Treatment of Food Selectivity in an Adult With Autism Spectrum Disorder

Abstract

Food selectivity is common among individuals with autism spectrum disorder (ASD). Left untreated, food selectivity may continue into adulthood, leading to social and nutritional deficits. Although effective treatments have been identified for young children and the school-aged population, these treatments may not be feasible with adults. As such, effective treatments for adults with ASD need to be identified. In this study, the participant was a 26-year-old male with ASD and a history of food selectivity. We compared two treatments that used differential reinforcement of alternative behaviors—one using positive reinforcement and the other negative reinforcement—to increase acceptance of novel fruits and vegetables. Both treatments resulted in increases in the acceptance of grapes and red bell peppers. When given the choice, the participant preferred the positive reinforcement contingency. Additionally, his food acceptance occurred in the presence of novel foods (e.g., carrots and lettuce) and settings. This study represents one of the only studies examining the treatment of food selectivity in adults with ASD.

Feeding disorders are common comorbid conditions associated with autism spectrum disorder (ASD; Hyman et al., 2020). Ledford and Gast (2006) estimated that up to 89% of individuals with ASD experience symptoms of at least one type of feeding challenge. One of the most common feeding disorders is food selectivity, which involves a limited food repertoire and a high-frequency single-food intake (Bandini et al., 2010; Mari-Bauset et al., 2014). Food selectivity can lead to greater risks of nutritional deficiencies in vitamin D, fiber, vitamin E, calcium, and protein (Bandini et al., 2010; Sharp et al., 2013, 2018; Zimmer et al., 2012). Food selectivity among individuals with ASD has been estimated to occur in as high as 85% of children sampled (Leader et al., 2020; Thullen & Bonsall, 2017). Although food selectivity is not exclusive to ASD or related disabilities, individuals with ASD tend to have a greater likelihood of being food selective (Bandini et al., 2010; Beighley et al., 2013; Nadon et al., 2011; Sharp et al., 2013).

Most of the research on treating food selectivity exhibited by individuals with ASD has focused on early childhood and school-aged children (Kuschner et al., 2015), yet adults with ASD often continue to exhibit food selectivity. Kuschner et al. (2015) surveyed adolescents and young adults with and without ASD. Participants involved in the study were 65 individuals diagnosed with ASD ranging from 12 to 28 years of age and 59 neurotypical adolescents ranging from 12 to 23 years of age. The participants with ASD were more likely to self-report food selectivity than the comparison group. These data suggest that food selectivity is not a problem unique to children with ASD.

Left untreated, food selectivity may persist through adulthood (Fodstad & Matson, 2008). Bandini et al. (2017) examined the progression of food selectivity in 18 children with ASD. They found that 65% of those meeting the criteria for high food selectivity during baseline (as defined as the number of foods eaten or food repertoire) had further restricted their repertoire by an average of 1.5 foods over an average of 6.4 years. Although they found that some feeding problems, such as food refusal, had improved at follow-up, individuals with high food selectivity at baseline showed no improvement. Instead, they had further limited their food repertoires.

Tanoue et al. (2016) conducted a longitudinal study of changes in food selectivity with 28 participants with ASD ranging from 3 to 18 years of age. They estimated the food repertoire of their participants based on parental interviews at the ages of 3, 6, 12, and 18 years. The number of foods consumed decreased or remained stable with approximately two thirds of their participants. Bandini et al. (2017) and Tanoue et al. suggested that food selectivity may likely persist, if not become more pervasive in some individuals, into adulthood. These data suggest that food selectivity among individuals with ASD is unlikely to resolve without treatment and therefore likely to continue into adulthood. Thus, food selectivity among adults with ASD should be a greater focus of feeding disorders research.

