Introduction

The relationship between autism spectrum disorders (ASDs) and eating disorders (EDs) has been widely studied in the last decades. However, within EDs, the focus has been mainly on anorexia nervosa (AN) [1,2,3]. In particular, ASDs and AN have been found to be often present in comorbidity [1], moreover, patients with AN scored higher than healthy controls (HC) and similar to subjects with ASDs at the autism–spectrum quotient, a widely used measure of autistic traits, suggesting difficulties with social skills, communication and flexibility in AN [2], finally, it has been reported that ASDs and AN shared some neuropsychological features, such as deficit in mental flexibility, in terms of difficulties with set-shifting and lack of central coherence (the difficulty in taking a global approach), deficit in theory of mind (the ability to infer the mental states of others) and impairment in emotional processing [3]. Despite the lack of studies conducted on EDs other than AN, the common neuropsychological profile found in ASDs and AN might be shared also by other subtypes of EDs, such as bulimia nervosa (BN) and binge-eating disorder (BED), and recent studies started to explore this aspect [4].

With the aim of evaluating the presence of eating concerns frequently seen in ASDs in late adolescents and adults without intellectual disabilities (also called subjects with high-functioning ASDs (HF-ASDs)), such as heightened sensitivity to sensory input related to food, hyper-selectivity in eating, rigid routines around mealtime (as where to eat and how cutlery is placed), and difficulties in enjoying company during a meal, a specific questionnaire named “the Swedish eating assessment for autism spectrum disorders” (SWEAA), has been recently developed [5]. It was implemented in three studies speck and colleagues [6] showed that participants with HF-ASDs, especially women, experienced several autistic-eating behaviours, scoring higher than HC at most of the SWEAA subscales, Karjalainen et al. [7] administered the SWEAA to a group of patients with AN, a group of subjects with ASDs and a group of HC, and found that, surprisingly, patients with AN not only scored higher than HC, but also than individuals with ASDs, however, both studies were limited by the only use of the SWEAA, which did not take into account eating disorders symptoms not characteristic of ASDs. Our group, in a recent study [8], showed that individuals with HF-ASDs, scored significantly higher than HC both at the SWEAA and at the eating attitude test (EAT-26) [9, 10], a screening questionnaire for eating disorders symptomatology, suggesting that individuals with HF-ASDs showed not only eating disturbances characteristic of ASDs, but also other eating disorders symptoms, such as distorted body image and tendency towards bulimic behaviours.

However, up to date, no study has ever performed a direct comparison between patients with EDs, individuals with HF-ASDs and HC, implementing measures aimed to detect both eating disturbances characteristic of ASDs and general eating disorders symptomatology.

In this study, we aimed to compare the results of the EAT-26 and the SWEAA in three groups of female participants, one with HF-ASDs, one with EDs, and one of HC. We decided to focus our attention only on female individuals, due to the well-known higher prevalence of eating disturbances in the female gender. Based on the above observations, we would expect both individuals with HF-ASDs and patients with EDs to score higher than HC, at both questionnaires. Moreover, analysing specific differences between EDs and HF-ASDs at the SWEAA and the EAT-26, would help clinicians to understand whether patients with EDs present a similar amount of autistic eating behaviours as subjects with ASDs, and, conversely, whether subjects with HF-ASDs reach the same level of eating disturbance symptomatology as EDs-diagnosed patients.

Methods

Thirty-four consecutive patients with a diagnosis of EDs were recruited from the day care Unit of San Paolo Hospital in Milan. All patients with EDs were diagnosed by a psychiatrist according to DSM-5 criteria. Within this group, 15 patients had a diagnosis of AN (restrictive type), seven patients of BN and 12 patients of BED. No participant with EDs had ever received a diagnosis of ASDs.

Thirty-four consecutive adult individuals with HF-ASDs were recruited from the tertiary level neuropsychiatric outpatient clinic of San Paolo Hospital in Milan. Each participant with HF-ASDs was diagnosed by a psychiatrist and a psychologist according to DSM-5 criteria and the module four of the autism diagnostic observation schedule-2nd version (ADOS-2). No participant with ASDs had ever received a diagnosis of EDs. Thirty-five HC were recruited amongst the general population through word of mouth. Their “healthy state” was determined through a specifically designed interview, including questions on current and past medical history.

All participants gave their written informed consent for the study. The Ethics Committee of San Paolo Hospital reviewed and approved the study protocol.

Exclusion criteria were: (1) age less than 18-years old, (2) inability to understand the instructions of the task, (3) psychotic disorders, (4) intellectual disabilities (I.Q. < 70), measured via the Wechsler adult intelligence scale—fourth edition (WAIS-IV), administered to each participant during the diagnostic process.

The assessment lasted approximately 30 min and included: (1) a demographic and clinical interview, (2) the EAT-26 [9, 10] (3) the SWEAA [5].

