Abstract
Objective
To address whether eating disorders (ED) or insulin omission (IOM) in adult persons living with type 1 diabetes (pwT1D) are associated with impaired glycaemic control.
Design
Cross-sectional analysis.
Settings
The French-Speaking Diabetes Society—Type 1 Diabetes Cohort (SFDT1) is an ongoing epidemiological cohort study that includes pwT1D in France who attend hospitals or private ambulatory diabetes centres.
Participants
Adult participants from the SFDT1 study, with data on ED and IOM. The current analysis was performed on data collected during the baseline visit in participants enrolled between December 2020 and March 2024.
Main outcome measures
Using the SCOFF, a self-reported questionnaire to screen for ED, and a single question on IOM to screen for IOM, we described four categories of pwT1D: no ED & no IOM, ED & no IOM, no ED & IOM and ED & IOM. We performed unadjusted and adjusted (for age, sex, diabetes duration, social vulnerability, smoking, alcohol status and insulin treatment) multinomial logistic regression models with the four categories as the outcome and glycaemic variables as explanatory variables, including continuous glucose monitoring (CGM) variables and HbA1c. No ED & no IOM was the reference outcome for all comparisons. We stratified each model by sex and fear of hypoglycaemia.
Results
We included 1113 participants, 51% males, median (IQR) age 38 (29–50) years, diabetes duration 21 (12–32) years. Prevalences were as follows: no ED & no IOM: 68% (n=758), ED & no IOM: 11% (n=124), no ED & IOM: 16% (n=177) and ED & IOM: 5% (n=54). With the fully adjusted model, and compared with the group no ED & no IOM, time in range (OR (95% CI) 0.5 (0.4 to 0.7)) and time below range (0.5 (0.3 to 0.8)) were inversely associated with ED & IOM. Moreover, time in range (0.4 (0.4 to 0.5)) was associated with IOM & no ED. Time above range (2.2 (1.6 to 2.9)), Glycaemic Risk Index (1.8 (1.3 to 2.5)), glucose monitoring indicator (2.2 (1.7 to 2.9)) and HbA1c (2.0 (1.5 to 2.5)) were directly associated with ED & IOM. We did not observe associations between CGM variables and ED & no IOM. Most associations were valid in both men and women. The associations were stronger in participants with a fear of hypoglycaemia. However, the associations remained even in people with a fear of hypoglycaemia.
Conclusions
Both ED and IOM are frequent in pwT1D, and IOM seems to be associated with impaired glycaemic control. As our analysis was cross-sectional, we cannot infer causality and cannot know whether IOM was a result of glycaemic control or the inverse (reverse causality). Our results suggest that IOM should be systematically screened in clinical practice. Further research is needed to better identify and care for EDs, with or without IOM, in T1D.
Trial registration number
Keywords: Primary Health Care, PSYCHIATRY, DIABETES & ENDOCRINOLOGY
STRENGTHS AND LIMITATIONS OF THIS STUDY.
We addressed whether the presence of an eating disorder independently of insulin omission (IOM) affects glycaemic control.
We used data from a large French national cohort on 1113 persons living with type 1 diabetes.
We controlled the models for the most relevant explanatory variables.
This analysis was cross-sectional, and therefore, it is not possible to infer causality.
The IOM question addressed all-cause IOM without precision on whether it was related to body image concerns or weight-loss intention.
Introduction
Type 1 diabetes (T1D) is a chronic disease that often starts early in life and is increasing in incidence.1 Preventing complications requires the best possible strategy for glycaemic control, as early and sustained as possible.2 Considerable progress has been made with technological innovations such as continuous glucose monitoring (CGM) and automated insulin delivery (AID), which have been shown to improve glycaemic control.3
The British Parliament issued a report entitled ‘Type 1 Diabetes and disordered eating: Parliamentary Inquiry’ in January 2024 to help prevent T1D and disordered eating and support those living with this disorder in recovering and living well with T1D. The authors state people feel abandoned due to the lack of research and integration in health services.4
Eating disorders (ED) are defined as behavioural conditions characterised by severe and persistent disturbance in eating behaviours and associated distressing thoughts and emotions about body weight and body image.5 EDs are psychiatric conditions that lead to the search for specific comorbidities (such as depression and bipolar disorder) to orientate patients towards specific treatments (drugs, cognitive and behavioural therapies …). In 33 studies in the general population (excluding persons living with type 1 diabetes (pwT1D)), lifetime ED weighted mean (ranges) was 8.4% (3.3%–18.6%) for women and 2.2% (0.8%–6.5%) for men.6 In pwT1D, ED prevalence varies between 2% and 10% and is probably more frequent in women and adolescents.7 IOM is not considered a psychiatric condition by the DSM-V (DSM-V, Diagnostic and Statistical Manual of Mental Disorders). In pwtT1D, a particular aspect is diabulimia, which combines ED and IOM.8
Disturbed eating behaviours are ED symptoms, such as restraint, binge, night eating and weight concerns, that do not entirely fulfil the criteria for ED diagnosis.9 Insulin omission (IOM) is frequent in pwT1D and is associated with a higher likelihood of diabetes complications and often with losing weight purposes.10 Disturbed eating behaviours are frequent in pwT1D, including IOM, and are associated with increased morbidity–mortality.9 11 The prevalence of disturbed eating behaviours is reported to be 8%–12% in boys and men and 27%–40% in girls and women living with T1D.12,14 These disorders are primarily studied in women and less commonly in men.
Among disturbed eating behaviours, diabulimia is defined as intentionally withholding insulin to result in weight loss.8 It is associated with an increased frequency of acute (ketoacidosis) and chronic diabetes-related complications and mortality (with a sixfold increase)15 16 and an increased HbA1c level.13 17 A person suffering from diabulimia can refrain from accessing primary and secondary care, sometimes ending up in emergency units.4
Glycaemic control assessed by CGM metrics may be affected by disturbed eating behaviours or ED, but this relationship has not been extensively studied. A secondary analysis showed an association of disturbed eating behaviour with time above range (TAR) in 90 adolescents with T1D.18 Two studies in adults with T1D (n=83–59) reported associations between IOM and higher mean glucose and TAR.19 20 Another study evaluated ED with the SCOFF questionnaire and showed no relationship between ED and CGM metrics in 198 adults with T1D.21 Although IOM is expected to be associated with deteriorated CGM metrics, there is no evidence of this association to our knowledge.21
The relationship between glycaemic control and ED, with or without IOM, remains debated. Other factors, such as fear of hypoglycaemia, can be associated with delayed prandial boluses22 and IOM.23 However, evidence of the role of hypoglycaemia in IOM or ED is scarce. In this context, we aimed to investigate whether glycaemic control (CGM metrics and HbA1c levels) was associated with ED or IOM in pwT1D. We also aimed to study the role of sex and the fear of hypoglycaemia in the association of glycaemic parameters with ED and IOM.
