Differential diagnosis of eating disorders with the use of classification trees (decision algorithm)

Abstract

Background: The aim of the study was to establish whether it is possible to make a correct diagnosis of various types of eating disorders on the basis of several variables. Method: A group of 213 females, 20 to 26 years of age, took part in the study. 92 women (control group) and 121 women (research group). We used the Eating Disorder Inventory (EDI), the Socio-cultural Attitudes Towards Appearance Questionnaire-3 (SATAQ-3) and a questionnaire of behavior towards the body. A selection of dependent variable predictors of disorder was made and a classification tree was developed. Results: People suffering from eating disorders differ from healthy individuals in the intensity of the variable "pattern for constructing emotional relationships on the basis of lack of trust and certainty in interpersonal relations". People suffering from various types of eating disorders differ - besides the BMI - on such psychological traits as fear of gaining weight and a restrictive strive to be slim, as well as dissatisfaction with the body. A differential diagnosis in line with a decision algorithm correctly identified healthy individuals in 100% of cases, bulimia in 78.72% of cases, binge eating disorder in 93.33% of cases, and anorexia in 86.36% of cases. Conclusions: Psychological predictors of eating disorders may be arranged in the form of a classification tree. It is possible to make an accurate differential diagnosis of eating disorders on the basis of results of measurements of six variables.