Convergent Validity of the Feeding and Eating in AutiSm Together Assessment (FEAST) Open Access

Comprehensive and validated assessments for use in clinical and research settings are needed to identify and address the complexities of feeding challenges specific to autistic children. Several caregiver-report instruments have been constructed to evaluate feeding and eating challenges; however, these are not specific to autism and primarily focus on problematic behaviors during mealtime, overlooking the more varied complexities of feeding challenges for autistic children.

Caregiver reports about feeding and mealtime challenges for autistic children reveal that there are multiple additional areas of concern that are relevant to clinical practice, suggesting that the challenges many families with autistic children face with respect to feeding are more heterogeneous than current assessments reflect (Aponte & Romanczyk, 2016; Bonsall et al., 2021; Leader et al., 2020; St. John & Ausderau, 2024). These concerns indicate a need for assessments that examine a variety of contributing factors to mealtime challenges that extend beyond behavior.

Comprehensive and specific assessment tools are essential to capture the variability and complexity of feeding challenges in autistic children. Feeding challenges are defined by a multitude of overlapping factors surrounding eating and mealtime, such as selective acceptance of foods, limited diet variety, sensory and behavioral challenges, difficulty with self-feeding, delayed oral motor abilities, increased caregiver responsibilities, and complex family routines (Bonsall et al., 2021; St. John & Ausderau, 2024). Autistic children frequently experience challenges across multiple areas with differential patterns, divergent trajectories, and varying clinical priorities (Ashley et al., 2020; Bonsall et al., 2021; St. John & Ausderau, 2024). Assessment tools must expand from the current behaviorally focused approach to adequately capture the heterogeneity of feeding challenges and guide clinical services for autistic children.

The Feeding and Eating in AutiSm Together assessment (FEAST) is a caregiver-report questionnaire developed by a research team at the University of Wisconsin–Madison. The purpose of the FEAST is to examine mealtime challenges in a variety of domains specific to autistic children and to provide a more holistic assessment tool for clinicians and researchers to measure overall feeding challenge severity. The FEAST is a two-part, 84-item measure that assesses multiple factors related to feeding and eating challenges for autistic children ages 2–12 yr. Part 1 of the FEAST contains four subscales, including (1) Sensory Processing, (2) Oral Motor Skills, (3) Behavior, and (4) Gastrointestinal Health (Ausderau et al., 2024). Previous studies have established the internal construct validity of the FEAST and these four subscales (Ausderau et al., 2024). The convergent validity of the FEAST in comparison with other commonly used feeding and eating assessments has yet to be established.

The Brief Assessment of Mealtime Behavior in Children (BAMBIC; Hendy et al., 2013) and the Screening Tool of Feeding Problems Applied to Children (STEP-CHILD; Seiverling et al., 2011) are well-validated, commonly used measures of mealtime behavior among autistic children (DeMand et al., 2015; Hendy et al., 2013; Seiverling et al., 2011, 2016). Therefore, they are an appropriate comparison against which to measure the convergent validity of the FEAST’s Total and subscale scores. The BAMBIC was designed for clinical use with clients who have been given a wide range of diagnoses (Hendy et al., 2013). It includes 10 questions that focus on three types of children feeding challenges: (1) Limited Variety, (2) Food Refusal, and (3) Disruptive Behavior. The STEP-CHILD is a commonly used pediatric feeding assessment that aims to identify a wider range of factors that influence child feeding differences (Seiverling et al., 2011). The STEP-CHILD is divided into six subscales: (1) Chewing Problems, (2) Rapid Eating, (3) Food Refusal, (4) Food Selectivity, (5) Vomiting, and (6) Stealing Food, which comprise a total of 15 items.

