Diabetes Self-Management Questionnaire (DSMQ)

Disordered eating in type 1 diabetes has been associated with ineffective self-management.8 We use the DSMQ to assess diabetes self-care activities on four subscales: Glucose Management, Dietary Control, Physical Activity, and Healthcare Use. Respondents rate how 16 different statements apply to their self-management regarding the last 8 weeks on a 4-point Likert scale. Higher scores indicate more effective diabetes self-management (range 0–48). The DSMQ (German version) was attested good psychometric qualities (eg, Cronbach’s α=0.84).15

GOLD score and Hypoglycaemia Fear Survey-II (HFS-II), short form

Episodes of hypoglycaemia can trigger disinhibited eating, which might be accompanied by feelings of guilt and shame.16 Unimpaired awareness of hypoglycaemia onset might prevent disinhibited eating as well as fear of hypoglycaemia, which is a diabetes-related psychosocial factor that has received limited attention in DEB research.4 To assess hypoglycaemia awareness, we use the GOLD score, which is a one-item self-report scale. Respondents are asked to rate their awareness of hypoglycaemia onset on a 7-point Likert scale from 1, ‘always aware of the onset of hypoglycaemia’ to 7, ‘never aware of the onset of hypoglycaemia’. A GOLD score≥4 indicates impaired hypoglycaemia awareness.17 To assess fear of hypoglycaemia, we use the Hypoglycaemia Fear Survey that comprises 33 items on two subscales (Worry and Avoidance). The German translation showed good psychometric qualities (eg, Cronbach’s α=0.96 for Worry subscale and Cronbach’s α=0.86 for Avoidance subscale).18 We decided to use the more economic HFS-II short form, a reliable and valid measure to assess the level of fear of hypoglycaemia with 11 items referring to the last 4 weeks.19 Answers are scored on a 5-point Likert scale, with higher sum scores representing higher levels of fear of hypoglycaemia (range 0–44).

Problem Areas in Diabetes Scale (PAID)

The PAID is a 20-item questionnaire assessing current diabetes-related distress, a key diabetes-related psychosocial factor that has been associated with DEB10 20 but has received limited research attention.4 The PAID items are rated on 5-point Likert scale and scores are summed and transformed into a total score (range 0–100). A higher total score indicates more severe diabetes-related distress, with a cut-off criterion of ≥40 for elevated diabetes-related distress. The German PAID translation was attested good psychometric quality (eg, Cronbach’s α=0.92).21

Diabetes Eating Problem Survey-Revised (DEPS-R)

The DEPS-R is a 16-item screening instrument for ED in type 1 diabetes, also covering underdosing or omission of insulin. The items are answered on a 6-point Likert scale, with higher sum scores indicating more frequent DEB in the last 4 weeks (range 0–80). A sum score≥20 indicates high ED risk. The German DEPS-R version was attested good internal consistency (Cronbach’s α=0.84) and showed significant correlations with HbA1c levels, Body Mass Index standard deviation scores and expert clinician reports.22 In addition to the DEPS-R sum score, we will use two single-item scores (item 4, ‘When I overeat, I don’t take enough insulin to cover the food’ and item 13, ‘After I overeat, I skip my next insulin dose’) to investigate the frequency of underdosing and omission of insulin.

International Physical Activity Questionnaire (IPAQ), short form

Physical activity has been shown to improve physical health and well-being in type 1 diabetes,23 thus might functioning as a protective factor for DEB. However, excessive exercise has also been associated with DEB in type 1 diabetes.24 We chose the IPAQ because it captures both physical activity in general as well as more intense exercise. The short IPAQ version consists of nine items, measuring the intensity (sedentary activity, walking, moderate activity, vigorous activity), duration (minutes) and frequency (days per week) of physical activity in the last 7 days or a regular week. A minimum of 150 min of activity per week indicates sufficient physical activity. The IPAQ was developed to obtain internationally comparable self-report data on physical activity and has acceptable psychometric properties (eg, test–retest reliability r=0.80).25

