Abstract
BACKGROUND: In pediatric settings, parents often raise concerns about possible gastrointestinal (GI) symptoms in autism spectrum disorder (ASD), yet the specificity of these concerns are not well studied.
OBJECTIVE: To conduct a meta-analysis of research investigating GI symptoms among children with ASD.
METHODS: We searched Medline, PsycINFO, and PubMed databases (1980–2012) in peer-reviewed journals. Analysis involved studies with a comparison group presenting quantitative data on GI symptoms using combinations of terms for ASD and GI indicators. The systematic search yielded 15 studies. We calculated effect sizes and 95% confidence intervals (CIs) using a random-effects model.
RESULTS: Children with ASD experience significantly more general GI symptoms than comparison groups, with a standardized mean difference of 0.82 (0.24) and a corresponding odds ratio (OR) of 4.42 (95% CI, 1.90–10.28). Analysis also indicated higher rates of diarrhea (OR, 3.63; 95% CI, 1.82–7.23), constipation (OR, 3.86; 95% CI, 2.23–6.71), and abdominal pain (OR, 2.45; 95% CI, 1.19–5.07).
CONCLUSIONS: Results indicate greater prevalence of GI symptoms among children with ASD compared with control children. Identified studies involved high methodological variability and lack of comprehensive data prohibited analysis of GI pathophysiologies (eg, gastroesophageal reflux) typically associated with organic etiologies, limiting conclusions about the underpinnings of the observed association. Future research must address critical questions about the causes and long-term impact of GI symptoms in ASD. Such analyses will require more systematic research and clinical activities, including improved diagnostic screening, standardized assessment, and exploration of potential moderators (eg, dietary restrictions).
- ADI-R —
- Autism Diagnostic Interview–Revised
- ADOS —
- Autism Diagnostic Observation Schedule
- ASD —
- autism spectrum disorder
- CARS —
- Childhood Autism Rating Scale
- CI —
- confidence interval
- DD —
- developmental delay
- ES —
- effect size
- GI —
- gastrointestinal
- MMR —
- measles-mumps-rubella
- OR —
- odds ratio
- SB —
- sibling
- SMD —
- standardized mean difference
- TD —
- typically developing
Autism spectrum disorder (ASD) is a neurodevelopmental condition that unfolds in the first few years of life and involves severe impairments in social interaction and communication, with concomitant restriction in interests and extreme attachment to routine or to repetitive or perseverative behaviors.1 Estimates of ASD in pediatric populations have climbed dramatically over the past decade, with ∼1 in every 88 children currently meeting diagnostic criteria in the United States.2 High prevalence combined with greater health care costs3 and caregiver burden4 associated with ASD intensifies the need to identify the etiology of ASD and develop interventions to remediate core and comorbid symptoms. Gastrointestinal (GI) dysfunction is frequently cited among children with ASD, and many causal and therapeutic hypotheses of ASD involve the GI system.5 This includes the idea that there is a specific GI pathology associated with ASD, triggered by abnormal immune function or elevated intestinal permeability. A great amount of controversy has surrounded this topic since publication and public awareness in 1998 naming a new pathologic entity, “autistic enterocolitis,”6 as responsible for developmental regression in 12 children after administration of the measles-mumps-rubella (MMR) vaccine. Ultimately, this research was retracted7 for several reasons, including questionable research practices, as found by the General Medical Counsel of the United Kingdom. An association between the MMR vaccine and ASD has not been supported in subsequent research8; however, questions remain about the status of GI system function in ASD. Additional theories have posited that children with ASD are at greater risk for gluten sensitivity,9 lactase deficiencies,10 and gene variants.11 Although the presence of a unique GI pathophysiology specific to ASDs has yet to be identified,5 elevated risk for GI symptoms in this population remains a critical issue in pediatric settings, because this population is significantly more likely to use GI agents and experience hospitalizations related to GI disturbance compared with peers.3
Concern about possible GI dysfunction in ASD is intensified by high rates of feeding concerns and consequent medical sequelae in ASD. Children with ASD have a fivefold elevated risk of developing a feeding problem compared with peers.12 Severe food selectivity (ie, eating only a narrow variety of foods) is the most common feeding concern documented among children with ASD, predominantly in the form of strong preferences for starches, snack foods, and processed foods and a bias against fruits, vegetables, and proteins.13 However, feeding concerns in ASD are often overlooked in relation to other areas of clinical concern, probably because selective eating patterns do not necessarily associate greater risk for compromised growth (eg, failure to thrive, declining growth velocity) that trigger clinical attention in pediatric settings.12,13 Evidence suggests that atypical patterns of intake in ASD place this population at risk for long-term nutritional or medical complications not captured by broad anthropometrics or analysis of overall energy intake, including vitamin and mineral deficiencies12 and compromised bone growth.14 Indeed, closer examination of nutrient intake in the ASD population indicates significant specific deficits (lower intake of calcium and protein) and a higher number of nutritional deficits overall among children with ASD.12 Food selectivity in ASD may also explain emerging evidence of a higher incidence of obesity,15 based on dietary patterns involving excessive consumption of processed snacks and calorie-dense foods. Etiological factors contributing to the pattern and prevalence of atypical intake in ASD remain elusive, but may involve pathophysiological processes in the GI tract. Although organic factors leading to difficult or painful eating, such as gastroesophageal reflux, gastroenteritis, and food allergies, often precipitate or play a role in the development of chronic feeding concerns in other pediatric populations,12 research has yet to identify a clear GI link to account for the emergence, maintenance, and topography of feeding problems associated with ASD.
