Psychological Medicine

Original Article

An empirical study of the typology of bulimia nervosa and its spectrum variants

a1 Department of Psychology, Wesleyan University, Middletown, CT, USA
a2 Department of Counseling and Applied Educational Psychology, Northeastern University, Boston, MA, USA
a3 Maryland Medical Research Institute, Baltimore, MD, USA
a4 Westat, Inc., Rockville, MD, USA
a5 Department of Cardiology, Cincinnati Children's Hospital, Cincinnati, OH, USA

Article author query
striegel-moore rh   [PubMed][Google Scholar] 
franko dl   [PubMed][Google Scholar] 
thompson d   [PubMed][Google Scholar] 
barton b   [PubMed][Google Scholar] 
schreiber gb   [PubMed][Google Scholar] 
daniels sr   [PubMed][Google Scholar] 


Background. There is an ongoing debate about the best way to classify eating disorders. This study examined potential subtypes of bulimia nervosa.

Method. Latent class analysis (LCA) was used to identify subtypes of bulimic symptomatology, utilizing data from 234 respondents in a cohort of black and white young women (n=2054). Participants were administered gated screening questions from the Eating Disorders Examination to determine DSM-IV diagnoses of eating disorders.

Results. A 3-class solution was judged best. Class 1, the ‘purger subtype’ (n=116), was characterized by vomiting, the use of fasting/diet pills, and relatively little bingeing. Class 2, the ‘binger subtype’ (n=97) comprised women who engaged in bingeing but minimal compensatory behaviors. Class 3, the ‘binge-purger subtype’ (n=21) had relatively high rates of all symptoms. Findings of differences between the three subtypes on validator variables and differences between the three subtypes compared to non-eating disorder groups suggest validity of the three bulimic subtypes identified in our analyses. Ethnicity and class membership were associated [χ2(3)=21·89, p<0·0001], reflecting a greater percentage of white women than black women in Class 1 and a greater percentage of black women than white women in Class 2.

Conclusions. LCA revealed one subtype that was similar to bulimia nervosa and two subtypes of bulimic symptomatology that did not meet criteria for bulimia nervosa yet appear to be clinically significant. Further study of the psychological correlates, course, and treatment response of these classes would be of clinical interest.

(Published Online October 12 2005)

c1 Wesleyan University, Department of Psychology, 207 High Street, Middletown, CT 06459-0408, USA. (Email: