Abstract
This study reports novel information taken from comparisons of children with autism spectrum disorder (ASD) with eating problems and those with ASD without eating problems with typically developing groups of children.
Eating problems are frequent in children with typical and atypical development (Sharp et al., 2013; Smith et al., 2020), and they seem to be significantly more prevalent in children with autism spectrum disorder (ASD) than in typically developing (TD) children or children with other neurodevelopmental disorders (Matson et al., 2009; Mayes & Zickgraf, 2019; Sharp et al., 2013). In TD children, eating problems are often thought to be part of the normal development (Bryant-Waugh, 2019). A recent study (Mayes & Zickgraf, 2019) showed the presence of atypical eating behaviors—namely, limited food preference or sensitivity to texture—in 4.8% of the TD sample (N = 313). In a study by Provost et al. (2010), parents described their children as picky eaters in one-quarter of the sample (n = 25). The prevalence of picky eaters was inconsistent in the literature studies, ranging from 14% to 50% in preschoolers and from 7% to 27% in later childhood (Cardona Cano et al., 2015). In most cases, childhood eating problems do not reach clinical significance and tend to resolve over time (Provost et al., 2010; Bryant-Waugh, 2019). Although some picky eaters (generally from socially disadvantaged backgrounds) continue to have problems at the age of 6 yr (about 4%), the majority of them (about 32%) show a remission of the atypical behaviors within 3 to 4 yr (Cardona Cano et al., 2015).
As stated earlier, eating problems are more common in children with ASD than in TD children or those with other developmental disorders. The prevalence of eating problems in ASD was estimated to reach up to 90% (Kodak & Piazza, 2008). A study by Mayes and Zickgraf (2019) with a sample of more than 2,000 children, of whom 1,462 were children with ASD, found a prevalence of 70.4% in children with ASD, compared with 13.1% in children with other neurodevelopmental disorders and 4.8% in TD children. Eating problems for children with ASD seem to persist beyond childhood, especially for children who have not received support such as parental counseling and psychoeducational behaviorally based interventions (Bandini et al., 2017; Kuschner et al., 2015; Page et al., 2022 ; Petitpierre et al., 2021; Suarez et al., 2014; Volkert & Piazza, 2012). Only Beighley et al. (2013) and Peverill et al. (2019) reported a remission of eating problems with age, except for a minority of the sample (Peverill et al., 2019).
Among the atypical eating behaviors in people with ASD, food selectivity appears to be the principal dimension involved (Beighley et al., 2013; Mayes & Zickgraf, 2019 ; Postorino et al., 2015; Twachtman-Reilly et al., 2008), with a median prevalence of 62% (Mayes & Zickgraf, 2019). Moreover, it appears to be the most change-resistant feature over time, persisting into late preadolescence (Bandini et al., 2017), although Beighley et al. (2013) found a remission across childhood, especially with Asperger’s syndrome. Selectivity includes not only a limited repertoire of accepted food or a single food intake but also food refusal, eating only one brand of food, and food preference that is based on color or shape. Other issues are related to anomalous sensitivity (e.g., hypersensitivity to food textures, food temperature, and hypo- or hyperresponsivity to environmental stimuli), to mealtime behaviors (e.g., difficulty remaining seated at the table all mealtime long, aggressive/disruptive behaviors, repetitiveness, rituals), and to eating behaviors (e.g., eating too little or too much, swallowing food without chewing, pocketing food without swallowing, pica; Hubbard et al., 2014; Margari et al., 2020 ; Marshall et al., 2014; Mayes & Zickgraf, 2019; Provost et al., 2010). Beyond the medical (e.g., gastrointestinal, cardiopulmonary, and neurological problems, kidney disease, and anatomical anomalies) and behavioral etiologies, environmental factors (e.g., inappropriate parental feeding practices, physical features, and presentation of food) also play an important role.
Sensory Impairments
Sensory problems are very common and pervasive in people with ASD (Brockevelt et al., 2013; Green et al., 2016 ; Leekam et al., 2007; McCormick et al., 2016; Suarez, 2012; Tomchek & Dunn, 2007; Watling et al., 2001), and, in some studies, approximately 90% of participants present with sensory symptoms in several sensory domains (Green et al., 2016 ; Leekam et al., 2007; Tomchek & Dunn, 2007). Those symptoms seem to persist over time—especially taste/smell and tactile sensitivity, among proximal domains, and auditory sensitivity, among distal domains (Leekam et al., 2007; McCormick et al., 2016). The following sensory dimensions seem to be the most affected: Tactile Sensitivity (Leekam et al., 2007; Tomchek & Dunn, 2007; Wiggins et al., 2009), Taste/Smell Sensitivity (Leekam et al., 2007; McCormick et al., 2016 ; Watling et al., 2001; Wiggins et al., 2009), and Auditory Sensitivity (McCormick et al., 2016 ; Tomchek & Dunn, 2007; Wiggins et al., 2009). Some authors found an association between sensory problems and the severity of autism (Ben-Sasson et al., 2009; Brockevelt et al., 2013; Sanz-Cervera et al., 2015), as well as restricted and stereotyped interests and behaviors (Kargas et al., 2015; Wiggins et al., 2009).