Research focusing on GI dysfunction and feeding concerns in ASD was recently raised to priority status by the National Institutes of Health Interagency Autism Coordinating Committee (http://iacc.hhs.gov/), which cited that in a network of 14 academic health centers throughout the United States, feeding and GI problems were reported in 50% of patients participating in treatment protocols. Moreover, a recent consensus report concluded that the available evidence suggests the likelihood of greater prevalence of GI symptoms and disorders in ASD while also noting that definitive conclusions and evidence-based recommendations regarding GI dysfunction in ASD were not possible because of the absence of high-quality clinical research data.5 This conclusion was based on a qualitative review of the literature involving a pool of 11 studies. Identified studies had wide methodological variability, with less than half of the studies including a comparison group and with data collection consisting primarily of parent reports, chart reviews, or study-specific questionnaires. This article has been cited 26 times in peer-reviewed publications (as of April 2013) and is probably a primary resource for pediatricians in approaching these issues, emphasizing the need for additional research in this area to guide clinical and research activities. Recent growth in the research literature, combined with the availability of quantitative procedures for synthesizing outcome data, present the opportunity for an updated analysis of the extant literature. The current review sought to survey the medical literature to identify studies using empirical methods to investigate GI diagnoses, signs, and symptoms among children with ASD and to summarize the evidence on the basis of both descriptive and meta-analytic procedures.
Methods
Search Strategy and Study Selection
Following the guidelines outlined by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement, we searched Medline, PsycINFO, and PubMed databases (January 1980 and September 2012) and conducted ancestral and online searches in English-language journals for eligible studies. The search parameters (Table 1) included combinations of key words regarding the target population (eg, “autism,” “autistic disorder”) paired systematically with GI indicators (eg, “digestion,” “gastritis”). We cross-checked search results and removed overlapping citations. References from relevant articles and previous reviews in this area5,16 were also searched for possible studies.
ASD and GI Key Words Used in Database Search
To be included in the meta-analysis, studies also had to meet the following criteria: the sample involved a pediatric population (birth to 18 years of age) with ASD; the study used a non-ASD comparison group without identified neurobehavioral delays to analyze GI problems (ie, typically developing peers or siblings); and the study presented data on GI symptoms either descriptively (eg, frequencies, percentages) or statistically (eg, P values, t scores). To avoid a known sampling bias, we excluded studies involving an ASD sample made up exclusively of children with known GI diagnoses17 or comparison groups made up exclusively of children without stomach or gut issues.18
Variables Coded, Data Extraction, and Reliability
Two authors (B.O.M., W.G.S.) screened all articles identified during the literature search and coded eligible studies by using a standardized checklist system (available on request). Data extracted during this process included descriptive information: study descriptors (eg, data collection, recruitment setting), participant demographic variables (ie, sample size, age, gender), composition of the comparison groups, ASD diagnostic procedures, GI assessment tools, and summary of findings. When it was reported, we coded the source of the ASD diagnosis, with possible categories including parent report, community provider using diagnostic criteria (eg, psychologist; pediatrician), chart review of diagnostic status, ASD rating scale (eg, Childhood Autism Rating Scale [CARS],19 Autism Diagnostic Interview–Revised [ADI-R]20), and Autism Diagnostic Observation Schedule (ADOS21). We also recorded whether a standardized measure of cognitive status or IQ was presented. Possible GI assessment tools included chart review, caregiver questionnaire, parent interview, physician evaluation, any relevant blood tests, stool studies, skin allergy tests, or endoscopy. Possible GI symptoms and diagnoses extracted included the range of terms listed in Table 1, with space provided to code for other medical concerns presented in articles (eg, feeding concerns).
