Sensitivity and Criterion Validity of ED100K Lifetime Eating Disorder Diagnoses: A Study Using a National Eating...

Epidemiological and genomic studies of eating disorders (ED) require scalable approaches to case ascertainment. The self-report ED100K questionnaire assesses lifetime ED diagnoses, has been translated into several languages, and has been used in large cohort studies, with data shared or planned for sharing through major repositories and the Psychiatric Genomics Consortium.

This retrospective diagnostic validation study linked ED100K data from the Eating Disorders Genetics Initiative-Sweden (EDGI-SE), a case-control cohort, with clinician-assigned diagnoses from the Swedish national quality register for EDs (Riksät). In the subset with Riksät-linked specialist care records (n = 6889), the sensitivity of ED100K-derived lifetime DSM-IV diagnoses, DSM-5 anorexia nervosa (AN), and broadly defined binge-eating behavior (BE-BROAD) was evaluated against clinician-interview diagnoses. In the full EDGI-SE cohort (N = 15,699), comprising community-ascertained cases and controls, criterion validity was assessed via associations with ED specialist care.

Sensitivity was high for AN (DSM-IV: 0.84, DSM-5: 0.96) and substantially lower for bulimia nervosa (BN; 0.61) and binge-eating disorder (BED; 0.34). Sensitivity for BE-BROAD was high (0.88), exceeding BN and BED. Most individuals with BN or BED in specialist care not identified with corresponding ED100K diagnoses were captured by broader ED100K classifications, including BE-BROAD and DSM-5 diagnoses. All ED100K diagnoses and BE-BROAD were strongly associated with ED specialist care, providing evidence for the criterion validity of ED100K ED diagnoses.

The ED100K shows strong sensitivity for lifetime AN and BE-BROAD. Lower sensitivity for BN and BED likely reflects challenges in retrospective self-report (e.g., recalling frequency and duration) and in capturing diagnostic features requiring greater subjective judgment. Broader binge-eating classifications may better capture cases missed by algorithm-based BN and BED diagnoses.

EDGI is a registered clinical trial (clinicaltrials.gov ID: NCT04378101).