Methods
Population
We performed a cross-sectional analysis of participants from the French-Speaking Diabetes Society—Type 1 Diabetes Cohort (SFDT1). The SFDT1 study aims to evaluate the cardiovascular risk in pwT1D. The SFDT1 has been previously described.24 To summarise, SFDT1 is an ongoing study that includes pwT1D in France who attend hospitals or private ambulatory diabetes centres. The SFDT1 study comprises data from self-reported questionnaires, face-to-face interviews, a physical examination, clinical assessments, blood samples and CGM measures. The current analysis was performed on data collected during the baseline visit in participants enrolled between December 2020 and March 2024.
Patient and public involvement
Patients were involved in the co-design of the SFDT1 cohort study (Collaboration with the Diabète Lab, FFD, France, 2020) but did not take part in the conduct, analysis, reporting or dissemination of this research.
Inclusion criteria
We included pwT1D aged 18 and above with available data on the questionnaire about ED and IOM and good-quality CGM data, defined as at least 70% of the data that should have been captured and the sum of time in range (TIR, % of time spent in the range of 70 and 180 mg/dL), time below range (TBR, % of time spent below 70 mg/dL) and TAR (% of time spent over 180 mg/dL) was between 98% and 102%.
Eating disorder and insulin omission assessment
To evaluate ED, we used the French version of the SCOFF questionnaire, which consists of five yes/no questions.25 26 This questionnaire showed a sensitivity of 94.6% and a specificity of 94.8% for ED diagnosis.25 ED was defined as ≥2 yes responses on the SCOFF questionnaire.26 The question about IOM was added to the five SCOFF questions and was the following: ‘Do you ever take less insulin than you should?’. If the answer to this question was ‘yes’, the participant was categorised as reporting IOM. The population completed the SCOFF and the IOM single-question at the baseline visit. We classified the participants’ ED and IOM status into four groups: ED & no IOM (ED without IOM), no ED & IOM (IOM without ED), ED & IOM and no ED & no IOM.
Glycaemic parameters
CGM metrics (TIR, TBR and TAR) calculated from the 2 week data were collected in the last 2 weeks before the baseline visit according to international recommendations.27 28 CGM data are collected on platforms used by clinicians in everyday practice. When patients came for inclusion visits, the two preceding weeks’ data were collected by a research assistant. Other CGM metrics were the glucose monitoring indicator (GMI), the Glycaemia Risk Index (GRI, 3 × % time spent below 54 mg/dL+2.4 × % TBR + (1.6 × % time spent over 250 mg/dL+0.8 × % TAR)29 and the coefficient of variation (CV, SD of glucose/mean glucose×100).30 Finally, the patient’s last HbA1c (mmol/mol) was reported.
Confounders
We included possible confounders not assumed to be colliders (influenced by the explanatory variable and the outcome) or mediators (in the causal pathway between the explanatory variable and the outcome). Therefore, we included the following individual characteristics: age, sex, diabetes duration, social vulnerability (EPICES score, with cut-off≥30.17),31 smoking (defined as current smoker vs former or never smoker), excessive alcohol drinking (any event of 5 or more drinks a day for at least 12 times a year) and mode of insulin administration (multiple injections, insulin pumps with open loop systems or AID).
Participants completed baseline questionnaires assessing their fear of hypoglycaemia (a short version of the Hypoglycaemia Fear Survey (HFS II),32 including the behaviour (HFS-B) and worry (HFS-W) scales. Both scales have 4 Likert items, each ranging from 0 to 4, and total scores range from 0 to 16. We used the median value as the threshold to define the fear of hypoglycaemia. The short version of the HFS II was answered at baseline.
We also measured the Audit of Diabetes Dependent Quality of Life-19 items (ADDQol19). The ADDQol19 was administered 1 month after baseline. An ADDQol19 in the lowest quartile indicates a low quality of life.33
Other variables for the description of the population
We described education (high school diploma or higher) and marital status (married or cohabiting vs neither). We also described treatment (multiple daily injections or insulin pumps alone or opened systems and AID, total insulin dose/kg), Body Mass Index (BMI) calculated as kg/m2 with measured height and weight, obesity (BMI≥30 kg/m2), binge drinking (≥4 drinks for women, or≥5 drinks for men during an occasion), percentage of time in vigorous physical activity using the short version of the IPAQ questionnaire,34 mean measured systolic and diastolic pressure in mm Hg, mean heart rate (bpm), mean total cholesterol and triglycerides.
We described retinopathy diagnosis (%) based on self-reporting or clinical data. Nephropathy (%) was defined as an albumin/creatinine ratio >30 mg/g or estimated glomerular filtration <60 mL/min per 1.73 m2. Neuropathy (%) was described as a Michigan Neuropathy Screening Instrument (MNSI) score >2.0318 or any evidence of autonomic neuropathy. Cardiovascular disease (%) was defined as any past event of acute coronary syndrome, angina, stroke, bypass cardiovascular surgery, coronary angioplasty, supra-aortic trunk surgery, arterial surgery of the lower limbs, non-traumatic amputation of the lower limbs, history of rhythm disturbances or hospitalisation for heart failure.
We reported episodes of severe hypoglycaemia in the preceding year (defined as hypoglycaemia events requiring the assistance of a third party) and episodes of ketoacidosis. Hypoglycaemia unawareness, measured at baseline, was described as a Gold score≥4.35 Diabetes distress was assessed 3 months after baseline using the Problem Areas In Diabetes (PAID) Scale,36 which is defined as a PAID score≥40.36 37 Treatment burden was assessed 2 months after baseline using the Treatment Burden Questionnaire, which comprises 15 questions.38 We defined burden with a cut-off of 59 (39% of the maximum value).