The purpose of this study was to examine whether the FEAST and its subscales can be validated against the BAMBIC and STEP-CHILD. The aim was to establish how child scores on the FEAST and its subscales relate to BAMBIC Total scores and STEP-CHILD Total Frequency scores within a large national sample. We hypothesized that there will be moderate or strong correlations between the FEAST Total scores and both BAMBIC Total scores and STEP-CHILD Total Frequency scores. Because of the BAMBIC’s behavioral focus, we hypothesized moderate or strong correlations between the FEAST Behavior subscale scores and BAMBIC Total scores and Food Refusal and Disruptive Behavior subscale scores. In addition, we hypothesized a moderate correlation between scores on the FEAST Sensory Processing subscale and those on the BAMBIC Limited Variety subscale. Moderate to strong correlations were hypothesized between FEAST Behavior subscale scores and the STEP-CHILD Food Refusal subscale scores. We also hypothesized moderate correlations between the FEAST Sensory Processing subscale scores and scores on the STEP-CHILD Food Selectivity subscale. Finally, we hypothesized that scores on the FEAST Oral Motor Skills subscale would be moderately correlated with the STEP-CHILD Chewing Problems subscale scores and have a small to moderate correlation with scores on the STEP-CHILD Vomiting subscale. Weak correlations were anticipated for the remaining tested relationships because the subscales across measures are designed to capture separate unique constructs, albeit all related to feeding.

This study had a cross-sectional design and aimed to determine, using a large national sample, the convergent validity of the FEAST by measuring the strength of correlation between participant scores on the FEAST and two well-validated, clinically used mealtime behavior measures (BAMBIC and STEP-CHILD).

Caregivers of 349 autistic children ages 2–12 yr completed the BAMBIC, STEP-CHILD, and FEAST as part of the Survey for Characterization of Feeding Challenges in Autistic Children–US, a national, cross-sectional study of caregivers (St. John & Ausderau, 2024). Surveys were collected between August 2016 and January 2017. The survey sample size was determined on the basis of the results of a statistical consultation, such that a sample size of 350–400 respondents was recommended to ensure valid and reliable survey results.

Survey respondents were recruited by means of the Interactive Autism Network (IAN; https://iancommunity.org) online research registry and self-selected to participate after receiving an email invitation. Participants were allowed to complete the survey over the course of several sessions using the same invitation link provided on the IAN. Electronic consent forms were made available to participants, and all survey procedures received approval from the University of Wisconsin–Madison Institutional Review Board (No. 2014–0945). Survey respondents were offered a $5 gift card in appreciation of their participation.

Participants were screened for inclusion and exclusion criteria with a questionnaire administered at the beginning of the survey. To participate, children needed to be between ages 2 yr and 12 yr, have been diagnosed with autism, and have caregiver-reported feeding challenges. Autism diagnoses were not confirmed as a part of this study; however, all prospective participants in the IAN registry have a recorded Social Communication Questionnaire (SCQ) score (Rutter et al., 2003). Only children with an SCQ score ≥15, which indicates a high likelihood of autism (Rutter et al., 2003), were recruited. In addition, a 2012 study authenticated caregiver-reported autism diagnoses among participants in the IAN database (Daniels et al., 2012). Participants were excluded if their caregiver reported a co-occurring diagnosis of childhood disintegrative disorder; fragile X syndrome; Rett syndrome; tuberous sclerosis; or significant visual, hearing, or physical impairments. After completing a screening to confirm that the inclusion criteria were met, participants were linked to the informed consent form and survey. Data from caregivers who completed the FEAST and both comparison measures were included in the analyses. All participant responses were downloaded and stored on an encrypted research server on a computer located in a locked university room.

A family demographics questionnaire, created specifically for the Survey for Characterization of Feeding Challenges in Autistic Children–US, was used to obtain information about the child and the family in four domains: (1) family characteristics (size and composition of a family), (2) child characteristics (age, gender, race/ethnicity), (3) child’s functioning level (physical, emotional, cognitive and learning performances), and (4) services the child receives (types of therapy). The questionnaire takes approximately 10–15 min to complete.

The Social Responsiveness Scale–Second Edition (SRS–2; Constantino & Gruber, 2012) was included in the Survey for Characterization of Feeding Challenges in Autistic Children–US to characterize child autistic traits. The SRS–2 is a validated caregiver-report measure that includes 65 items and asks about children’s responses in social settings.