Sensory problems are classified into three subtypes: (1) hyperresponsiveness, characterized by excessive behavioral reactions to sensory stimuli; (2) hyporesponsiveness, characterized by behavioral underreactivity to sensory stimuli; and (3) sensory seeking, characterized by specific interests in intense sensory experiences (Miller et al., 2007). It was found that hyper- and hyporesponsiveness patterns can also coexist in people with ASD (Ausderau et al., 2014 ; Baranek et al., 2006; Ben-Sasson et al., 2007, 2009; Lane et al., 2010), especially in children showing a generalized sensory impairment (Ausderau et al., 2014). A recent update of Ben-Sasson et al. (2019) showed significant differences between children with ASD and TD children in all types of sensory reactions (hyperresponsiveness, hyporesponsiveness, and sensory seeking). Some physiological research suggests that the exaggerated reactions to textures or noises might be caused by a low threshold for one or multiple sensation channels. On the contrary, in underresponsivity, a high threshold has been hypothesized (Suarez, 2012).
Association Between Eating Problems and Sensory Features in Children With Autism Spectrum Disorder
Most of the authors have found an association between eating and sensory problems in people with autism (see Cermak et al., 2010, for a narrative review of the scientific literature over a 25-yr period; Chistol et al., 2018; Johnson et al., 2014; Kral et al., 2015; Lane et al., 2014; Nadon et al., 2011; Suarez, 2012; Wang et al., 2019; Zobel-Lachiusa et al., 2015). To the best of our knowledge, only two studies have not (Aponte & Romanczyk, 2016;Schreck & Williams, 2006). The decision to eat or reject a piece of food seems to be strongly mediated by sensory processing (Petitpierre et al., 2021). This mediation is thought to be dysfunctional in people with ASD, probably because of the difficulty in creating brain templates that allow people to react in accordance to the sensory input flow (Rozenkrantz et al., 2015). Therefore, the behavioral responses to food might be extreme (either rejection or exclusive preference; Petitpierre et al., 2021).
According to Cermak et al. (2010) and Suarez (2012), a particular role is played by the tactile and oral overresponsivity in determining food selectivity as well as difficulties with food textures. An association with food refusal and lower intake of vegetables was also found (Chistol et al., 2018).
Lane et al. (2014) found a strong correlation between taste/smell sensitivity and food selectivity, but they found no association with tactile sensitivity. The olfactory defensiveness could affect a person’s refusal of food that has a certain smell but also the avoidance of eating in cafeterias or restaurants (Cermak et al., 2010). Zulkifli and Rahman (2021) found associations between food selectivity and tactile and taste/smell sensitivities, and auditory filtering, and an association between food refusal and tactile and movement sensitivities.
Taken together, the aforementioned studies showed clear evidence that impaired sensory processing is associated with feeding and eating problems. However, not all people with ASD have feeding or eating difficulties. Therefore, the question is, Are there any differences in the sensory profiles of children with ASD who have eating problems (ASD–W) as compared with children with ASD who do not have eating problems (ASD–WO)? Are there any differences between them and their TD peers? Are there age-related differences in the eating and sensory features of both children with ASD–W and those with ASD–WO?
To the best of our knowledge, only few studies compared children with ASD–W to those with ASD–WO on these domains (Crippa et al., 2022; Panerai et al., 2020; Prosperi et al., 2021), but no studies have dealt with the comparison of both ASD subgroups with TD peers.
The objectives of our study were (1) to investigate the differences between children with ASD and TD children with regard to eating and sensory characteristics and to confirm whether these differences are more marked in children with ASD–W than in those with ASD–WO; (2) to detect any associations between sensory and eating behaviors and any most involved sensorial dimensions; and (3) to search for any age-related differences in sensory and eating characteristics in children with ASD.
