MMPI was used to identify psychopathological factors that might predict which patients are at risk of dropping out of our cognitive behavioral therapy with a behavior protocol governing privileges for inpatient anorectic patients. The overall dropout rate of 32% found in our study was very similar to the rate of 31–33% documented by three recent studies [1
]. As for demographic and clinical factors, some researchers have demonstrated that AN patients of the binge eating/purging type [3
] of higher age at admission [18
] and with a longer duration of illness [16
] had a significantly higher rate of dropout from inpatient treatment. In contrast, we did not find differences in the subtype of AN, age at admission, or duration of illness between the dropouts and completers, consistent with the data of Zeeck et al. [4
] for AN inpatients. The finding of no difference in the subtype of AN between completers and dropouts might be explained by our treatment strategy. When binge/purging behavior is found, we quickly counsel patients and implement a behavioral therapeutic intervention, by which they are encouraged to confront and overcome their problems. This may contribute to a reduction of patient resistance to change and reinforcement of the therapeutic relation for treatment, thus reducing premature treatment termination.
The psychopathology of dropouts included social and emotional alienation, lack of ego mastery, emotional instability, and an antisocial attitude. These characteristics make it difficult for patients to fit in well in an environment in which interpersonal relationships are important, such as during a hospital stay. Patients with an antisocial attitude often tend to break the rules of the hospital ward and easily act out. Also, patients predisposed to dropping out can be expected to have difficulty adjusting to a strict behavior protocol governing privileges because of their psychopathological characteristics. As a result, they have decreased motivation, leave the hospital without permission, attempt suicide, or shoplift, which leads them to drop out of inpatient treatment. Treatment based on cognitive behavioral therapy with a behavior protocol governing privileges was successful for many of our patients. However, it should be carefully adopted for anorectic patients who exhibit the psychopathological elements identified in this study because these patients have great difficulty accepting strict rules that limit their behavior. Moreover, treatment regimens must be developed to meet the specific needs of those patients. More family problems were found in dropouts who left because of therapist initiative than in either completers or dropouts who left because of patient initiative. Cooperation of the family is important to successful treatment, however family function is decreased under the conditions of severe family problems, which may result in a disincentive for treatment. Shoplifting and the theft of food accounted for almost half of our therapist initiated premature termination. We previously reported that the "hypomania" and "over-controlled hostility" subscales of the MMPI were risk factors for stealing behavior in AN patients [19
]. "Over-controlled hostility" was not but "hypomania" was a psychopathological feature of our dropouts. "Hypomania" shows emotional instability and, especially when leaning toward the hypomanic state, a patient can be expected to have difficulty adjusting to a strict treatment protocol that includes limitations on behavior and communication and may easily start acting out. Careful consideration is required, therefore, in embracing this treatment program for patients with high "hypomania" scores. Patients terminated by their own initiative mainly because of decreased motivation for treatment, which included acting out in ways such as leaving the hospital without permission. To prevent dropout from inpatient treatment, we need to sufficiently explain the meaning of the behavior protocol governing privileges to both the patient and their family before admission and to confirm their motivation for treatment.
This study has some limitations. First, we did not evaluate all possible personality traits because we did not conduct a structured interview; therefore, the presence of personality disturbance was not assessed. Second, our treatment was done in a general ward of a university hospital, whereas most of the patients in European and American studies were treated in a unit specific for eating disorders. Because of the difference of environments, our results for premature termination may not be comparable to those of other studies. Third, although data showing reduced patient motivation was drawn from descriptions found in medical records, no quantitative measurement of motivation was done using a psychometrical test. Therefore, we can not conclusively report the extent to which a reduction of motivation influenced premature termination. Fourth, our results did not determine the prognosis of patents after discharge because we examined only dropout from inpatient treatment. It will be necessary to do more research on the long-term outcome of cognitive behavioral therapy with a behavior protocol governing privileges for AN patients.