As mentioned, the FEAST is a two-part, 84-item caregiver-report questionnaire used to measure mealtime and feeding challenges for autistic children ages 2–12 yr. It takes an estimated 15 min to complete. The FEAST was developed to identify underlying challenges related to feeding behaviors and guide interventions services. FEAST items were developed by a team of experts, including researchers, clinicians, and parents of autistic children, who used a combination of an extensive literature search; clinical expertise; lived expertise; and preliminary findings from previous research, including caregiver interviews about feeding challenges (Ausderau et al., 2024). For a full description of the development and initial validation of the FEAST, see (Ausderau et al., 2024)

Part 1 of the FEAST aims to identify the severity and underlying factors contributing to a child’s feeding challenges. It includes Items 1–32, which are rated on a Likert scale that ranges from 1 (almost never) to 5 (almost always). Part 2, which includes Items 33–84, was designed for clinical utility and aims to provide a description of the child’s current challenges in their environmental context and inform treatment planning. Part 2 does not affect the Total score or the subscale scores. The FEAST provides a summary feeding challenge severity score (FEAST Total) and four subscale scores across four areas of feeding challenges: (1) Sensory Processing, (2) Oral Motor Skills, (3) Behavior, and (4) Gastrointestinal Health.

A factor analysis of the FEAST demonstrated good construct validity of the four subscales, with a high internal consistency (α = .84; BLINDED). An ordinal confirmatory factor analysis of the responses of 396 primary caregivers of autistic children revealed good fit of 32 questionnaire items from the four FEAST predetermined areas. The between-factor correlation matrix showed that the Pearson correlations among the four factors (i.e., Sensory Processing, Oral Motor Skills, Behavior, and Gastrointestinal Health) were positive and small to medium, as hypothesized (Ausderau et al., 2024). The FEAST has demonstrated construct validity with strong internal consistency, good unidimensional models supporting the four-factor model, and overall good model fit indices for the measurement of the four underlying constructs (Ausderau et al., 2024).

The BAMBIC is a caregiver-report questionnaire that measures feeding problems among children with three diagnostic conditions, including autism; children other developmental disability diagnoses; and children with no additional diagnoses but who present with feeding challenges. Hendy et al. (2013) reported excellent psychometric properties, including internal validity and convergent validity. The BAMBIC includes 10 items and examines children’s mealtime behavior as assessed with three subscales: (1) Limited Variety, (2) Food Refusal, and (3) Disruptive Behavior. The BAMBIC Total score and subscale scores were used in the analysis in this study.

The STEP-CHILD is a validated caregiver-report measure for children who have challenges with eating and feeding. It evolved from the 23-item Screening Tool of Feeding Problems (Matson & Kuhn, 2001). The STEP-CHILD is used to evaluate feeding challenges in children ages 24 mo–18 yr. It consists of 15 items that assess six domains of mealtime behaviors, each including 2–3 items: (1) Chewing Problems (3 items), (2) Rapid Eating (3 items), (3) Food Refusal (3 items), (4) Food Selectivity (2 items), (5) Vomiting (2 items), and (6) Stealing Food (2 items; Seiverling et al., 2011). Each item is scored on the basis of severity and frequency. Summed scores for severity (STEP-CHILD Total Severity) and frequency (STEP-CHILD Total Frequency) of mealtime behaviors were calculated. Item response options on the FEAST are most similar to the STEP-CHILD Total Frequency item response options. For this reason, STEP-CHILD Total Frequency was chosen as a comparison measure for this analysis. STEP-CHILD subscale scores were also used in the analyses.

Participants’ responses were scored according to each measure’s respective manual during the data analysis of the Survey for Characterization of Feeding Challenges in Autistic Children–US. Surveys were administered using university-hosted Qualtrics software (Qualtrics Labs, 2016). Responses were deidentified and assigned a participant identification number. Deidentified participant data from the survey were analyzed using R (Version 4.4.1; R Core Team, 2024), R Studio (RStudio Team, 2020), and R packages dplyr (Wickham et al., 2023) and ggplot2 (Wickham, 2016).

We investigated the relationships among participants’ FEAST Total and FEAST subscale scores, BAMBIC Total and BAMBIC subscale scores, and STEP-CHILD Total Frequency and STEP-CHILD subscale scores using Pearson product–moment correlation coefficients. Bonferroni correction was applied to account for multiple correlations. Probability (p) values ≤.0009 were considered statistically significant. For the purposes of this study, coefficients between .5 and .7 were considered moderate correlations, and coefficients >.7 were considered strong correlations. However, cutoffs to interpret correlation coefficients are variable, and the strength of the relationship should be interpreted in the broader context of the study (Schober et al., 2018). We performed preliminary analyses to ensure no violations of the assumptions of linearity, normality, or homoscedasticity. Scatterplots were created to check for outliers.

