Abstract
The objective of this study was to examine food selectivity in children with autism spectrum disorders longitudinally. Additionally explored were the stability of the relationship between food selectivity and sensory over-responsivity from time 1 to time 2 and the association between food selectivity and restricted and repetitive behavior at time 2. A total of 52 parents of children with autism were surveyed approximately 20 months after completing an initial questionnaire. First and second surveys each contained identical parent-response item to categorize food selectivity level and a scale to measure sensory over-responsivity. A new scale to measure restricted and repetitive behaviors was added at time 2. Results comparing time 1 to time 2 indicated no change in food selectivity level and a stable, significant relationship between food selectivity and sensory over-responsivity. The measure of restrictive and repetitive behavior (time 2) was found to significantly predict membership in the severe food selectivity group. However, when sensory over-responsivity and both restricted and repetitive behaviors were included in the regression model, only sensory over-responsivity significantly predicted severe food selectivity. These results support conclusions about the chronicity of food selectivity in young children with autism and the consistent relationship between food selectivity and sensory over-responsivity.
Keywords
Introduction
Food selectivity is a problem in a majority of children with autism spectrum disorders (ASD) (Schreck and Williams, 2006), which can have serious nutrition (Sharp et al., 2013; Zimmer et al., 2011) and family quality of life implications (Rogers et al., 2011; Suarez et al., in press). Despite the prevalence of food selectivity in this population, little is known about possible causative factors associated with this problem and whether children with ASD outgrow food refusal behavior. Greater understanding of food selectivity is needed in order to inform treatment protocols that target this issue and aim to ameliorate the negative nutritional and quality of life consequences for children with ASD and their families.
Food selectivity has been defined as the acceptance of a limited number of foods as part of the child’s regular diet (Bandini et al., 2010). Some estimates of the co-occurrence of food selectivity and autism range as high as 72% (Schreck and Williams, 2006). A critical review of empirical research on food selectivity in children with autism by Matson and Fodstad (2009) led the authors to conclude that feeding problems in children with ASD are frequent, under-recognized, and in need of further research.
The consequences of food selectivity for children with ASD and their families can be significant. Some children with autism eat as few as five foods (Suarez et al., 2012) and some refuse whole food groups (Sharp et al., 2013). This can have serious nutritional implications (Dovey et al., 2008; Zimmer et al., 2011). For example, a meta-analysis by Sharp et al. (2013) found children with ASD were at significant risk of serious nutritional deficits (e.g. lower intake of protein and calcium). Children with autism have been found to refuse to eat fruits and vegetables (Bandini et al., 2010) and may show preference for foods that are “fatty” or sugary” (Cornish, 1998). This lends credence to the emerging evidence for greater rates of obesity in children with ASD (Egan et al., 2013). In general, it is reasonable to conclude that food selectivity can impact nutrition, and potentially health, for children with ASD.
Food selectivity in children with ASD also has implications for quality of life in families. The communication, social, and restrictive and repetitive behavior (RRB) deficits that are hallmarks of ASD (American Psychiatric Association (APA), 2000) make parenting a child with ASD profoundly challenging (e.g. Lyons et al., 2010; Smith et al., 2010). In some children with ASD who also have food selectivity, these challenges are compounded by mealtime behavior that further disrupts the family routine. In a qualitative study by Suarez et al. (in press) mothers of children with autism and food selectivity described the stress they experienced at mealtime when their child had a severe negative behavioral reaction to food and would not sit peacefully at the table. Food selectivity can decrease mealtime satisfaction (Fulkerson et al., 2008) adding to the difficulty of raising a child with ASD.
Despite the negative implications of food selectivity in children with ASD, little is known about the progression of food selectivity over time. A review of the literature showed that to date, food selectivity in children with ASD has not been studied longitudinally to determine if these children add foods naturally to their diets as they age. However, several authors have completed cross-sectional work to look at the association between age of the child and food selectivity that shed doubt on the expectation that food selectivity will lessen with age. For example, using parent questionnaires to compare the food acceptance of children with autism compared to typically developing children, Bandini et al. (2010) found that the age of the child was not significantly correlated with having a limited food repertoire in either group. The authors concluded that the commonly held belief that children would outgrow their food selectivity was not supported by their cross-sectional findings. In the study preceding this longitudinal follow-up (Suarez et al., 2012), age was not a significant predictor of food selectivity level in the perceptions of caregiver participants surveyed. The next step, which is addressed in the current longitudinal study, is to look at changes in food selectivity over time in the same children.
