Eating Disorder Diagnostic Crossover: A Meta-Analysis

19/08/2026

AI Summary

Between-syndrome diagnostic crossovers are rare (0% to 13%), with BED to BN showing the highest observed rate.
Within-syndrome shifts and transitions to OSFED/EDNOS are common (17% to 25%); AN subtype shift from restrictive to binge-purging 23%.
Diagnostic criteria and assessment method significantly moderate crossover rates, supporting repeated specialist, interview-based reassessment in clinical practice.

AI Summary

This meta-analysis aims to quantify the frequency and direction of diagnostic crossover among EDs (both within-syndrome and between-syndrome), to evaluate study-level moderators of crossover rates, and to conduct a formal appraisal of the quality of the included studies.

Mixed-effects models tested diagnostic criteria, follow-up duration, treatment setting, diagnostic method, and demographics as moderators.

RESULTS: Between-syndrome crossovers (e.g., bulimia nervosa [BN] to anorexia nervosa [AN]) were rare, ranging from 0% (e.g., AN to binge eating disorder [BED]) to 13% (e.g., BED to BN), whereas within-syndrome shifts such as AN subtype shifts (23% from restrictive to binge-purging type) and transitions into other specified feeding or eating disorder (OSFED)/eating disorder not otherwise specified (EDNOS) were more common (17% to 25% across syndromes).

Basic summary

Int J Eat Disord. 2026 Aug 17. doi: 10.1002/eat.70175. Online ahead of print.

ABSTRACT

OBJECTIVE: Diagnostic crossover among eating disorders (EDs) is clinically important, yet rates are inconsistently quantified. This meta-analysis aims to quantify the frequency and direction of diagnostic crossover among EDs (both within-syndrome and between-syndrome), to evaluate study-level moderators of crossover rates, and to conduct a formal appraisal of the quality of the included studies.

METHOD: A literature search was conducted following PRISMA 2020 guidelines, including all relevant articles on PsycINFO, Pubmed/Medline, and ProQuest Dissertations & Theses Global databases from inception to June 10, 2026. Twenty-seven studies (24 samples; 287 effect sizes) were synthesized using random-effects models. Mixed-effects models tested diagnostic criteria, follow-up duration, treatment setting, diagnostic method, and demographics as moderators.

DISCUSSION: Diagnostic instability reflects within-AN shifts and OSFED/EDNOS transitions more than between-syndrome crossovers, underscoring the value of repeated specialist, interview-based reassessment.

PMID:42608954 | DOI:10.1002/eat.70175

Document this CPD