Introduction

Type I diabetes (TID) among teenagers is a well-established risk factor for eating disorders (ED). The risk of developing eating disorders (ED) in these teenagers is 2.5 times higher than healthy peers; in other words, for each teenager without TID developing an ED, 2–3 adolescents with TID develop the same disorder [1, 2]. According to Gagnon et al. [3], depressive symptoms and negative coping strategies work as predictors of ED development, with a significant impact on the mental health, social life, and quality of life of these adolescents [3, 4]. These may be exemplified by feelings of exclusion of social life, self-control difficulties, self-criticism, and great concern with body image [3, 5]. This last concern is one of the motives that drive the reduction or omission of insulin doses following binge eating by these adolescents [6,7,8]. This behavior is described as diabulimia and frequently results in poor metabolic control and microvascular complications in the long run, leading to nephropathies, diabetic foot, and hospitalizations [9]. It has been demonstrated that those who restrict the insulin doses have their lives shortened by 7 years when compared to those who do not engage in this practice. Despite the increased risk of diabulimia [10], most nutritional assessments in TID teenagers do not regularly include the investigation of disturbed eating behaviors and preoccupations (e.g., eating, weight and shape concerns, and eating restraint) commonly associated with ED, focusing mostly on dietary patterns. However, little is known about how much the dietary patterns of adolescents with TID are related to disturbed eating behaviors and concerns.

The dietary intake of adolescents with TID has an impact on the metabolic control (glycaemic control and lipid profile) [9, 11, 12]. In general, adolescents with TID show a low adherence to the food pyramid guide recommendation [11] (overby), mostly to fruits, vegetables, milk, and dairy products [11, 13,14,15,16]. The literature in the field supports an increased intake that overrides the recommendation of high total and fat saturated and cholesterol by adolescents with TID [11, 13, 17,18,19]. This pattern of higher intake of fat is commonly used by TID adolescents as a mean to manage their insulin doses, since these foods, such as bacon, cheese, and meat, do not contain carbohydrates [20, 21].

The dietary pattern of teenagers with TID and risk of EDs has been object of a study by Tse et al. [22]. These authors found that those considered to be at higher risk for ED presented a lower quality of diet, lower satiety, and higher intake of total and saturated fat in comparison with those at lower risk for ED [22]. This pattern of low quality of diet has also been described for adolescents without TID [23, 24]. Together with these important healthy issues, food is one of the central topics of their lives and, most of the times, acts to limit spontaneity and social life [5]. However, studies that investigate the relationship between ED psychopathology and dietary intake in adolescents with or without TID have not been identified.

Given this gap in the literature, the goal of this article is to investigate an association between dietary patterns and the presence of ED psychopathology in adolescents with TID and to compare it with a non-diabetic control group of teenagers. Due to existing concerns with body changes secondary to diabetes and its treatment, we hypothesized that TID adolescents would consume a healthier diet (more fruits, milk, vegetables, meat, and less sweets, oil, and fat), and would present a higher level of ED psychopathology in comparison with their peers without TID.

Methods

Study design

This is a case-control study, which compares adolescents with TID (case group) to health peers (control group), matched for age at the rate of 1 case: 2 controls.

Setting

Adolescents with TID were recruited at the Endocrinology Clinic of the Federal University of São Paulo (UNIFESP) from March 2010 to May 2014 (these adolescents represent a subsample of a larger study with children and adolescents with TID)—[25]. Control group were recruited from public schools in the southern city of São Paulo, between October 2013 and October 2014.

Participants

This study was approved by the Ethics Committee of the Federal University of São Paulo—UNIFESP/EPM and the adolescents that agreed to participate, as well as their parents or guardians, have the consent form signed. The sample consisted of 50 adolescents with TID and 100 adolescents without TID. Both groups were considered as inclusion criteria: female, aged between 11 and 16 years, agree and sign the consent form. Specific inclusion criteria for each group included: (1) case group (TID): diagnosed by their endocrinologists with TID, duration of disease at least 6 months, do not present any cognitive impairment, mental illness, or dietary restrictions due to other clinical conditions such as celiac disease, dyslipidemia, liver disease, kidney disease, hypertension, and anemia and (2) control group (CG): without type 1 or 2 diabetes mellitus or other clinical conditions, as cited to case group.

Data sources/measurement

Anthropometric data

Weight and height were measured by portable anthropometric scales (weighing machine and stadiometer) used to calculate the Body Mass Index (BMI = kg/m²). Participants were wearing light clothing and no shoes. The nutritional status classification was made by the growth curves of the World Health Organization for adolescents [26] which consider the BMI for their age on: eutrophic between the 3rd percentile and 85, underweight under 3rd percentile, overweight between 85 and 97 percentiles, and obesity above 97th percentile.