An additional missing piece in developing a full understanding of food selectivity in children with ASD is identifying associated factors that distinguish children with food selectivity from those without. This information may eventually lead to the identification of causes of food selectivity and can inform treatment. Two factors that have been suggested in association with food selectivity among children with ASD are sensory over-responsivity (SOR) (Cermak et al., 2010; Suarez et al., 2012) and RRBs (Schreck et al., 2004; Zimmer et al., 2012).
Sensory over-responsivity (SOR) is a subcategory of sensory processing disorder. It is defined as an extreme over-reaction to sensation from any of the seven sensory systems (i.e. tactile, vestibular, auditory, proprioceptive, gustatory, olfactory, and visual) (Miller et al., 2007). The presence of sensory processing disorders in some children in the autism population, and SOR specifically, is generally accepted (e.g. Ben-Sasson et al., 2007; Kern et al., 2006, 2007). In the initial study preceding this longitudinal follow-up, Suarez et al. (2012) found that children with severe food selectivity (i.e. those who accepted less than 10 foods) and moderate food selectivity (i.e. those who accepted 11–20 foods) had significantly higher scores on a measure of SOR than children who accepted 21 or more foods. As others have suggested (Cermak et al., 2010), the hypothesis driving this study was that a pattern of SOR may contribute to hypersensitivity to food textures, resulting in food selectivity. Thus, the current study was designed to examine the stability of the relationship between SOR and food selectivity over time contributing to understanding of possible factors that may underlie food selectivity.
RRB also has been hypothesized to be a possible cause of food selectivity among children with autism (Schreck et al., 2004; Zimmer et al., 2012); however, to the authors’ knowledge, this has yet to be explored empirically. In anecdotal support of this hypothesis, Schreck et al. (2004) noted that children with autism often have been described as requiring a particular presentation of food items. For example, forms of RRB related to mealtime behavior include needing to use the same utensil or dish for every meal, requiring that food be prepared in a particular way, or insisting on food being a certain brand. RRB may be related to food selectivity in that the insistence on sameness or consistency of the texture or taste of a food is a primary factor that contributes to the child’s willingness to accept a particular food. For example, a processed cheese slice will be the same texture and taste every time it is unwrapped and eaten, whereas a slice of cantaloupe may be different depending on the season or grower. Children with higher scores on a measure of RRB might be more likely to have severe levels of food selectivity. This hypothesis was tested empirically in this research.
Some prior researchers have identified a significant association between SOR and RRB, suggesting a need to consider both variables in investigations of food selectivity. In a prior study that used parent questionnaires to measure the relationship between report of symptoms of sensory processing disorder and RRB, Boyd et al. (2009) found that the Sensory Questionnaire composite score was correlated moderately with stereotypy and compulsions on the RRB measure. Of note, these researchers did not look specifically at SOR but at the larger, general category of sensory processing disorder. In a series of two related studies, Joosten et al. (2009) and Joosten and Bundy (2010) used parent and teacher rating scales to explore the relationship between SOR, RRB, and anxiety symptoms in children with autism. First, using a motivation assessment scale, they determined that children with autism and intellectual disability were more likely to engage in stereotypical and repetitive behavior to alleviate anxiety and escape a situation than children who exhibited intellectual disability alone (Joosten et al., 2009). Then in a second paper, Joosten and Bundy (2010) added a measure of sensory processing. They found that children with ASD and intellectual disability, who used stereotypical and repetitive behavior (i.e. RRB) in the first study to alleviate anxiety, were significantly more over-responsive to sensation and had more extreme sensation avoidance behaviors than children with intellectual disability alone. The authors concluded that low thresholds for sensation (i.e. SOR) may contribute to increased anxiety, which children attempt to alleviate through stereotyped and repetitive behavior.
It is possible that the severity of RRB in a child with ASD is related to food selectivity largely through its relationship to higher levels of SOR. In theory, this form of food selectivity would be aimed at relieving the discomfort associated with the experience of adverse sensation. Consistent with this theory, the child’s discomfort with the sensory experience of eating and resulting anxiety could predict levels of food selectivity in the child. This posited link between sensory discomfort and food selectivity may parallel the cause for feeding difficulty in some children without autism, where pathology causes pain during feeding (e.g. allergies, gastroesophageal reflux disease (GERD)) often leading to food refusal (Morris and Klein, 2000).
