The study team enrolled 143 adolescents with diabetes aged 10–19 years who attended a pediatric endocrinology outpatient clinic between 10 February and 26 April 2024. The mean age was 14 ± 2.3 years, and 56.6% of participants were female. Nearly half of the cohort (49.7%) had a diabetes duration of 1–5 years. Investigators obtained laboratory HbA1c values from clinical records during the same visit and measured BMI. Eating attitudes were assessed using the Turkish version of the EAT-26 via a face-to-face questionnaire that took approximately 5–6 minutes per participant.
Data indicated a mean EAT-26 total score of 16.62 ± 8.30 for the sample, which falls below the commonly used screening cutoff of 20 points. The study team reported subscale means as follows: Dieting 10.39 ± 5.58, Bulimia/food preoccupation 2.33 ± 3.03, and Oral Control 3.79 ± 3.08. The authors stated that a total score ≥20 is typically considered a threshold warranting further assessment; they emphasized that the EAT-26 provides screening information and does not establish a clinical diagnosis.
The investigators classified glycaemic control using ISPAD thresholds: good control as HbA1c <7.5% and poor control as ≥7.5%. Analyses found no significant difference between mean EAT-26 total score and glycaemic control categories. The multivariable linear regression—performed with HbA1c as the dependent variable and including EAT-26 total score alongside age, gender, BMI z-score, diabetes duration, dietary adherence, physical activity, and meal skipping as independent variables—did not identify an independent association between EAT-26 total score and HbA1c. Authors reported these null findings directly and did not infer causality given the study design.
Although the total EAT-26 score did not track with glycaemic control, the study team observed differences in EAT-26 total and subscale scores according to selected demographic and clinical characteristics. Specifically, variations were seen by BMI, by treatment modality, and by self-reported treatment adherence. The authors reported that higher scores on the Dieting subscale correlated with body-image–related concerns in this sample. Exact subgroup statistics and pairwise comparison results were reported in the full dataset accessed via the article; the summary in the source stresses these between-group differences without claiming that they mediate HbA1c outcomes.
Internal consistency for the full EAT-26 in this sample produced a Cronbach’s α of 0.688. When examined by subscale, α was 0.662 for Dieting, 0.534 for Bulimia/food preoccupation, and 0.549 for Oral Control. The authors interpreted the total-scale reliability as acceptable, while noting lower internal consistency for certain subscales. The scoring approach followed established EAT-26 conventions: items 1–25 scored with higher frequency indicating higher points, and item 26 scored in the reverse direction. Cutoffs for subscales were noted in the methods (Dieting 18; Bulimia 4; Oral Control 5), and the authors clarified that these thresholds serve as screening indicators rather than diagnostic criteria.
Statistical analysis relied on nonparametric methods because variables deviated from normality. The study team used Kruskal–Wallis tests with Dunn–Bonferroni post hoc comparisons for group differences, Spearman correlation for continuous variables, and multivariable linear regression for adjusted analyses. Significance was set at p<0.05. The University of Health Sciences ethics committee approved the research (Decision No: 23–705). Written informed consent or guardian consent plus adolescent assent was obtained according to age. The authors stated that the underlying dataset is not publicly available because participant consent did not include data sharing and the dataset contains potentially identifiable health information; data may be available to qualified researchers on request, subject to ethics committee approval and applicable regulations.
Limitations and unanswered questions reported by the authors
The authors acknowledged several constraints inherent to the report. The cross-sectional design prevents causal inference between eating attitudes and glycaemic control. Some EAT-26 subscales showed limited internal consistency in this sample, which may affect subscale-level interpretation. The dataset is not publicly shared due to consent restrictions, limiting secondary analysis. The authors did not report long-term follow-up or intervention data, and therefore cannot address how reported eating attitudes might change over time or influence future metabolic outcomes.