Reliability and validity of a parent‐reported screening tool for disordered eating in children and young people with type 1 diabetes

There is a high prevalence and complex overlap between type 1 diabetes (T1D) and disordered eating. However, screening for disordered eating in children and young people (CYP) with T1D is not routinely conducted, with reluctance reported by both professionals and parents. This study aimed to validate a parent‐reported version of a validated disordered eating screening tool for CYP with T1D (the Diabetes Eating Problems Survey‐Revised; DEPS‐R).


| INTRODUCTION
A recent position statement highlighted multiple gaps in the evidence base relating to the mental health of people with diabetes, one of which is to understand how and why people with diabetes develop eating disorders (ED) and how best to provide care for these individuals. 1The comorbidity between type 1 diabetes (T1D) and disordered eating is often observed during adolescence, with research demonstrating an increased prevalence of both clinical EDs and disordered eating in children and young people (CYP) with T1D than those without. 2Although the rates of disordered eating and ED in CYP with T1D have varied between studies, an incidence range of between 3.8% and 27.5% has been identified, which increases to 38%-40% if insulin omission is included in these criteria. 3t is important to establish detailed definitions and an understanding of disordered eating in people with T1D to enable identification and early intervention. 1Effective and appropriate screening for disordered eating can be challenging 4 ; some parents report concern that asking about the presence of disordered eating may provide information on how to engage in disordered eating behaviours of which CYP might not be aware (e.g.omitting insulin). 5Furthermore, within clinical settings, screening for ED is rarely carried out by diabetes healthcare professionals, with a lack of skills, confidence and time to raise the issue of disordered eating with their patients cited as explanatory reasons. 6espite this lack of routine screening, eating disorder tools have been validated for self-report in CYP with T1D, 7 whereas no parent-reported screening or outcome measures have been validated for use in this population.Parent-report measures may be preferrable as CYP with ED have been found to demonstrate greater levels of denial and lower levels of desire to receive help. 8 Parent-reports can also be beneficial for early detection of symptoms, as parents are often the first to notice signs of ED and engage with appropriate treatment sooner. 9urthermore, having parental reports inherently increases parental involvement.In turn, this can increase parental knowledge about disordered eating and self-efficacy, which benefits family engagement in the treatment of ED. 10 Taken together, a parent-reported measure may be a suitable alternative to identify maladaptive eating behaviours in CYP, who benefit from early detection and input in their recovery. 11One validated measure that assesses disordered eating within CYP with T1D is the Diabetes Eating Problem Survey Revised. 7It has been found to have excellent internal consistency with an adolescent population; however, a parent-reported version of the DEPS-R has not yet been validated, which was the aim of this study.To demonstrate convergent validity, it was hypothesised that the parent-and CYPreported DEPS-R would be significantly and positively correlated.To demonstrate construct validity, it was hypothesised that parent-reported DEPS-R would significantly correlate with relevant child eating behaviours, parent-reported diabetes-related emotional distress, parent-reported well-being, parent-reported history of ED and parent-reported concerns of disordered eating for their CYP, in addition to child BMI and HbA1c.

| Design and procedure
This study sought to examine the reliability and validity of the parent-reported DEPS-R.The data utilised baseline data from a larger study 12,13 where validation was a secondary aim.The larger study was a trial of an intervention aimed at parents to prevent disordered eating in their CYP aged 11-14 years with T1D.During baseline assessment, parents completed outcome measures, which included an adapted parent-reported DEPS-R.The CYP of parents were invited to complete the validated DEPS-R 8 ; their participation was optional, and their parents' participation children and young people, disordered eating, parents, reliability, screening, type 1 diabetes, validity What's new?
• Early identification and intervention of disordered eating in children and young people (CYP) with type 1 diabetes (T1D) is important given CYP's increased risk.• We found that a parent-adapted version of the Diabetes Eating Problem Survey-Revised (DEPS-R) indicated significant validity and reliability on a number of dimensions.• Parent-reported DEPS-R may prove to be an acceptable alternative to screening for disordered eating in CYP with T1D.
was not contingent on their child's.The parent-reported DEPS-R utilised the same questions as the original DEPS-R but were re-worded to be about their child as opposed to themselves (e.g.'Losing weight is an important goal to me', became 'Losing weight is an important goal to my child').Two co-applicants of the funded study were mothers of children with T1D and provided insight into developing and reviewing the parent-reported DEPS-R.