In order to gain a more complete understanding of food selectivity and ASD, this study was designed to explore the relationships between SOR, RRB, and food selectivity. It was conducted by asking parents to respond to questions regarding the three constructs for their children with autism at a second point in time. That is, this study is the second in a series with the same participants from a study reported by Suarez et al. (2012). This longitudinal design made it possible to explore how parents perceived food selectivity for their child with ASD at two points in time and to contribute to answering questions about whether and how much such children’s food choices might change during their early years.
Two primary aims of this study were: (1) to examine whether young children with ASD previously shown to exhibit food selectivity maintain their food selectivity level as they age, or alternatively eat more or fewer foods, and (2) to evaluate the stability of the relationship between food selectivity and SOR scores over time. By adding the measure of RRB, the contribution of this factor to the prediction of severe food selectivity also could be examined.
Methods
This study used an electronic survey methodology to gather the perceptions of the caregivers of children with ASD who participated in the earlier survey (Suarez et al., 2012). The earlier survey included a question about the child’s food acceptance and a 19-item scale to measure SOR. It also included an opportunity for parents to provide an email address if they would be willing to be contacted for a follow-up survey. The survey for this study (time 2) contained 65 questions, including the same question used to characterize the child’s food selectivity level, the 19-item scale to measure SOR, and an additional 43-item scale to measure RRB. The first sample was obtained through the Autism Speaks Interactive Research Network, which emailed a participation invitation to randomly selected parents of children with autism in their database. A total of 141 parents responded to the first survey. Their responses are described by Suarez et al. (2012). A subgroup of the first sample (n = 114) indicated that they would be willing to be contacted again and provided an email address. This group of 114 parents was contacted again 20 months later, and 52 responded to the invitation to complete the second survey. Human Subjects Institutional Review Board at Western Michigan University approved the study protocol for the initial and follow-up survey.
Instrumentation
Demographic information was gathered during the original study (Suarez et al., 2012), and participants for the current study (time 2) were asked to provide their names to ensure the repeated measures match to the original study (time 1) data. Demographic data included the gender, education level, race, and ethnicity of the caregiver. Information about the gender and age of the child was also collected. Parents were asked to indicate whether their children had been diagnosed with ASD, and, in addition, to provide the source of the diagnosis (e.g. pediatrician, psychologist, or interdisciplinary team).
Defining food selectivity—Parent Questionnaire
Identical to the first survey, parents were asked to select a category representing how many foods their child accepts as part of his or her regular diet (i.e. less than 5, 6–10, 11–20, 21–30, and 30+). Classification of food selectivity level, as operationalized in the original research (Suarez et al., 2012), was defined categorically (i.e. severely selective as acceptance of less than 10 foods, moderately selective as acceptance of 10–20 different foods, and typically selective as acceptance of 21 or more different foods). This operationalization was supported by results of the study by Cornish (1998), who analyzed nutritional adequacy at these three levels and found that children eating less than 20 different foods are at risk for nutritional deficits; Cornish also identified severe nutritional concerns for children who eat less than 8 foods. In addition to categorization of food selectivity, the parent participants were asked if their children had ever received feeding treatment either at time 1 or time 2 or between them. This question was asked to identify whether any changes in food selectivity possibly could be attributed to treatment.
Defining SOR scale
As developed for the original study (Suarez et al., 2012), the SOR scale was based on cross-referencing of items from two clinical respected sources (i.e. Dunn, 1999; Miller, 2006). This scale incorporated 19 items, which asked about examples of tactile, visual, auditory, and vestibular processing problems.
RRBs Scale
A measurement of RRB was added to study 2 using the Repetitive Behaviors Scale–Revised (RBS-R; Bodfish et al., 1999). This rating scale, which is based on informant report, has six subscales that measure the type and severity of restrictive and repetitive behaviors (i.e. stereotyped, self-injurious, compulsive, ritualistic, sameness, and restricted behaviors). Scores may range from 0 (no dysfunction) to 129 (highest level of restrictive and repetitive behaviors). Several authors have validated this scale for use with the ASD population (Lam and Aman, 2007; Mirenda et al., 2010).
Data analysis
Statistical analysis included a Wilcoxon signed-rank test to examine changes in parental reports of their children’s food selectivity from time 1 to time 2. Using a one-way analysis of variance (ANOVA), differences in SOR scores of the three food selectivity levels were examined. Logistic binary regression was used to determine if SOR scores and RRB scores could predict unique variance in severe food selectivity when entered into the same model. Finally, logistic binary regression was used to see if SOR scores at time 1 could predict membership in the severe food selectivity level at time 2.