The double-entered data allowed for κ coefficients (for categorical variables) and intraclass correlations (for interval and continuous variables) to be calculated on all extracted information. The mean κ agreement for categorical data was 82% (range: 52%–100%). The overall intraclass correlation for interval and continuous data was 0.97 (range: 0.62–1). Coder agreement exceeded the 80% standard widely adopted and recommended during quantitative synthesis of research.22
Statistical Analysis
To calculate the effect size (ES), we used means (SDs) or frequencies (percentages), or if necessary, we estimated the ES from test statistics (eg, χ2, t tests). When summary statistics were not presented in an article, we attempted to contact the primary author via e-mail. The primary goal of the meta-analysis was to determine the overall difference in GI symptoms between children with and without ASD. For each GI concern, we calculated a separate ES estimate across studies. For studies involving multiple comparison groups (eg, ASD versus typically developing [TD] or sibling [SB]), we pooled the comparison groups, producing an overall effect size comparing children with and without ASD. This was the case in only 2 studies.23,24 Our exclusion criteria also excluded subgroups of children with other medical concerns or developmental delays (DDs), although descriptive information (eg, age range, sample size) regarding DD samples is presented in Table 3 for reference. We did not conduct an analysis of potential moderators (eg, age, gender, diagnostic status) given the small number of available studies and the lack of descriptive data presented in the articles. Finally, for a variable to be included in the analysis, we required at least 6 studies presenting data on a GI symptom, in line with recommended standards for systematic reviews.25,26
We entered and analyzed data using Comprehensive Meta-Analysis 2.27 We converted all ESs to standardized mean difference (SMDs) using a random-effects model.28 A positive SMD (SMD > 0) indicated more GI-related concerns in children with ASD than in the comparison group. We evaluated SMD magnitude using conventional standards (0.2 = small, 0.5 = medium, 0.9 = large).29 To aid in clinical interpretation of outcomes, we also calculated the corresponding odds ratio (OR) with 95% CIs, with values reflecting the odds of a child with ASD having a GI symptom compared with a child without ASD.
To assess heterogeneity within subgroups and between studies, we calculated ESs and associated 95% CIs for each subgroup. In addition, we used the Q test to formally determine whether heterogeneity was present. To assess the robustness of our results, we conducted a sensitivity analysis to determine how sensitive the combined estimate was to any 1 study by repeatedly calculating the overall ES with 1 study omitted per iteration and compared the results with the overall study effect. We analyzed the threat of possible publication bias to the validity of the obtained outcomes using the funnel plot,30 failsafe N,31 and the trim and fill.32 The failsafe N method determines the number of additional “negative” studies (eg studies showing no difference between ASD and non-ASD groups) needed to reduce the overall test to nonsignificance. A small number of negative studies indicate possible publication bias. The trim and fill method is a nonparametric method that first estimates the number of “missing studies” in a meta-analysis and then determines what effect they would have on the outcome had they been present all along. If the added studies significantly change the test result, then publication bias is possible.
Results
Characteristics of Studies and Participants
The search yielded 15 articles out of a pool of 961 possible studies. Figure 1 presents a flow diagram of the screening and identification process. All articles were published since 2000, with 5 (33%) since 2010. Only 4 of the 11 studies (36%) identified by Buie et al5 met inclusion criteria, resulting in 11 unique articles (73%) in the review. Table 2 provides a summary of methods by study. Research designs included prospective studies (53%), as well as studies mining existing databases33–36 or medical charts.37–39 In terms of confirming diagnostic status, 3 studies (20%) used either the ADI-R or ADOS; 3 (20%) used the CARS. Six (40%) relied exclusively on chart review or clinical provider, 1 study (7%) involved parent report, and the remaining 2 studies (13%) did not identify the source of ASD diagnosis.
Flow diagram of included and excluded studies.