Recent data from the Centers for Disease Control and Prevention estimate the population of adults with ASD in the United States to be more than 5.4 million (Dietz et al., 2020). Although robust research on the effective treatment of food selectivity exists for young children and the school-aged population (Sharp et al., 2010; Volkert & Vaz, 2010), there is a dearth of research supporting effective treatments for the adult population with ASD (Chawner et al., 2019). Research with adults is particularly important because as individuals age, treatments that have been found to be effective for young children, such as those that include extinction procedures, may become untenable. Additionally, adults with ASD will likely have a longer reinforcement history of refusing a variety of foods, thus likely increasing the resistance to treatment (St. Peter Pipkin & Vollmer, 2009).

Given the number of adults with ASD in the United States and the likelihood that the behavior may worsen over time, it is imperative that effective treatments for food selectivity exhibited by adults with ASD be treated to ameliorate the deleterious effects of prolonged food selectivity. Therefore, the purpose of this study was to evaluate behavioral treatments to increase the acceptance of novel foods with a focus on fruits and vegetables in an adult with ASD. We compared two differential reinforcement of alternative (DRA) procedures to create treatment options.

Method

Participant and Setting

The participant, Jason, was a 26-year-old male diagnosed with ASD. He attended a university-based employment support center. Jason lived at home with his family and commuted to the center Monday through Friday. He was able to independently travel by public transportation (bus) to and from the university-based center and home, and he navigated the university on the campus buses. He had a highly developed repertoire of expressive and receptive language skills and was able to speak in full sentences. He followed oral and written directions, was able to converse with his peers, engaged in amateur competitive sports, and read when given a book or newspaper.

Jason was employed in an on-campus dining hall 5 days a week and was supported by a behavioral technician until he mastered all his job responsibilities, at which point he worked independently. His job responsibilities included cleaning tables and entrance doors, wiping down food stations, and serving food. Outside of his work hours, Jason engaged in various community-based leisure activities with his behavioral technician, such as playing table tennis, pool, and air hockey, as well as shooting basketball hoops.

According to family and self-reports, Jason had a long history of food selectivity, specifically a limited food repertoire. They reported that Jason had a history of rejecting novel or different foods. Jason’s daily diet consisted of a select few items such as hamburgers, chicken nuggets, and peanut butter and jelly sandwiches since age 2. Staff reported that he only consumed peanut butter and jelly sandwiches or peanut butter crackers for lunch at work. Jason had no history of prior treatment for food selectivity by health care providers. His parents and the clinical staff supporting Jason expressed concerns about his food selectivity. He had taken a daily multivitamin since he was 2 years of age, and his parents reported he had no history of gastrointestinal symptoms such as constipation, diarrhea, or upset stomach. Additionally, Jason had no history of difficulty chewing or swallowing foods and maintained adequate chewing and swallowing throughout the study. Jason’s BMI was 29.6.

All sessions were conducted by staff at the university-based employment support center except the last two sessions, which were conducted at the campus dining hall. The room was equipped with a small table and chair. Additionally, staff members had clipboards with data sheets.

Response Measurement and Interobserver Agreement

Frequency data for all foods except the side salad were collected on food acceptance during each session using a pen and paper. Acceptance of the side salad was scored as a dichotomous response (yes = 1 or no = 0). Acceptance of all foods except the side salad was defined as Jason independently depositing the target food in his mouth. Acceptance of the side salad was scored when the bowl was empty of all lettuce, bell peppers, and carrots. If some or all of the side salad remained in Jason’s bowl, it was not considered acceptance.

Checks for a clean mouth were not used as Jason was an adult. However, he likely consumed all accepted foods because he never spit out food, nor did he appear to pack food, and his speech was clear following sessions. Jason accepted the grapes whole and took small bites of each piece of the other target foods. All target foods were presented on a plate, except the side salad, which was presented to Jason in a bowl. The same bowl was used across settings.

A second observer collected data during 33% of sessions. Each session consisted of one meal. Interobserver agreement was calculated by dividing the smaller amount of food accepted by the larger amount of food accepted and multiplying the result by 100. Interobserver agreement was 94% (range 86%–100%).

Experimental Design

A combined alternating-treatments and changing-criterion design was used to compare the efficacy of the two treatments and demonstrate experimental control.