Eating habits

Strict dieting, dietary restrictions (eg, only ‘healthy’ foods are allowed) and emotional dysregulation are central components in models for development and maintenance of DEB in type 1 diabetes (eg,6). Therefore, we chose to include pattern of eating (ie, frequency of meals and snacks), eating style, as well as emotional eating (eating as a strategy to cope with emotional distress) in the DEBBI study. We further plan to use these variables to identify possible DEB subtypes. To assess pattern of eating, two questions of the German version of the Eating Disorder Examination (EDE) interview26 were adapted. Participants are asked to rate meals and snacks of the last 4 weeks (breakfast, mid-morning snack, lunch, mid-afternoon snack, evening meal, evening snack, nocturnal eating) on a 7-point Likert scale (0, ‘meal/snack not eaten’ to 6, ‘meal/snack eating every day’). In case of nocturnal eating episodes, participants are further asked if impairment of awareness or recall occurred. To assess healthy eating style, we use a 16-item questionnaire by König et al.27 The questionnaire presents statements on food preferences and healthy eating behaviours (eg, ‘I do not eat fast food’, ‘I eat a lot of fruit and fresh vegetables’). Items are answered on a 7-point Likert scale, with a higher total score representing a healthier eating style (range 16–112).27 To assess emotional eating, we use a subscale of the Intuitive Eating Scale-2 (IES-2). The IES-2 examines intuitive eating, which has been described as reliance on internal cues of hunger and satiety and low preoccupation with food.28 The subscale ‘Eating for physical rather than emotional reasons’ (EPR) consists of eight items which are answered on a 5-point Likert scale. Higher mean values indicate less emotional eating (range 1–5). The EPR subscale was attested good psychometric properties (eg, Cronbach’s α=0.91).28

Eating Disorder Examination-Questionnaire (EDE-Q), short form

The EDE-Q is a reliable and valid ED questionnaire based on the EDE interview. It complements the diabetes-specific DEPS-R questionnaire by measuring the frequency and severity of (diabetes unspecific) ED symptoms, for example body dissatisfaction, which has been associated with DEB.20 29 The short form (EDE-Q8) consists of eight items, two from each of the subscales Restraint, Eating Concern, Weight Concern and Shape Concern. The items refer to the last 28 days and are scored on a 7-point Likert scale. A mean global ED psychopathology score and mean subscale scores can be calculated (range 0–6). Higher mean scores indicate more frequent and/or severe ED symptoms. The German EDE-Q8 version was attested good psychometric properties (global score; eg, Cronbach’s α=0.93).30

Perceived Stress Scale (PSS-10) and locus of control (IE-4)

It is hypothesised that the psychological burden associated with a chronic medical condition like type 1 diabetes enhances vulnerability for DEB and feelings of loss of control.4 31 The latter may lead to insulin misuse as a mean to regain control,4 which may also be associated with locus of control. We include perceived stress and locus of control as a way to picture psychological burden, and to inform possible subtypes. We will further be able to differentiate between diabetes-specific and diabetes-unspecific levels of stress associated with DEB. We use a German adaption of the PSS-1032 to measure diabetes-unspecific levels of stress, comprising 10 items (on two subscales, Perceived Helplessness and Perceived Self-Efficacy) referring to the last month. Respondents are asked to rate their feelings and thoughts on a 5-point Likert scale. Higher total scores reflect greater stress levels (range 10–50). The German adaption was attested good psychometric quality (total score; eg, Cronbach’s α=0.88 in a nonclinical and α=0.89 in a clinical sample).32 To assess internal and external locus of control, we use the German IE-4 scale that comprises four items on two subscales (Internal vs External Locus of Control).33 The items are answered on a 5-point Likert scale. Mean scores for each subscale can be computed and compared with reference values for the German population (range 1–5). The questionnaire was attested satisfactory psychometric properties (eg, test–retest reliability r=0.56 for Internal and r=0.64 for External Locus of Control).33

