CONTEXT: Children with autism spectrum disorder (ASD) frequently use special diets or receive nutritional supplements to treat ASD symptoms.
OBJECTIVES: Our objective was to evaluate the effectiveness and safety of dietary interventions or nutritional supplements in ASD.
DATA SOURCES: Databases, including Medline and PsycINFO.
STUDY SELECTION: Two investigators independently screened studies against predetermined criteria.
DATA EXTRACTION: One investigator extracted data with review by a second investigator. Investigators independently assessed the risk of bias and strength of evidence (SOE) (ie, confidence in the estimate of effects).
RESULTS: Nineteen randomized controlled trials (RCTs), 4 with a low risk of bias, evaluated supplements or variations of the gluten/casein-free diet and other dietary approaches. Populations, interventions, and outcomes varied. Ω-3 supplementation did not affect challenging behaviors and was associated with minimal harms (low SOE). Two RCTs of different digestive enzymes reported mixed effects on symptom severity (insufficient SOE). Studies of other supplements (methyl B12, levocarnitine) reported some improvements in symptom severity (insufficient SOE). Studies evaluating gluten/casein-free diets reported some parent-rated improvements in communication and challenging behaviors; however, data were inadequate to make conclusions about the body of evidence (insufficient SOE). Studies of gluten- or casein-containing challenge foods reported no effects on behavior or gastrointestinal symptoms with challenge foods (insufficient SOE); 1 RCT reported no effects of camel’s milk on ASD severity (insufficient SOE). Harms were disparate.
LIMITATIONS: Studies were small and short-term, and there were few fully categorized populations or concomitant interventions.
CONCLUSIONS: There is little evidence to support the use of nutritional supplements or dietary therapies for children with ASD.
- ASD —
- autism spectrum disorder
- CGI —
- Clinical Global Impression
- GFCF —
- RCT —
- randomized controlled trial
- SOE —
- strength of the evidence
Autism spectrum disorder (ASD) is characterized by impairments in social interaction, communication, and behavior as well as sensory challenges. Substantial evidence supports benefits of specific behavioral, educational, and some pharmacologic interventions for children with ASD. However, given the limits of available treatments in improving core and associated ASD symptoms, substantial resource challenges in accessing evidence-based treatment approaches, and perceptions regarding lessened risks of treatment, many families, if not a majority of families, pursue dietary and nutritional approaches as components of treatment.1–11 Given limitations in the existing research base, families and providers alike often struggle to understand the safety and potential benefit of such approaches.
Proponents of the frequently used gluten/casein-free (GFCF) diet posit varied theories (eg, excess opioid peptide levels) regarding why individuals with ASD may have altered metabolism of gluten or casein proteins that may negatively affect behavior.12–14 Evidence to support specific theories, however, is lacking.15,16 Studies have also explored differences in nutrient status in children with and without ASD and potential correlations with ASD symptoms as well as the effects of vitamin supplementation. The results of these studies have been inconclusive.3,17–28
Despite limited evidence and limited understanding of the potential mechanisms underlying variations in nutrition and metabolism that may affect behavior, specialized or restricted diets and nutritional supplementation are frequently used treatments in children with ASD. In the present review, a component of an Agency for Healthcare Research and Quality–commissioned update of a comparative effectiveness review of therapies for children with ASD conducted by the Vanderbilt Evidence-based Practice Center,29 we examine the evidence specifically for nutritional or dietary interventions in children with ASD. The full comparative effectiveness review update30 and review protocol (PROSPERO registry number: CRD42016033941) are available at www.effectivehealthcare.ahrq.gov.
Search Strategy and Study Selection
We searched the Medline database via PubMed, Embase, and the Cochrane Library from January 2010 to September 2016 using a combination of controlled vocabulary and key terms related to interventions for ASD (eg, autism, ASD, therapy). We note that the original review,31 which the current report updates, included studies published from January 2000 to 2011. We also hand-searched the reference lists of included articles and recent reviews addressing ASD therapies to identify potentially relevant articles.
We developed inclusion criteria in consultation with an expert panel of clinicians and researchers (Table 1). We included comparative study designs (eg, randomized controlled trials [RCTs] and prospective or retrospective cohort studies) and studies published in English. We required that eligible RCTs have a total minimum sample size of 10. We required a higher minimum sample size (n = 20) for other comparative studies because they typically have fewer controls for bias than RCTs.