Missing data
We detected missing data and applied multiple imputations using the chained equation approach with the R-package ‘Mice’.39 We decided how many imputations to perform based on the maximum percentage of missing data.40 We modelled missing data and selected the best predictors for each variable in our data set using the ‘Quickpred’ function in the R package Mice. We pooled the estimates and calculated the CIs according to Rubin’s rules.41
Statistical analysis
We described the variables as mean (SD), median (IQR) or numbers (%) if they were continuous and normally distributed, continuous and non-normally distributed or categorical, respectively.
We compared the four groups’ characteristics using non-parametric statistics (Kruskal-Wallis test for numerical variables and χ2 test for categorical variables).
We performed multinomial logistic regression analysis (‘nnet’ R package) to examine the association between the ED and IOM status (outcome) with glycaemic variables (explanatory variables). No ED/no IOM was the reference outcome for all comparisons. The explanatory variables were standardised by subtracting the mean and dividing by the SD. We computed three types of models with increasing integration of adjusting factors: model 0 was to test the crude association between each explanatory variable and the outcome. Model 1 was further adjusted for age, sex and diabetes duration, and model 2 was model 1 further adjusted for social vulnerability, smoking, alcohol status and insulin treatment. With model 2, we tested interactions with sex and fear of hypoglycaemia.
Results
From 2294 adult participants in the SFDT1 study with the SCOFF questionnaire available, we included 1113 persons (median (IQR) age of 38 (29, 50) and 49% of women) who fulfilled the inclusion criteria (online supplemental figure 1). We identified missing data. The highest missing data rate was for albuminuria (35.5%); therefore, we imputed 40 data sets using 20 iterations (online supplemental table 1).
Table 1 shows the characteristics of the sample population. The prevalence of any ED or IOM was 32% (n=355): 11% (n=124) reported ED & no IOM, 16% (n=177) no ED & IOM and 5% (n=54) ED & IOM. Compared with participants with no ED & no IOM (68%, n=758), those in the three other categories were younger, had shorter diabetes duration, were more likely to be women, were more socially vulnerable and were less likely to cohabit or be married (table 1, section A). Participants with ED (with or without IOM) had a lower education level. By sex, 39.4% of women had ED or IOM (ED & no IOM: 15.1%, IOM & no ED: 18.1%, ED & IOM: 6.2%) versus 24.6% of men (ED & no IOM: 7.3%, IOM & no ED: 13.8%, ED & IOM: 3.5%).
Table 1. Characteristics of the sample according to the eating disorder and insulin omission categories.
| Variables | All (n=1113) | No ED & no IOM (n=758) | ED & no IOM (n=124) | No ED & IOM (n=177) | ED & IOM (n=54) |
P value |
|---|---|---|---|---|---|---|
| Individual characteristics | ||||||
| Age, years (median (IQR)) | 38 (29, 50) | 40 (30, 52) | 36 (29, 46) | 35 (28, 47) | 32 (25, 42) | <0.001 |
| Women, N (%) | 548 (49) | 332 (44) | 83 (67) | 99 (56) | 34 (63) | <0.001 |
| Diabetes duration, years (median (IQR)) | 20 (12, 32) | 22 (13, 33) | 18 (11, 29) | 18 (10, 29) | 17 (6, 23) | <0.001 |
| BMI, kg/m² (mean (SD)) | 25.9 (4.7) | 25.8 (4.7) | 27.5 (4.8) | 25.4 (4.8) | 26.3 (4.7) | 0.368 |
| Social vulnerability, N (%) | 240 (22) | 130 (17) | 40 (32) | 49 (28) | 21 (39) | <0.001 |
| Higher education, N (%) | 752 (68) | 516 (68) | 81 (65) | 120 (68) | 35 (65)* | 0.001 |
| Not married, no cohabiting, N (%) | 371 (33) | 229 (30) | 46 (37) | 72 (41) | 24 (44)* | <0.001 |
| Diabetes treatment | ||||||
| Multiple daily injections, N (%) | 412 (37) | 278 (37) | 42 (34) | 70 (40) | 22 (41) | <0.001 |
| Insulin pumps alone or with open loop systems, N (%) | 511 (46) | 338 (45) | 54 (44) | 90 (51) | 29 (54) | <0.001 |
| Hybrid closed-loop systems, N (%) | 190 (17) | 142 (19) | 28 (23) | 17 (10) | 3 (6) | <0.001 |
| Total insulin dose, U/kg/day (mean (SD)) | 0.58 (0.26) | 0.58 (0.26) | 0.58 (0.26) | 0.58 (0.26) | 0.60 (0.27) | 0.836 |
| Glycaemic control | ||||||
| Time in range, % of the time between 70 and 180 mg/dL (mean (SD)) | 59 (16) | 62 (16) | 62 (15) | 48 (16) | 49 (17) | <0.001 |
| Time above range, % of the time >180 mg/dL(mean (SD)) | 36 (17) | 33 (16) | 34 (15) | 48 (18) | 48 (18) | <0.001 |
| Time below range, % of the time <70 mg/dL (median (IQR)) | 3 (1, 6) | 3 (1, 6) | 3 (1, 6) | 2 (1, 5) | 0 (2, 3) | <0.001 |
| Glycaemia Risk Index (GRI), pp (mean (SD)) | 57 (37, 83) | 52 (35, 75) | 50 (35, 69) | 83 (59, 100) | 79 (62, 100) | <0.001 |
| Coefficient variation, % (mean (SD)) | 38 (8) | 38 (9) | 38 (9) | 39 (8) | 39 (9) | 0.01 |
| Glucose management indicator, % (mean (SD)) | 7.3 (0.8) | 7.2 (0.8) | 7.2 (0.8) | 7.9 (0.8) | 7.9 (0.8) | <0.001 |
| HbA1c, mmol/mol, (mean (SD)) | 58 (12) | 56 (11) | 56 (11) | 64 (11) | 67 (11) | <0.001 |
The p values were calculated using the Kruskal-Wallis test to compare any difference in the four categories with continuous outcomes and the χ2 test to compare categorical variables.
BMI, body mass index; ED, eating disorder; IOM, insulin omission.