Brief COPE

A relationship between dysfunctional coping strategies, like avoidance, and DEB in type 1 diabetes has been demonstrated.9 We assess coping strategies with a German adaption of the Brief COPE,34 a 28-item questionnaire comprising 14 scales that can be summarised into four latent coping scales (Focus on Positive, Support Coping, Active Coping, Evasive Coping). The items refer to usual thoughts and actions while facing a difficult situation. Items are answered on a 4-point Likert scale. Mean coping scale scores represent the degree to which the participant has been engaging in that coping strategy (range 1–4). The German adaption showed satisfactory psychometric properties (eg, Cronbach’s α=0.76 for Focus on Positive, α=0.81 for Active Coping, α=0.76 for Support Coping, α=0.70 for Evasive Coping).34

Brief Patient Health Questionnaire (PHQ-9) and Staite-Trait-Anxiety Inventory (STAI), subscale trait anxiety

Mood dysregulation and negative affect are central components of models in DEB in type 1 diabetes,8 20 29 and bidirectional associations between depressive symptoms, anxiety and DEB have been demonstrated.35 36 In the DEBBI study, depressive symptoms are assessed with the German version of the 9-item depression module from the Patient Health Questionnaire.37 The items are scored on a 4-point Likert scale, referring to the frequency of depressive symptoms in the last 2 weeks. Higher total scores represent more frequent/severe depressive symptoms (total score range 0–27). A total score ≥5, ≥10, ≥15 represents mild, moderate and severe depression, respectively. The German PHQ-9 version was attested satisfactory psychometric properties (eg, Cronbach’s α=0.87).37 To assess anxiety, we use the trait subscale of the well-established German version of the STAI.38 The trait questionnaire consists of 20 items and asks respondents to rate how they generally feel on a 4-point Likert scale. Higher total scores represent greater trait anxiety (range 20–80). The trait subscale has good psychometric quality (eg, Cronbach’s α=0.90).38

Rosenberg’s Self-Esteem Scale

To measure self-esteem, a factor that has been proposed to endorse the development and maintenance of DEB in type 1 diabetes,6 8 we use the revised German adaptation of Rosenberg’s Self-Esteem Scale.39 Participants are asked to rate 10 items on a 4-point Likert scale. Higher total scores (range 0–30) indicate higher global self-esteem. The German revised version was attested good psychometric properties (eg, Cronbach’s α=0.84).39

Multidimensional Perfectionism Scale (MPS-F)

Perfectionism has been discussed to promote frustration in diabetes self-management and, thus, as a part of a dysfunctional scheme for self-evaluation, can lead to DEB like dietary restriction, binge-eating and insulin omission.6 29 To measure perfectionism, we use three subscales of the German version of the MPS-F that were attested satisfactory construct validity.40 Our adaption consists of 20 items measuring perfectionisms on the subscales Personal Standards, Concern over Mistakes and Doubts about Action. Participants indicate their answers on a 6-point Likert scale. Higher total scores represent greater perfectionism (range 20–120). The three selected subscales were attested satisfactory internal consistency (Personal standards, Cronbach’s α=0.84; Concern over Mistakes, Cronbach’s α=0.89; Doubts about Action, Cronbach’s α=0.70).40

Self-generated items

We use self-generated items to gather information on diabetes duration, insulin therapy, insulin doses, HbA1c levels, diabetes technology use, glucose fluctuation, participation in structured patient education programmes, acute and long-term diabetes complications, hypoglycaemia-related disinhibited eating (four questions adapted after16), insulin overdosing (which might be practiced to be ‘allowed’ to eat sweets to treat hypoglycaemia), number of diets, snacking, tobacco and alcohol consumption, demographics, weight, height and comorbid physical/mental health diagnosis and treatment.

To investigate glucose fluctuation, participants using continuous glucose monitoring are asked to enter the mean time they spent in moderate-to-severe hypoglycaemia (% of glucose values<55 mg/dL or 3.1 mmol/L), in mild hypoglycaemia (% of glucose values 55–70 mg/dL or 3.1–3.9 mmol/L), in range (% of glucose values 70–180 mg/dL or 3.9–10 mmol/L) and in hyperglycaemia (% of glucose values>180 mg/dL or 10 mmol/L) in the last 30 days.