Behavior Research and Therapy, 29, 71-83.

Powers P. (1997). Management of patients with comorbid medical conditions. In D. M. Garner & P. E. Garfinkel (Eds.), Handbook of treatment for eating disorders (pp.424-436), New York: Guilford Press.

Santonastaso, P., Zanetti, T., Sala, A., Favaretto, G., Vidotto, G., Favaro, A. (1996). Prevalence of eating disorders in Italy: A survey on a sample of 16-year of female students. Psychotherapy and Psychosomatics, 65, 158-162.

Sundgot-Borgen, J. (1994). Risk and trigger factors for the development of eating disorders in female elite athletes. Med Sci Sports Exerc 26:414-419.

Vetrone, G., Cuzzolaro, M. & Antonozzi, I. (1997). Clinical and subthreshold eating disorders: Case detection in adolescent schoolgirls. Eating and Weight Disorders. 1, 22-33.