Concerning treatments, the IOM (with or without ED) groups were more on multiple daily insulin injections. The ED & no IOM group used AID more frequently (table 1, section B).
BMI did not differ between groups and was 25.9 (SD 4.7) kg/m² in the total sample. However, the prevalence of obesity was significantly higher among the ED groups (with or without IOM). The categories differed in the chronic complication rate. Retinopathy and nephropathy were more prevalent in IOM without ED, neuropathy was more frequent in ED without IOM, and cardiovascular diseases were less frequent in IOM without ED (online supplemental table 2, sections A and B).
The CGM data differed significantly among the categories. The no ED & no IOM and the ED & no IOM groups had the highest TIR and TBR values and the lowest TAR, GRI, GMI and HbA1c levels compared with the no ED & IOM and the ED & IOM groups (table 1, section C).
The no ED & no IOM participants had the lowest frequency of fear of hypoglycaemia compared with the other groups (online supplemental table 2, section D). The ED & no IOM and the ED & IOM groups observed the highest levels of hypoglycaemia unawareness. Concerning treatments, the IOM (with or without ED) groups were more on multiple daily insulin injections. The ED & no IOM group used AID more frequently (table 1, section B).
Quality of life, diabetes distress and burden differed between groups, being higher in IOM & no ED and very much higher in ED with or without IOM. Quality of life was impaired in either ED or IOM and ED & IOM categories (online supplemental table 2, sections E, F, G).
Figure 1 describes the observed phenotypes of each group. The no ED & no IOM group is characterised by older age, higher diabetes duration and high TIR and TBR. They have lower levels of social vulnerability and HbA1c. The group ED & no IOM have high TIR and TBR. They are younger and have higher social vulnerability. The no ED & IOM group has very high TAR and CV, high HbA1c, social vulnerability and low TBR and TIR. The ED & IOM group is characterised by a very high HbA1c, TAR, social vulnerability, a high CV and younger age and a low TIR.
Figure 1. Phenotypes of people with T1D according to eating disorders (ED) and insulin omission (IOM). The ED–IOM groups are no ED & no IOM (in green), ED & no IOM (in blue), no ED & IOM (in orange) and ED & IOM (in red). Each variable’s values (numerical labels in the plot) are the mean of each variable in its ED–IOM group and are plotted as maximum to minimum mean. CV, coefficient of variation; pp, per point; TAR, time above range; TBR, time below range; TIR, time in range; y, years.
Figure 2 describes the association of glycaemic variables as explanatory variables of ED and IOM groups with the no ED & no IOM group as the reference level, where data were adjusted by age, diabetes duration, sex, social vulnerability, smoking, alcohol status and insulin treatment (fully adjusted model 2). Table 2 describes the non-adjusted and adjusted models. No association was found between glycaemic control variables and the ED & no IOM group. Most glycaemic variables were significantly associated with no ED & IOM and ED & IOM, showing a negative association with TIR and TBR and a positive association with TAR, GMI, GRI and HbA1c. The strength of the association decreased from model 0 to model 1 and from model 1 to model 2. In most cases, the difference between model 1 and model 2 was greater, suggesting that the main confounders were in model 2.
Figure 2. Association of glycaemic variables with ED and IOM groups (n=1113). The plot represents multinomial logistic regressions of each glycaemic parameter as an explanatory variable of the ED and IOM groups. The ED and IOM status are ED & no IOM (in blue), no ED & IOM (in orange) and ED & IOM (in red). The reference level was the group no ED & no IOM, represented by the vertical green line. The variables were standardised. The models were adjusted for age, diabetes duration, sex, social vulnerability, smoking, alcohol status and insulin treatment. CV, coefficient of variation; ED, eating disorder; GMI, glucose monitoring indicator; GRI, Glycaemia Risk Index; IOM, insulin omission; TAR, time above range; TBR, time below range; TIR, time in range.
Table 2. Association of ED and IOM with glycaemic variables.
| Main explanatory variable | Models | ED & no IOM (n=124) | IOM & no ED (n=177) | ED & IOM (n=54) |
|---|---|---|---|---|
| TIR (%) | Model 0 | 0.9 (0.8, 1.2) | 0.4 (0.4, 0.5) | 0.4 (0.3, 0.6) |
| Model 1 | 1.0 (0.8, 1.2) | 0.4 (0.4, 0.5) | 0.5 (0.3, 0.6) | |
| Model 2 | 1.0 (0.8, 1.3) | 0.4 (0.4, 0.5) | 0.5 (0.4, 0.7) | |
| TBR (%) | Model 0 | 0.9 (0.8, 1.1) | 0.8 (0.6, 0.9) | 0.5 (0.3, 0.8) |
| Model 1 | 0.9 (0.8, 1.1) | 0.8 (0.6, 1.0) | 0.5 (0.3, 0.8) | |
| Model 2 | 0.9 (0.8, 1.1) | 0.8 (0.6, 1.0) | 0.5 (0.3, 0.8) | |
| TAR (%) | Model 0 | 1.1 (0.9, 1.3) | 2.4 (2.0, 2.8) | 2.4 (1.8, 3.1) |
| Model 1 | 1.1 (0.9, 1.3) | 2.3 (2.0, 2.8) | 2.3 (2.1, 3.1) | |
| Model 2 | 1.0 (0.8, 1.3) | 2.3 (1.9, 2.7) | 2.2 (1.6, 2.9) | |
| GRI (pp) | Model 0 | 0.9 (0.8, 1.1) | 2.4 (1.9, 2.9) | 2.0 (1.5, 2.8) |
| Model 1 | 0.9 (0.8, 1.1) | 2.3 (1.9, 2.8) | 2.0 (1.4, 2.7) | |
| Model 2 | 0.9 (0.7, 1.1) | 2.3 (1.9, 2.8) | 1.8 (1.3, 2.5) | |
| CV (%) | Model 0 | 1.0 (0.8, 1.2) | 1.3 (1.1, 1.5) | 1.2 (1.8, 1.6) |
| Model 1 | 1.0 (0.8, 1.2) | 1.3 (1.1, 1.5) | 1.1 (1.8, 1.5) | |
| Model 2 | 0.9 (0.8, 1.2) | 1.2 (1.0, 1.5) | 1.1 (1.8, 1.5) | |
| GMI (%) | Model 0 | 1.1 (0.9, 1.4) | 2.3 (2.0, 2.8) | 2.5 (1.9, 3.2) |
| Model 1 | 1.1 (0.8, 1.4) | 2.3 (1.9, 2.8) | 2.4 (1.8, 3.1) | |
| Model 2 | 1.0 (0.8, 1.3) | 2.2 (1.9, 2.7) | 2.2 (1.7, 2.9) | |
| HbA1c (mmol/mol) | Model 0 | 1.1 (0.8, 1.3) | 2.0 (1.7, 2.4) | 2.3 (1.8, 2.9) |
| Model 1 | 1.0 (0.8, 1.3) | 2.0 (1.6, 2.3) | 2.2 (1.7, 2.7) | |
| Model 2 | 0.9 (0.7, 1.2) | 1.9 (1.6, 2.2) | 2.0 (1.5, 2.5) |
Model 0 was to test the crude association between each explanatory variable and the outcome (ED and IOM status), Model 1 was further adjusted for age, sex, and diabetes duration, and Model 2 was Model 1 further adjusted for social vulnerability, smoking, alcohol status, and insulin treatment. The reference level was the group No ED & No IOM
CV, coefficient of variation; ED, eating disorder; GMI, glucose monitoring indicator; GRI, Glycaemia Risk Index; IOM, insulin omission; TAR, time above range; TBR, time below range; TIR, time in range.
