Brief Report: Association Between Behavioral Features and Gastrointestinal Problems Among Children with Autism Spectrum Disorder
Recent reports suggest certain behaviors among children with autism spectrum disorders (ASD) may indicate underlying gastro-intestinal (GI) problems, and that the presence of these behaviors may help alert primary care providers to the need to evaluate a child with ASD for GI problems. The purpose of this population-based study of 487 children with ASD, including 35 (7.2%) with a medically documented history of GI problems, was to compare behavioral features of children with and without a history of GI problems. Unusual sleeping or eating habits and oppositional behavior were significantly associated with GI problems. These behaviors, however, were frequent in both children with and without GI problems, suggesting they may have limited utility in a screening capacity for GI problems.
KeywordsAutism spectrum disorderGastrointestinal
Autism spectrum disorder (ASD) encompasses a group of developmental disorders with a range of behavioral presentations and likely diverse etiologic factors (Newschaffer et al. 2007). A number of clinical and epidemiological studies have suggested that children with ASD are at increased risk for gastro-intestinal (GI) problems (Ibrahim et al. 2009), and some have suggested that certain behavioral problems observed in children with ASD may be indicative of a child’s response to, or attempt to communicate the discomfort of, an underlying GI problem (Horvath et al. 1999; Williams et al. 2010; Bauman 2010). Specific behavior problems proposed as possible expressions of GI distress include sleep disturbances, stereotypic or repetitive behaviors, self-injurious behaviors, aggression, oppositional behavior, irritability or mood disturbances, and tantrums. A recent pediatric consensus report called for additional research on the association between “problem behaviors” and GI problems, and for the development of a screen for GI problems in children with ASD (Buie et al. 2010).
The purpose of this brief report is to determine, in a large, population-based sample of 8 year-old children with ASD, whether the behavioral characteristics specified above occur more frequently among those who have been diagnosed with a GI problem than those without a medically documented history of GI problems. For comparison, we also evaluate the frequency, in those with and without a history of GI problems, of other behavioral characteristics that are common in children with ASD but that have not been hypothesized to be potential expressions of GI problems.
Study Population and Design
We implemented a cross-sectional study of children who were 8 years of age in 2006 and met the case definition for ASD through the Centers for Disease Control and Prevention’s Autism and Developmental Disabilities Monitoring (ADDM) Network. The ADDM Network is a multi-site, population-based, autism surveillance system wherein the surveillance case definition for ASD is not entirely dependent upon previous clinical diagnoses (Centers for Disease Control Prevention (CDC) 2009).
Although most sites participating in the ADDM Network incorporate information from both medical and educational records in determining ASD case status, some sites’ data collection was limited to information from medical or clinical records. To ensure that assessment for GI problems was possible, the study sample was restricted to children whose records contained an evaluation from a medical doctor. The frequency of reported GI problems varied across sites, and sites that relied heavily on educational records tended to report few to no instances of GI problems. To minimize potential confounding by ADDM Network site, the present analysis includes data collected from three sites where at least 5% of eligible children were diagnosed with a GI problem. In total, 487 out of 619 children with ASD from three sites (Alabama, Pennsylvania, and Wisconsin) had been evaluated by a medical doctor and were included in the analysis. While the ADDM Network utilizes both healthcare and educational sources for surveillance, the three sites included in this analysis relied exclusively on healthcare sources for records (Centers for Disease Control Prevention (CDC) 2009).
Surveillance Ascertainment of ASD
For surveillance purposes, children were classified as having an ASD if they displayed behaviors documented in evaluation records that were consistent with the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition, Text Revision (DSM-IV-TR) criteria (American Psychological Association (APA) 2000) for autistic disorder, PDD-NOS, or Asperger disorder at any time through age 8 years. Children suspected of having ASD were identified by screening evaluations from qualified professionals (e.g., including pediatricians, psychiatrists, nurses, speech therapists, psychologists, occupational therapists, and others). Children whose medical records were associated with an International Criteria for Diagnosis, 9th Revision (ICD-9) code for child neurodevelopmental disorders (e.g., 299.0 for autistic disorder or 314.0 for attention deficit disorder) were reviewed. Demographic data, descriptions of behaviors, diagnostic summaries, psychometric test results, and information about co-occurring disorders or disabilities were collected and entered into a centralized composite record and reviewed by trained clinicians according to a specified protocol to determine case status and document non-ASD diagnoses and associated features (e.g., abnormalities in sleeping). The protocol was approved by the institutional review board at each respective surveillance site.