The EAT-26 is a 26 item, self-report, questionnaire designed to assess symptoms and concerns characteristic of eating disorders, validated in English [9] and in Italian [10] on samples of patients with eating disorders. Participants are asked to reply to each item on a six-point Likert scale ranging from “never” to “always”. A total score and three subscales (dieting, bulimia and food preoccupation and oral control) were calculated following authors’ instructions [9].

The SWEAA is the only self-report validated measure to assess the presence of eating disorders symptoms specifically in a population of adults with ASD without intellectual disabilities [5]. It is composed by 65 items, to be replied to on a five point Likert scale, ranging from 0 (never) to 4 (always). A total score was calculated by averaging the first 60 items, and multiplying the obtained mean by 25, following authors’ instructions. The following subscales were also calculated: (1) perception, (2) motor control, (3) purchase of food, (4) eating behaviour, (5) mealtime surrounding, (6) social situation at mealtime, (7) other behaviour associated with disturbed eating, (8) hunger/satiety, (9) simultaneous capacity, (10) pica (9) autism quotient (AQ, averaging items 61–65, not included in the total score). For further information about the two questionnaires, see Supplementary materials.

Statistical analyses

Statistical analyses were performed with SPSS v.26 [11].

To assess differences between the three groups at the EDs questionnaires, a series of univariate and multivariate ANCOVA was run, with “group” as factor, body mass index (BMI) as continuous covariate, and, separately, as dependent variable: (1) the total score of EAT-26: (2) the EAT-26 subscales’ scores, (3) the total score of SWEAA, (4) the SWEAA subscales’ scores. Bonferroni post-hoc analysis was used to verify specific differences between two of the three groups. Significance level was set at α = 0.05 and all tests were two-tailed.

Results

We did not have any missing data. Groups were balanced for age and BMI (all p > 0.05). A significant difference emerged between groups at the EAT-26 total score (p < 0.001), with patients with EDs scoring higher than subjects with HF-ASDs (p = 0.008) and HC (p < 0.001) and with subjects with HF-ASDs scoring higher than HC (p = 0.008) (Fig. 1) BMI did not have any significant effect (p = 0.062). Groups were different at all the EAT-26 subscales: (1) dieting (p < 0.001), with both patients with EDs and subjects with HF-ASDs scoring higher than HC (p < 0.001 and p = 0.010, respectively), (2) bulimia and food preoccupation (p < 0.001) with patients with EDs scoring higher than subjects with HF-ASDs (p = 0.013) and HC (p < 0.001) and with subjects with HF-ASDs scoring higher than HC (p = 0.016), (3) oral control (p < 0.001), with patients with EDs scoring higher than both subjects with HF-ASDs (p = 0.010) and HC (p < 0.001). BMI had a significant effect only on the subscale oral control (p < 0.001).

Fig. 1
figure 1

EAT-26 and SWEAA scores in the three experimental groups: patients with eating disorders, subjects with high-functioning autism spectrum disorder and healthy controls. EAT-26 eating attitude test-26, SWEAA Swedish eating assessment for autism spectrum disorders, EDs eating disorders, HF-ASDs  high-functioning autism spectrum disorder, HC healthy controls (*p < 0.05)

Full size image

A significant difference emerged between groups at the SWEAA total score (p < 0.001), with both patients with EDs and subjects with HF-ASDs scoring higher than HC (both p < 0.001) no differences between subjects with HF-ASDs and patients with EDs emerged (p = 0.901) (Fig. 1) BMI did not have any significant effect (p = 0.346). In particular, groups were different at the following SWEAA subscales: (1) perception (p < 0.001), with subjects with HF-ASDs scoring higher than both patients with EDs (p = 0.004) and HC (p < 0.001), (2) motor control (p = 0.005), with subjects with HF-ASDs scoring higher than HC (p = 0.004), (3) purchase of food (p = 0.027) with subjects with HF-ASDs scoring higher than HC (p = 0.045), (4) eating behaviour (p < 0.001) with both patients with EDs and subjects with HF-ASDs scoring higher than HC (both p < 0.001), (5) mealtime surrounding (p < 0.001) with both patients with EDs and subjects with HF-ASDs scoring higher than HC (both p < 0.001), (6) social situation at mealtime (p < 0.001) with subjects with HF-ASDs scoring higher than HC (p < 0.001), (7) other behaviour associated with disturbed eating (p < 0.001) with patients with EDs scoring higher than HC (p < 0.001), (8) hunger/satiety (p < 0.001) with patients with EDs scoring higher than HC (p < 0.001), (9) simultaneous capacity (p = 0.002) with subjects with HF-ASD scoring higher than HC (p = 0.001) (10) AQ (p < 0.001) with subjects with HF-ASD scoring higher than both patients with EDs and HC (both p < 0.001), and patients with EDs scoring higher than HC (p = 0.002). BMI did not have a significant effect on any of the SWEAA subscales.

Demographics and psychometric assessment of the three groups are reported in Table 1.