Data Extraction and Analysis
One investigator extracted data regarding study design, descriptions of study populations, intervention and comparison groups, and baseline and outcome data using a standardized form. A second investigator independently verified the accuracy of the extraction and made revisions as needed. Significant heterogeneity in interventions and outcomes reported precluded meta-analysis; thus, we synthesized studies qualitatively.
Assessment of Study Risk of Bias and Strength of Evidence
Two investigators independently evaluated the overall methodologic risk of bias of individual studies using the ASD-specific assessment approach developed and used in previous reviews of interventions for ASD.29,32,33 Senior reviewers resolved discrepancies in risk of bias assessment, and we used an approach described in the full review34 to determine low, moderate, or high risk of bias ratings.
Assessment of the strength of the evidence (SOE) reflects the confidence that we have in the stability of treatment effects in the face of future research. The degree of confidence that the observed effect of an intervention is unlikely to change in additional research, the SOE, is presented as insufficient, low, moderate, or high. Assessments are based on consideration of study limitations, consistency in the direction of the effect, directness in measuring intended outcomes, precision of effect, and reporting bias.35 We determined the strength of evidence separately for major intervention-outcome pairs using a prespecified approach described in detail in the full review.34
Our searches (conducted for the broader systematic review update30) identified 6583 citations, of which 19 RCTs (reported in multiple publications) met inclusion criteria and addressed diet or nutritional therapies (Fig 1).36–58 Seventeen of these studies were published after the completion of our initial review of therapies for children with ASD,29 and 2 were included in the previous review.49,50 Four RCTs had low risk of bias,38,41,43,47 10 had moderate,36,37,39,40,46,49–51,53,56–58 and 5 had high risk.42,44,45,48,52,54,55 Table 2 outlines study characteristics and risk-of-bias assessments. Study treatment durations ranged from 7 days to 2 years, and sample sizes ranged from 12 to 92 (total N = 732). Follow-up occurred immediately posttreatment in all studies.
Ω-3 Fatty Acid Supplementation
Little evidence supports the effectiveness of Ω-3 supplementation to improve core or associated ASD symptoms. Three RCTs of Ω-3s versus placebo (low41 and moderate39,40 risk of bias) reported no significant group differences on most measures of challenging behavior, communication, language, and adaptive behavior.39–41 One study reported significantly improved scores in the placebo group compared with the Ω-3 group in externalizing behaviors after 6 months of treatment,39 and another reported a significant improvement in parent ratings of stereotypy and lethargy in children receiving Ω-3 supplements compared with those receiving placebo; teacher ratings were not significantly different.40 Another RCT (moderate risk of bias) of dietary docosahexanoic acid (DHA) supplementation versus placebo reported improvement in parent-rated social skills in children receiving placebo versus those receiving docosahexanoic acid, whereas teachers rated communication as more improved in the treatment group compared with placebo.37 Scores on other measures did not differ significantly between groups. Supplemental Table 3 reports outcome data for all studies.
Digestive Enzyme Supplementation
Evidence is inadequate to assess the effects of short-term digestive enzyme supplements. Two RCTs (moderate risk of bias) addressed digestive enzyme supplements compared with placebo: 1 evaluated a proteolytic enzyme supplement (Peptizyde)49 and the other a digestive enzyme supplement (Neo-Digestin)36; both supplements contained papain and pepsin or peptidase. The Peptizyde RCT reported no significant differences in measures of behavior, sleep quality, or gastrointestinal symptoms, and no significant differences in adverse effects.49 In a 3-month trial of Neo-Digestin versus placebo, symptom severity scores improved significantly in the treatment group compared with placebo.36
Two RCTs (low38 and moderate58 risk of bias) addressed methyl B12 supplementation. Clinical Global Impression (CGI) scores improved significantly in the methyl B12 group in 1 RCT (effect size = 0.84, P = .005), but studies reported few other significant group differences in measures of behavior or communication.58 Two RCTs of levocarnitine (moderate51 and high42 risk of bias) reported improvements in symptom severity in the levocarnitine group compared with placebo, but scores on other behavioral measures or adverse effects did not differ between groups.42 In the second RCT, symptom severity did not differ between groups after 6 months of treatment.51
Data to assess the effects of GFCF diets are limited because dietary approaches and outcome measures varied among studies as did control diets and monitoring of adherence to GFCF diets. Four RCTs (in multiple publications) compared GFCF diets to either an unaltered diet44,45,54–57 or a low-sugar diet (total N across studies = 82).53 One RCT (moderate risk of bias) reported no significant differences between groups on measures of development or behavior, although the control group improved significantly from baseline on visual reception, withdrawal, aggression, and attention subscales (P values < .05) Another crossover RCT (moderate risk of bias) similarly reported no statistically significant differences between groups on measures of symptom severity or language, although parents of 7 of the 15 children participating in the study reported improvements in language.56,57 In a retrospective analysis of videotapes recorded during the study period, investigators found no significant group differences in verbal communication between children in the diet or control groups or between children whose parents reported language improvements after the study period and those whose parents did not.