Description of Experimental Characteristics and Assessment Methods by Study
Assessment of GI status involved either caregiver report (73%) or medical chart review (27%). Diarrhea was the most common symptom assessed, followed by general GI concerns. This latter category primarily involved denoting the percentage of children experiencing ≥1 GI clinical element based on a composite of possible symptoms23,33,36–41; 3 studies broadly asked caregivers whether there were concerns about their child’s bowels,42 whether their child experienced chronic or ongoing GI problems,23 or whether their child had a history of GI dysfunction.43 Nine studies (60%) included information outside the core symptoms captured by our coding sheet, such as recurrent fever,39 feeding or dietary concerns,37,41,42 or behavioral concerns (eg, sleep, sudden aggressive behavior)40 or simply included a broad category of “other GI problems.”36
The pool of studies involved a total sample of 2215 children with ASD (Table 3). All studies reported information on age (ie, mean, SD, range). Two studies37,38 involved a longitudinal design assessing the incidence of GI symptoms at fixed time periods; both also presented an overall composite of symptoms, which was used for the current analysis. Eight studies (53%) reported matching or statistical equivalence for age across groups; 4 (27%) noted equivalence in the proportion of boys to girls.
Description of Participants
Overall Measure of ES
Four variables (ie, general GI concerns, diarrhea, constipation, abdominal pain) met the 6-study threshold for inclusion in the analysis. Table 4 presents ES estimates calculated by using random effects models. Overall test for heterogeneity of study ESs was statistically significant for all 4 outcome measures, indicating that the random effects model was appropriate. Analysis indicated higher levels of GI symptoms in all 4 areas among children with ASD when compared with children without ASD. OR estimates for reports of general GI symptoms ranged from 0.39 to 48.25 (Fig 2), with an overall OR of 4.42, suggesting that the odds of GI symptoms in children with ASD are 4 times more prevalent than for children without ASD. The corresponding SMD was large and statistically significant (P < .001). OR estimates for diarrhea ranged from 0.60 to 19.80 (Fig 3), with an overall OR of 3.63, suggesting a greater than threefold increase in the prevalence of diarrhea in children with ASD. A similar trend was detected for constipation, with OR estimates ranging from 0.86 to 32.87 (Fig 4) and an overall OR of 3.86. The corresponding SMD for both variables was large and statistically significant (P < .001). Finally, OR estimates for abdominal pain ranged from 0.93 to 7.91 (Fig 5), with an overall OR of 2.45. The corresponding SMD represented a medium and statistically significant (P < .05) effect size.
ESs, 95% Confidence Limits, and Within-Group Tests for Heterogeneity for Studies Included in the Meta-analysis for GI Symptoms
Forest plot of general GI concerns with 95% CIs. aOR outside of range.
Forest plot of diarrhea with 95% CIs.
Forest plot of constipation with 95% CIs. aOR outside of range.
Forest plot of abdominal pain with 95% CIs.
Sensitivity Analysis, Publication Bias, and Reliability of Results
Sensitivity analysis involved visual inspection of CIs for the overall ES after each study was removed one at a time. No study significantly altered the overall mean ES estimates for each of the GI variables included in the analysis. Visual inspection of the funnel plots indicated no potential publication bias for the outcomes general GI concerns and abdominal pain. The funnel plots for constipation and diarrhea suggested potential publication bias. However, for both of these outcomes, the failsafe N analysis indicated that there would need to be at least 140 and 230 published studies, respectively, with nonsignificant findings related to each of the outcomes to change the current ES to nonsignificant. Additionally, Duval and Tweedie’s trim and fill method suggested only imputing 2 missing studies for the outcomes constipation and diarrhea. Imputing the 2 studies did not significantly change the ES or the conclusion. The failsafe N was 282 for general GI concerns and 25 for abdominal pain. This evidence lends credence to the robustness of our findings.
Discussion
This meta-analysis represents the first rigorous evaluation of evidence regarding GI symptoms in children with ASD, quantifying the past 32 years of research by using standardized data collection, strict inclusion criteria, and multiple statistical tools to ensure the most accurate assessment of current knowledge on this topic, and it involved a total of 15 studies on 2215 children with ASD. The results indicate greater risk of general GI symptoms among children with ASD than in those without ASD, indicating that this population may be more prone to specific symptoms of abdominal pain, constipation, and diarrhea. By conventional standards, findings reflect a large ES, corresponding to a greater than threefold elevated risk of general GI concerns, constipation, and diarrhea between children with and without ASD and a “medium” effect size corresponding with a greater than twofold elevated risk of abdominal pain. This pattern of results corroborates previous anecdotal reports and descriptive studies suggesting greater GI complaints in ASD while emphasizing the need for more evidence-based data to support best standards of care.