Procedure

Prior to treatment, probes were conducted to determine whether Jason would accept novel fruits such as apples and grapes. Five probes were conducted, once per day, just before lunch. He refused to accept the food during 100% of probes.

Jason’s parents provided us with a list of foods that Jason did not eat but were commonly eaten by other members of the family and would be readily available at home. The foods initially targeted for intervention were grapes, baby carrots, and red bell peppers. The terminal goal was approximately 237 mL of grapes and baby carrots, as well as one whole red bell pepper. The serving sizes of each target food were based on U.S. Department of Agriculture guidelines (SNAP Education Connection, n.d.). The red bell peppers were cut into 12 slices, approximately 0.6 cm by 10 cm each. Grapes and baby carrots were presented whole. Although Jason’s parents reported that he had never eaten baby carrots, he accepted up to 11 whole baby carrots during baseline. Therefore, we did not implement treatment for baby carrots.

Preference Assessment

Prior to evaluating treatments, we conducted multiple-stimulus without-replacement (MSWO) preference assessments (DeLeon & Iwata, 1996) to identify likely positive and negative reinforcers. All stimuli selected for the positive reinforcer preference assessment were identified through participant interview. Activities selected to include as potential negative reinforcers were those that were commonly available in the center, such as washing dishes, vacuuming, and cleaning the table.

Potential negative reinforcers were identified by writing 10 different tasks on individual index cards specifying the number of tasks to complete or amount of time to complete each task. Each of the 10 cards was placed on the table in a line with approximately 5 cm between them. Jason was asked to read the cards and let us know which one he wanted to do the most. Once a selection was made, he was asked to engage in the task that was written on the selected card. Once he completed the task, he was asked to make another selection until all 10 selections were chosen. The MSWO for negative reinforcers was conducted three times.

Jason demonstrated selection bias (i.e., selecting items on the left side) during the MSWO for positive reinforcers, which necessitated the use of a modified paired-choice preference assessment (Fisher et al., 1992) in which items were presented on the same plate with one item at the bottom of the plate. In total, 10 items were presented during the preference assessment for potential positive reinforcers: 8 edibles and 2 activities. Initial preference assessments showed that croissants were his most preferred edible item (positive reinforcer) and washing dishes was his least preferred activity (negative reinforcer).

Following the preference assessments, we compared two treatments, positive reinforcement and negative reinforcement contingencies, to ensure Jason and his parents had treatment options. At the beginning of each week, a morning and afternoon schedule of sessions that staff were to conduct was posted. The first session of the day was selected based on a coin toss. The second session of the day was counterbalanced with the morning session. Morning sessions were conducted approximately 3 hr after Jason had accepted a small breakfast. Afternoon sessions were conducted approximately 4 hr after lunch. Novel probes for the side salad (at the dining hall only) were presented at his regular lunchtime just before he ate lunch.

Baseline

During baseline, the experimenter asked Jason to take a seat at the table and sat next to him. The experimenter placed slightly more than the terminal amount of food in front of him. The experimenter provided Jason with the following instruction: “You can eat as many [target food] as you want; you don’t have to eat any at all.” If Jason engaged in vocal or gestural refusal (e.g., a statement such as “never had it, never will” or a flat palm with fingers pointed up at chest level), the food was removed and the session was terminated. Alternatively, the session was terminated after he accepted all of the food or after 5 min passed. No beverages were present during the session. Jason was given the option of using a fork for the side salad only, as the other target foods were finger foods. The experimenters did not prompt Jason to eat any of the target foods at any time throughout the study, and conversation during sessions was minimal, as Jason was typically eating.

Treatment

Treatment was similar to baseline except that we compared two variations of DRA procedures: positive reinforcement and negative reinforcement. Treatment sessions alternated semirandomly between positive and negative reinforcement contingencies each day—one in the morning and one in the afternoon.