| Measures
Four standardised questionnaires were completed by parent participants (the parent-reported DEPS-R, the Child Eating Behaviour Questionnaire (CEBQ 14 ), the Problem Areas in Diabetes-Parent Revised (PAID-PR 15 ) measure and the Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS 16 ) and one questionnaire (the DEPS-R 7 ) was completed by CYP participants.All measures were completed online.

| Data analysis
Statistical analyses were performed using IBS SPSS version 28.Internal consistency of the total scores for the parent-reported DEPS-R was assessed using Cronbachs' α, with a value of >0.7 suggesting good internal consistency. 17Item-total correlations were analysed using reliability analyses to measure the correlation between individual items on the parent-reported DEPS-R and the overall measure.Convergent validity between the parentreported and the CYP-reported versions of the DEPS-R was assessed through intraclass correlation coefficient (ICC) analysis.ICC estimates and their 95% confidence intervals were calculated based on an average measurement, absolute agreement and a two-way random-effects model. 18Construct validity was assessed through correlations between the parent-reported DEPS-R and participant demographics, key clinical information and total scores on validated measures with similar or relevant constructs, including the PAID-PR, the WEMWBS and CEBQ subscales.
To analyse missing data, Missing Completely at Random (MCAR) analysis was conducted with the parent-and CYP-reported versions of the DEPS-R demonstrating significant MCAR results.This is likely because one item from each questionnaire was initially omitted from the online version of the questionnaire, meaning only a third of the sample responded (item 14 on the parent-reported DEPS-R 'I feel that my child's eating is out of control' and item 12 on the CYP-reported DEPS-R, 'Other people tell me to take better care of my diabetes').As per questionnaire developer recommendations, data for these items were imputed using means from other items on the DEPS-R.
The overall sample size for this study exceeded calculations for all intended analyses, as assessed using a priori and post-hoc power calculations for Cronbach's alpha, ICC and construct validity analyses.The number of parent participants exceeded all minimum sample sizes, and the number of CYP participants exceeded or met the minimum sample sizes of all tests, except for Cronbach's α reliability analysis (see Supplementary File 1 for full calculations).

| Internal consistency
The CYP-reported DEPS-R demonstrated good internal consistency with a Cronbach's α of 0.87 for the total score.9 ('I try to keep my blood sugar high so that I will lose weight') and Item 10 ('I eat in a way to get ketones') were removed from the internal consistency analysis as they demonstrated zero variance (i.e., all participants reported 0 on this item).The parent-reported DEPS-R had good internal consistency with a Cronbach's α of 0.89 for the total score.Item 8 ('My child makes themselves vomit') of the parentreported DEPS-R was removed from the internal consistency analysis, as it also demonstrated zero variance.Whilst there was a difference in the internal consistency of the parent-reported DEPS-R for male (Cronbach's α = 0.82) and female CYP (Cronbach's α = 0.91), both were deemed to be high, suggesting equivalence of the construct across gender.

| Convergent validity
ICC analysis was undertaken to assess the inter-rater reliability between parent and CYP participants' responses on the DEPS-R.Based on total scores, a good degree of reliability was found between the parent-reported DEPS-R and CYP-reported DEPS-R.The average ICC was 0.746, with a 95% CI from 0.554 to 0.855 (F(50,50) = 3.930, p < 0.001), indicating moderate to good inter-rater reliability between these measures.Item-to-item ICC analyses were completed to assess whether there was significant parent-and CYP-reported agreement on specific items of the DEPS-R (Table 2).There was moderate to good reliability between parent-and CYP-report on 10 out of the 14 items analysed (ICC ranged from 0.455 to 0.815, all ps <0.05).Unsurprisingly, no significant correlations were observed between CYP-and parent-reported on the three items with zero variance (parent-reported DEPS-R item 8 ('My child makes themselves vomit') and CYP-reported DEPS-R items 9 ('I try to keep my blood sugar high so that I will lose weight') and 10 ('I eat in a way to get ketones').