Online supplemental figure 2 shows the influence of sex. It describes the association of glycaemic variables with ED and IOM categories, with the no ED & no IOM group as the reference level stratified by sex. The associations of ED & IOM with glycaemic parameters are maintained. The glycaemic control parameters in both sexes were not associated with ED & no IOM.
TIR, TAR, GRI, GMI and HbA1c were associated with no ED & IOM in both sexes, while TBR was only associated with females but not males. TBR, TAR, GMI and HbA1c were associated with ED & IOM in both sexes. The interaction terms explanatory variable-sex were significant for GMI in the group ED & IOM (p value=0.03).
Online supplemental figure 3 shows the association of glycaemic variables and categories stratified by the median of the fear of hypoglycaemia (total score ranged from 0 to 30 pp, the median was 11.1 pp). Compared with the lower fear of hypoglycaemia group (n=558), the group with a higher fear of hypoglycaemia (n=555) showed stronger associations with TIR, TBR, TAR, GRI, GMI and HbA1c. Interaction terms of fear of hypoglycaemia with TIR (p value=0.017), TAR (p value=0.046), GRI (p value=0.046), GMI (p value=0.027) and HbA1c (p value=0.036) before stratification were significant for ED & no IOM group. Fear of hypoglycaemia was directly associated with IOM (OR (95% CI) 1.07 (1.04 to 1.10). However, these differences do not make a clinical difference because even in the pwT1D with the lowest fear of hypoglycaemia, no ED & IOM and ED & IOM were associated with worse glycaemic control.
Discussion
This study showed that ED and IOM in adults with T1D are highly prevalent, as well as in men. Poor CGM parameters and HbA1c levels were observed in pwT1D with IOM per se, regardless of their ED status. Socially vulnerable, younger participants with shorter diabetes duration and suboptimal glycaemic control were more often represented in the ED & IOM group. BMI was the same among all ED–IOM categories, while diabetes distress was much higher in persons with ED with or without IOM. However, participants with ED display a high level of distress and should be screened and specially cared for. Compared with the group with no ED & no IOM, in participants with no ED & IOM and ED & IOM, we observed an inverse association with TIR and TBR and a direct association with TAR, GRI, GMI and HbA1c. The associations between glycaemic control parameters and ED and IOM groups persisted after controlling for key confounders. Further stratification by sex or fear of hypoglycaemia did not materially change the findings.
ED, with or without IOM, was more frequent in women. However, 44% of men declared either an IOM or ED. Among men, 10.8% had ED with or without IOM, which is higher than in the general population.42 When IOM was present without ED, the association with all parameters of glycaemic control remained in both sexes, except for TBR (not associated in men). In persons with ED & IOM, the association with all parameters of glycaemic control remained in both sexes, except for GMI, whose association with ED & IOM is stronger in women.
We observed that the ‘no ED & no IOM’ population was the oldest and had the most prolonged diabetes duration. However, the associations remained significant after adjustment by age and duration (models 1 and 2).
We observed a trend of an association between poor glycaemic control (TIR and HbA1c) and ED & no IOM among participants with a higher fear of hypoglycaemia. Although this trend was non-significant, it can suggest a higher risk of having poorer glycaemic control in people with an ED and fear of hypoglycaemia.
Treatment burden was higher in persons with IOM and more so if there was an ED. We cannot rule out the possibility that treatment burden is a confounder in the association between IOM and glycaemic control, since unfortunately, we only have data on treatment burden for a restricted subsample (482 participants, 43% of the total sample).
Results from the literature on ED and IOM status, and their association with glycaemic control as measured by CGM metrics, are less conclusive. A review of three studies41 suggests that disturbed eating behaviour, as measured by the Diabetes Eating Problem Survey—Revised, is associated with a high TAR and a lower TIR. However, these were small studies (n=59–90) with short-time CGM evaluations (3–7 days, sometimes blinded to patients). One more recent study confirms this relationship.43 One further study evaluated ED, depicted by the validated SCOFF questionnaire, and showed no relationship with CGM metrics in 198 pwT1D treated with subcutaneous insulin infusion with adequate glycaemic control.21
Therefore, we decided to differentiate ED from the IOM. Diabulimia is an important issue, but it combines both ED and IOM in its definition.11 Our results highlight the central role of IOM, regardless of the reason why people omit insulin. Indeed, IOM played a central role after adjusting for other covariates of IOM, such as the type of treatment and stratifying by fear of hypoglycaemia. IOM may be the more proximate behaviour driving the associations with impaired glycaemic control. At the same time, EDs per se did not impact CGM data or HbA1c. Few studies have evaluated ED (without IOM) and CGM data. Albaladejo et al21 found no difference in CGM between patients with positive and negative SCOFF scores. However, the participants had adequate glycaemic control (HbA1c 53 mmol/mol) and were all treated with subcutaneous insulin infusion.
