S
Soveriaclinical platform
Log InStart Free →

Eating disorder screening: EAT-26 for first-line screening, EDE-Q for in-depth assessment

8 min read
May 13, 2026
For Everyone
Galmiche et al., Am J Clin Nutr, 2019; meta-analysis of 94 studies, lifetime prevalence
8,4%
of women in the general population experience at least one eating disorder in their lifetime. Among men — 2.2%. Most cases go undiagnosed: the classical picture of anorexia or bulimia is rarer than atypical forms and binge eating disorder, which are easy to miss without structured screening.
Lifetime prevalence of EDs by type and population
Any ED (women)
8,4%
BED (binge eating)
3,5%
Any ED (men)
2,2%
Bulimia Nervosa
1,5%
Anorexia Nervosa
0,9%

What EDs are and why diagnosis gets missed

EDs are a heterogeneous group of disorders unified by disturbed eating patterns and associated psychological dysregulation. DSM-5 distinguishes six main diagnoses: Anorexia Nervosa (AN), Bulimia Nervosa (BN), Binge Eating Disorder (BED), ARFID, OSFED (other specified), and USFED (unspecified). The ICD-11 structure is close, with minor differences.

The central clinical underdiagnosis pattern is the stereotype "ED = a young thin girl". Reality, per Galmiche et al. (2019, *Am J Clin Nutr*, meta-analysis of 94 studies): lifetime prevalence of any ED is 8.4% in women and 2.2% in men; point prevalence rose from 3.5% (2000–2006) to 7.8% (2013–2018). Binge Eating Disorder at 3.5% lifetime is the most common of all EDs and equally prevalent in men and women. Atypical anorexia (anorexic psychopathology at normal or elevated weight) — a common OSFED form especially easy to miss.

Key fact

EDs are not defined by weight. Anorexia nervosa can occur at normal weight (atypical AN — a formal DSM-5 category). Binge eating disorder is often accompanied by obesity, but it isn't "overeating" — it is a structural psychiatric disorder marked by compulsivity and emotional dysregulation. Hudson et al. (2007) in the NCS-R showed: only a minority of people with EDs ever received treatment specifically for the ED. Structured screening is the only reliable way to identify cases that don't fit the stereotype.

EAT-26: first-line screening

EAT-26 (Eating Attitudes Test-26, Garner, Olmsted, Bohr & Garfinkel 1982, *Psychological Medicine*) — 26 items, mostly with a 0–3 Likert scale. Total range 0–78. A cut-off of ≥ 20 indicates risk warranting further evaluation. The scale is freely distributed by the author for clinical and research use.

Psychometrics. Cronbach α across international samples typically runs 0.86–0.94. Sensitivity and specificity at the ≥ 20 cut-off vary widely by population: in clinical samples, Constaín et al. (2014, 2016) showed sensitivity up to 100% and specificity 98% in Colombian male and female clinical groups; in community samples, Nunes et al. (2005) on Brazilian women found sensitivity of only 40%. Kang et al. (2017) on a Chinese clinical sample reported sensitivity 0.66–0.68. The conclusion: EAT-26 is a strong detection tool in clinical contexts, but in universal screening (schools, primary care) it produces substantial false negatives and benefits from a two-stage approach.

Three empirically derived subscales: Dieting (dietary patterns), Bulimia and Food Preoccupation (bulimic and food-related thoughts), and Oral Control (control over eating). Subscale scores give a profile, but clinical decisions always rest on total score plus structured diagnostic assessment.

  • 26 items, completion time 5–10 minutes
  • Total score 0–78, cut-off ≥ 20 — risk threshold
  • 3 subscales: Dieting, Bulimia/Preoccupation, Oral Control
  • Sensitivity varies 40–100% across populations — clinical context matters
  • Freely distributed by the author (eat-26.com)

How to read EAT-26: the ≥20 cut-off and its limits

Standard cut-off interpretation:

  • 0–9: normal / non-clinical range
  • 10–19: mild concern with eating patterns; monitor in the presence of other risk factors
  • ≥ 20: clinically significant ED risk; requires further structured assessment (EDE-Q + clinical interview)

A core clinical caveat: an EAT-26 cut-off does not diagnose the ED type. A high score can come from anorexia nervosa, bulimia nervosa, OSFED, or BED — and the instrument covers BED least well (binge episodes are not represented separately). So a positive EAT-26 is a signal to move on to more specific instruments, not a final answer. Conversely: a negative EAT-26 does not rule out an ED, especially atypical anorexia or BED — clinical screening should run in parallel.

EDE-Q: in-depth DSM-5 assessment

EDE-Q (Eating Disorder Examination Questionnaire, Fairburn & Beglin 1994) — 28 items, assessing behavioral patterns over the past 28 days. Unlike EAT-26, EDE-Q covers all key DSM-5 diagnostic criteria: restraint over eating, compensatory behaviors, objective binge-eating episodes, and body image disturbance.

Four subscales: Restraint, Eating Concern, Shape Concern, Weight Concern. Each scale runs 0–6 (mean of items). The Global score is the mean across all 28 items. Clinical cut-offs vary 2.3–4.0 depending on population. Psychometrics are stable: Mond et al. (2004) on 5,255 Australian women yielded Cronbach α = 0.93 for the Global score and test-retest correlations of 0.66–0.94 over 2 weeks. Berg et al. (2012) reviewed EDE/EDE-Q and confirmed a stable factor structure across most samples.

Soveria · Measurement-Based Care
Start practicing MBC — from your first session
14 days free. Soveria integrates into your practice — no method change required.
Try for free

The key difference from EAT-26 — EDE-Q gives the frequency of specific behaviors over 28 days: objective binge episodes, compensatory strategies, excessive exercise. This makes it possible to apply the DSM-5 algorithm to differentiate anorexia / bulimia / BED / OSFED. EAT-26 doesn't offer this.

EDE-Q is available in Soveria. Completion takes 10–15 minutes. The platform automatically computes the four subscales and Global score, and reports behavioral-episode frequencies separately for clinical interpretation.

Screening vs diagnosis: where the boundary lies

The boundary between screening and diagnosis is a central methodological anchor in ED work. Screening (EAT-26 or EDE-Q as self-report) provides a signal: "there is probably a problem". The diagnosis is made through a structured clinical interview verifying DSM-5/ICD-11 criteria, duration, frequency, functional impact, and medical status.

The gold-standard diagnostic interview is the EDE (Eating Disorder Examination, Cooper & Fairburn 1987, current version 17.0D) — the interviewer-administered counterpart of EDE-Q. It takes 60–75 minutes and requires training. The alternative is SCID-5-ED (eating disorder module of SCID-5).

Why specifically a structured interview? Because EDs include behaviors patients frequently conceal: compensatory episodes, dietary restriction, body image disturbance. Self-report scales can systematically underestimate these. The EDE adds probe questions, clarifications, and calibration to standardized definitions — something no questionnaire alone provides.

Comorbidity and why AN has the highest mortality of any mental disorder

EDs almost never exist in isolation. Per Hudson et al. (2007, NCS-R, n = 9,282), comorbidity with other psychiatric disorders for anorexia nervosa, bulimia nervosa, and BED runs 56–94% across categories: depression, anxiety disorders, impulse-control disorders, substance use. The implication: a positive EAT-26 or EDE-Q should always be accompanied by comorbidity screening (PHQ-9, GAD-7 at minimum; PCL-5/ITQ when a traumatic background is suspected).

A separate clinical reality is medical status. Anorexia nervosa has the highest mortality of any psychiatric disorder: a standardized mortality ratio of 5.86 (Arcelus et al. 2011, meta-analysis). That means people with AN have a death rate roughly six times higher than the general population of the same age and sex. About one-fifth of AN deaths are suicides; the rest are medical complications of prolonged starvation and compensatory behaviors.

Practical takeaway

A positive ED screen does not mean "book 8 sessions of CBT". It is a signal for a multidisciplinary assessment: psychotherapist + physician (physical status, ECG when AN is suspected, lab markers when purging is involved) + dietitian where appropriate. AN with significant weight loss often requires medical care as the first priority; psychotherapy works effectively only after physical stabilization.

How to use in practice

Bottom-line guidance — five typical workflows for private practice.

  • Initial appointment with suspected ED: EDE-Q (available in Soveria) + comorbidity screening (PHQ-9, GAD-7). When positive → referral for structured EDE interview or an ED specialist with medical evaluation
  • Universal screening in primary care or university clinics: EAT-26 as a first-line filter; positives → EDE-Q or interview. Keep in mind sensitivity ranges from 40% to 100% by population — a two-stage approach is needed
  • Tracking treatment progress in ED therapy: EDE-Q every 4–8 weeks — following change across the four subscales and behavioral-episode frequencies
  • Client self-seeking an "ED test": EAT-26 as online screening — but with a mandatory recommendation to consult a specialist on a positive result. Self-screening without subsequent clinical interpretation is methodologically incorrect
  • Does not replace: medical evaluation (physical status, ECG, lab markers) — that is the physician's domain, not the psychologist's

A brief note on Russian-language adaptations. EAT-26 and EDE-Q are used in Russian clinical practice via working translations; formally published psychometric validations on Russian-speaking samples are not currently indexed in PubMed. For practice this means: use the instruments anchored to international psychometric data, always confirm interpretation in clinical context, and avoid claims of "officially RF-validated version" — that fact is not in the published literature.

The core principle — self-report tells you "there is something to discuss"; a structured interview plus medical evaluation make the diagnosis. EAT-26 and EDE-Q are the first and second steps in differential assessment, not the third and not the last. In Soveria, EDE-Q runs as a self-report with automatic four-subscale breakdown and behavioral-episode frequencies — this turns "let's talk about how you've been eating this month" into specific numbers that can anchor clinical decisions.

Soveria Digest
Every week.
Only what works.

Articles on evidence-based therapy, platform updates, and clinical instrument reviews. No noise.

Weekly · Unsubscribe anytime
Start Free →