| Children and young people's eating behaviours
As parent-reported DEPS-R increased, reports of CYP eating behaviours that may be early precursors to eating disorders or obesity significantly increased, as detected by the CEBQ subscales (Cronbach's α for eight subscales ranged from 0.77 to 0.94).The CEBQ subscales that were positively correlated with the parent-reported DEPS-R included food responsiveness (r = 0.484, p < 0.001), emotional overeating (r = 0.622, p < 0.001) and food fussiness (r = 0.273, p = 0.010).

| Diabetes distress, quality of life and well-being
As parent-reported DEPS-R increased, parents' PAID-PR scores (Cronbach's α = 0.93) also significantly increased (r = 0.484, p < 0.001).There was a significant positive relationship between parent-reported quality of life and DEPS-R (r = 0.315, p = 0.003), which suggests that as scores for CYP disordered eating increased, overall quality of life was more greatly impaired.There was also a significant negative relationship between parent-reports of DEPS-R and the mental well-being of parents, as indicated by the WEMWBS (Cronbach's α = 0.93), so as CYP disordered eating increased, parental well-being decreased (r = −0.216,p = 0.042).

| Parent worry
There was a significant correlation found between parentreported DEPS-R and whether parents worried or not about their CYP's eating behaviours (r = −0.437,p < 0.001).This indicates that parents who reported concern about their child's eating behaviours scored higher on the measure of CYP disordered eating, demonstrating known group validity.

| Demographic and clinical characteristics
In relation to demographic variables, parent-reported DEPS-R was significantly related to identifying as a female parent (r = 0.258, p = 0.015) or having a female child (r = 0.220, p = 0.039).No other significant relationships Note: ICC values interpreted as follows: <0.5 poor reliability; 0.5-0.75moderate reliability; 0.75-0.9good reliability; >0.9 excellent reliability.
were found between parents or CYP age.When exploring relationships with clinical there was a positive, moderately sized relationship between parent-reported DEPS-R and CYP BMI (r = 0.537, p < 0.001).No significant effects were observed between parent-reported DEPS-R and HbA1c, the number of unexpected hospital visits, missed diabetes clinic appointments, experience of diabetic ketoacidosis (DKA) or whether parents had experienced or been diagnosed with an eating disorder.