We can discuss the reasons why pwT1D omit insulin injections. The question about IOM did not specify if it was related to weight or body image control. However, IOM is associated with deteriorated glycaemic control, whether or not there is an ED and regardless of the fear of hypoglycaemia. Among the reasons for IOM, the cost of the insulin treatment is not a problem in our study, as the French social insurance system covers it. Forgetfulness has not been assessed. We cannot rule out the possibility that treatment burden is a confounder in the association between IOM and glycaemic control, but it was analysed only in a small subsample.
Strengths and limitations
The main strengths of this study are that this analysis is based on the SFDT1 study, which includes a large and diverse sample of pwT1D, combining an in-depth description of glycaemic control with CGM parameters, an independent evaluation of ED and IOM, and extensive phenotypic data. Our study also provides results in many men with T1D, which is essential because most literature concerns women. We report correlations with key patient-reported outcomes, such as diabetes distress, treatment burden and quality of life.
Some limitations of this study include reliance on the SCOFF questionnaire for ED screening. Moreover, it is a self-reported tool, and as such, does not validate the diagnosis, which should be performed by a medical interview. As with any self-reported measure, it is likely to introduce recall bias. Therefore, some authors consider that a positive SCOFF refers to likely ED21 rather than ED diagnosis. The recall bias is the same for IOM as it was self-reported. In people with normal or high BMI (here, the prevalence of a BMI higher than 30 kg/m² was 18%, very close to that in the general French population, 17%), the validity of SCOFF is questionable. Indeed, some items of the SCOFF may be misunderstood by a person living with obesity, for example. ‘Do you believe yourself to be fat when others say you are too thin?’.44 Therefore, the prevalence of ED in this study could be underestimated. However, the 16% total prevalence of ED (11% ED without IOM plus 5% with IOM) is within the 10% to 40% prevalence previously described in the literature.11 Particularly in a French online survey of students aged 18 to 25 (n=3508), assessed 14 months after the beginning of the COVID-19 pandemic, the ED prevalence was 51.6% in women and 31.9% in men.45 Another limitation was that we did not include adolescents, as there were very few in the SFDT1 study at the time of the analysis. Finally, we could not include some confounders, such as the burden of treatment, because they were available in a very restricted subsample of the study participants.
Perspectives
The British parliamentary report was written by Rt. Hon Sir George Howarth MP and Rt. Hon Theresa May MP4 in January 2024 and suggested that more attention should be paid to pwT1D and ED. Based on our new findings, we recommend systematically searching for IOM in persons with altered CGM parameters or an elevated HbA1c. We also recommend that ED be systematically screened for, despite the shame and culprit patients may present, even though it does not seem to affect glycaemic control, but because we show that diabetes distress is high and, as such, increases the risk of future complications and altered quality of life.46
Diabetes care should be offered using a mix of competencies from the somatic and psychological fields. These evaluations should be performed on any pwT1D, including males and persons with normal or high BMI, to fight the representation of ED as affecting a young, lean woman. Finally, another perspective to consider is the likely role of the treatment in managing glucose in people with EDI or IOM, particularly the AID devices.
Conclusion
We analysed data from a large sample of pwT1D and found that ED, IOM or a combination of both is frequent. We discovered that EDs are also frequent in males with T1D. IOM, but not ED alone, is associated with impaired glycaemic control. As our analysis was cross-sectional, we cannot infer causality and cannot know whether IOM was a result of glycaemic control or the inverse (reverse causality). These results highlight the need for ED and IOM screening to prevent diabetes complications and improve the quality of life in pwT1D.
Supplementary material
Acknowledgements
We would like to express our deepest gratitude to the participants of the SFDT1 study for their valuable contribution. Their participation is instrumental in the progress of type 1 diabetes-related research. Additionally, we thank the SFDT1 study group for their diligent efforts and dedication to the study. Their collective expertise and commitment have played a critical role in ensuring a high-quality standard for the SFDT1 data (https://cohorte-sfdt1.jimdosite.com/). The authors thank the regional clinical research coordinators and data managers for their role in ensuring high-quality data collection.
Footnotes
Funding: SFDT1 is supported by the cohort sponsor, the Foundation Francophone pour la Recherche sur le Diabète (FFRD), with institutional support from the Société Francophone du Diabète (SFD) and by research grants from Breakthrough T1D and iCare4CVD, as well as contributions from the following partners: Aide aux Jeunes Diabétiques (AJD), Fédération Française des Diabétiques (FFD), Lilly, Abbott, Air Liquide Healthcare, Novo Nordisk, Sanofi, Insulet, Medtronic, Dexcom, Ypsomed and Lifescan. The funders had no role in the study design, data collection, analysis or interpretation; manuscript preparation; or the decision to publish.
Prepublication history and additional supplemental material for this paper are available online. To view these files, please visit the journal online (https://doi.org/10.1136/bmjopen-2025-104542).
Provenance and peer review: Not commissioned; externally peer reviewed.
Patient consent for publication: Not applicable.
Ethics approval: This study involves human participants. The study was approved by the ethics committee CPP Ouest V-RENNES (No ID-RCB:2019A01681-56) in December 2019. Participants gave informed consent to participate in the study before taking part.
Data availability free text: Data used for this analysis are available for academic researchers on request submitted to the scientific committee of SFDT1: cohorte.sfdt1@gmail.com. R scripts created for this analysis are available on request to the corresponding author (guy.fagherazzi@lih.lu)
Map disclaimer: The depiction of boundaries on this map does not imply the expression of any opinion whatsoever on the part of BMJ (or any member of its group) concerning the legal status of any country, territory, jurisdiction or area or of its authorities. This map is provided without any warranty of any kind, either express or implied.
Patient and public involvement: Patients and/or the public were involved in the design, or conduct, or reporting or dissemination plans of this research. Refer to the Methods section for further details.
Data availability statement
Data are available upon reasonable request.