One RCT (high risk of bias) reported significant parent-rated improvements in communication, resistance to communication, social isolation, repetitive or challenging behavior, and overall impairment in children on a GFCF diet compared with those on a usual diet (P values ≤ .007).54,55 Children on the GFCF diet also improved significantly on tests of cognitive skills, motor skills, verbal and social communication, anxiety, and reaction to changes in environment and routine compared with control children (P values < .05). Another high risk of bias RCT with 24-month follow-up of participants reported few differences in behavioral measures between children on a GFCF diet and those with no dietary restrictions44,45; ASD symptoms improved significantly in participants in the GFCF diet group versus the no diet group at 12 months, but were not different on any measure in a subset of participants followed for 24 months.
Other Dietary Approaches
One RCT (high risk of bias) compared a gluten-free diet to a usual diet and reported significant improvements in gastrointestinal symptoms (stomachache, bloating, constipation) from baseline in the gluten-free diet group, but not in the control group. Diarrhea did not improve significantly in either group. Stereotyped behavior and communication improved significantly in the gluten-free group compared with control children (P values ≤ .005).52 Another small RCT (low risk of bias) comparing a gluten- and dairy-free diet with a diet including both gluten and dairy reported no significant group differences in challenging behavior (hyperactivity, irritability, inattention).47
Two small RCTs (low59 and moderate46 risk of bias) evaluated the “challenges” of gluten- or casein-containing foods, but the evidence is inadequate to determine if short-term gluten- or casein-containing foods affect ASD symptoms or gastrointestinal function. One RCT that randomized children who were maintaining GFCF diets to foods with gluten, gluten and casein, or placebo foods reported no significant group differences in measures of challenging behaviors or measures of sleep quality and stool frequency.43 Another RCT (moderate risk of bias) assessing the effects of introducing gluten/casein-containing foods versus placebo foods similarly reported no significant effects of added gluten or casein on behavior or gastrointestinal symptoms.46 Finally, a single RCT (high risk of bias) compared boiled or raw camel’s milk with cow’s milk and reported no significant differences in ASD severity between groups after 2 weeks of treatment.48
Studies that reported harms either reported no significant difference between the intervention group and the control group, or reported 0 harms for each group. Harms were disparate and the clinical significance was generally difficult to assess (Supplemental Table 4).
Ω-3 fatty acid supplementation and placebo did not affect challenging behaviors and was associated with minimal harms. Our confidence in these conclusions is low (low SOE). Despite the number of RCTs with low or moderate risk of bias addressing other supplements, data were inadequate to make conclusions about all clinical efficacy and harms outcomes because only a few small studies addressed each supplement (insufficient SOE). Similarly, although multiple RCTs evaluated variations of a GFCF diet, studies addressed different outcomes and different approaches to restricted and control diets; thus, data were inadequate to make conclusions about the body of evidence (insufficient SOE) or about other dietary interventions (challenge foods, camel’s milk).
Despite their widespread reported use, little evidence supports the effectiveness of nutritional supplements or the GFCF diet for improving ASD symptoms. Harms reported in studies were generally considered mild, but the long-term effects of these therapies are not well understood. Although the conduct of studies generally improved from those reported in our 2011 review, evidence remains insufficient for most interventions given the small sample sizes, lack of longer term follow-up, and heterogeneous agents and populations. Few studies assessed the effect of concomitant behavioral or other therapies, although many children with ASD receive multiple interventions.