| DISCUSSION
This study aimed to evaluate the reliability and validity of a parent-reported screening tool for disordered eating in CYP with T1D.This measure demonstrated good internal consistency.It established convergent validity through good inter-rater reliability between the CYP-and parent-reported DEPS-R and construct validity via significant relationships with factors examining similar constructs; these include certain food behaviours, diabetes-related distress, overall mental well-being, parent-rated quality of life, parent worry, parent and CYP gender and CYP BMI.There were no significant relationships found between the parent-reported DEPS-R and some expected constructs, such as parents' histories of ED, and elements of CYP diabetes care management, such as HbA1c, the number of unexpected hospital visits, missed diabetes appointments and DKA episodes.The good internal consistency of this measure is in keeping with the findings from previous validation studies. 7,19owever, the reliability of this measure is a novel finding, as previous parent-reported measures of disordered eating in non-T1D populations have failed to demonstrate good reliability and validity between parent and child reports. 20hese differences may be in part explained as parents of CYP with T1D, especially given the age recruited within this study, are more likely to be involved in their child's diabetes care and have more knowledge of their child's eating behaviours than parents of non-T1D/older adolescents.
Given that eating disorders tend to develop during midto-late adolescence, it would be useful to explore whether parent-reported screening becomes less reliable as children age and transition to more independent self-management.
The associations between the parent-reported DEPS-R and certain food behaviours fit with previous literature, as food responsiveness, emotional overeating and food fussiness have been linked to the development of different eating disorders. 15,21These behaviours may prove to be particularly important indicators given the secretive nature and shame surrounding eating disorders, meaning parents may not be fully cognisant of their children's thoughts and feelings as identified within the DEPS-R.The relationships with parents' well-being and quality of life were in line with previous literature. 22Additionally, the more worried parents were about their child's eating behaviours, the more they reported disordered eating behaviours in the parent-reported DEPS-R, which aligns with research demonstrating that parent concern is an effective early detection tool. 23The associations found with gender also follow a pattern seen in the wider literature, indicating that females are more likely to engage with problematic eating behaviours and attitudes than males. 24urthermore, mothers are typically the primary caregivers of children with T1D, 25 so they may be more aware of signs  of disordered eating in their children.Significant associations between BMI and reporting of CYP disordered eating also mirror previous findings 26 ; therefore, this measure may accurately identify potential signs of disordered eating that impact or have been impacted by CYPs' BMI.
The finding that there were no relationships between parent-reports of CYP disordered eating and CYP diabetes care management was surprising, as these factors are considered precursors to developing probable clinical eating disorders. 27However, the responsibility of attending appointments and involvement in T1D management may rest with the parents of young adolescents, as assessed in this sample, which may mean better appointment attendance, better adherence to the T1D regimen and less DKA episodes. 28Additionally, the young CYP in this sample may be less aware of strategies to restrict weight, such as omitting insulin, which could lead to poorer diabetic management and consequences.This lack of awareness may also explain the variance observed on CYP-reported DEPS-R items relating to keeping high blood sugar and ketones for the purpose of weight loss.This is the first study that has validated a parent-reported screening measure for disordered eating in CYP with T1D.This provides a stepped change to our knowledge, providing insight into how parents can provide reliable and accurate information in relation to their child's eating behaviours and what factors mediate this accord between CYP and parents.One limitation is that one item on both the parent-and CYP-reported measures was erroneously missed from data collection for part of the sample.While the means for missing responses were imputed as per developer guidelines, a more robust approach may have been to utilise multiple imputations.Consequently, the findings for this adapted measure and the correlations for these items should be interpreted with caution.Future research should review these findings with all items completed to see if reliability and validity are maintained.Additionally, these findings can only be applied to predominantly white, married parents of 11-14-year-olds, which limits the generalisability of this adapted screening measure; findings should be replicated in a more diverse sample.
The parent-reported DEPS-R can be used for screening of CYP disordered eating behaviours within clinical health and therapeutic settings, providing early detection of disordered eating habits.This would allow for more accurate prevalence rates, forecasting the provision of services and early intervention, potentially limiting the progression of disordered eating and improving outcomes.Whilst establishing a clinical cut-off for the parent-reported DEPS-R is not possible due to the sample size and measures used, the original DEPS-R developers recommend scores of 20 or above as indicative of disordered eating behaviours requiring further conversations and possible referral. 7However, without established national or international treatments, there is a risk of HCPs detecting problems but not being able to act on them, which may cause reluctance to screen. 5In the absence of treatment pathways, the parent-reported DEPS-R could be used in conjunction with the CYP-reported DEPS-R to provide opportunities for supportive conversations about disordered eating between clinicians and families that are vital for ED prevention. 29Finally, the item-toitem correlations provide insight into which areas of disordered eating parents are most accurately reporting on behalf of their child.This may encourage diabetes clinicians to explore these topics with parents informally, which may prompt further exploration.
It would be clinically prudent to assess responsiveness to change in the DEPS-R following targeted intervention, given that the results here are associative and some eating disorder cognitions are potentially driven by parental and family behaviours/relationships, meaning we cannot assume causation.Additionally, the reliability and validity of this measure in different age groups of CYP, such as those aged 14-17 years or under 11 years -to explore differences across life stages.As CYP age and become increasingly responsible for their diabetes care, parents may be less able to reliably report on their child's disordered eating.Furthermore, research into parental involvement in diabetes management and diet is recommended for identifying mediating factors for the accordance between parents and their children.Future studies may wish to complete a development study, including factor analyses, to determine the validity of the items on this parent questionnaire; a shorter, more exact measure may be more clinically useful and less burdensome in limited settings.To conclude, this study provides novel and encouraging evidence for the reliability and validity of the parent-reported DEPS-R for CYP with T1D.This adapted screening tool may provide clinical use for early detection of disordered eating in CYP with T1D and support further research in the development and validation of this adapted measure for eating disorders.

2.2 | Participants and consent
Demographic and clinical characteristics of parents and children and young people.
T A B L E 1 Intra-correlation coefficients for DEPS-R item-to-item correlations between parent and adolescent reports.
T A B L E 2 Construct validity analyses-Spearman's and bootstrapped point-biserial correlations of parent-reported DEPS-R score with relevant measures, and participant demographic and clinical characteristics.Higher scores indicate greater well-being.Higher scores indicate greater impairment in quality of life.
b Higher scores indicate greater problems.c d e Higher scores indicate less concern.