References
- 1.Gregory GA, Robinson TIG, Linklater SE, et al. Global incidence, prevalence and mortality of type 1 diabetes in 2021 with projection to 2040: a modelling study. Lancet Diabetes Endocrinol. 2022;10:741–60. doi: 10.1016/S2213-8587(22)00218-2. [DOI] [PubMed] [Google Scholar]
- 2.Lachin JM, Nathan DM, DCCT/EDIC Research Group Understanding Metabolic Memory: The Prolonged Influence of Glycemia During the Diabetes Control and Complications Trial (DCCT) on Future Risks of Complications During the Study of the Epidemiology of Diabetes Interventions and Complications (EDIC) Diabetes Care. 2021;44:2216–24.:dc203097. doi: 10.2337/dc20-3097. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Boughton CK, Hovorka R. The role of automated insulin delivery technology in diabetes. Diabetologia. 2024;67:2034–44. doi: 10.1007/s00125-024-06165-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.The Diabetes Times; 2022. [29-Oct-2024]. New parliamentary inquiry into type 1 diabetes and eating disorders launched.https://diabetestimes.co.uk/new-parliamentary-inquiry-into-type-1-diabetes-and-eating-disorders-launched/ Available. Accessed. [Google Scholar]
- 5.American Psychiatric Association. Diagnostic and Statistical Manual of Mental Disorders (DSM-5. American Psychiatric Publishing; 2021. [Google Scholar]
- 6.Galmiche M, Déchelotte P, Lambert G, et al. Prevalence of eating disorders over the 2000-2018 period: a systematic literature review. Am J Clin Nutr. 2019;109:1402–13. doi: 10.1093/ajcn/nqy342. [DOI] [PubMed] [Google Scholar]
- 7.Mannucci E, Rotella F, Ricca V, et al. Eating disorders in patients with type 1 diabetes: a meta-analysis. J Endocrinol Invest. 2005;28:417–9. doi: 10.1007/BF03347221. [DOI] [PubMed] [Google Scholar]
- 8.Kınık MF, Gönüllü FV, Vatansever Z, et al. Diabulimia, a Type I diabetes mellitus-specific eating disorder. Turk Pediatri Ars. 2017;52:46–9. doi: 10.5152/TurkPediatriArs.2017.2366. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Quick VM, Byrd-Bredbenner C. Disturbed eating behaviours and associated psychographic characteristics of college students. J Hum Nutr Diet. 2013;26 Suppl 1:53–63. doi: 10.1111/jhn.12060. [DOI] [PubMed] [Google Scholar]
- 10.Goddard G, Oxlad M. Caring for individuals with Type 1 Diabetes Mellitus who restrict and omit insulin for weight control: Evidence-based guidance for healthcare professionals. Diabetes Res Clin Pract. 2022;185:109783. doi: 10.1016/j.diabres.2022.109783. [DOI] [PubMed] [Google Scholar]
- 11.Dean YE, Motawea KR, Aslam M, et al. Association Between Type 1 Diabetes Mellitus and Eating Disorders: A Systematic Review and Meta-Analysis. Endocrinol Diabetes Metab . 2024;7:e473. doi: 10.1002/edm2.473. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Nip ASY, Reboussin BA, Dabelea D, et al. Disordered Eating Behaviors in Youth and Young Adults With Type 1 or Type 2 Diabetes Receiving Insulin Therapy: The SEARCH for Diabetes in Youth Study. Diabetes Care. 2019;42:859–66. doi: 10.2337/dc18-2420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Troncone A, Affuso G, Cascella C, et al. Prevalence of disordered eating behaviors in adolescents with type 1 diabetes: Results of multicenter Italian nationwide study. Int J Eat Disord. 2022;55:1108–19. doi: 10.1002/eat.23764. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Wisting L, Frøisland DH, Skrivarhaug T, et al. Disturbed Eating Behavior and Omission of Insulin in Adolescents Receiving Intensified Insulin Treatment. Diabetes Care. 2013;36:3382–7. doi: 10.2337/dc13-0431. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Gibbings NK, Kurdyak PA, Colton PA, et al. Diabetic Ketoacidosis and Mortality in People With Type 1 Diabetes and Eating Disorders. Diabetes Care. 2021;44:1783–7. doi: 10.2337/dc21-0517. [DOI] [PubMed] [Google Scholar]
- 16.Goebel-Fabbri AE, Fikkan J, Franko DL, et al. Insulin restriction and associated morbidity and mortality in women with type 1 diabetes. Diabetes Care. 2008;31:415–9. doi: 10.2337/dc07-2026. [DOI] [PubMed] [Google Scholar]
- 17.Scheuing N, Bartus B, Berger G, et al. Clinical characteristics and outcome of 467 patients with a clinically recognized eating disorder identified among 52,215 patients with type 1 diabetes: a multicenter German/Austrian study. Diabetes Care. 2014;37:1581–9. doi: 10.2337/dc13-2156. [DOI] [PubMed] [Google Scholar]
- 18.Eisenberg Colman MH, Quick VM, Lipsky LM, et al. Disordered Eating Behaviors Are Not Increased by an Intervention to Improve Diet Quality but Are Associated With Poorer Glycemic Control Among Youth With Type 1 Diabetes. Diabetes Care. 2018;41:869–75. doi: 10.2337/dc17-0090. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Merwin RM, Dmitrieva NO, Honeycutt LK, et al. Momentary Predictors of Insulin Restriction Among Adults With Type 1 Diabetes and Eating Disorder Symptomatology. Diabetes Care. 2015;38:2025–32. doi: 10.2337/dc15-0753. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Merwin RM, Moskovich AA, Honeycutt LK, et al. Time of Day When Type 1 Diabetes Patients With Eating Disorder Symptoms Most Commonly Restrict Insulin. Psychosom Med. 2018;80:222–9. doi: 10.1097/PSY.0000000000000550. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Albaladejo L, Périnet-Marquet P, Buis C, et al. High prevalence with no gender difference of likely eating disorders in type 1 mellitus diabetes on insulin pump. Diabetes Res Clin Pract. 2023;199:110630. doi: 10.1016/j.diabres.2023.110630. [DOI] [PubMed] [Google Scholar]
- 22.Annuzzi G, Triggiani R, De Angelis R, et al. Delayed prandial insulin boluses are an important determinant of blood glucose control and relate to fear of hypoglycemia in people with type 1 diabetes on advanced technologies. J Diabetes Complications. 2024;38:108689. doi: 10.1016/j.jdiacomp.2024.108689. [DOI] [PubMed] [Google Scholar]