During the positive DRA reinforcement contingency condition (DRA+), the experimenter sat next to Jason and placed slightly more than the terminal amount of food on a plate or in a bowl directly in front of him. We presented more than the terminal amount of food in the event that Jason accepted a greater amount of food than the prespecified criterion. Half of a croissant was placed approximately 15 cm away from the target food. The experimenter provided the following instruction: “You can eat as many [target food] as you want, or you don’t have to eat any at all, but you need to eat [quantity] to get the croissant.” The session ended if Jason (a) accepted the target amount of food and declared either vocally or gesturally that he was done, (b) engaged in vocal or physical refusal, or (c) let 5 min pass without having accepted the specified amount of food. We did not immediately terminate the session when he accepted the prespecified amount of the target food to allow him the opportunity to exceed the criterion. If he accepted the prespecified amount of the target food and indicated he was finished, the experimenter provided him immediate access to the croissant and brief praise (e.g., “Great job!”). If Jason did not meet the reinforcement criterion, no feedback was provided and the croissant was removed.

The negative DRA reinforcement contingency condition (DRA−) was similar to the DRA+ condition, except that we provided negative reinforcement contingent on consuming the prespecified amount of the target food. The experimenter placed slightly more than the terminal amount of food on a plate or in a bowl and an index card with “Wash 3 dishes” written on it in front of Jason. No other food was present. The experimenter told Jason, “You can eat as many [target food] as you want, or you don’t have to eat any at all, but you have to eat [quantity] or wash three dishes.” If Jason accepted the prespecified amount of food and declared vocally or gesturally that he was done, he did not have to wash the dishes. If Jason engaged in verbal or physical refusal or let 5 min pass without having accepted the prespecified amount of food, he was prompted to wash dishes (“Please go into the kitchen and wash the dishes.”). He always complied with this request.

We conducted a final phase in which we provided a concurrent choice between DRA+ and DRA−. The croissant (DRA+) and index card (DRA−) were placed directly in front of Jason on the table. Jason was asked, “Do you want to work for this [experimenter pointed to the half croissant on a plate on the table] or this [experimenter pointed to the index card with ‘Wash 3 dishes’ written on it] today?” Once he selected the contingency, we started the respective treatment session.

The criterion was increased by two food pieces after three consecutive sessions in which the prespecified goal was met. The terminal goal was approximately 237 mL of grapes or one whole red bell pepper.

Novel Food and Setting

Once Jason met the criteria for accepting grapes and red bell peppers, the experimenters probed a novel food: side salad. The side salad initially consisted of a bowl of romaine lettuce (the novel food), four red bell pepper slices, and four baby carrots. One session was conducted in the morning and one in the afternoon. Jason was asked to sit at the table, and the side salad was placed in front of him. He was offered dressing, but he declined. A fork was made available, although he did not always use it. Side salad probes were conducted under both DRA+ and DRA− conditions, and he had to accept the entire bowl of side salad to receive reinforcement. Jason was given the instruction “You can eat as much as you want, or you don’t have to eat any, but you need to eat the entire salad to [get the ice cream sandwich / not wash six dishes].” Once two sessions of each reinforcement contingency were conducted at the employment support center, experimenters probed the side salad in a new environment (dining hall, his place of employment) with novel foods (spring salad mix and shredded carrots). Probes at the dining hall were conducted once per day, immediately before Jason’s regular lunchtime. Jason was asked to make his own salad from the salad bar.

Results

Figure 1 shows the number of grapes accepted during the DRA+ condition (top panel) and DRA− condition (bottom panel). Acceptance remained low during baseline, with Jason accepting between zero and two grapes per session. Once treatments were implemented, Jason accepted more grapes during each treatment condition, eventually reaching the terminal goal of 12 grapes (approximately 237 mL). There were two sessions during the DRA− condition in which Jason exceeded the predetermined goal (Sessions 8 and 16). During these sessions, the experimenter neglected to state the correct instructions (i.e., did not state the predetermined goal). Additionally, experimenters erroneously increased the criterion (i.e., too few sessions or too many sessions) during several phases in both the DRA+ and DRA− conditions. Despite these treatment integrity errors, his behavior increased with each criterion shift and remained stable at each level until the criterion changed in both the DRA+ (top panel) and DRA− (bottom panel) conditions. During the concurrent-choice phase, Jason consistently chose the DRA+ reinforcement contingency. Finally, the results transferred across various settings, including home and work. He continued to accept grapes at the 1-month follow-up, and his caregivers reported that he had continued eating grapes at home.