In total, 349 participants completed the FEAST, BAMBIC, and STEP-CHILD as part of the Survey for Characterization of Feeding Challenges in Autistic Children–US. Participant characteristics are presented in Table 1.

We calculated Pearson product–moment correlation coefficients between participant scores to establish the correlations between the FEAST Total and subscale scores and scores on the BAMBIC and STEP-CHILD. Assumptions of linearity and normality were met.

Table 2 shows the correlations between the FEAST Total and subscale scores and the BAMBIC Total and subscale scores. Significant moderate positive correlations were identified between FEAST Total scores and BAMBIC Total scores. The FEAST Behavior subscale scores demonstrated moderate positive correlations with the BAMBIC Total, Food Refusal subscale, and Disruptive Behavior subscale scores. These results confirm the hypothesis that BAMBIC Total scores and subscale scores would have the strongest correlations with FEAST Behavior subscale scores because of the overall behavioral focus of the BAMBIC. The correlations between the FEAST Total and subscale scores and scores on the BAMBIC Limited Variety subscale were weak; the strongest correlation was with the FEAST Sensory Processing subscale rather than the FEAST Behavior subscale. The FEAST Gastrointestinal Health subscale was minimally correlated with the BAMBIC Total and subscale scores, indicating a probable divergence in the constructs measured by the FEAST Gastrointestinal Health items and the BAMBIC.

Table 3 shows the correlations between the FEAST Total and subscale scores and the STEP-CHILD Total Frequency and subscale scores. Significant moderate positive correlations were identified between the STEP-CHILD Total Frequency scores and FEAST Total scores and scores on the Behavior subscale and the Oral Motor Skills subscale. As expected, the FEAST Behavior subscale demonstrated significant moderate correlations with STEP-CHILD Total Frequency scores and STEP-CHILD Food Refusal subscale scores. A significant moderate positive correlation was identified between FEAST Oral Motor Skills subscale scores and the STEP-CHILD Chewing Problems subscale scores, indicating that the two subscales are capturing similar feeding constructs. Similarly, with respect to the correlations between the FEAST subscales and BAMBIC subscales, the FEAST Sensory Processing subscale was most closely related to the STEP-CHILD Food Selectivity subscale, and the FEAST Gastrointestinal Health subscale had weak or nonsignificant relationships with all STEP-CHILD scores.

The results of this study provide evidence for the convergent validity of the FEAST with existing feeding measures. Overall, the results are consistent with our hypotheses and demonstrate the strongest relationships between Total and subscale scores that measure similar constructs related to eating and mealtime. Significant moderate positive correlations among FEAST Total scores BAMBIC Total scores, and STEP-CHILD Total Frequency scores demonstrate that the content of the FEAST measures the construct of feeding and eating challenge severity similarly to well-validated measures of child feeding problems. These results substantiate the FEAST as a valid assessment for evaluating mealtime challenges. Differential patterns of relationships among the subscale scores on the FEAST, BAMBIC, and STEP-CHILD demonstrate the overlap and divergence in the feeding-related constructs measured by each subscale.

The strongest correlation was between scores on the FEAST Behavior subscale and BAMBIC Total scores. This was expected given that 8 out of 10 items on the BAMBIC ask caregivers to rate the frequency of problematic behaviors associated with mealtimes. The relationship between the FEAST Behavior subscale and the STEP-CHILD Total Frequency scores also was moderate. This result also was expected because multiple items on the STEP-CHILD also ask caregivers to rate the frequency of problematic mealtime behaviors (e.g., “Child steals or attempts to steal food”). The correlations among the FEAST Behavior subscale scores, BAMBIC Total score, and STEP-CHILD Total Frequency score suggest that mealtime behaviors captured by the latter two’s Total scores are distinctly captured by the FEAST Behavior subscale. Mealtime behaviors are essential to consider in feeding evaluation; however, because of the heterogeneity of feeding and eating challenges for autistic children (Leader et al., 2020; St John & Ausderau, 2024), additional factors that can influence mealtime participation for autistic children (e.g., their sensory responses and oral motor skills) need to be considered. A comprehensive evaluation of mealtime behavior and additional mealtime and feeding factors can support targeted treatment planning beyond behavior management.