At a minimum, these results reinforce expert consensus5 that parents and health care providers should be educated about possible GI symptoms and pathophysiology in children with ASD, and children suspected of having possible GI disorders should be screened accordingly. In addition, the magnitude of the observed association combined with difficulties identifying and studying GI dysfunction in ASD warrants the adoption of a lower referral threshold by practitioners for evaluation and treatment by a gastroenterologist if an underlying problem is suspected. Children with ASD often present with limited verbal communication, and as a result, their symptom presentation may be unusual compared with that of their peers.5 For example, the emergence or exacerbation of problem behaviors, such as aggression, self-injury, sleep disturbance, or irritability, without clear environmental influence (ie, antecedents or consequences), may be the only indication of an underlying GI problem.44 Unfortunately, lack of evidence in this area has prevented the development of evidenced-based guidelines to help physicians navigate the diagnostic and early intervention process.45 This includes how to best identify possible subsets of children most at risk and how to modify clinical practice with consideration to the unique combination of behavioral, neurologic, or medical issues in ASD. For example, the guidelines on managing constipation in ASD are the same as those for assessment and treatment of all constipated children published by the North American Society of Pediatric Gastroenterology, Hepatology, and Nutrition, with added notes regarding possible modifications based on characteristics of the diagnosis. Modifications mainly underscore the need for greater flexibility when working with children with ASD given common obstacles, such as rejecting medicines “because of the flavor.”45
Findings regarding greater GI symptoms in ASD should be considered in the context of several limitations. The pool of studies included in the current review was small, and only 4 out of 15 possible GI problems met the 6-study threshold for inclusion in the meta-analysis. Insufficient data were available to determine whether GI symptoms often linked with an organic pathology, such as gastroesophageal reflux, gastroenteritis, food allergies, and inflammatory bowel disease, are more common among children with ASD. Questions also remain about the relative contribution of behavioral factors, such as toileting and feeding problems, to the observed association between diarrhea, constipation, and abdominal pain in ASD. Estimates suggest that ∼95% of childhood constipation may be functional, without an underlying physiologic cause,46 and many children with ASD present with nonorganic toileting problems that may precipitate or play a role in the development of constipation, including absent or delayed acquisition of bowel training47 and higher rates of problem behaviors related to changes in toileting routine.48 Fecal retention in ASD may also occur secondary to difficulty with sensory stimuli, sensory processing, and motor problems, leading to altered gastrointestinal motility and defecation physiology.49 It is also possible that elevated rates of constipation may be related to the ubiquity of food selectivity in this population, as the dietary patterns often associated with ASD involve high intake of processed food and lack fiber-containing fruits and vegetables, which provide a natural laxative effect and decrease intestinal transit time.50 Thus, the interweaving in children with ASD of behavior, diet, alterations in feeding choices, perceived improvements or decrements in function by parents and caregivers with alterations in diet, all point toward an important and clinically relevant role for broad-based, longitudinal, unbiased studies of feeding patterns, GI symptoms, and behavior.
Close appraisal of the available literature also indicates the need for greater methodological rigor. The summarized research was based largely on parent report and medical chart reviews, and no studies included confirmation of GI problems by a third-party care provider such as a physician. Definitions of GI symptoms also varied among included studies. For example, descriptions of diarrhea ranged from simply diarrhea24,33,40,43,51 to persistent diarrhea,42 chronic diarrhea,39,52 frequent diarrhea/colitis,35 and a history of diarrhea.34 One study focused on stool frequency and consistency,41 defined as >4 weeks involving painless passage of ≥3 large unformed stools per day, and another provided specific definitions of GI symptoms that would lead to this diagnosis (ie, enteritis, colitis, gastroenteritis, or loose stool).37 Studies using chart review or existing databases often had unique operational definitions, which could affect detection or prevalence estimates. In the future, prospective controlled population-based studies with a physician evaluation and established definitions should be pursued to increase standardization and minimize measurement discrepancies across studies. This will probably necessitate a standardized measure focusing on GI issues among children with ASD, because no instruments exist to guide clinical and research activities. Ideally, this instrument would be developed using the methods described in the Food and Drug Administration Guidance53 to serve as an endpoint for randomized clinical trials and comparative effectiveness studies while increasing early detection to support best standards of care. In addition, the use of a toileting diary focusing on stools per day paired with characterization by the descriptive and visual Bristol Stool Scale54 could provide a more detailed, standardized examination of possible GI symptoms, such as diarrhea and constipation. In the absence of reliable and valid assessment, it is difficult to compare the severity of GI symptoms across samples and to develop and evaluate effective interventions. It will also be imperative to better characterize samples using diagnostic measures, such as ADOS21 and ADI,20 that have been standards of best practice in research for more than a decade. Only 40% of identified studies used a standardized assessment to confirm ASD status, which reflects the limitations of our knowledge of GI symptoms and disorders in ASD.
