Screening for Eating Disorders in Women with T1D: Validating the DEPS-R
Severina Haugvik, Mareille H.C.L. Hennekes, Maartje de Wit, Elena Toschi, Christopher D. Desjardins, Torild Skrivarhaug, Knut Dahl-Jørgensen, Eric Stice and Line Wisting
Can the Diabetes Eating Problem Survey-Revised (DEPS-R) reliably identify eating disorder diagnosis in women with type 1 diabetes? Diabetic Medicine 2026;43:e70207. Epub 2026 Jan 8.
Women with type 1 diabetes (T1D) face an elevated risk of disordered eating behaviours (DEB) and eating disorders (ED), which are associated with serious diabetes complications including increased mortality risk from insulin omission. The Diabetes Eating Problem Survey-Revised (DEPS-R) is the most widely used diabetes-specific screening tool and is recommended by international guidelines, yet its ability to identify a formal DSM-5 eating disorder diagnosis had never been validated against a clinical diagnostic interview using ROC analysis. This study, involving researchers from Amsterdam UMC, Oslo University Hospital, Stanford University, and Joslin Diabetes Center, addresses that gap using data from a large multinational trial.
Baseline data from 293 women aged 14–35 years with T1D and body image concerns, enrolled in the multinational Diabetes Body Project randomised controlled trial across Norway, the Netherlands, and the United States, were analysed. The DEPS-R was evaluated against the semi-structured Eating Disorder Diagnostic Interview (EDDI), which establishes DSM-5 diagnoses. ROC analysis, univariate logistic regression, and two-sample t-tests were performed.
Key POINTS:
- The DEPS-R shows good overall accuracy: ROC analysis yielded an area under the curve (AUC) of 0.82, meaning the tool correctly identified eating disorder status in 82% of cases.
- The current cutoff of ≥20 has high sensitivity but moderate specificity: The established threshold correctly identified 87.5% of women with an ED (sensitivity), but also flagged many without one, with a specificity of only 60.4% and a false positive rate of 39.6%.
- Alternative cutoffs may be considered: A threshold of ≥25 offered the best balance between sensitivity (75.0%) and specificity (77.9%), while ≥17 maximised sensitivity (96.8%) at the cost of identifying half the sample as positive screens.
- Twelve of sixteen items were significantly associated with an eating disorder diagnosis: The strongest predictors were items relating to intentional hyperglycaemia for weight loss (item 9), self-induced vomiting (item 8), insulin omission after overeating (item 13), loss of control over eating (item 14), alternating between restriction and bingeing (item 15), and eating in secrecy (item 5).
- A stepwise clinical approach is proposed: Items 14, 15, and 5 — which address broader eating behaviours — may serve as accessible entry questions in clinical consultations, while items 9, 8, and 13 — reflecting more severe compensatory behaviours — can help confirm pathology and guide referral decisions.
- High DEPS-R scores are associated with poorer clinical outcomes: Women scoring ≥20 showed significantly higher diabetes distress, lower diabetes-specific quality of life, higher HbA1c, lower time-in-range, and greater body dissatisfaction compared to those scoring below the threshold.
This is the first study to validate the DEPS-R against a formal DSM-5 eating disorder diagnosis using ROC analysis in a large multinational sample of young women with T1D. The findings confirm the DEPS-R as a useful screening tool with good overall accuracy and high sensitivity, while highlighting its moderate specificity as a limitation. The identification of specific items most strongly associated with eating disorder diagnosis offers a practical framework for clinicians to integrate structured conversations about disordered eating into routine diabetes care, reducing unnecessary referrals while ensuring those most at risk are identified and supported.
Concluding, the authors state
"The DEPS-R showed good performance to identify correct ED status. .......... Collectively, the findings expand the knowledge of the DEPS-R and suggest potential implications for clinical practise." -
Please click here for the full text pdf.