Results
Of the 141 respondents to the first survey (time one), 52 completed the second survey as well. This was a response rate from the original sample of 37%. Of the 52 participants who filled out this follow-up survey, 29.1% reported that their child ate less than 10 foods as part of their regular diet (defined as severe food selectivity), 36.4% reported that their child ate 11–20 foods (defined as moderate selectivity), and 34.5% reported that their child ate 21+ foods (defined as typical selectivity). Table 1 describes the demographic characteristics of participants at each food selectivity level at time 2. As in the original study (Suarez et al., 2012), the subsample in this follow-up survey was mostly Caucasian (96%) and the children with autism were mostly boys (88%).
Description of participants by food selectivity level.
ASD: autism spectrum disorder.
The amount of time between the initial survey and this longitudinal follow-up ranged from 11–21 months with a mean of 20.24 months. At time 1, the age of the children ranged from 41 to 107 months, with a mean of 79 months. Consistent with the time between surveys, the chronological ages of the children at time 2 ranged from 55 to 128 months, with a mean of 100 months. Similar to study 1 (Suarez et al., 2012), using ANOVA, there were no significant differences in the age of the child depending on the child’s membership in one of the three food selectivity categories, F(2, 51) = .045, p = .956 (see means and standard deviations (SDs) in Table 2).
Food selectivity level at times 1 and 2 with corresponding SOR scores.
SOR: sensory over-responsivity; ASD: autism spectrum disorder.
Changes in food selectivity level from time 1 to time 2 were examined as well. Table 2 indicates the parent report of percentage of children who accepted the same, more, or less foods from time 1 to time 2, with corresponding SOR scores at each time period. The results showed that the majority of children fell into the same food selectivity category at times 1 and 2. The percentages of children who were rated (at time 1–time 2) in severe–severe categories (21.8%), moderate–moderate categories (16.4%), and typical–typical categories (21.9%) totaled to 60.1% of the total sample. In all, 20% of parents surveyed selected a food-range choice that indicated that their child accepted fewer foods at time 2 than at time 1 (moderate–severe 5.5%, typical–moderate 12.7%, and typical–severe 1.8%), and only 12.8% selected a choice that indicated that their child accepted more foods at time 2 than at time 1 (severe–moderate 7.3% and moderate–typical 5.5%). Using the Wilcoxon signed-rank test, there was no significant difference in food selectivity level between time 1 and time 2 (p = .275)—approximately 20 months later. During this time period, only 2 of the 52 caregivers reported that their child had received treatment for feeding issues. One of the two participants who had received treatment changed from being rated as having severe to moderate food selectivity; the other started in severe and stayed in severe.
Next, the SOR scores reported at time 1 and time 2 were tested for significant differences using a paired t-test. The overall mean SOR score from time one for participants in the longitudinal sample was 50.38 (range: 20–92), and the mean SOR for time two was 47.58 (range: 20–93). Higher SOR scores equate to parent report of more symptoms related to SOR. Using paired samples t-test, there was not a significant change in SOR scores from time 1 to time 2, t(54) = 1.85, p = .07.
The stability of the association between SOR scores and food selectivity level was examined next using ANOVA. Figure 1 illustrates the SOR scores at times 1 and 2 for children as a function of how their parents reported their food selectivity at time 2 (i.e. in severe, moderate, or typical food selectivity groups). Table 3 contains the SOR means and standard deviations by food selectivity category at time 2. Similar to the initial study, there was a significant difference in SOR scores between children with different food selectivity levels, F(2, 54) = 9.81, p < .001. Specifically, Bonferroni post hoc testing revealed that SOR scores for children classified as having severe selectivity at time two were significantly higher than for children who were classified as having moderate or typical selectivity (severe and moderate: p = .045; severe and typical: p <.001), but those classified as showing moderate selectivity at time 2 were not significantly different than those classified as typical (p = .135).
SOR scores for each food selectivity level at time 2.
Summary of means, standard deviations, and confidence intervals for SOR and RBS-R scores at each food selectivity level.
CI = confidence interval; SOR: sensory over-responsivity; RBS-R: Repetitive Behaviors Scale–Revised.
A new variable included in the longitudinal follow-up study, which was measured only at time two and not at time one, was RRB. Means and standard deviations for the RBS-R are shown in Table 3. The RBS-R was found to differ significantly for children based on their food selectivity levels at time 2, F(2, 54) = 425, p = .019. Specifically, Bonferroni post hoc analysis showed that children with severe food selectivity had significantly higher scores (indicative of greater RRB) than children who were typically selective (severe and moderate: p = .383; severe and typical: p = .016; moderate and typical: p = .446).