Fig. 1.

Acceptance of Grapes. Note. The top panel shows the positive differential reinforcement of alternative behavior (DRA+) condition, and the bottom panel shows the negative (DRA−) condition. The baseline, choice, and follow-up data are the same across panels. The horizontal dashed lines are the criterion for that phase. Please note that this is a combined changing-criterion and alternating-treatments single-case design. For the sake of clarity, we separated the two treatments across panels

Figure 2 shows the total number of red bell pepper slices accepted during the DRA+ (top panel) and DRA− (bottom panel) conditions. Acceptance remained variable during baseline. Acceptance of red bell peppers decreased from baseline during the initial implementation of the DRA+ and DRA− conditions. Therefore, we interviewed Jason again and reassessed his preferences for both positive and negative reinforcers using identical procedures to the initial preference assessments. We identified a small ice cream sandwich, similar in caloric value to the previous edible reinforcer, as the new positive reinforcer. While assessing for negative reinforcers, we doubled the magnitude of each task (10 push-ups instead of 5, etc.) and identified that washing dishes remained Jason’s least preferred activity following an increase from three to six dishes.

Fig. 2.

Acceptance of Red Bell Peppers and Salad. Note. The top panel shows the positive differential reinforcement of alternative behavior (DRA+) condition, and the bottom panel shows the negative (DRA−) condition. Red bell pepper acceptance is shown in the left y-axis, and salad acceptance is shown in the right y-axis. The baseline, choice, and follow-up data are the same across panels. The horizontal dashed lines are the criterion for that phase. Please note that this is a combined changing-criterion and alternating-treatments single-case design. For the sake of clarity, we separated the two treatments across panels. BL = baseline

Changes in the reinforcers resulted in an increase in the acceptance of red bell peppers when the DRA+ and DRA− conditions were reintroduced. Jason reached the terminal goal of 12 red bell pepper slices (one whole red bell pepper) in both conditions. Experimenters erroneously increased the criterion (i.e., too few sessions or too many sessions) during several phases in both the DRA+ and DRA− conditions. Despite these treatment integrity errors, his acceptance increased with each criterion shift and remained stable at each level until the criterion changed in both the DRA+ (top panel) and DRA− (bottom panel) conditions. During the DRA− condition, when we increased the criterion to six red bell pepper slices, food acceptance decreased to zero. It was not until Jason advanced to eight slices of red bell pepper in the DRA+ condition that he accepted them again during the DRA− condition. This suggests that the DRA+ condition may have been slightly more effective than the DRA− condition. During the concurrent-choice phase, Jason consistently chose the DRA+ condition. Results transferred across various settings, including his home and work environments.

Jason successfully accepted a novel food (side salad) during the DRA+ and DRA− reinforcement conditions at the employment support center. He also accepted a side salad with additional novel foods (spring salad mix and shredded carrots) at the dining hall.

Discussion

These results showed that two variations of DRA with positive and negative reinforcement contingencies were effective in increasing food acceptance in an adult with ASD who had a long history of food selectivity. Prior to the intervention, the participant had a limited food repertoire. Treatment resulted in the acceptance of healthier foods: grapes and red bell peppers. Treatment effects maintained for grape acceptance for at least 1 month. Additionally, treatment effects transferred to novel foods such as carrots and lettuce across settings. There was little difference across treatment conditions in terms of the number of sessions necessary to reach the terminal goals. However, when presented with a choice between treatments, Jason exclusively selected DRA+.

Previous research on the treatment of food selectivity has generally focused on children with ASD. Our results extend previous food selectivity research by demonstrating that it is possible to treat food selectivity exhibited by an adult with ASD who has a long history of eating a limited diet. The effects of treatment also transferred to an untrained food (side salad) and a novel setting (dining hall). Treatments were implemented by a dozen behavioral technicians across sessions, which may have helped to promote the transfer of stimulus control across foods and settings.