The breadth of inquiry should also be expanded to consider possible deviations in the establishment and maintenance of the gut microbiome in ASD and the relative contribution of early feeding practices to overall gut health and acceptance of new feeding tastes and textures. Studies of fecal DNA extracts have found certain bacterial clusters overrepresented in children with ASD and gastrointestinal complaints compared with children with similar GI complaints but typical neurobehavioral development (see Mulle et al55 for review). Provisional evidence also suggests that children with ASD are at greater risk for suboptimal breastfeeding, including late initiation and shorter duration of exclusive breastfeeding.56,57 In addition to promoting optimal nutrition, breast milk assists in the development of the gastrointestinal tract, pancreas and endocrine system, and related mucosal defenses,58 and it is therefore possible that suboptimal breastfeeding may result in atypical colonization of the gut microbiome in ASD.59 In turn, changes in the gut microbiome may help explain anecdotal reports of improvement in behavioral functioning in response to dietary changes if such changes serve a probiotic function and improve symptoms of irritable bowel syndrome (eg, bloating, abdominal pain, flatulence) among certain children with ASD.60 Such reports help propagate interest in the use of dietary manipulation (eg, gluten- and casein-free diets) as an ASD-focused treatment.55 Dietary interventions, including the gluten- and casein-free diets, nutritional supplements, enzymes, and antimicrobial agents, have not been substantiated by empirical investigation,61 presenting a clear need to investigate how possible deviations in the GI tract in ASD relate to current dietary recommendations being promoted and implemented in the ASD community.
It is clear that greater clinical and research scrutiny is needed to increase awareness on this topic and thus support development of the best standards of care. Previous controversy surrounding the MMR vaccine and proposed causal link between ASD and infection of the GI tract probably deterred investigators from dedicating resources to examine GI functioning in this population while fostering uncertainty in the ASD community regarding the validity of this line of inquiry. Based on the available data, this meta-analysis indicates a greater risk for general GI concerns, constipation, diarrhea, and abdominal pain in ASD; however, conclusions about the nature and etiology of the observed associations are tentative at best. In addition, data on the prevalence of other GI symptoms (eg, gastroesophageal reflux, food allergies) typically associated with organic pathology are insufficient. As a result, the most logical conclusions remain that rates of other GI pathophysiology in ASD are similar to those observed in the general population, and there is no evidence suggesting a unique GI pathology in ASD.5 Additional research is needed to elucidate the etiology, prevalence, topography, and remediation of GI problems in ASD, with consideration of the potential interwoven contributions of factors such as immune abnormalities, mucosal barrier dysfunction, gastrointestinal motility, feeding and toileting concerns, and the gut microbiome.
Footnotes
- Accepted February 12, 2014.
- Address correspondence to William G. Sharp, PhD, Pediatric Feeding Disorders Program, The Marcus Autism Center, 1920 Briarcliff Road, Atlanta, GA 30329. E-mail: wgsharp{at}emory.edu
Dr McElhanon participated in the study conception, design, identification of studies, study selection, data collection, data extraction, data analysis, data interpretation, and drafting and revision of the article; Dr McCracken participated in data analysis, data interpretation, and drafting and revision of the article; Dr Karpen participated in the study conception, data interpretation, and critical revision of the article for important intellectual content; Dr Sharp participated in the study conception, design, identification of studies, study selection, data collection, data extraction, data analysis, data interpretation, and drafting and revision of the article; and all authors approved the final manuscript as submitted.
FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.
FUNDING: No external funding.
POTENTIAL CONFLICT OF INTEREST: Dr Sharp is a member of the Nutricia Advisory Board on GI Concerns in Children With ASD. The other authors have indicated they have no potential conflicts of interest to disclose.
References
- Copyright © 2014 by the American Academy of Pediatrics