Because there were significant differences in food selectivity level associated with both SOR and RBS-R scores in separate bivariate analyses, both were entered into a logistic regression model to determine if each contributed significantly and uniquely to being categorized as showing severe food selectivity (vs moderate or typical) and to examine the interaction between the two constructs (SOR and RRB). When SOR and RBS-R scores were placed in the regression model separately, both were able to significantly predict severe food selectivity (SOR: p = .001; RBS-R: p = .020). However, when entered into the model together, only the SOR scores could significantly predict membership in the severe food selectivity category (SOR: p = .022; RBS-R: p = .782).
Finally, logistic regression was used to analyze whether SOR scores at time one, approximately 20 months earlier, could predict membership in the severe food selectivity category at time two. This analysis also showed a significant relationship (ExpB = 1.82, p = .012). Thus, for every 10-point increase in the original SOR score, the odds were 1.82 greater that the child would eat less than 10 foods as part of his or her regular diet almost 2 years later.
Discussion
This study was a longitudinal investigation of changes in food selectivity in children with ASD as reported by their parents at two points in time, approximately 20 months apart. A primary question was whether the significant relationship between food selectivity and SOR that was found at time 1 (with higher SOR associated with moderate and severe food selectivity) (Suarez et al., 2012) would be maintained at time 2. In this follow-up study, an additional construct, RRB, was examined to investigate whether it might also be associated with food selectivity, as children with ASD often are particular about different aspects of mealtime (e.g. need to use a particular plate or only eat particular colored foods) (Schreck and Williams, 2006). Food selectivity could represent simply another example of RRB.
One question was whether food selectivity could be expected to decrease simply as a function of age. The results indicate that the age of the child is not significantly associated with the child’s membership in food selectivity level. Few children (n = 8; 12.8%) increased the number of foods that they ate over the 20-month study period, and there was not a significant difference in food selectivity level from time 1 to time 2. If all the children were toddlers or of preschool-age at both time points, one could attribute this to their young age. However, the children in this study showed a wide range of ages. At time 1, the children’s ages ranged from 41 to 107 months (3;5 (years;months) to 8;11, mean 6;7); at time 2, they ranged from 55 to 128 months (4;7 to 10;8, mean 8;4). This suggests that when children with ASD demonstrate food selectivity, it is likely to be a persistent issue that is unlikely to get better simply over time.
These findings provide evidence that food selectivity in children with ASD is a chronic problem that is unlikely to change, at least without treatment. In line with the findings of no change over an almost 2-year period that could be associated with age, several cross-sectional studies of children with autism (Bandini et al., 2010; Suarez et al., 2012) have provided evidence showing a lack of relationship between food selectivity and the age of the child. This appears to be the first longitudinal study, however, to look at this relationship in the same children over time. This study points to the possibility that food selectivity is not something that children between the ages of 3 and 11 years will likely outgrow with the passage of time alone. Treatment of food selectivity and associated factors may be necessary for children to add new foods to their diet. Efficacy of treatment for food selectivity is an important direction for future research.
Only two parents reported that their children received any feeding treatment over the period of time between study one and study 2. One of these children went from severe to moderate food selectivity and one started in severe and stayed in severe. The type, frequency, and duration of treatment in these cases are unknown. The fact that only two of the 16 children whose parents report they eat less than 10 foods as part of their regular diet suggests that, as Matson and Fodstad (2009) concluded based on their literature review, food selectivity may be an under-recognized and under-treated component of the autism spectrum. This may also be important to note due to the possible nutritional implications of this behavior (Cornish, 1998; Zimmer et al., 2012), with health implications for children with ASD. Some evidence in the literature supports that interdisciplinary feeding programs, including behavioral components, can increase dietary diversity in children with autism (Laud et al., 2009).
In this longitudinal study, SOR scores also did not change significantly from time 1 to time 2. Similar to study 1 (Suarez et al., 2012), there was a significant difference in SOR scores for children in the severe and typical food selectivity groups; however, in contrast to study 1, moderate and typical food selectivity levels did not have significantly different SOR scores. Finally, SOR scores from time 1 could significantly predict membership in severe food selectivity category 20 months later at time 2.