- 23.Polonsky WH, Anderson BJ, Lohrer PA, et al. Insulin omission in women with IDDM. Diabetes Care. 1994;17:1178–85. doi: 10.2337/diacare.17.10.1178. [DOI] [PubMed] [Google Scholar]
- 24.Riveline JP, Vergés B, Detournay B, et al. Design of a prospective, longitudinal cohort of people living with type 1 diabetes exploring factors associated with the residual cardiovascular risk and other diabetes-related complications: The SFDT1 study. Diabetes Metab. 2022;48:101306. doi: 10.1016/j.diabet.2021.101306. [DOI] [PubMed] [Google Scholar]
- 25.Garcia FD, Grigioni S, Chelali S, et al. Validation of the French version of SCOFF questionnaire for screening of eating disorders among adults. World J Biol Psychiatry. 2010;11:888–93. doi: 10.3109/15622975.2010.483251. [DOI] [PubMed] [Google Scholar]
- 26.Morgan JF, Reid F, Lacey JH. The SCOFF questionnaire: assessment of a new screening tool for eating disorders. BMJ. 1999;319:1467–8. doi: 10.1136/bmj.319.7223.1467. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Bergenstal RM, Ahmann AJ, Bailey T, et al. Recommendations for standardizing glucose reporting and analysis to optimize clinical decision making in diabetes: the ambulatory glucose profile. J Diabetes Sci Technol. 2013;7:562–78. doi: 10.1177/193229681300700234. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Battelino T, Alexander CM, Amiel SA, et al. Continuous glucose monitoring and metrics for clinical trials: an international consensus statement. Lancet Diabetes Endocrinol. 2023;11:42–57. doi: 10.1016/S2213-8587(22)00319-9. [DOI] [PubMed] [Google Scholar]
- 29.Klonoff DC, Wang J, Rodbard D, et al. A Glycemia Risk Index (GRI) of Hypoglycemia and Hyperglycemia for Continuous Glucose Monitoring Validated by Clinician Ratings. J Diabetes Sci Technol. 2023;17:1226–42. doi: 10.1177/19322968221085273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Monnier L, Colette C, Wojtusciszyn A, et al. Toward Defining the Threshold Between Low and High Glucose Variability in Diabetes. Diabetes Care. 2017;40:832–8. doi: 10.2337/dc16-1769. [DOI] [PubMed] [Google Scholar]
- 31.Labbe E, Blanquet M, Gerbaud L, et al. A new reliable index to measure individual deprivation: the EPICES score. Eur J Public Health. 2015;25:604–9. doi: 10.1093/eurpub/cku231. [DOI] [PubMed] [Google Scholar]
- 32.Gonder-Frederick LA, Schmidt KM, Vajda KA, et al. Psychometric properties of the hypoglycemia fear survey-ii for adults with type 1 diabetes. Diabetes Care. 2011;34:801–6. doi: 10.2337/dc10-1343. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Bradley C, Todd C, Gorton T, et al. The development of an individualized questionnaire measure of perceived impact of diabetes on quality of life: the ADDQoL. Qual Life Res. 1999;8:79–91. doi: 10.1023/a:1026485130100. [DOI] [PubMed] [Google Scholar]
- 34.Bassett DR., Jr International physical activity questionnaire: 12-country reliability and validity. Med Sci Sports Exerc. 2003;35:1396. doi: 10.1249/01.MSS.0000078923.96621.1D. [DOI] [PubMed] [Google Scholar]
- 35.Gold AE, MacLeod KM, Frier BM. Frequency of severe hypoglycemia in patients with type I diabetes with impaired awareness of hypoglycemia. Diabetes Care. 1994;17:697–703. doi: 10.2337/diacare.17.7.697. [DOI] [PubMed] [Google Scholar]
- 36.Polonsky WH, Anderson BJ, Lohrer PA, et al. Assessment of diabetes-related distress. Diabetes Care. 1995;18:754–60. doi: 10.2337/diacare.18.6.754. [DOI] [PubMed] [Google Scholar]
- 37.de Wit M, Pouwer F, Snoek FJ. How to identify clinically significant diabetes distress using the Problem Areas in Diabetes (PAID) scale in adults with diabetes treated in primary or secondary care? Evidence for new cut points based on latent class analyses. BMJ Open. 2022;12:e056304. doi: 10.1136/bmjopen-2021-056304. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Tran V-T, Montori VM, Eton DT, et al. Development and description of measurement properties of an instrument to assess treatment burden among patients with multiple chronic conditions. BMC Med. 2012;10:68. doi: 10.1186/1741-7015-10-68. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Buuren S, mice G-O. Multivariate Imputation by Chained Equations inR. J Stat Softw. 2011 doi: 10.18637/jss.v045.i03. [DOI] [Google Scholar]
- 40.White IR, Royston P, Wood AM. Multiple imputation using chained equations: Issues and guidance for practice. Stat Med. 2011;30:377–99. doi: 10.1002/sim.4067. [DOI] [PubMed] [Google Scholar]
- 41.Rubin DB. Multiple Imputation for Nonresponse in Surveys. John Wiley & Sons; 2004. [Google Scholar]
- 42.Gorrell S, Murray SB. Eating Disorders in Males. Child Adolesc Psychiatr Clin N Am. 2019;28:641–51. doi: 10.1016/j.chc.2019.05.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Propper-Lewinsohn T, Shalitin S, Gillon-Keren M, et al. Glycemic Variability and Disordered Eating Among Adolescents and Young Adults with Type 1 Diabetes: The Role of Disinhibited Eating. Diabetes Technol Ther . 2025;27:113–20. doi: 10.1089/dia.2024.0267. [DOI] [PubMed] [Google Scholar]
- 44.Coop A, Clark A, Morgan J, et al. The use and misuse of the SCOFF screening measure over two decades: a systematic literature review. Eat Weight Disord. 2024;29:29. doi: 10.1007/s40519-024-01656-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Tavolacci M-P, Ladner J, Dechelotte P. COVID-19 Pandemic and Eating Disorders among University Students. Nutrients. 2021;13:4294. doi: 10.3390/nu13124294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Poos S, Faerovitch M, Pinto C, et al. The role of diabetes distress in Diabulimia. J Eat Disord. 2023;11:213. doi: 10.1186/s40337-023-00924-7. [DOI] [PMC free article] [PubMed] [Google Scholar]