Table 1 Demographics and psychometric assessment of the three groups
Full size table

Discussion

In this study, we aimed to directly compare the scores of the EAT-26, a questionnaire commonly used as a screening tool to detect eating disorders symptomatology, and the SWEAA, a questionnaire developed to uncover the presence of eating behaviours frequently seen in ASDs [5], in three experimental groups of women, namely patients with EDs, subjects with HF-ASDs and HC.

Our results showed that, at the EAT-26, female patients with EDs scored significantly higher than female individuals with HF-ASDs, and both of them scored higher than HC, suggesting that, although women with HF-ASDs reported high scores at an eating disorder screening questionnaire, they did not reach the level of severity of EDs-diagnosed patients. This trend was similar also for the two EAT-26 subscales bulimia and food preoccupation and oral control, respectively, assessing: (1) the tendency towards bulimic behaviours and the time spent thinking about food, (2) the self-control of eating and social pressure around weight, which was also negatively correlated to BMI on the contrary, at the subscale dieting, which evaluated the tendency to avoid fattening food and shape preoccupations [9], individuals with HF-ASDs behaved more similar to patients with EDs than HC.

On the other hand, no differences between individuals with HF-ASDs and patients with EDs emerged at the SWEAA total score, where they both scored higher than HC; in other words, individuals with HF-ASDs and patients with EDs presented a similar amount of autistic eating behaviours, which was higher than the one observed in HC (Fig. 1). Moreover, no significant differences emerged between EDs and HF-ASDs groups at most of the SWEAA subscales; in particular, it is worth noticing that, at the SWEAA subscale AQ, not only individuals with HF-ASDs scored higher than both patients with EDs and HCs, as expected, but also patients with EDs scored higher than HC, in line with the recent literature [2]. BMI did not affect any of the SWEAA subscales, providing an additional evidence that the SWEAA detected eating disturbances specific of ASDs and, thus, not directly related to EDs [8]. Our results are apparently in contrast with the study by Karleenjen and colleagues [7], which showed that patients with AN presented a higher amount of eating autistic-like behaviours than the ASDs group itself. However, this discrepancy might be due to the fact that, here, we took into account a broader range of EDs, with the aim of exploring the possible overlap between ASDs and EDs in a wider perspective, including patients with BN and BED [4]. Future studies, including larger groups of patients with AN, BN and BED, and comparing them to HF-ASDs groups, are needed, to assess whether the reported amount of autistic eating behaviours is specific of one eating disturbance or can be generalised to all EDs. Most importantly, in the study by Karleenjen et al. [7] (as well as in the one by Spek et al. [6]), it was not possible to rule out the possibility that individuals with ASDs would have scored similarly to an EDs-diagnosed group at a general screening questionnaire for EDs symptomatology. In this study, we made a step further, showing that, although the HF-ASDs group scored lower than the EDs group at the EAT-26 (and therefore, did not show the same level of EDs symptomatology), the two groups presented similar levels of autistic eating behaviours.

On one hand, if, in our previous paper [8], we hypothesised that high scores at the SWEAA of individuals with HF-ASDs should be interpreted by the clinician as an alarm for the possible presence of other eating disturbances, these data suggested that participants with HF-ASDs did not reach the pathological level of EDs symptomatology as patients with EDs did.

On the other hand, our results are in line with the recent model by Brede et al. [12] focussing on AN and restrictive eating behaviours, they proposed a model where specific autism-related processes might give rise to and maintain restrictive eating behaviours, underlying AN, in autistic individuals. Our data showed that, in patients with EDs, autistic eating behaviours were present as frequently as in individuals with HF-ASDs, and therefore we could speculate that these behaviours might have a role in the maintaining of the ED itself. Their role in the development of a manifest EDs symptomatology should be further investigated through longitudinal studies, although some evidences have already suggested the presence of autistic traits before the onset of EDs [12, 13].

We acknowledge the limitations of our study. First, the small sample size, which rendered unreliable any group comparison between the different EDs (AN, BN and BED), second, we did not take into account the psychopharmacological therapy of participants, which might have biased the results. Third, we only included female participants in our sample.

In conclusion, our results give a further preliminary evidence of the overlap between autistic traits and EDs symptomatology overall, and should be taken into account in the definition of a shared model between EDs and ASDs, which, ultimately, might lead to the development of an official clinical guidance in the recognition and treatment of EDs and ASDs comorbidity.

What is already known on this subject?

Previous studies showed that patients with anorexia nervosa and individuals with autism spectrum disorder (ASDs) share common neuropsychological deficits and social and emotional difficulties, along with disturbed eating behaviours. However, a direct comparison between patients with eating disorders (EDs) of all types, individuals with high functioning ASDs and healthy controls (HC) at measures detecting both eating disorders symptoms and eating disturbances characteristic of autism was missing in the literature.

What this study adds?

Our study showed that individuals with HF-ASDs did not reach the same level of EDs symptomatology as patients with EDs. However, patients with EDs did not seem to present a different amount of eating behaviours characteristic of ASDs than HF-ASDs. Both groups (patients with EDs and with HF-ASD) scored higher than HC at both scales. These findings might also have some important therapeutic implications.