Abstract
Studies using parent-report measures have described the high prevalence of food selectivity in children with autism spectrum disorder (ASD). However, few studies have documented food acceptance in a controlled laboratory environment. The objective of this study was to compare laboratory food acceptance in children with ASD with that of children with typical development (TD). In addition, the relationships between food acceptance and the child’s age, sensory processing pattern, and autism severity were explored. Results indicate that children with autism (n = 31) accepted fewer foods in the laboratory environment than the children with TD (n = 21) and that food acceptance was related to age but not to ASD severity. In addition, sensory processing scores were associated with food acceptance for the combined ASD and TD groups. Results are discussed in the context of the literature. This information has the potential to support evaluation and treatment of food selectivity.
Food selectivity is common in children and disproportionately affects children with autism spectrum disorder (ASD). Estimates of feeding issues are as high as 25% in children who have typical development (TD; Manikam & Perman, 2000) and 70%–80% in children with ASD (Schreck & Williams, 2006; see also Bandini et al., 2010). Children with food selectivity refuse a large proportion of foods offered and restrict themselves to a very narrow repertoire of foods (Bandini et al., 2010). In addition, children with food selectivity often have negative physiological responses and disruptive behavior when presented with nonpreferred foods (Howe & Wang, 2013; Williams, Gibbons, & Schreck, 2005). As a result of food selectivity, families often experience decreased satisfaction with meals (Rogers, Magill-Evans, & Rempel, 2012; Suarez, Atchison, & Lagerwey, 2014), and children are at risk for nutrient inadequacies that can have long-term health implications (Pinhas-Hamiel & Zeitler, 2005; Suarez & Crinion, 2014; Urakami et al., 2009).
Several hypotheses have been drawn to explain the underlying mechanisms of food selectivity, particularly for children with ASD. For example, food selectivity may be tied to the age of the child (Beighley, Matson, Rieske, & Adams, 2013), to autism severity (Postorino et al., 2015), or to symptomatology characteristic of autism such as restricted and repetitive behaviors (Matson & Fodstad, 2009; Suarez, Nelson, & Curtis, 2014). In addition, a growing body of evidence indicates that food selectivity in children with autism may be linked to sensory processing disorder and, more specifically, to a pattern of behavioral overresponse to sensation (Cermak, Curtin, & Bandini, 2010; Smith, Roux, Naidoo, & Venter, 2005, Suarez, Nelson, & Curtis, 2014). It is possible that children with ASD and food selectivity have an overresponse, referred to as a low threshold, for sensation that leads them to be very sensitive and sometimes avoidant of typically innocuous sensory experiences that cause them discomfort (Dunn, 2014). Children with food selectivity may choose “safe” foods that they do not anticipate being a tactile, taste, or smell challenge to their very sensitive systems and may avoid foods they suspect may cause discomfort. Although evidence of the link between a low threshold for sensation (i.e., hyperreactivity, defensiveness, overresponsivity) and food selectivity in children with autism is growing, continuing to develop an understanding of possible factors underlying food selectivity could lead to more refined and effective treatment.
The vast majority of research on food selectivity has been done using parent-report measures related to eating behaviors in the home (Bandini et al., 2010; Schreck, Williams, & Smith, 2004; Williams et al., 2005). Only one study that included direct observation of food acceptance in children with autism and food selectivity was identified: Ahearn, Castine, Nault, and Green (2001) exposed 30 children with autism to 12 different foods and found that few had a high level of food acceptance (measured in number of bites accepted). However, this study did not include a comparison group of typically developing children or evaluate any of the factors, such as age, autism severity, or sensory processing differences, that may have been related to food acceptance. There is a need to evaluate and compare the actual food acceptance of children with ASD with the food acceptance of typically developing children using direct observation to supplement information from parent report.
In this study, I evaluated food acceptance in a laboratory environment in children with ASD compared with typically developing children. In addition, I sought to place food acceptance behavior into the context of each child’s sensory processing patterns. The research questions were as follows:
Is there a difference in food acceptance between children with an ASD diagnosis and children with TD?
