Gluten Free and Autism

Kelley L. Harrison, MA, BCBA, LBA-KS
Thomas Zane, PhD, BCBA-D
Department of Applied Behavioral Science, University of Kansas

Parents of children with autism often feel a sense of urgency to find a treatment that will help their children realize their fullest potential and live an independent life. This sometimes makes families anxious to try new treatments that tout promising claims, and makes them impatient to wait on scientific evidence to support these claims. One such widely used intervention is the gluten-free and casein-free (GFCF) diet, which completely eliminates gluten (found in wheat products) and casein (found in dairy products) from the individual’s diet (Elder, 2008). In an online survey conducted in 2006, up to 27% of parents of children with autism reported using a GFCF diet (Green et al., 2006). Additionally, Perrin et al. (2012) conducted a large-scale survey including over 3,000 individuals with autism spectrum disorder (ASD) and found 17% of the individuals with ASD used special diets of some kind, with the most popular of these diets being a GFCF diet.

What is the Conceptual and Scientific Link Between ASD and GFCF Diets?

The exact etiology of autism is still unknown, leading researchers to consider many possible factors, diet being one. Food restriction as a strategy for changing behavior is not new, with research dating back to the 1920s. The “opioid-excess theory” (Panksepp, 1979) or “leaky-gut syndrome” (Shattock & Whiteley, 2002) has been one of the predominant conceptualizations supporting the role of food and diet in autism symptomology. Panksepp (1979) proposed that symptoms of autism may be due to excessive opioid activity. Shattock and Whiteley (2002) proposed that children with autism have abnormal leakage from their gut (due to increased intestinal permeability). Both gluten and casein are broken down in the gut into compounds with opioid properties. Therefore, some suggest that the abnormal leakage from the gut allows the opioids to pass into the central nervous system and produce increased brain opioid activity and ultimately, disrupt brain function (Christison & Ivany, 2006). Much of the research surrounding the biological connections of gluten and casein and autism symptoms stems from an article by Wakefield and colleagues (1998) in which the authors suggested three potential environmental links to autism: 1) a potential link between behavior abnormalities and variations in gastrointestinal functioning, 2) a potential link between autism and the Mumps, Measles, and Rubella (MMR) vaccination, and 3) a potential link between autism and a deficiency in vitamin B12. However, due to undisclosed financial interests of the authors, significant inaccuracies in the paper, and what eventually was shown to be deliberate fraud, Wakefield and colleagues were found guilty of ethical violations and scientific misrepresentation. The findings of the research were eventually rejected and the article was pulled from the journal.

Other researchers investigating the opioid-excess theory often produced conflicting results. Support for this theory is drawn from reports of increased intestinal permeability as well as increased urinary peptide levels. For example, D’Eufemia and colleagues (1996) found abnormal intestinal permeability in 43% of their tested population with autism, but in none of their control population. Contrary to this study, Robertson and colleagues (2008) found no gastrointestinal differences between children with intellectual or developmental disabilities and typically-developing children. Likewise, Reichelt, Knivsberg, and Nodland (1991) found increased urinary peptide levels in children with ASD, but Cass and colleagues (2008) found no significant differences in urinary peptide levels between children with ASD and typically-developing children.

Conceptually, a GFCF diet may seem like a logical treatment for ASD because it eliminates two compounds with opioid properties (i.e., gluten and casein) from the diet. If individuals with ASD do indeed have increased opioid levels and increased intestinal permeability (the combination of which may lead to a disruption in brain function), then eliminating food with opioid properties should decrease opioid levels and, in turn, eliminate or at least decrease disruption in brain function. However, these suggested gastrointestinal differences (i.e., increased opioid levels and increased intestinal permeability) in individuals with ASD are not consistently observed, thus diminishing the conceptual and scientific link between ASD and GFCF diets.

What Does Research Have to Say About ASD and GFCF Diets?

Research on specifically restricting gluten and casein in the diet of children with autism is relatively limited, despite its popularity as a treatment for autism (Elder 2008). Proponents of a GFCF diet suggest benefits across a wide range of symptoms related to autism, with changes in social engagement and verbal skills being the most commonly noted (Christison & Ivany, 2006). However, support for the GFCF diet comes primarily from clinicians, parents, and educators and is based largely on self-report, not direct measurement (Seung, Rogalski, Shankar, & Elder, 2007), thus limiting the validity of the findings.