For descriptive purposes, we calculated the distribution of the sample by sex, race or ethnicity, and ADDM autism classification (Autistic Disorder vs. PDD-NOS). Intellectual disability (ID) was classified when information on ID was available and the most recent IQ score was less than 70. Descriptions of seizure-like activity were also collected. Cerebral palsy was monitored in two sites (Wisconsin and Alabama).
We identified eight behavioral features cited in a recent pediatric consensus report that may be indicative of GI problems among children with ASD (Buie et al. 2010) which had analogous measures in the ADDM data set: abnormalities in sleeping; stereotyped and repetitive motor mannerisms; self-injurious behaviors; abnormal eating habits, abnormalities in mood or affect; argumentative, oppositional, defiant, or destructive behaviors; aggression; and temper tantrums. These behaviors were coded according to ADDM Network methodology, using verbatim descriptions of the behavior from the evaluations. To determine whether children with ASD and GI problems simply have more documented ASD-related behaviors of any kind, we selected an additional six behaviors that would not seem to be related to GI discomfort: oblivious to other children; lack of imaginative play; lack of or excessive fear; insistence on sameness; delayed motor milestones; and abnormal cognitive development (e.g., documentation of uneven, scattered, or “savant” skills; adaptive skills with at least one standard deviation between subtests).
The ADDM data include verbatim descriptions of non-ASD diagnoses or conclusion statements made by the examiner conducting the evaluation. We defined GI problems using many of the terms mentioned by the consensus report (Buie et al. 2010) (constipation, abdominal pain, diarrhea, encopresis, gastroesophageal reflux disease (GERD), gastritis, abdominal bloating, disaccharidase deficiencies, inflammation of GI tract, abnormalities of the enteric nervous system, functional abdominal pain, irritable bowel syndrome (IBS), flatulence, Celiac disease). Additional searches were performed to identify instances of abbreviations or misspellings, and all matches were visually inspected. All evaluations and data were thoroughly reviewed regardless of whether a previous evaluation indicated a particular behavior.
We performed cross-tabulations and Chi-square tests of the significance of differences in the frequency of descriptive and behavioral characteristics by presence of GI problems, among the 487 children with ASD. Before combining data from the three ADDM sites, we examined the association within each site and found no evidence of heterogeneity across sites. We also computed prevalence ratios with 95% confidence intervals to describe the magnitude of associations between behavioral characteristics and history of GI problems, and estimated the overall sensitivity and positive predictive value of the selected behavioral characteristics as indicators of or screening items for GI problems among children with ASD.
Descriptive characteristics of the study sample, stratified by presence or absence of GI problems
% With GI problems
(N = 35)
%Without GI problems
(N = 452)
Co-occurring intellectual disability
Co-occurring cerebral palsyb
Co-occurring seizure-like activity
Percent of children with GI problems by presence or absence of selected behavioral characteristics
Number (%) with behavior (out of 487)
% With GI problems
Prevalence ratioa (95% CI)
Behavioral characteristics hypothesized to be potential expressions of GI problems among children with ASD
3.1 (1.5, 6.4)
1.4 (0.7, 2.8)
Self injurious behavior
1.4 (0.8, 2.7)
Abnormal eating habits
2.7 (1.2, 6.4)
1.7 (0.8, 3.7)
2.2 (0.9, 5.6)
2.5 (1.0, 6.3)
2.2 (1.0, 4.9)
Has at least 1 of the above behaviors
Other behavioral characteristics common in children with ASD
Oblivious to other children
Lacking imaginative play
0.6 (0.2, 1.7)
Lack of or excessive fear
1.5 (0.8, 2.9)
Insistence on sameness
0.3 (0.1, 1.4)
Delayed motor milestones
5.1 (1.2, 20.1)
Abnormal or uneven cognitive development
0.6 (0.3, 1.4)
Notably, nearly all of the children with ASD, including all 35 with a documented history of GI problems and 446 (98.7%) of others, exhibited at least one of the behavior problems hypothesized to be potential indicators of GI distress. For this reason, these behaviors would not be useful as a potential screen for GI problems; though with a cut-off of ≥ one item the sensitivity would be 100%, the positive predictive value would be only 7.2% and virtually all children with ASD would potentially be referred for GI evaluations. Increasing the number of behaviors needed to “screen positive” for GI problems from one to five increased the positive predictive value modestly, from 7.2 to 9.4%, but only at a cost to sensitivity, which concomitantly declined from 100 to 80%.
