[Bulimia and epidemiology]

Despite the multiplication of the epidemiological studies related to bulimia, the prevalence of the syndrome remains difficult to estimate, due to the variance of the results. 1 to 17% of the students will be bulimic. Numerous methodological imperfections may explain this disparity. In the studies the populations differentiate in accordance with age, social or economical level or ethnic group. Furthermore many of them were not randomly selected. Additionally, a few studies draw a comparison between the respondents and the sample from which they were drawn. This comparison is very important when the response rates are low. These two factors weaken the representation of the studies. All but one of the studies evaluated subjects by means of self-report questionnaires. The terms used are often not precise enough and there is a risk that individual subjects will interpret them idiosyncratically. Moreover, a lot of these questionnaires were pioneered in the studies, and reliability and validity data are lacking. The screening instruments "BITE" and "BULIT" have been formally evaluated, although they have not been used for the epidemiological evaluations. The BEQ has a good reliability, but has a low rate of identifying false positives. Studies using these self-report questionnaires operationalize the DSM III, the DSM III-R or the Russell criterias. As far as bulimia is concerned, there exist a small relationship between DSM III and DSM III-R. Furthermore, the DSM III criteria for bulimia have been criticized for being overinclusive. The inclusion in the DSM III-R of "methods of weight control" and "a persistent overconcern with shape and weight" contribute to decrease the prevalence of the disease.(ABSTRACT TRUNCATED AT 250 WORDS)