Researchers investigating the behavioral effects of a GFCF diet on children with autism produce conflicting results. Knivsberg and colleagues have conducted several studies evaluating the effects of GFCF diets on behavior as well as biological markers of individuals with autism. Among the most experimentally rigorous, Knivsberg, et al. (2002) compared a test group of children who received a GFCF diet to a control group of children who experienced no change in diet. Both test and control groups consisted of children with autism and abnormal urinary peptide patterns. A single-blind design was used and behavior measurements were collected using validated measurement instruments in baseline and then again one year later. Results suggested significantly greater improvement in the diet group as compared to the control group across ratings of attention, social and emotional behaviors, communicative ability, cognitive factors, sensory/motor behaviors, and total impairment. However, this study (like most of the research on GFCF diets) was limited because parents, teachers, and participants were not blind to the treatment conditions, and the primary data were subjective in nature and derived from interviews with the parents.

Other researchers have completed studies with better experimental rigor to address some of these experimental limitations. For example, Elder and colleagues (2006) completed a double-blind test that included direct behavioral observations, biological tests (i.e., urinary peptide levels), and validated measurement instruments (i.e., Childhood Autism Rating Scale; CARS). This study addressed the limitation of parents, teachers, and participants knowing the treatment conditions by using a double-blind test. It also addressed the limitation of relying on interviews for detecting behavior change by directly measuring differences in behavior using behavioral observations before and after implementing the GFCF diet. With these limitations addressed, Elder et al. (2006) found no significant differences between the test and control group across the direct behavior observations, the urinary peptide levels, and the CARS assessment.

More recently, Hyman and colleagues (2016) conducted a double-blind test in which 14 children with autism were placed on a GFCF diet. Following an extended period of time (i.e., 2-6 weeks) on a GFCF diet, food items containing gluten and/or casein were introduced into the participants’ diet once a week for 12 weeks, unbeknownst to the participants. Data were collected on three domains that are often identified by proponents of GFCF diets as targets of the diet. Specifically, data were collected on physiologic functioning (i.e., bowel movements), challenging behaviors (i.e., sleep, activity, attention) and behaviors associated with ASD (i.e., sensory motor behavior, social relationships, affectual reactions, sensory responses, language). Data were collected on observed behaviors using well-established rating scales. Like Elder and colleagues (2006), Hyman and colleagues found no statistically significant differences in physiologic functioning, behavior problems, or ASD symptoms when children followed a GFCF diet as compared to when gluten and casein were reintroduced into their diet.

In addition to studies that utilize group design, Irvin (2006) utilized a within-subject design to evaluate the effects of the GFCF diet on the rate of problem behavior exhibited by a 12-year-old boy diagnosed with autism and an intellectual disability. This design allows for rigorous experimental control because it controls for biological, environmental, and/or behavioral differences that may exist across participants in a group design. Specifically, Irvin conducted an assessment of problem behavior under various environmental conditions within two diet phases (i.e., GFCF and regular diet). Results suggested that problem behavior did not decrease during the GFCF diet phase as compared to the regular diet phase. That is, the absence of gluten and casein in this participant’s diet did not result in a decrease in problem behavior. Interestingly, Irvin also reported a marked increase in meal refusal during the GFCF diet phase. This is important to note because GFCF diets are suggested to be associated with health risks due to the decreased nutrient intake (e.g., protein; Hediger et al., 2008). Any health risks due to a decreased nutrient intake are likely to be exacerbated if meals in general are refused.

Although these more recent studies begin to address the limitations of previous research, they are still limited in terms of using relatively small sample sizes and not assessing the impact of the diet over a long period of time. Some research has suggested behavior change as a result of the GFCF diet does not occur immediately and instead occurs after several months of exposure to the diet (Seung et al., 2007). Nonetheless, when major limitations to experimental rigor are addressed (e.g., participants are unaware of treatment phase, behavioral changes are based on direct observation), research tends to suggest GFCF diets do not treat symptoms associated with ASD.

Financial Cost and Health Risks Associated With GFCF Diets

In addition to the mixed research results, there are other factors to consider when evaluating GFCF diets as a treatment for ASD. Stevens & Rashid (2008) estimated that, on average, it costs 242% more to buy gluten-free foods as compared to its gluten-containing counterparts. The price difference ranges from 32% among meats to 455% among soups and sauces. Additionally, placing a child on a GFCF diet may be socially stigmatizing because the child cannot eat the same food as his or her peers (Mulloy, et al., 2009).