Although this was not a treatment study, understanding factors associated with food selectivity may lead to more effective treatment protocols. The significant relationship between SOR scores and food selectivity, in which higher levels of over-responsivity are associated with fewer foods accepted, appears to be stable over time. Also, that severe food selectivity at time 1 could be predicted by SOR scores 20 months earlier highlights the fact that children with behaviors indicative of SOR should be screened for food selectivity. Early identification of food selectivity may allow for families to receive support for this issue.
A second variable that has been hypothesized in the literature to be related to food selectivity is RRB. In this study, we found significant differences between RRB-R scores in the severe and typical selectivity and severe and moderate selectivity groups but not in the moderate and typical selectivity groups. Also, in this study, scores on a scale for measuring RRB predicted membership in the severe food selectivity group when entered alone in a regression model. However, this hypothesis was not supported when both the RRB and the SOR scores were entered into the same model. In this case, the variance in the RRB score could be explained by the child’s SOR score, suggesting a close relationship between these two factors, with prominence for the SOR variable and no significant unique variance contributed by the RRB to food selectivity classification. This may be evidence supporting the relationship between SOR and RRB, as proposed by Joosten and Bundy (2010). Consistent with this theoretical explanation, it is possible that the presence of SOR in children with autism leads to discomfort with many sensory experiences, including eating. One potential explanation is that this discomfort may lead to the adoption of RRB in an attempt to alleviate anxiety due to the disturbing sensations associated with SOR. In relationship to food, children may experience sensory discomfort with eating and, therefore, may begin to restrict their diet to foods they deem “safe.” This discomfort then could lead to anxiety around eating and RRB (e.g. insisting on particular presentation of food, refusing to trying new foods) to attempt to make mealtime predictable and reduce anxiety. However, the causal nature of these relationships can only be hypothesized at this time because anxiety measures were not part of the current experimental design. Research is needed to investigate them further.
Because neither food selectivity nor SOR scores changed significantly over time, it was not possible to observe whether changes in one preceded changes in the other, allowing causal inferences. This also provides a direction for future research. There is a need to investigate how treatment for SOR in children who have SOR and food selectivity may influence the child’s ability to add new foods to his/her diet. Any corresponding changes in RRB and anxiety also should be considered in order to understand the possible roots of food selectivity in the autism population and causal directions. Such investigations should also include observations at more than two points in time.
Several limitations of this study must be addressed when interpreting the results. Similar to study one (Suarez et al., 2012), the sample respondents were mostly well educated white mothers which might not be fully representative of families in the ASD population. An attrition rate of 63% introduces additional concerns about generalizability. Also, the ASD diagnosis (per parent report) was not independently verified and parents provided a subjective report of their child’s behavior related to all key variables. In addition, the SOR scale for the original study was developed by the researcher and, although it cross referenced questions from existing scales and was subjected to expert panel review, it has not been evaluated psychometrically for validity and reliability. Finally, parents reported the number of foods their child ate as part of their regular diet. This introduces subjectivity into the categorization of food selectivity levels. A direction for future research includes assessing the validity of this type of measure in comparison with things like direct observation of food refusal, and/or food diaries or inventories. Despite these limitations, this study provides important insights into food selectivity and ASD that can provide a catalyst for future studies.
In summary, this study provides longitudinal information about the course of food selectivity in children with autism and the relationship between SOR, RRB, and food selectivity. Food selectivity did not change in individual children over time, pointing to the chronicity of this problem for children with autism. Also, the relationship between food selectivity and SOR appears to be stable. RRB did not independently contribute to prediction of food selectivity when considered in a regression model with SOR. Severe food selectivity at time 2 could be predicted with SOR scores 20 months earlier. Finally, in this study, very few children with autism and food selectivity had been treated for that issue, so it was not possible to investigate whether the condition would be more likely to change with treatment.
Food selectivity may be an under-recognized and under-treated problem in the autism population (Matson and Fodstad, 2009). Research on treatment for food selectivity that considers SOR could provide information about the causal relationship between these two comorbid factors in children with autism. Finally, these results suggest that the treatment for food selectivity that considers SOR may be more effective than the treatment for the food refusal behavior alone. Treatment packages that include antecedent manipulation (Najdowski et al., 2012) and systematic desensitization (Koegal et al., 2004) may provide avenues to empathetically address the discomfort associated with eating that may be experienced by children with ASD. It is a hypothesis that should be tested in future research.
Footnotes
Acknowledgements
This study was made possible through the recruitment assistance of the Interactive Autism Network (IAN) Project at the Kennedy Krieger Institute, Baltimore.
Funding
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.