These findings generally align with conclusions in recent reviews addressing specific diets or supplements. One Cochrane review evaluating Ω-3 fatty acids reported no evidence for effects on social interaction, communication, hyperactivity, or stereotypy.27
Another review of GFCF diets included 32 studies, typically with high risk of bias, and noted scarce evidence for GFCF diets, with positive effects reported only in lower quality studies.60
Even without a clear evidence base documenting safety and efficacy, many families of children with ASD use diet and nutritional approaches.61,62 Parents have cited better alignment with their personal views as well as perceived fewer side effects than conventional medications as reasons for using “complementary or alternative” therapies, including restricted diets and nutritional supplements.5–9 Caregivers have also reported making treatment decisions about such therapies without a clinician’s input, noting a perceived unwillingness to consider potential benefits among clinicians, even in the face of few evidence-based effective therapies.63 These findings continue to highlight the need for shared decision-making among providers and families, including understanding of family motivations for using specific therapies and discussion of balancing potential benefits with potential risks and resource and time costs to families.64,
Limitations of the Review Process
We included studies published in English only and did not include unpublished data. Although our preliminary scan of non-English publications identified few potential eligible studies, we recognize that some nutritional supplements may have been studied only in non-Western countries and will not be addressed in the current review. We also included only comparative studies of medical interventions with at least 10 children with ASD. Given the heterogeneity in treatment regimens, outcomes addressed in each study, and patient populations, we were limited in our ability to meta-analyze findings or identify potential subgroups that may respond more favorably to specific treatments. Finally, we used a nonvalidated tool to assess risk of bias; this tool evaluates constructs similar to those assessed by organizations such as the Cochrane Collaboration, with the addition of ASD-specific domains.
Overall, studies of nutritional supplements or specialized diets were typically small and short-term (<6 months) and provided little evidence regarding the potential effects of these approaches. Several agents were addressed in single studies, which limit conclusions about their effects. These findings can help to inform shared caregiver and clinician decision-making about therapies for children with ASD.
Dr Shanthi Krishnaswami and Ms Jessica Kimber contributed to the data extraction. We thank the full research team and of the Agency for Healthcare Research and Quality task order officers and associate editor for their input.
- Accepted February 10, 2017.
- Address correspondence to Zachary E. Warren, PhD, Departments of Pediatrics and Psychiatry, Vanderbilt Kennedy Center, Vanderbilt University Medical Center, PMB 40, 230 Appleton Pl, Nashville, TN 37203. E-mail:
This manuscript was derived from a systematic review conducted by the Vanderbilt Evidence-based Practice Center, “Medical Therapies for Children with Autism Spectrum Disorder (ASD)—an Update,” which will be published in full on the Agency for Healthcare Research and Quality Web site.
This project was funded under contract HHSA209201500003I from the Agency for Healthcare Research and Quality, US Department of Health and Human Services. The authors of this manuscript are responsible for its content. Statements in the manuscript should not be construed as endorsement by the Agency for Healthcare Research and Quality or the US Department of Health and Human Services. Agency for Healthcare Research and Quality retains a license to display, reproduce, and distribute the data and the report from which this manuscript was derived under the terms of the agency’s contract with the author.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: This project was supported by the Agency for Healthcare Research and Quality (contract HHSA290201500003I).
POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.
COMPANION PAPER: A companion to this article can be found online at: www.pediatrics.org/cgi/doi/10.1542/peds.2017-0730.
- Perrin JM,
- Coury DL,
- Hyman SL,
- Cole L,
- Reynolds AM,
- Clemons T
- Goin-Kochel RP,
- Myers BJ,
- Mackintosh VH
- Warren Z,
- Veenstra-VanderWeele J,
- Stone W; Vanderbilt Evidence-based Practice Center, et al
- Weitlauf AS,
- Sathe NA,
- McPheeters ML,
- Warren Z; Vanderbilt Evidence-based Practice Center
- Weitlauf AS,
- McPheeters ML,
- Peters B; Vanderbilt Evidence-based Practice Center, et al
- Lounds Taylor J,
- Dove D,
- Veenstra-VanderWeele J; Vanderbilt Evidence-based Practice Center, et al
- Williamson E,
- Sathe NA,
- Andrews JC, et al
- ↵Methods guide for effectiveness and comparative effectiveness reviews. AHRQ Publication No. 10(14)-EHC063-EF. Available at: http://effectivehealthcare.ahrq.gov/index.cfm/search-for-guides-reviews-and-reports/?pageaction=displayproduct&productid=318.gov. Accessed April 29, 2017
- Pedersen L,
- Parlar S,
- Kvist K,
- Whiteley P,
- Shattock P
- Fahmy SF,
- El-hamamsy MH,
- Zaki OK,
- Badary OA
- Johnson CR,
- Handen BL,
- Zimmer M,
- Sacco K,
- Turner K
- Seung H,
- Rogalski Y,
- Shankar M,
- Elder J
- Hebert EB
- Committee on Children with Disabilities, American Academy of Pediatrics
- Copyright © 2017 by the American Academy of Pediatrics