We intervened on Jason’s food selectivity because of parental and staff concerns. Additionally, we relied on parental interviews to identify the target foods. When implementing a feeding program, one should ensure all precautions are taken to address any medical issues such as food allergies. Clinicians should strongly consider consulting a dietician prior to and throughout the treatment of individuals for guidance on appropriate target foods that may directly address possible nutritional deficits presented by an individual’s limited food repertoire, as well as appropriate serving sizes to target. Clinicians should maintain a safe, controlled environment for treatment when introducing novel foods when evaluating food selectivity.

Curtin et al. (2015) found that parents of individuals with ASD experienced higher levels of spousal stress during mealtime than parents of typically developing individuals. Prior to the study, Jason’s caregivers expressed concern for his well-being, social impact, and quality of life due to his limited food repertoire. Since the implementation of the current study, his caregivers reported that he continued to accept novel foods into his repertoire and that while attending a social function for work, Jason accepted novel food items with his peers. Future research should include additional measures of well-being for participants and their caregivers.

We used a changing-criterion design to gradually increase food acceptance of novel foods. This meant that many sessions were required to reach the terminal criteria. It may be the case that we did not need all the steps. Therefore, future research should incorporate terminal-criterion probes to determine whether all the criterion changes are necessary. Additionally, future studies may want to include maintenance data and a social validity measurement for stronger results.

Although the staff rotation may have also been beneficial for transferring the treatment effects, it may have been a limitation to treatment integrity. Treatment integrity errors were made when increasing the criterion, at times drifting away from the proper steps of the protocol as it was written. There were multiple phases in which too few or too many sessions relative to the changing criterion were conducted. This frequent staff rotation may have made it challenging to convey the necessary criterion to all staff members. Despite these treatment integrity failures, the treatment continued to be effective, which may suggest an increase in the preference for or a decrease in the aversiveness of the foods.

There were two sessions during the DRA− condition with grapes in which the experimenter issued incorrect instructions—the experimenter failed to state the predetermined number of grapes to consume. During these two sessions, Jason far exceeded the goal for that phase. These data suggest that instructional control may have played a role in the increase in food acceptance. However, the reinforcement contingency was still important because staff and parents had repeatedly instructed Jason to consume new foods in the past with no success. Future research should evaluate the influence of instructional control on responding.

Another limitation was that we used the same foods across the treatment comparison (DRA+ vs. DRA−). We did this to control for possible extraneous effects of different food preferences. However, by using the same foods across treatment conditions, any differences in treatment efficacy across conditions may have been obscured because of carryover effects. Therefore, future studies should consider using different foods across treatment comparisons.

A final limitation of this study was that there was only one participant. Given the paucity of research on the treatment of food selectivity exhibited by adults with ASD, additional research with adults will be important to identify a range of effective treatments.

The ability to consume a variety of foods has several benefits, such as improving and maintaining overall health and nutrition. Malnourishment can precipitate a multitude of medical conditions, as well as increase stress and fatigue. Another benefit of increasing the acceptance of novel foods to include fruits and vegetables is that it may facilitate socialization at novel restaurants, social gatherings, and work functions. The ability to consume novel foods may help decrease stress in social situations where eating a wide variety of foods is typically expected and sometimes required. Food selectivity is a disruptive and potentially dangerous behavior that should not be overlooked, particularly in adults with ASD.

Implications for Practice

• It is important to recognize that food selectivity is a problem for adults with ASD.

• Differential reinforcement can be an effective treatment for food selectivity exhibited by individuals with a long history of restricted food-related repertoires.

• Treatment effects may generalize to novel foods in novel settings.

• Providing participants with concurrent treatment options is feasible and practical.

Availability of data and material

Available upon request.

Declarations

Conflict of interest

We have no known conflicts of interest to disclose.

Ethical approval

This study was approved by the university institutional review board.

Informed consent

The participant consented to participation.