Finally, GFCF diets may be associated with health risks because they are often associated with diminished essential nutrients. Hediger and colleagues (2008) found significantly reduced bone thickness (which leads to increased fracture risk) in boys on a casein-free diet as compared to boys on a minimally restricted or unrestricted diet. The authors suggested this may be due to the decreased calcium and vitamin D intake associated with a casein-free diet. Similarly, Arnold, Hyman, Mooney, and Kirby (2003) found that children on GFCF diets have significantly poorer protein nutrition than children on unrestricted diets, potentially placing developing brains at risk for protein malnutrition. In a comprehensive review of research regarding feeding problems and nutrient status of children with ASD, Sharp and colleagues (2013) found children with ASD are significantly more likely to experience feeding problems as compared to their peers. Additionally, a nutrient analysis indicated a significantly lower intake of calcium and protein for children with ASD in general. This is potentially significant because children with ASD are likely to exhibit behavioral rigidity which often leads to a preference for snack foods and the consumption of fewer vegetables and fruits as compared to typically-developing peers (Foxx & Mulick, 2016). Thus, a child with ASD may already be at risk for detrimental health side effects associated with a diet low in calcium, vitamin D, and protein (Sharp et al., 2013). A child with ASD who also consumes a GFCF diet may essentially double his or her risk for detrimental health side-effects associated with a diet low in essential nutrients.

Future Research

If one felt compelled to investigate these diets further, future research may focus on: a) using a larger sample size for group studies, b) conducting studies utilizing within-subjects designs to control for individual differences often seen across individuals with ASD, c) conducting longer baseline and test periods to ensure the diet is in place long enough to capture any results, and d) continuing to utilize direct behavior measurements and double-blind tests (Elder, 2008). Additionally, future research may consider whether positive outcomes associated with GFCF diets are due to the absence of gluten and casein or simply due to a more nutritious diet that may be associated with a diet free of gluten and casein. Because children with ASD are likely to consume a limited diet (that includes a preference for snack foods), a GFCF diet may simply result in the consumption of a more nutritiously balanced diet because it likely eliminates foods low in fiber or high in fat and increases the consumption of fruits and vegetables. Additionally, it may eliminate excessive intake of dairy products which can lead to constipation. It is possible that positive behavior change may be observed simply because the individual is consuming a healthier diet and thus, feels better physically (e.g., more energy, eliminate constipation; Foxx & Mulick, 2016).

What is the bottom line?

If any child is allergic to gluten or casein, then the GFCF diet should be considered. However, without a clear medical necessity, until further research suggests a strong causal relationship between the GFCF diet and specific and measurable positive changes in socially important targets of individuals with autism, it should not be used to treat this disorder (Mulloy et al., 2009). The biological findings surrounding the opioid-excess theory are limited and mixed. The behavioral changes resulting from the use of a GFCF diet with children with autism are limited and mixed. The financial cost and associated health risks further suggest avoiding a GFCF diet as a treatment for autism. Diet changes should always be decided upon and closely monitored by a physician. Additionally, parents choosing to use this treatment should weigh all costs associated with the diet (e.g., money, time, ability to evaluate effects, health status of child, social stigmatization) against the uncertain benefits.


Arnold, G. L., Hyman, S. L., Mooney, R. A., & Kirby, R. S. (2003). Plasma amino acids profiles in children with autism: Potential risk of nutritional deficiencies. Journal of Autism and Developmental Disorders, 33, 449-454.

Cass, H., Grigras, P., March, J., McKendrick, I., O’Hare, A. E., Owen, L., & Pollin, C. (2008). Absence of urinary opioid peptides in children with autism. Archives of Disease in Childhood, 93, 745-750.

Christison, G. W., & Ivany, K. (2006). Elimination diets in autism spectrum disorders: Any wheat amidst the chaff? Developmental and Behavioral Pediatrics, 27, 162-171.

D’Eufemia, R., Celli, M., Finocchiaro, R., Pacifico, L., Viozzi, L., Zaccagnini, M., … Giardini, O. (1996). Abnormal intestinal permeability in children with autism. Acta Pediatrics, 85, 1076-1079.

Elder, J. H. (2008). The gluten-free, casein-free diet in autism: An overview with clinical implications. Nutrition in Clinical Practice, 23, 583-588.