Do children with ASD eat the same number of fruits, vegetables, dairy items, proteins, and snacks as children with TD?
Is age or autism severity related to number of foods accepted in children with ASD?
Are sensory processing scores, and specifically items related to a low threshold for sensation, associated with food acceptance?
Method
This study was part of a larger laboratory study of food acceptance in the Western Michigan University (WMU) Brain Research and Interdisciplinary Neurosciences lab. The first phase of the study involved completion by parents of an online questionnaire. At the conclusion of this questionnaire, parents were given a pictorial description of the laboratory portion of the study to allow them a greater understanding of what was involved and whether their child would be a good candidate. Then, child participants were invited to the lab and asked to “take a bite” of 16 different foods presented one at a time. Foods presented were chosen from among foods identified as commonly offered to and eaten by children with autism in the United States (Suarez & Crinion, 2014).
Foods offered represented different food groups and included four fruits (e.g., apple), four vegetables (e.g., carrot), three dairy items (e.g., yogurt), two proteins (i.e., chicken strip, hot dog), two snack items (e.g., cracker), and one unusual food (spicy chickpea). Children with a gluten-free diet were given a gluten-free substitute for the cracker items. During the food presentation, children were given no other verbal encouragement to eat the food. Food acceptance was documented, and documentation was validated immediately after each session using a video recording of the session. Foods were presented one at a time approximately 8 s after the child refused or swallowed the previously offered food. The WMU institutional review board approved the study protocol.
Participants
Fifty-two participants (31 with ASD, 21 with TD) ages 4–14 yr completed the study. Study participants with an ASD diagnosis were recruited through local clinics and autism support groups. Children were included in the ASD group if their parents reported that their child had a diagnosis of autism from a reputable source (i.e., school system [Wu et al., 2016], neuropsychologist) and this diagnosis was confirmed with the Social Responsiveness Scale (SRS; Constantino & Gruber, 2005). Children in the TD group were recruited through word of mouth and snowball sampling. These children were included in the study if their parents denied any diagnosis that affected learning or behavior.
Instrumentation
The caregivers of each child participant filled out a demographic questionnaire that contained questions including the child’s date of birth and gender. In addition, several other instruments were used to collect autism and sensory information.
Social Responsiveness Scale.
The SRS was used to confirm the diagnosis of the ASD group and to evaluate the relationship between autism severity and food selectivity. The SRS is a questionnaire that contains 65 items to evaluate the autism-related behaviors of children ages 4–18 yr. This instrument has been found to have substantial agreement (.70) with the Autism Diagnostic Interview–Revised (Constantino et al., 2003). In addition, interrater reliability for the SRS was high (.80). This instrument was found to be a valid and feasible measure of autistic traits (Duvall et al., 2007).
Short Sensory Profile.
The Short Sensory Profile (SSP; Dunn, 1999) was used to measure sensory processing; this 38-item parent-report questionnaire is a shorter version of the original Sensory Profile (Dunn, 1999). The SSP contains a summary that is used for interpretation of the measure. This summary details results (total scores) from the individual profile sections that allow determination of patterns of sensory processing in respondents. The profile sections include tactile sensitivity, taste/smell sensitivity, movement sensitivity, underresponsive/seeks sensation, auditory filtering, low energy/weak, and visual/auditory sensitivity. This instrument has been validated for measuring sensory processing (Dunn, 1999) and has been used widely with children who have ASD (Ben-Sasson et al., 2009). It has high internal validity (.90–.95; Dunn, 1999) and identifies differences in approximately 90% of children with ASD compared with control children (Dunn, Myles, & Orr, 2002). Also, items on the scale have been used to measure the presence of a low threshold for sensation (Mazurek et al., 2012; McIntosh, Miller, Shyu, & Hagerman, 1999). For the purpose of this study, the tactile sensitivity, taste/smell sensitivity, movement sensitivity, and visual/auditory sensitivity section totals were added together to quantify low threshold.
Laboratory Food Acceptance.
A data observation sheet to record laboratory food acceptance was prepared with foods randomly ordered for each child. After each food was presented, an observer marked whether the child accepted the food. Accepted was defined as independently putting the food in the mouth and swallowing, and refused was defined as not swallowing the food. Foods accepted were tallied for each child.
Analysis
Descriptive statistics were used to capture the sample demographics. Differences between the groups of children with ASD and children with TD were explored using independent t tests. Pearson’s correlations were used to examine relationships among age in months, autism severity, sensory processing, and food acceptance.
Results
Sample Characteristics
The mean age of the children with ASD was 9.03 yr (standard deviation [SD] = 3.21) and that of the children in the TD group, 8.91 yr (SD = 3.19). The difference between the ages of the two groups was not significant, t(52) = 0.204, p = .839. Of the 31 children with ASD, 3% were Black, 82% were White and non-Hispanic, 9% were Hispanic, and 6% did not identify their racial or ethnic background; 90% were male. Of the 21 children in the TD group, 20% were Black, 60% were White and non-Hispanic, 10% were Hispanic, and 10% did not identify their racial or ethnic background; 55% were male. On the SRS, all the children whose parents reported that they had ASD met the autism threshold, with 7% in the moderate group and 50% in the severe group.
Laboratory Food Acceptance
First, I explored the relationship between number of laboratory foods accepted and child’s age in months. There was a significant relationship between foods accepted and age in the children with ASD, r = .528, p = .002 (two tailed), but not in the TD group, r = .150, p = .529 (two tailed). Next, I compared the number of foods accepted by the children with ASD and the children in the TD group, including the number of fruits, vegetables, dairy items, proteins, and snack items. Table 1 contains the means and SDs for each of these items as well as results of the independent t tests. Levene’s test for equality of variances (Portney & Watkins, 2015) was significant for all foods tested. Therefore, values for equal variances not assumed are reported. Children with an ASD diagnosis accepted significantly fewer total foods and significantly fewer foods in each category with the exception of snack foods.
Number of Foods Accepted by Children With ASD (n = 31) Compared With Children With TD (n = 21)
Note. ASD = autism spectrum disorder; M = mean; SD = standard deviation; TD = typical development.
Autism Severity and Sensory Processing
Children whose parents reported their child had autism completed the SRS. The raw scores (higher scores indicate greater autism severity) for each child were correlated with the number of foods the child accepted in the laboratory. The correlation between SRS raw score and number of foods accepted was not significant, r = −.144, p = .44.
Next, SSP items were correlated with the number of foods accepted for the combined ASD + TD groups. SSP items included the low threshold items and the underresponsive/seeks sensation, auditory filtering, and low energy/weak items, as well as the total SSP score. On the SSP, lower scores indicate more sensory processing differences, with cutoff values for children with definite difference, probable difference, and typical performance compared with a normative sample of typically developing children. Table 2 contains the r and p values and means and SDs for this analysis. A significant relationship was found between all SSP items and number of foods accepted in the lab for all items except the low energy/weak items. As the number of foods a child accepted in the lab declined, the trend toward greater sensory processing difference increased. Because of lack of sufficient power (Suarez, Nelson, & Curtis, 2012), this relationship was not explored in the separate ASD and TD groups.
Relationships Between Number of Foods Accepted and Short Sensory Profile Scores for Combined ASD + TD Groups (N = 52)
Note. ASD = autism spectrum disorder; M = mean; SD = standard deviation; TD = typical development.
Discussion
This study compared the food acceptance of children with ASD with that of children with TD in a laboratory setting. In addition, the relationships between food acceptance and age, autism severity, and sensory processing patterns were explored. Children in the ASD group were observed to eat significantly fewer foods in total and in all food categories except snack foods than TD children. A significant relationship was found between foods accepted and age in the ASD group only, and no relationship was found between foods accepted and severity of autism symptoms in this group. Significant relationships were found between all SSP items, except low energy/weak items, and foods accepted in the combined ASD + TD groups.
It is no surprise that children with ASD ate significantly fewer foods in the laboratory setting than TD children. Many studies using parent-report measures have documented that children with ASD have food selectivity at much greater rates than children with TD (Bandini et al., 2010; Curtin et al., 2015; Schreck et al., 2004). However, few studies have used direct observation to compare food acceptance behaviors of children with ASD with those of children with TD. Evaluation of food refusal behavior in a laboratory setting has the advantages of eliminating possible parent bias and providing a single controlled setting in which to offer food in a standardized way. This testing environment allows researchers to rule out the influence of confounding characteristics of the family’s mealtime environment and interaction for a more objective measurement of food acceptance behavior.
In this laboratory study, food acceptance was related to age in children with ASD but not in children with TD. This finding of a relationship between age and food acceptance in children with ASD contrasts with those of other studies that have refuted this link (Suarez, Nelson, & Curtis, 2012, 2014) and may be explained, in part, by the nature of participation in a laboratory experiment. Children with ASD have difficulty with changes in routine, and in this study, they needed to manage not only to eat food but also to eat in an unfamiliar environment with unfamiliar people. The connection I found between food acceptance and age in children with ASD may relate to an increased ability to manage new, nonroutine experiences and the possibility of greater food acceptance as they age.
Food acceptance and severity of ASD symptoms were not related. This finding contrasts with that of a study by Postorino and colleagues (2015) that demonstrated significantly higher scores on the SRS for a group of children with food selectivity and ASD compared with a group of children with ASD alone. The difference in findings between that study and the current one may be accounted for by the different method of classification of food selectivity. In the current study, the number of foods accepted in the laboratory protocol was directly compared with scores on the SRS. In Postorino and colleagues’ study, food selectivity was a categorical classification, and children whose parents reported that their child had “food refusal” and “high-frequency single-food intake” were included in the food selectivity group. The need is clear for a valid and reliable instrument that provides universal diagnostic criteria for food selectivity so comparisons between studies can be made to advance the evidence base.
On the basis of a review of previous research, I expected that food acceptance for children with ASD would be significantly associated with sensory processing differences, specifically those reflected in scores on the low-threshold items of the SSP (Cermak et al., 2010; Suarez, Nelson, & Curtis, 2014). In this study, this expectation held true, even when children with TD were included in the analysis. The number of foods accepted for the total group (ASD + TD) was significantly associated with the low threshold, underresponsive/seeks sensation, auditory filtering, and total SSP items. A larger sample size would be needed to investigate these relationships separately in the ASD and TD groups, an important direction for future research. It would also be interesting to learn whether children with TD and food selectivity also have differences in sensory processing.
Limitations and Directions for Future Research
This study has several limitations. It lacked adequate power to allow investigation of potential associations between food acceptance and sensory processing differences in the ASD and TD groups separately. The findings are valuable for informing sample sizes in future studies and point to the need to look more closely at the relationship between sensory processing and food selectivity in children with TD.
Another limitation of this study is that the children with TD were not screened for ASD. However, care was taken to document parent denial of any diagnosis that would influence learning and behavior, and children who had such diagnoses were excluded from the study. In addition, the parents of children tested in the lab were not required to withhold a meal until after participation, and the foods presented may or may not have been part of each child’s regular diet. These factors could have influenced willingness to eat the foods presented.
Finally, children with ASD have difficulty with novel experiences and changes in routine. For this reason, eating in the lab may have affected them more than children with TD. These issues should be addressed in future studies.
Implications for Occupational Therapy Practice
Occupational therapy practitioners play a key role in the treatment of mealtime dysfunction, including food selectivity, in children with and without comorbid developmental disabilities. These findings have the following implications for clinical practice in this area:
This study replicates the finding of decreased food acceptance in children with ASD compared with those with TD in a controlled laboratory setting.
Practitioners need to routinely screen for food selectivity in children with ASD. In addition, sensory processing differences may be a factor related to food selectivity in children without developmental disabilities. This potential relationship requires further investigation.
Measures of sensory processing should be included in assessment of food selectivity for all groups of children who experience food refusal.
Footnotes
Acknowledgments
This work was supported by a grant from the Faculty Research and Creative Activities Award, Western Michigan University. I thank the families who participated in this study and the many occupational therapy students who assisted with recruitment and data collection.