Study selection

A total of 7298 records were identified through electronic database searches (Fig. 1). After removing 1611 duplicates, 5687 titles and abstracts were screened, of which 691 full-text articles underwent review. Of these, 630 did not meet eligibility criteria, leaving 61 studies included in the final synthesis.

Fig. 1

Flowchart of included studies. DEBs, disordered eating behaviours

Population characteristics

The 61 studies included 111,208 participants across more than 22 countries (Table 2, ESM Table 3). Of these, 27,419 people had type 1 diabetes, 70,071 people had no diabetes diagnosis, 867 people had type 2 diabetes, and 11,910 people had T1DE. Additionally, 821 people lived with an eating disorder but no diabetes, 45 were medical outpatients and 75 were healthcare professionals.

Table 2 Summary of included studies

Gender/sex data were reported in 60 studies; 18 included only female participants. In total, 76% of participants were women. One study [28] did not report gender/sex, and two studies [29, 30] reported gender/sex for participants with type 1 diabetes but not for healthcare professionals (ESM Table 3).

Study characteristics

Of the 61 studies, 59 were observational and two were experimental. Observational designs included 28 cross-sectional, 11 cohort and seven case–control studies, as well as seven qualitative, three mixed-methods and three ecological momentary assessment (EMA) studies. One quasi-experimental study and one feasibility randomised controlled study were also included. One study spanned 22 countries, predominantly the USA and the UK. Eighteen studies were conducted in the USA, 11 in the UK, six in Australia, four in Norway, two each in France, the Netherlands, Ireland, Japan, Germany, Portugal and Turkey, and one each in Brazil, Canada, Greece, Italy, Morocco, Spain, Poland, Republic of Korea and Finland. Publication dates ranged from 1989 to 2025 (ESM Table 3).

Quality appraisal

Overall study quality was good, with internal validity generally higher than external validity (ESM Table 3). Qualitative studies scored well; six had a low risk of bias for external validity and one was unclear. For the 54 quantitative studies, external validity varied: about 33% had a low risk of bias, 59% an unclear risk and 7% a high risk. Internal validity was generally stronger than external validity, with 48% showing a low risk of bias, 48% an unclear risk and 4% a high risk. Cross-sectional and cohort studies displayed a wide range of quality, with most meeting several validity criteria but some showing unclear risks due to limited representativeness or potential confounding. Case–control studies scored well, with five having good internal and external validity. Mixed-methods and EMA studies were generally well conducted, although applying appraisal checklists not designed specifically for mixed methods introduced interpretive bias.

Definition and classification of T1DE

In total, the 61 included studies used 29 different approaches to identify or define T1DE (Fig. 2). These approaches included validated and non-validated questionnaires, clinical assessments, diagnostic classification systems clinical records and patient self-diagnosis. Approximately 40% of the studies (n=24) applied multiple approaches [4, 10, 31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52] to supplement their primary screening measure for T1DE. For example, Harrison et al [46] combined a validated diabetes and disordered eating questionnaire with a clinical assessment.

Fig. 2

Methods and tools used to define disordered eating in people with type 1 diabetes. BITE, Bulimic Investigatory Test of Edinburgh; BULIT-R, Bulimia Test – Revised; CDRS, Contour Drawing Rating Scale; DEBQ, Dutch Eating Behavior Questionnaire; DEPS-R, Diabetes Eating Problem Survey – Revised; DOS, Düsseldorf Orthorexie Scale; EDCBQ, Eating Disorder Compensatory Behaviour Questionnaire; EDE-Q, Eating Disorders Examination Questionnaire; EDI, Eating Disorder Inventory; EAT-26, Eating Attitudes 26; ED, eating disorder; EDI-3RC, Eating Disorder Inventory Risk Composite; LENTCA, Logistic System for Nosographic Evaluation of Eating Disorders; m-SCOFF, modified Sick, Control, One stone, Fat, Food; ORTO-15, Ortorexie-15; SCID-P, Structured Clinical Interview for DSM Disorders; SCOFF, Sick, Control, One stone, Fat, and Food

While some studies employed diabetes-specific questionnaires, others adapted general eating disorder questionnaires that were not standardised for people with type 1 diabetes. Researchers sometimes combined a standardised questionnaire with clinical interviews or extracted diagnoses from medical records. In other studies, participants self-reported T1DE symptoms or identified themselves as having ‘diabulimia’, which is not formally recognised as a distinct diagnosis. The different classification approaches are described in the following sections.

Validated questionnaires

The Diabetes Eating Problem Survey – Revised (DEPS-R) [53] was the most frequently used questionnaire, used in 24 studies [4, 7, 35, 38, 46, 48,49,50,51,52, 54,55,56,57,58,59,60,61,62,63,64,65,66,67]. Authors reported that the DEPS-R had good reliability and validity and captured T1DE behaviours more accurately than general eating disorder questionnaires [68]. Some studies adopted the DEPS-R as their primary measure; others paired it with additional questionnaires or clinical assessment frameworks.

The Eating Disorder Examination Questionnaire (EDE-Q) [69] was used in 14 studies [31,32,33, 35, 40, 43, 47, 49,50,51,52, 60, 70, 71], sometimes alongside DSM or ICD criteria. The EDE-Q and its interview-based version, the EDE, were selected because they are validated for eating disorders in the general population and have good internal consistency in populations with type 1 diabetes [69]. When paired with clinician-based interviews, authors believed that the EDE-Q detected subclinical symptoms.

The Bulimia Test – Revised (BULIT-R) [72], a tool assessing bulimic symptomatology that is grounded in DSM-3-R criteria for bulimia nervosa, was used in five studies [10, 31,32,33, 41].

The Eating Disorder Inventory (EDI) [73] and its variations (EDI-3RC, EDI-2) were employed in five studies [10, 36, 39, 44, 74], with authors selecting them for their robust psychometric properties.

Other validated questionnaires adapted from the general eating disorder literature included the Eating Attitudes Test (EAT-26) [14], applied in five studies [54, 75,76,77,78], and the Sick, Control, One stone, Fat, Food (SCOFF) questionnaire [79], used in two studies [34, 80]. Five more validated questionnaires were used in one study each: the Bulimic Investigatory Test of Edinburgh (BITE) [45], the Appearance Schemas Inventory [37], the Dutch Eating Behaviour Questionnaire [81], the Düsseldorf Orthorexie Scale (DOS) [43], and the Contour Drawing Rating Scale (CDRS) [47].

Non-validated questionnaires

Non-validated questionnaires, each used in a single study, included a modified version of SCOFF (m-SCOFF) [34], Ortorexie-15 (ORTO-15) [43] and the Eating Disorder Compensatory Behaviour Questionnaire [36]. These were employed to capture aspects not fully represented in validated tools, such as body image distortions, orthorexia or compensatory behaviours such as insulin restriction.

Author-generated questionnaires

In eight studies, authors developed their own questionnaires to assess behaviours or cognitions specific to T1DE [9, 37,