Method
Participants
A total of 165 children were recruited: 117 with ASD (95 boys, 22 girls), chronological age, M = 66.93 mo (SD = 28.45) and 48 TD (32 boys, 16 girls), M chronological age = 66.85 mo (SD = 32.55). Children with ASD were recruited over a 2-yr period (2019–2020) in three centers that are highly specialized in the diagnosis and treatment of ASD. (In accordance with a specific legislation decree issued by the Regional Government of Sicily, a number of autism centers have been established in different areas of the Sicilian Region. These centers can provide social and health care services to people with ASD under the authorization of the Regional Health Council, which consequently covers the costs of health care services provided). All participants received diagnoses in accordance with the criteria of the Diagnostic and Statistical Manual of Mental Disorders (5th ed.; DSM–5; American Psychiatric Association, 2013). The diagnoses were further confirmed by administering the Autism Diagnostic Observation Schedule (2nd ed.; Lord et al., 2012) or the Childhood Autism Rating Scale (2nd ed.; Schopler et al., 2010). The severity of the disorder was classified into three levels (1 to 3) according to the DSM–5. Ten children (8.6%) showed a severity level of 1 (requiring support), 34 (29%) showed a severity level of 2 (requiring substantial support), and 71 (61%) showed a severity level of 3 (requiring very substantial support). Eighty-seven (74.36%) had comorbid intellectual disability, 15 (12.8%) had normal intellectual functioning, and 13 (11%) had borderline intellectual functioning. Eighty-eight children (75%) showed language impairments. Parents’ educational level was classified into high (both college graduates, or one college graduate and one with a high school diploma; n = 25), medium (both with a high school diploma, or one with a high school diploma and one with a middle school diploma; n = 65), and low (both with a middle school diploma or lower; n = 18). The parents’ socioeconomic levels were classified as high (both professionals or managers, or one professional or manager and one employee; n = 10), medium (both employees, one employee and one worker, one professional or manager and one worker, or one professional or manager; n = 29), low (one employee, two workers, or one worker; n = 57), and no income from work (both unemployed; n = 9).
On the basis of the cutoff (≥34) of the Brief Autism Mealtime Behaviors Inventory (BAMBI; DeMand et al., 2015; Lukens & Linscheid, 2008), the ASD group was divided into two subgroups: children with ASD–W (n = 43; 36.7%) and children with ASD–WO (n = 74; 63%).
TD children were recruited from the local community public schools. Their parents’ educational levels were classified as high (n = 20), medium (n = 35), and low (n = 5). Their parents’ socioeconomic levels were evaluated as high (n = 9), medium (n = 17), low (n = 17), and no income (n = 3).
We obtained written informed consent from the parents of all the children recruited in this study, after obtaining approval from the local ethics committee. The intervention was conducted according to the criteria of the Declaration of Helsinki (World Medical Association, 2013).
Assessment Instruments
The BAMBI (DeMand et al., 2015; Lukens & Linscheid, 2008) is a parental report questionnaire designed to detect mealtime behavior problems on a Likert scale ranging from 1 (never/rarely) to 5 (always). The higher the total score, the more eating problems there are. DeMand et al. (2015) explored the psychometric properties of the BAMBI in a sample of 273 children with ASD, ages 2 to 11 yr. Four main factors, derived from 15 items, were proposed: Food Selectivity, Disruptive Mealtime Behaviors, Food Refusal, and Mealtime Rigidity. The internal consistency (Cronbach’s α) of the scale was .835, and it was specifically as follows for the four factors: Food Selectivity, α = .82; Disruptive Mealtime Behaviors, α = .69; Food Refusal, α = .82; and Mealtime Rigidity, α = .67. A cutoff total score of 34 was found, with sensitivity of .76 and specificity of .24, totaling 81% of children with ASD as identified.
The Sensory Experience Questionnaire (SEQ; Baranek et al., 2006; Little et al., 2011) is a brief (30-question) questionnaire to caregivers on the frequency of the child’s responses to sensory stimuli in the natural context. The item score is assigned by using a Likert scale ranging from 1 (almost never) to 5 (almost always). The higher the scores, the greater the sensory problems. The SEQ measures the Hyperresponsiveness (exaggerated or avoidant response) and Hyporesponsiveness (lack of or delayed response) to social and nonsocial stimuli (Baranek et al., 2013). Both a total score and four subscale scores are provided by the tool (Social and Nonsocial Hyper- and Hyporesponsiveness). High internal consistency (Cronbach’s α = .80) and test–retest reliability (intraclass correlation coefficient = .92) were reported by Little et al. (2011) on a sample of 358 questionnaires.
The Short Sensory Profile (SSP; Dunn, 1999) is a caregiver questionnaire including 38 items scored on a Likert scale ranging from 1 (always) to 5 (never). Lower scores indicate more atypical sensory responses. A total score and seven subsection scores can be obtained: Tactile Sensitivity, Taste/Smell Sensitivity, Movement Sensitivity, Underresponsive/Seeks Sensation, Auditory Filtering, Low Energy/Weak, and Visual/Auditory Sensitivity. On the basis of the cutoff points available, the total score and each of the subsection scores can be classified into one of the three following categories: Typical Performance, Probable Difference, and Definite Difference. The SSP appeared to be useful in differentiating children with sensory impairments from those without sensory impairments (discriminant validity, >95%); the internal reliabilities ranged from .70 to .90, and the internal validity correlations for the subsections ranged from .25 to .76 (Dunn, 1999; McIntosh et al., 1999).
Study Design
We conducted a nonrandomized comparative study. The three questionnaires were administered as parent interviews by clinical psychologists of the three Sicilian centers participating in the study, as part of the diagnostic procedures.
Data Analysis
For statistical calculations, we used IBM SPSS Statistics (Version 20).
As a preliminary step, to examine chronological age differences in ASD and TD, we used Student’s t test (Student, 1908). To search for any differences in gender, number of children with chronological ages ≤6 yr or >6 yr, parental education, and socioeconomic levels of children with ASD versus TD children, we used the chi-square test (Pearson, 1900). Moreover, we computed the means, standard deviations, skewness, and kurtosis for the analysis of the BAMBI, SEQ, and SSP score distributions (both the total and subsection scores for each questionnaire).
Then, to analyze the differences between the three groups (ASD–W, ASD–WO, and TD) and the differences between the age groups (≤6 yr vs. >6 yr) in relation to BAMBI, SEQ, and SSP scores, we ran a series of 3 × 2 multivariate analyses of variance (MANOVAs; Bock, 1985). The statistical significance of differences between the ASD–W, ASD–WO, and TD groups for each dimension of the BAMBI, SEQ, and SSP was evaluated with post hoc Bonferroni’s correction (Dunn, 1961). We used η2 (Cohen, 1973) as a measure of the effect size: η2 < .06 indicates a small effect, .06 ≤ η2 ≤ .14 indicates a medium effect, and η2 > .14 indicates a large effect size.
Finally, we calculated the Pearson’s correlation coefficient to investigate the association between sensory (SSP and SEQ) and eating (BAMBI) scores (Mukaka, 2012). Differences in correlation coefficients between children with ASD and TD children were tested using the Fisher’s z method.
Results
Preliminary Analyses
No statistically significant differences were found between children with ASD and TD children in the following domains: chronological age, t(163) = .015, p = .99; gender, χ2(1) with Yates’s correction = 3.27, p = .07; frequency of children with chronological age ≤6 yr or >6 yr, χ2(1) with Yates’s correction = 0.07, p = .79; parental education, χ2(2) with Yates correction = 5.07, p = .06; and socioeconomic levels, χ2(3) with Yates’s correction = 5.6, p = .13. Therefore, the two groups turned out to be comparable and properly balanced.
The BAMBI, SEQ, and SSP scores were normally distributed with skewness and kurtosis values ranging approximately from −1.0 to 1.0 (Muthén & Kaplan, 1985), with exceptions for Food Refusal on the BAMBI and the Low Energy/Weak dimension on the SSP (Table 1).
Preliminary Descriptive Analyses
Note. BAMBI = Brief Autism Mealtime Behaviors Inventory; HyperR = Hyperresponsiveness; HypoR = Hyporesponsiveness; SEQ = Sensory Experience Questionnaire; SSP = Short Sensory Profile; UnderR = underresponsive.
Differences in Sensory and Eating Profiles
Differences between children with ASD–W, children with ASD–WO, and TD children with regard to sensory and eating profiles are indicated in Table 2 and are also described in the following text for each questionnaire.
Differences Between Children With ASD–W, Children With ASD–WO, and TD Children on BAMBI, SEQ, and SSP Scores
Note. Different subscript letters within a row indicate means that differ from each other (p < .05). ASD–W = autism spectrum disorder with eating problems; ASD–WO = autism spectrum disorder without eating problems; BAMBI = Brief Autism Mealtime Behaviors Inventory; SEQ = Sensory Experience Questionnaire; SSP = Short Sensory Profile; TD = typically developing.
***p < .001.
Brief Autism Mealtime Behaviors Inventory
The multivariate effects of group—Wilks’s λ = .38, F(8, 312) = 23.97, p < .001, η2 = .38—and the Group × Age interaction—Wilks’s λ = .86, F(14, 304) = 2.93, p = .004, η2 = .07—were significant, whereas age turned out to be not significant. The univariate effects of group were significant on all BAMBI dimensions. We found that children with ASD–W showed significant differences in all BAMBI dimensions, using Bonferroni’s post hoc test in the comparison with both TD children and those with ASD–WO (Table 2). Furthermore, we found statistically significant univariate effects for the Group × Age interaction in Disruptive Mealtime Behaviors, F(2, 159) = 4.07, p = .019, η2 = .05; and Mealtime Rigidity, F(2, 159) = 6.57, p = .002, η2 = .08.
Sensory Experience Questionnaire
The multivariate effects of group, Wilks’s λ = .56, F(8, 312) = 13.05, p < .001, η2 = .25; and age, Wilks’s λ = .92, F(4, 156) = 3.26, p = .013, η2 = .08; were significant. The effect of the Group × Age interaction was not as significant, Wilks’s λ = .88, F(8, 312) = 2.48, p =. 013, η2 = .06. The univariate effects of group were significant on all dimensions of the SEQ. Bonferroni’s post hoc test detected significant differences between children with ASD (both ASD–W and ASD–WO) and TD children in all dimensions of SEQ (Table 2). The univariate effects of age showed significant differences only for the Hyporesponsiveness dimension across the social context, F(1, 159) = 5.9, p = .016, η2 = .036; and across the nonsocial context, F(2, 159) = 4.9, p = .029, η2 = .03;, with small effect sizes.
Short Sensory Profile
Only the multivariate effect of group was statistically significant, Wilks’s λ = .51; F(14, 304) = 8.55, p < .001, η2 = .28. The multivariate effects of both age and the Group × Age interaction were not significant. Table 2 shows the univariate effects of group, which are statistically significant for all SSP dimensions, with medium-to-large effect sizes, except for Movement Sensitivity. Bonferroni’s post hoc test detected significant differences among the ASD–W, ASD–WO, and TD groups in Tactile Sensitivity, Underresponsive/Seeks Sensation, and Auditory Filtering. Children with ASD–W showed the lowest mean scores in all of the aforementioned dimensions. Children with ASD–W differed significantly from those with ASD–WO and from TD children in the Taste/Smell Sensitivity dimension, whereas children with ASD–WO and TD children showed no differences. Both children with ASD–W and those with ASD–WO differed significantly from TD children (but not from each other) in the Low Energy/Weak and Visual/Auditory Sensitivity dimensions.
Correlation Analysis
Pearson’s correlation coefficients showed similar patterns of association between the BAMBI, SEQ, and SSP total scores for children with ASD and TD children, with all p values < .001 (for the BAMBI and SEQ: r ASD = .32, and r TD = .34; for the BAMBI and SSP: r ASD = −.45, and r TD = −.46; and for the SEQ and SSP: r ASD = −0.66, and r TD = −.63). Overall, these correlations turned out to be moderate for both children with ASD and TD children.
Tables 3 and 4 show the results of the correlation analyses between all BAMBI and SEQ or SSP dimensions, respectively. As shown in Table 3, an association between Social Hyporesponsiveness and Food Refusal was found for both children with ASD and TD children. The other patterns of association were different in the two groups: In children with ASD, both Social and Nonsocial Hyperresponsiveness were predominantly associated with some BAMBI dimensions. Social Hyporesponsiveness was associated with Food Refusal. No association was found between Nonsocial Hyporesponsiveness and BAMBI dimensions. In the TD group, both Social and Nonsocial Hyporesponsiveness and Nonsocial Hyperresponsiveness were associated with the BAMBI dimensions. For children with ASD, no associations were found between the SEQ dimensions and Food Selectivity, whereas for TD children, Nonsocial Hyperresponsiveness was associated with Food Selectivity. Table 4 shows a strong association between the Taste/Smell Sensitivity and Underresponsive/Seeks Sensation dimensions and the BAMBI categories in both the ASD and TD groups. The association between the BAMBI categories and the Tactile Sensitivity dimension was more marked in ASD. Auditory Filtering was involved only for children with ASD, as well as Movement Sensitivity in TD children. The Low Energy/Weak and Visual Auditory Sensitivity dimensions were not associated with BAMBI, either in children with ASD or in TD children.
Pearson Correlation Coefficients Between BAMBI and SEQ Dimensions in Children With ASD and TD Children
Note. For ASD, n = 117; for TD, n = 48. ASD = autism spectrum disorder; BAMBI = Brief Autism Mealtime Behaviors Inventory; SEQ = Sensory Experience Questionnaire; TD = typically developing.
*p < .05. **p < .01. ***p < .001.
Pearson Correlation Coefficients Between BAMBI and SSP Dimensions in Children With ASD and TD Children
Note. For ASD, n = 117; for TD, n = 48. ASD = autism spectrum disorder; BAMBI = Brief Autism Mealtime Behaviors Inventory; SSP = Short Sensory Profile; TD = typically developing.
*p < .05. **p < .01. ***p < .001.
Discussion
Atypical sensory features and eating problems have been reported to be common in people with ASD, with a higher prevalence than in TD peers (Petitpierre et al., 2021). An association between sensory and eating problems was found by some authors (Cermak et al., 2010; Chistol et al., 2018; Johnson et al., 2014; Kral et al., 2015; Lane et al., 2014; Nadon et al., 2011; Suarez, 2012; Wang et al., 2019; Zobel-Lachiusa et al., 2015). Only recently have some studies assessed the presence of differences in sensory and eating characteristics in people with ASD–W and people with ASD–WO (Crippa et al., 2022 ; Panerai et al., 2020; Prosperi et al., 2021). Our opinion is that this is a crucial topic to improve existing understanding about this field and obtain further knowledge to guide intervention programs tailored to individual sensory and eating difficulties and strengths.
In this section, the statistically significant differences between groups is discussed only in the case of moderate to large effect sizes, because of their greater meaningfulness.
Among participants with ASD, 36.7% showed eating problems on the BAMBI. This percentage is slightly higher than that found in our previous study (Panerai et al., 2020), but it still remains lower than that reported in other studies (Kodak & Piazza, 2008; Ledford & Gast, 2006; Sharp et al., 2013 ; Twachtman-Reilly et al., 2008), probably because of the differences in the tools used. On the contrary, it is similar to the percentage reported in studies where the same questionnaire was used (Catino et al., 2019; Prosperi et al., 2021). In accordance with the studies published earlier (Beighley et al., 2013; Mayes & Zickgraf, 2019; Postorino et al., 2015; Twachtman-Reilly et al., 2008), the Food Selectivity dimension was the most consistently represented in our ASD–W group (on average, 70% of the maximum possible score vs. 45.5% in the ASD–WO group and 43.5% in the TD group), but the other BAMBI dimensions are also considerably represented.
The MANOVA showed that the whole sample with ASD differed from the TD children in all dimensions of the BAMBI. Moreover, the ASD–W group differed from both the ASD–WO and TD groups. However, the ASD–WO and TD groups did not differ from each other (Table 2). These results are consistent with those of previous studies (Panerai et al., 2020; Prosperi et al., 2021). Additional information derived from our results is that children with ASD–WO and TD children did not differ from each other in the BAMBI dimensions. Therefore, it can be assumed that eating behaviors in children with ASD–WO are similar to those of their TD peers. This condition may represent a strength in children with ASD–WO.
Age did not determine any within-group differences; therefore, eating problems can be assumed to be very similar in both TD preschoolers and school-age children and those with ASD. However, because no longitudinal observations have been carried out, we can only hypothesize (but not know) that eating difficulties would remain the same over time in each child. Similarly, we cannot establish the level of impact of ASD severity on these results. A study by Pham et al. (2020) reported an association between ASD severity and eating difficulties, whereas other studies did not (Prosperi et al., 2017; Smith et al., 2020). Future studies are needed, with longitudinal observations and larger samples, representative of all severity levels, to better describe the trajectories of eating problems, the behavioral manifestations that are more or less permanent over time, and the role of treatments of specific eating problems in modifying these trajectories.
As for the sensory features of our sample, the MANOVA of the SEQ highlighted significant differences relating to the factors group (large effect size) and age (medium effect size). As reported in Table 2, no differences were found between children with ASD–W and those with ASD–WO on either the hypo- or the hyperresponsiveness dimensions. On the contrary, statistically significant differences were found between children with ASD–W and those with ASD–WO versus TD children on all of the SEQ dimensions, with large effect sizes especially for the social hyporesponsiveness dimension. These results are consistent with those of Ben-Sasson et al. (2019) and Feldman et al. (2020), who found significant differences between children with ASD and TD children in overreactivity, underreactivity, and sensory seeking.
With regard to the SSP results, children with ASD–W appeared to be more widely and markedly impaired than children with ASD–WO, thus confirming that eating behaviors in ASD are affected by multiple sensory experiences (Panerai et al., 2020).