Challenging behavior is often a caregiver’s first indication that a child is experiencing a feeding problem, and therefore it is a core component of feeding and eating assessments. Although items on the BAMBIC and STEP-CHILD may capture some aspects of sensory preferences, oral motor skills, or gastrointestinal health, they do so indirectly, by measuring the behaviors that may be the result of underlying challenges in these areas, or with few items. Observed behavioral responses should be analyzed to identify the underlying factors driving them (e.g., a sensory aversion to the presented food, child’s internal hunger or satiation cues, gastrointestinal reflux). The additional information obtained by the four unique subscales of the FEAST may provide a more complete evaluation of mealtime challenges for autistic children so that the underlying needs can be addressed to support the child’s and family’s goals.

The FEAST was designed to directly measure these additional contributing feeding domains, including 22 items that measure the constructs of sensory processing (n = 8), oral motor skills (n = 8), and gastrointestinal health (n = 6; Ausderau et al., 2024). The results of this study identified differential patterns in the correlations between the FEAST subscale scores and the BAMBIC and STEP-CHILD; weak or insignificant correlations between subscales suggest that the respective subscales are measuring unique constructs. Divergence between scales was most clearly observed in the correlations between scores on the FEAST Gastrointestinal Health subscale and all scores for both the BAMBIC and STEP-CHILD. The identified small significant and largely insignificant correlations suggest that the FEAST Gastrointestinal Health subscale is measuring a unique construct compared with the BAMBIC and STEP-CHILD. Measurement of gastrointestinal and health concerns in combination with feeding behaviors is essential given the high co-occurrence of gastrointestinal symptoms in autistic children (Babinska et al., 2020; Leader et al., 2020, 2022). Underlying gastrointestinal difficulties can be a major barrier to positive eating and mealtime experiences because of discomfort experienced during or after eating.

Correlations among the FEAST Oral Motor Skills subscale and BAMBIC and STEP-CHILD Total and subscale scores were primarily small and positive. One expected exception is the moderate correlation among scores on the FEAST Oral Motor Skills subscale, the STEP-CHILD Total Frequency score, and scores on the STEP-CHILD Chewing Problems subscale. The STEP-CHILD Chewing Problems subscale consists of three items focused on chewing, swallowing, and regurgitating food, whereas the BAMBIC does not include questions about chewing, swallowing, or regurgitation. Therefore, the STEP-CHILD may be a more direct measure of oral motor abilities than the BAMBIC. The weak correlations between scores on the FEAST Oral Motor Skills, Gastrointestinal Health, and Sensory Processing subscales in relation to the comparison measures suggest that the FEAST provides a more direct evaluation of factors that contribute to mealtime challenges that might not be as well captured by the BAMBIC or STEP-CHILD.

In addition to observable behaviors, feeding challenges are related to a number of factors, including sensory processing and perception, medical concerns, cognitive and adaptive skills, and caregiver stress (Esposito et al., 2023; Page et al., 2022). Addressing mealtime and feeding challenges in families with autistic children requires consideration of the family system and an expansion from behavior-only approaches (Matthews et al., 2025). There currently is a lack of validated measures that capture a multifaceted assessment of feeding challenges in autistic children and inform expanded intervention approaches.

The FEAST fills an important gap in the clinical evaluation of autistic children experiencing feeding challenges by considering these challenges across factors. To be specific, the FEAST makes a unique contribution by adding the evaluation of oral motor skills and gastrointestinal health in addition to capturing mealtime behavior and sensory factors. The FEAST provides occupational therapy practitioners with an autism-specific, comprehensive, validated assessment tool to guide assessment and intervention planning. The findings from this study demonstrate that the FEAST captures the construct of feeding behaviors as well as other validated measures do, and the additional areas of feeding challenge captured by the FEAST (sensory processing, oral motor skills, and gastrointestinal health) are distinct from the constructs measured by other available tools.