Economic scale

For this evaluation, questionnaire of Economic Classification Brazil the Brazilian Association of Research Companies [27] was used. This scale graduate economic class (A1, A2, B1, B2, C1, C2, D, and E) based on ownership items (i.e., television, radio, automobile, etc.) and education level of the head of household. Economic classes are divided according to the minimum criterion of family income, with A1 class as better economic benefits and E as the lowest economic resource.

Food frequency questionnaire to assess intake of the food groups included in the food guide pyramid adapted for adolescents (FFQ-FP)

To assess dietary intake and the number of servings of food groups, the food frequency questionnaire (FFQ-PP) was used, which is based on the food guide pyramid. This instrument was developed and validated in Brazil by Martinez et al. [28]. It has a list of 50 foods and divided into two parts: the first with favorite foods by adolescents (treats, snacks, chocolates, ice cream, etc.) and the second with the food groups of the Brazilian food guide pyramid [29]. It represents 97.7% of the calories consumed and 90% of the nutrient intake of a day. For the present study, the diet option of food that did not have sugar in its composition (i.e., cookies, chocolate, powdered chocolate, etc.) was included. FFQ-FP provides nutritional information such as energy intake, fiber, distribution of macronutrients (proteins, lipids, and carbohydrates), and micronutrients (calcium and iron). In addition, the FFQ-FP assess the number of servings of food groups [(1) bread, cereal, rice and pasta; (2) fruits; (3) vegetables; (4) milk, yogurt and cheese; (5) meat, poultry, fish and eggs; (6) dry beans; (7) fat and oils; (8) sweets]. The software used to analyze the data was the Diet Pro 5.0 [30].

Child-eating disorders examination (12th edition)

The Child-Eating Disorders Examination (EDE-Ch) [31], a modified version based on the Eating Disorders Examination (EDE) (for adults), was used to identify risk eating behaviors in children and adolescents. This interview is considered the gold standard for measuring the severity of psychopathology and to diagnose eating disorders based on DSM-IV and ICD-10 [32]. The EDE has good validity and reliability for the four-dimensional subscales: restraint eating, eating, weight, and shape concern [31]. In addition, it also evaluates four compensatory methods for weight control: self-induced vomiting, use of laxatives, use of diuretics, and excessive physical activity. The modifications of the Ch-EDE version were made to make more concrete questions and some items were restructured to better capture the participant’s intent, than the behavior itself [31]. Some questions that assess the importance of shape and weight and three ways of excessive food intake were also introduced: objective bulimic episodes, subjective bulimic episodes, and excessive intake. Each score ranges from 0 to 6. Higher scores indicate greater behaviors severity and, therefore, a higher risk for eating disorders [33]. All data collected were about the last 28 days and the previous 3 months.

Study size

The sample size calculation was based on the percentage of total calories from carbohydrate and the study of Helgeson et al. [17] was used as a reference. An average of 49.5% (SD 7.04) of total calories from carbohydrates in diabetic adolescents and of 56.53% (SD 6.43) in controls was reported by authors. Based on these parameters, the estimate sample size, for significance level of p = 0.05 and a statistical power of 90%, is 20 teenagers in the diabetic group and 20 adolescents in the control group. To ensure a good effect and to detect differences between groups, 150 participants were included in a rate of 1 case (TID): 2 controls (participants without TID).

Data analysis

Our exploratory analysis started by evaluating distributions, frequencies, and percentages for each of the numeric and categorical variables. Categorical variables were evaluated for near-zero variation [34]. Extensive graphical displays were used for both univariate analysis and bivariate associations, accompanied by broader tests such as maximal information coefficient [35]. Pearson correlation coefficients and tests as well as non-negative matrix factorization [36] were used to measure the unadjusted association between ED psychopathology and dietary intake. Generalized linear models with a Gaussian family distribution [37] were then used to evaluate the association between ED psychopathology as outcomes and variables representing dietary intake. Unadjusted models involved the association between individual ED psychopathology and dietary intake, whereas adjusted scores measured the association between each ED psychopathology and the set of all dietary intake, age, and BMI.

Results

Participants were balanced in relation with BMI, although age was significantly higher among those with diabetes than those without diabetes. Participants with diabetes had dietary intake with a larger percentage of bread, cereal, rice and pasta, vegetable, milk, yogurt and cheese, fat and oil groups, and fibers while having a far smaller percentage of their diet involving the sweets group. Finally, in relation with ED psychopathology, no significant differences were observed between diabetic and non-diabetic participants (Table 1).

Table 1 Patient sample characteristics
Full size table

When evaluating the unadjusted association between ED psychopathology and dietary intake, dietary intake rich in fibers was significantly associated with both the global and eating concern scores among diabetic participants. In contrast, among non-diabetic participants, increased meat, poultry, fish, and eggs group and decreased bread, cereal, rice, and pasta group consumption were more significantly associated with higher ED psychopathology (Table 2). Aligned with these findings, the heatmap representation demonstrates a similar pattern of association, but adding that the intake of meat, poultry, fish and eggs, bread, cereal, rice, and pasta and fat and oil group eating patterns are clustered as a single behavior when taking into account the combined diabetic and non-diabetic samples (Fig. 1).