Statistically significant differences were found between children with ASD–W and both children with ASD–WO and TD children (Table 2) in all dimensions of the SSP (with the exception of Movement Sensitivity). Our findings showed that people with ASD–W were more impaired in the Taste/Smell Sensitivity, Tactile Sensitivity, Underresponsive/Seeks Sensation, and Auditory Filtering dimensions. On the contrary, scores on the Low Energy/Weak and Visual/Auditory Sensitivity dimensions were not more marked for children with ASD–W, compared with the scores for children with ASD–WO, and statistically significant differences were found in a comparison with TD children only. The Taste/Smell Sensitivity dimension appeared to be peculiar of the ASD–W group, showing significant differences in comparison with both the ASD–WO and TD groups (conversely, the ASD–WO and TD groups did not differ from each other). Taste/Smell Sensitivity seemed to be the dimension that affects more strongly the eating behaviors, and in some studies (Bennetto et al., 2007; Boudjarane et al., 2017), a cortical dysfunction was hypothesized, with an involvement of the primary olfactory cortex (Baum et al., 2015) for threshold and smell identification and an involvement of the primary gustatory cortex for taste reactivity (Avery et al., 2018). However, these hypotheses need to be further investigated. The Tactile Sensitivity dimension is associated with food consistency and texture preference (Cappellotto & Olsen, 2021), dislike of certain foods, neophobia (defined as the avoidance or the reluctance to eat new foods), and pickiness (Farrow & Coulthard, 2012; Nederkoorn et al., 2015). The Auditory Filtering dimension might be related to the conviviality and social communication during meals. Also, the sounds of food or the eating environment may influence the child’s acceptance of certain foods. The Underresponsive/Seeks Sensation dimension might be related to inattention and a lack of motivation for engagement in the social relation during meals while being related, at the same time, to focusing on a search for more intense nonsocial stimuli.
The detailed analysis of the associations between the dimensions of eating behaviors and the sensory sensitivities revealed some differences between children with ASD and TD children (Tables 3 and 4). The discussion is limited only to correlations with a medium to large effect size (r = .30 and r ≥ .50, respectively; Cohen, 1988). Taken together, the results in Table 3 showed multiple correlations in both children with ASD and TD children. In children with ASD, the Social dimensions of Hyper- and Hyporesponsiveness were associated with Food Refusal (both) and Disruptive Mealtime Behaviors (only Hyperresponsiveness). Therefore, the presence of a social hyper- or hyporesponsiveness could represent a risk for the development of eating problems in children with ASD. Nonsocial Hyperresponsiveness was correlated with Mealtime Rigidity (with a medium-sized correlation), and with all the other BAMBI dimensions (with a small-sized correlation). A study by Crippa et al. (2022) found that higher sensory sensitivity to environmental stimuli was related to a greater risk for developing eating problems in children with higher levels of ASD features. In our study, on the basis of SEQ results, we succeeded in identifying Nonsocial Hyperresponsiveness as being specifically associated with Mealtime Rigidity in ASD. However, the association with the other dimensions of eating problems should be further investigated.
In TD children, associations between the BAMBI dimensions and Nonsocial responsiveness (both Nonsocial Hyper- and Hyporesponsiveness, but especially Hyperresponsiveness), as well as associations between the BAMBI dimensions and Social Hyperresponsiveness, were found. On the basis of the results obtained, it may be summarized that, in children with ASD, the two dimensions of Social Responsiveness were mainly associated with eating problems; therefore, the social aspect linked to mealtime seemed to be more involved with eating problems. On the contrary, in TD children, responsiveness to nonsocial stimuli seemed to be more involved.
Most of the past studies have focused on the Food Selectivity dimension. In contrast, this study has taken into consideration all the eating dimensions. In other studies, some associations were found between the Food Selectivity and Taste/Smell Sensitivity dimensions (Lane et al., 2014; Zulkifli & Rahman, 2021), Tactile Sensitivity (Cermak et al., 2010; Suarez, 2012; Zulkifli & Rahman, 2021), and Auditory Filtering (Zulkifli & Rahman, 2021). Our study confirmed a strong correlation between the Food Selectivity and Taste/Smell Sensitivity dimensions in children with ASD. Instead, the association between Food Selectivity and the other two sensory dimensions turned out to be weak. Food refusal, in previous studies, was associated with taste/smell (Cermak et al., 2010) and tactile (Zulkifli & Rahman, 2021) sensitivities, whereas, in our study, a strong association was found also with the Underresponsive/Seeks Sensation and Auditory Filtering dimensions. Moreover, Disruptive Mealtime Behaviors and Mealtime Rigidity were associated also with Taste/Smell Sensitivity, and the Disruptive Mealtime Behaviors dimension was also associated with Tactile Sensitivity, Auditory Filtering, and Underresponsive/Seeks Sensation (the disinterest in environmental stimuli or the exclusive search for only specific stimuli may be hypothesized as leading the child to a rejection of foods or to other disruptive behaviors, such as spitting out food or running away from the table). Overall, Taste/Smell Sensitivity appeared to be the main sensory dimension involved in eating problems, being associated with all BAMBI dimensions in children with ASD and with three out of four dimensions in TD children.