FEAST Total and subscale scores can help occupational therapy practitioners identify areas to prioritize in intervention, and the additional FEAST Part 2 items can guide child- and family-specific goal setting and treatment planning. For example, a child may present with family-reported disruptive mealtime behavior and, on evaluation, the FEAST Oral Motor Skills subscale is noted to have the highest score, indicating significant feeding challenges in oral motor skills. An occupational therapy practitioner may thus adjust the treatment plan to address oral motor challenges as an underlying antecedent to observed behavioral responses during mealtime.

The findings of this analysis should be applied to clinical practice with consideration of the study’s strengths and limitations. The FEAST is a valuable clinical tool; however, as with any standardized questionnaire it may not capture all relevant factors related to a child’s feeding challenges, such as the family’s food security, cultural practices, and feeding priorities. The FEAST should be used in combination with clinician interviews and other evaluation tools in the context of a family-centered evaluation. Another limitation of this study is that the sample was recruited using a research registry, and participants had self-selected to take part in the Survey for Characterization of Feeding Challenges in Autistic Children–US. Although this allowed a large national sample, the demographics of the sample population were found to be moderately homogeneous, which may limit the generalizability of these findings. Feeding and mealtime practices are culturally embedded; therefore, validation of the FEAST with additional, diverse samples is warranted. In addition, the use of a research registry may have skewed the results toward people who are more motivated to participate in research studies, namely, families of higher socioeconomic status, older autistic children, and those who use public health insurance (Kalb et al., 2019). Although families with autistic children ages 2 to 12 yr were included in the sample, the mean participant age was 8.66 yr; therefore, developmental differences across the sample’s age range should be taken into consideration when applying the results of this study in practice. Future research should further establish the validity of the FEAST with a broader sample, including nonautistic children, and aim to determine scoring interpretation guides and cutoff scores.

This study has the following implications for occupational therapy practice:

• ▪ Occupational therapy practitioners are experts on how functional performances are shaped by interactions among the person, environment, and occupation. Unlike other assessments, such as the BAMBIC, which primarily measure mealtime behaviors, the FEAST addresses health, environmental, and occupational involvement to provide a more comprehensive picture of eating and feeding challenges for autistic children and their families, which may be particularly useful for occupational therapy practitioners.

• ▪ The FEAST provides occupational therapy practitioners with a validated, autism-specific assessment tool for feeding challenges, including subscale measurements that can inform child-centered intervention planning. The FEAST adds a unique opportunity to consider gastrointestinal health concerns when evaluating feeding challenges in autistic children. The addition of Gastrointestinal Health subscale scores may serve as a catalyst for interdisciplinary collaboration between occupational therapy practitioners and gastrointestinal specialists to provide coordinated treatment.

• ▪ Successful collaboration between clinicians and families begins with evaluation. The FEAST directly measures additional constructs that are not captured by existing assessments. Clinicians using this tool can confidently identify which constructs (sensory, behavior, gastrointestinal health, and oral motor skills) are contributing to a child’s difficulties with feeding. This understanding can allow them to better design feeding interventions that target specific challenge areas and partner with caregivers to improve their experiences and enable mealtime participation of their autistic children in everyday life.

• ▪ The FEAST is a caregiver-report measure and is not intended to be a diagnostic tool. For clinical practice, observation used in conjunction with the FEAST may provide clinicians with a better understanding of caregiver concerns and children’s eating habits. In addition, the FEAST has more items relative to the BAMBIC and the STEP-CHILD, and therefore it takes longer to complete. The FEAST may be most appropriate to administer when a feeding or eating challenge is identified by caregivers and less appropriate for routine developmental screenings.

The FEAST assessment is a recently developed, validated, and autism-specific feeding assessment. The results of this study demonstrate that the FEAST has adequate convergent validity with two other validated feeding questionnaires, the BAMBIC and STEP-CHILD. The findings support that the FEAST expands on available assessment of mealtime behavior with subscale measurement of sensory processing, oral motor skills, and gastrointestinal health. The FEAST assessment provides clinicians, researchers, and caregivers with a more comprehensive assessment of feeding and eating challenges for families with autistic children than previous assessments.

The FEAST is available and free for clinicians to use. Please contact the authors to request access.