Table 2 Correlation coefficients between eating disorder scores and dietary patterns, unadjusted
Full size table
Fig. 1
figure 1

Heatmap demonstrating the association as well as clustering patterns among eating disorders scores and dietary patterns in the overall sample. Heatmap: G1 (bread, cereal, rice and pasta group), G2 (fruits group), G3 (vegetables group), G4 (milk, yogurt and cheese group), G5 (meat, poultry, fish and eggs group), G6 (dry beans group), G7 (fat and oils group), and G8 (sweets group)

Full size image

When evaluating the adjusted association between dietary intake and ED psychopathology, higher fiber intake was no longer significantly associated with eating disorder scores among diabetic participants. In contrast, a higher meat, poultry, fish, and egg group intake a low bread, cereal, rice, and pasta group intake were still statistically associated with higher ED psychopathology among non-diabetics participants (Table 3). These patterns are demonstrated in scatterplots for the overall sample through the negative slope of lines associated with bread, cereal, rice and pasta intake and positive slope for meat, poultry, fish, and egg group intake (Fig. 2).

Table 3 Association between eating disorder scores and dietary patterns, adjusted
Full size table
Fig. 2
figure 2

Scatterplot and regression lines demonstrating the association between eating disorder scores and dietary patterns in overall sample. Scatterplot: G1 (bread, cereal, rice and pasta group), G2 (fruits group), G3 (vegetables group), G4 (milk, yogurt and cheese group), G5 (meat, poultry, fish and eggs group), G6 (dry beans group), G7 (fat and oils group), and G8 (sweets group)

Full size image

Discussion

This study investigates the relationship between risky ED psychopathology and dietary intake of adolescents with and without TID. As expected, the dietary intake of diabetic adolescents differed from that of non-diabetic adolescents in our sample; TID adolescents showed higher consumption of bread, cereal, rice and pasta, vegetables, milk, yogurt and cheese, fat and oil groups, and fiber, while non-diabetic adolescents presented higher consumption of sweets. Levels of ED psychopathology were not different between groups, but correlations between dietary intake and ED psychopathology were found (for unadjusted models): a positive association between intake of fiber and ED psychopathology in the diabetic group, and intake of meat, poultry, fish, and eggs and ED psychopathology in the non-diabetic group. In addition, a negative correlation was found between the intake of bread, cereal, rice and pasta, and ED psychopathology among non-diabetic adolescents. However, when models are adjusted for BMI and age, only the associations observed among non-diabetic adolescents remain. In other words, these findings suggest that, among TID adolescents, an increased intake of fiber raises a red flag to the presence of increased ED psychopathology. In the same way, among non-diabetic adolescents, a diet that combines increased intake of meats, poultry, fish, and eggs and reduced intake of breads, cereals, rice, and pasta also raises a red flag for a greater risk of higher ED psychopathology.

The main risk factors related to the development of ED in the general population include a history of dieting or dietary restraint, increased BMI, and⁄or body dissatisfaction and female gender [38, 39]. Individuals with TID may be even more susceptible than the general population to develop an ED [1, 2]. It is known that eating behaviors of adolescents with TID are different from that of adolescents without the condition, differences that are mainly due to the dietary demands of their treatment to achieve metabolic control. Besides that, there is increased risk of insulin-related weight gain (difficulty to lose weight once TID is diagnosed) and associated body dissatisfaction [40], as well as effects of diabetes on psychological development [41]. Binge eating and insulin under-dosing was found to be the most common weight loss methods used by adolescents with TID [10, 40, 42, 43]. Moreover, increased physical and mental health comorbidity risks are present in individuals with TID [6, 44,45,46], and when combined with ED, this risk may increase and become of major clinical importance.

This study found that higher ED psychopathology was associated with high meat consumption (high protein diet) and low carbohydrates consumption (low-carb diet) among adolescents without TID. High protein diets promote increased energy expenditure by raising postprandial thermogenesis and metabolism at rest [47]. Besides, this diet pattern may promote a high satiety and less energy intake, because when amino acids concentrations are elevated, it produces anorexigenic hormones and neuropeptides, such as leptin, ghrelin, and pro-opiomelanocortin (POMC), and also decreases neuropeptide Y (NPY) and Agouti-related peptide (Agrp), both orexigenic neuropeptides. Furthermore, substrate oxidation changes, and more fat is oxidized than is consumed, resulting in a negative fat balance [47,48,49,50,51,52,53,54,55]. This diet aims to reduce body fat and weight, and to maintain lean body mass [47, 56, 57]. Moreover, this high protein diet may impact significantly eating behavior that is driven by reward as, compared to meals with the standard amounts of protein, high protein meals reduce the activation of brain regions typically associated with craving for food and food reward [58, 59]. In addition, high protein meals compared with meals with the standard amounts of protein show greater reduction in eating snacks that are rich in fats and/or sugars at night [59]. Thus, diets with high protein and low carbohydrate intake provide resources for adolescents to regulate their dietary intake to control their weight and shape.