We found a significantly marked difference between children with ASD and TD children in the Auditory Filtering dimension: In children with ASD, it was associated with the BAMBI dimensions, especially with Disruptive Mealtime Behaviors and Food Refusal, whereas no association was found in TD children. Therefore, only in children with ASD did the Auditory Filtering seemed to mediate eating behaviors.
Taken together, the correlation analyses showed four main sensory dimensions associated with eating behaviors in ASD (Taste/Smell Sensitivity, Tactile Sensitivity, Underresponsive/Seeks Sensation, and Auditory Filtering), and three main sensory dimensions in TD children (Taste/Smell Sensitivity, Underresponsive/Seeks Sensation, and Movement Sensitivity). The Low Energy/Weak and Visual Auditory Sensitivity dimensions were not associated with any BAMBI dimensions, either in children with ASD or TD children. Although these two dimensions were impaired in children with ASD–W and those with ASD–WO (Table 2), they seemed to have no relationship with eating problems.
This study has some strengths and limitations. Among the strengths, we can consider the use of standardized tools and the clear identification of the ASD–W subgroup on the basis of a specific cutoff value. Another strength is the comparison of children with ASD–W and those with ASD–WO with TD controls, that, to the best of our knowledge, has not been investigated thus far. Finally, the multivariate analysis allowed us to thoroughly analyze the data relating to the global effects of a number of factors—namely, group, age, and the interaction between them—on the eating and sensory dimensions.
Our study also presented with some limitations to be taken into account: The sample consisted mostly of children with ASD who required substantial or very substantial support; therefore, our results are not generalizable to children with ASD with a lower level of severity. Therefore, a larger ASD group that is more representative of all severity levels would be greatly recommended. It would be also interesting to observe the trend of eating and sensory anomalies over time by including in the sample multiple age groups, from infancy to adulthood.
Second, in our study, no other clinical groups were included. In future studies, it would advisable to include, for example, peers with intellectual disability or other neurodevelopmental disorders, to increase the understanding of whether different clinical populations show different sensory and eating profiles.
Third, the parents of the children included in our study were administered three questionnaires; their answers may have been affected by a certain degree of subjectivity.
Finally, we believe that future studies should also focus on designing treatment models to examine the eating and sensory difficulties of people with ASD, and the possibility of modifying their eating and sensory trajectories, as well as guarantee stable improvements over time.
Implications for Occupational Therapy Practice
What kind of information can be drawn from our study and taken into consideration by occupational therapists in view of the treatment? Consider the following: ▪ Children with ASD–W present with a global impairment in eating behaviors and several impairments in sensory responsiveness, mainly, Taste/Smell Sensitivity (specific only to the ASD–W group), Tactile Sensitivity, Underresponsive/Seeks Sensation, and Auditory Filtering. These dimensions are also associated with all BAMBI subdomains. ▪ Children with ASD–WO and TD children do not differ from each other in eating behaviors. ▪ The whole sample with ASD presents with sensory hyper- and hyporesponsiveness. ▪ The eating and sensory behaviors of children with ASD–W and those with ASD–WO do not seem to change with age.
Given these findings, this study has the following implications for occupational therapy practice: ▪ Occupational therapists should be particularly scrupulous and accurate in the evaluation of eating and sensory behaviors of children with ASD, so as not to omit important elements that are essential for intervention planning. ▪ Given the strong association between eating and sensory behaviors in children with ASD, the intervention program must integrate all the sensory components, especially taste, smell, touch, and hearing. An environment enriched with sensory stimuli might become more pleasant and, consequently, reduce the automatic search for sensations. As a part of a multidisciplinary team, the occupational therapists may also contribute to support parents, providing behavioral strategies for improving child behaviors at mealtime.
Conclusion
Taken together, the results obtained in our study showed significant differences between the eating and sensory profiles of children with ASD and those of TD children—especially those of children with ASD–W, who appear to be the most impaired. Moreover, early treatment of eating difficulties seems to be the best choice, because those difficulties hardly resolve spontaneously. Early multidimensional, integrated treatments may avoid the chronicization, or prolongation, of impairments; reduce risks in the child’s development; and improve sensory processes; and, consequently, more adaptive responses during meals.
Footnotes
Acknowledgments
This study was funded by the Italian Ministry of Health (Ricerca Corrente 2022–2024). We acknowledge the valuable contribution of all families who joined the study. We are also grateful to the professional team of the social cooperative “I Corrieri dell’Oasi” (psychologists, pedagogues, professional educators, speech therapists, psychomotor therapists, rehabilitation therapists, occupational therapists, art teachers, and health care assistants) and to Rosa Di Giorgio for her support in editing and revising the manuscript of the article.