Elder, J. H., Shankar, M., Shuster, J., Theriaque, D., Burns, S., & Sherril, L. (2006). The gluten-free, casein-free diet in autism: Results of a preliminary double blind clinical trial. Journal of Autism and Developmental Disabilities, 36, 413-420.

Foxx, R. M., & Mulick, J. A. (2016). Controversial therapies for autism and intellectual disabilities: Fad, fashion, and science in professional practice. New York, NY: Routledge.

Green, V. A., Pituch, K. A., Itchon, J., Choi, A., O’Reilly, M., & Sigafoos, J. (2006). Internet survey of treatments used by parents of children with autism. Research in Developmental Disabilities, 27, 0-84.

Hediger, M. L., England, L. J., Molloy, C. A., Yu, K. F., Manning-Courtney, P., & Mills, J. L. (2008). Reduced bone cortical thickness in boys with autism or autism spectrum disorder. Journal of Autism and Developmental Disorders, 38, 848-856.

Hyman, S. L., Stewart, P. A., Foley, J., Cain, U., Peck, R., Morris, D. D., … Smith, T. (2016). The gluten-free/casein-free diet: A double-blind challenge trial in children with autism. Journal of Autism and Developmental Disorders, 46, 205-220.

Irvin, D. S. (2006). Using analog assessment procedures for determining the effects of a gluten-free and casein-free diet on rate of problem behavior for an adolescent with autism. Behavioral Interventions, 21, 281-286.

Knivsberg, A. M., Reichelt, K. L., Holen, T., & Dodland, M. (2002). A randomized, controlled study of dietary intervention on autistic syndromes. Nutritional Neuroscience, 5, 251-261.

Mulloy, A., Lang, R., O’Reilly, M., Sigafoos, J., Lancioni, G., & Rispoli, M. (2009). Gluten-free and casein-free diets in the treatment of autism spectrum disorders: A systematic review. Research in Autism Spectrum Disorders, 4, 1-12.

Panksepp, J. (1979). A neurochemical theory of autism. Trends in Neuroscience, 2, 174-177.

Perrin, J. M., Coury, D. L., Syman, S. L., Cole, L., Reynolds, A. M., & Clemons, T. (2012). Complementary and alternative medicine use in a large pediatric autism sample. Pediatrics, 130, 877-882.

Robertson, M., Sigalet, D., Holst, J., Meddings, J., Wood, J., & Sharkey, K. (2008). Intestinal permeability and glucagon-like peptide-2 in children with autism: A controlled pilot study. Journal of Autism and Developmental Disorders, 38, 1066-1071.

Reichelt, K., Knivsberg A., Lind, G., & Nodland, M. (1991). Probable etiology and possible treatment of childhood autism. Brain Dysfunction, 4, 308-319.

Seung, H., Rogalski, Y., Shankar, M., & Elder, J. (2007). The gluten-free and casein-free diet and autism: Communication outcomes from a preliminary double-blind clinical trial. Journal of Medical Speech-Language Pathology, 15, 337-345.

Sharp, W. G., Berry, R. C., McCracken, C., Nuhu, N. N., Marvel, E., Saulnier, C. A., … Jaquess, D. L. (2013). Feeding problems and nutrient intake in children with autism spectrum disorders: A meta-analysis and comprehensive review of the literature. Journal of Autism and Developmental Disorders, 43, 2159-2173.

Shattock, P., & Whiteley, P. (2002). Biochemical aspects in autism spectrum disorders: Updating the opioid-excess theory and presenting new opportunities for biomedical intervention. Expert Opinion on Therapeutic Targets, 6, 175-183.

Stevens, L., & Rashid, M. (2008). Gluten-free and regular foods: A cost comparison. Canadian Journal of Dietetic Practice and Research, 69, 147-150.

Wakefield, A. J., Murch, S. H., Anthony, A., Linnell, J., Casson, D. M., Malik, M., … Walker-Smith, J. A. (1998). Ileal-lymphoid-nodular hyperplasia, non-specific colitis, and pervasive developmental disorder in children. Lancet, 351, 637-641.

Citation for this article:

Harrison K. L., & Zane, T. (2017). Is there science behind that? Gluten-free and casein-free diets. Science in Autism Treatment, 14(2), 32-36.

Print Friendly, PDF & Email