Although findings were weaker among adolescents with TID, high fiber intake was associated with higher levels of ED psychopathology. TID patients are already concerned with their food intake, and the control of their diet is a routine for these adolescents. However, this last statement is not always true. These adolescents have some difficulties to adhere to nutritional recommendations. Our findings suggest that the TID adolescents present a higher intake of fiber when compared to adolescents without diabetes. However, both groups did not reach the recommendations for this nutrient in their diet (data not shown), i.e., 26 g [60]. In addition, the daily amount of carbohydrate intake is also determined for these patients [61,62,63]. Nevertheless, some of them decrease the intake of this nutrient consciously to minimize weight gain [18]. A higher intake of fiber may be the most appropriate way to manage glycaemic control and maintain a healthy diet in an attempt to reduce weight. These findings are in accordance with studies that found an association of high fiber diets with ED behaviors [38, 64, 65]. Misra and colleagues [66] found a higher intake of fiber in individuals with anorexia nervosa compared to healthy adolescents and that leptin levels correlated inversely with the intake of total fiber and with the intake of both soluble fiber and insoluble fiber [66]. Dietary fiber is known to lower the energetic density of the diet mainly by binding water, and water contributes to weight but not to energy [64, 65, 67]. Fibers act by slowing the absorption of the nutrients in the small intestine and mainly by reducing cholesterol absorption [68]. Furthermore, fiber intake promotes prolonged satiety thus limiting the intake of energy within a meal and delaying the feeling of hunger; it also benefits body composition by ameliorating the lipid profile; finally, it improves the glycaemic control and insulin sensitivity due to its low glycaemic index [68,69,70,71,72]. Therefore, as observed in healthy adolescents, teenagers with diabetes also control their weight and shape through their food intake. Thus, health professionals that care for these patients should be alert when the above-mentioned red flags in their diet are raised.

Limitations of this study include the potential unrepresentativeness of our samples of adolescents with and without TID, which were consisted of adolescents attending one tertiary university diabetes center and students from two schools of São Paulo city. Moreover, although the Child-EDE is a gold standard diagnostic instrument of ED in children and adolescents, it is a questionnaire that demands an average of 1 h or more to be applied, and this long duration may affect responses given by the adolescents as they get tired along the interview. In addition, it is possible that children and adolescents present values of scores within the normal range (in subscales and global score of Child-EDE), but present a diagnosis of ED [31]. Another limitation of this study is the use of the food frequency questionnaire in children and adolescent, who may not pay sufficient attention to the food or food portions they eat routinely. However, this food frequency questionnaire presents a quantitative method with specific number of foods and beverages, categories of responses that indicates the usual frequency of consume, and the size of a common portion, which facilitates the answers of the participant. This instrument was built according to the guide of the food pyramid for teenagers [28]. Finally, the manipulation of insulin doses to reduce or control weight gain was not an outcome measure of the present study. Nevertheless, we did find three girls (6%) of our sample who would not adhere to the insulin treatment as recommended, and would reduce or omit insulin doses, but reasons for presenting this behavior were not explored. One of these girls was overweight, showed body image dissatisfaction, and was diagnosed with an Eating Disorder Not Specified (DSM-IV). For some researchers [73], most of all patients omit insulin due to some psychological issues, not always it is due shape and weight concern. Therefore, it can be hypothesized that insulin omission, which leads to weight loss, is more closely connected to mood regulation and self-harm behavior than to weight and shape issues. Future studies should explore further this topic.

The risk of increased ED psychopathology in adolescents with TID differs from those without the disease in relation with their dietary intake. Adolescents with TID have a subtle red flag for risk of increased ED psychopathology, which demands careful attention by health professionals. This can be achieved by the application of screening instruments and complete diagnostic investigation in adolescents who present this dietary pattern and behavioral profiles that suggest risk of developing an ED. In addition, health professionals should pay more attention to the language to instruct about and discuss diabetes care practices (requirements of diet and blood glucose, perfectionism of decision-making, being in control and reach the objective) to prevent ED [5]. Adolescents with ED and TID need more treatment investment when compared to those without diabetes, as they demonstrate lower motivational levels, perseverance, persistence, and self-esteem that result in higher dropouts and poor outcomes [73]. Research is needed to replicate our findings among larger samples of different cultures and eating habits, since concerns with eating, weight, and shape change according to the society that the individuals live.