This study provides some support for the hypothesized association between selected behavioral characteristics in children with ASD and the occurrence of GI problems. Even with the relatively small size of the population-based sample available, the study found significant positive associations for several behaviors hypothesized to be expressions of GI problems in children with ASD. At the same time, it did not find such associations for most of the control behaviors examined, suggesting that children with ASD and GI problems were not simply more likely to be described as having more of any type of symptoms or behaviors than children with ASD and no history of GI problems.
The unexpected finding of a significant positive association between GI problems and delayed motor milestones, which we had not hypothesized to be associated with GI problems—or a possible expression of GI discomfort—is interesting in light of two other incidental observations in this study, which we had not hypothesized. These two incidental findings included the associations between GI problems and both CP and seizure disorders. Overall, the observed associations between GI problems and delayed milestones, CP and seizure disorders lend support to the idea that a sub-set of children with ASD might suffer from underlying neurological and/or immunological dysfunction that affects multiple organ systems and functions, including those that are GI related. They are also consistent with other studies showing associations between GI problems and epilepsy (Gobbi et al. 1992) as well as the severity of CP (Erkin et al. 2010).
Perhaps the most important contribution of this study is the finding that the behavioral characteristics hypothesized to be expressions of GI problems are very common in children with ASD, yet not specific to those with GI problems. As a result, the presence of these behaviors would not be useful on their own for screening or identifying children requiring GI evaluation.
Although GI problems may contribute to selected behaviors in some children with ASD (Ibrahim et al. 2009), most children with ASD who exhibit these behaviors did not have a medically documented history of GI problems.
Limitations and Future Directions
The reliance of this study on information extracted from medical records is both a strength and a limitation. It is likely that this data source allowed accurate identification of GI problems that were relatively persistent and severe enough to require medical attention among children with ASD in the populations under surveillance. At the same time, medical records likely under-identify many of the less severe and less persistent GI problems that are identified in studies based on parental report. In general, studies relying on medical records to identify GI problems (Niehus and Lord 2006; Mouridsen et al. 2010; Taylor et al. 2002) report lower frequencies of GI problems than studies relying on parental report or clinical examinations of referred samples (Richler et al. 2006; Wang et al. 2011; Valicenti-McDermott et al. 2008) Although studies based on direct clinical examinations or parental report may have better opportunities to ascertain GI problems than those based on records, the results of such studies may be generalizable only to patients of specific academic medical centers or to voluntary participants in an online autism registry. In contrast, the results of the present records-based study are generalizable to children in the population meeting diagnostic criteria for ASD.
Although all three sites represented in this study followed the same protocol for extracting information and requesting records from medical sources, variability by ADDM site is possible. The relatively small number of children with GI problems limited our statistical power to explore site differences. Without a comparison group of children without an ASD, this study does not provide information about whether GI problems are more frequent in children with ASD than age-matched controls without ASD.
The calls for population-based studies to better understand the relationship between ASD and GI problems present a methodological challenge. Population-based studies such as ours provide greater generalizability and representativeness of ASD in the population, yet the tasks of measuring specific behaviors and systematically evaluating GI problems are more easily accomplished in clinic-based studies (which are likely to suffer from referral or other biases). A strength of the ADDM Network methodology is that it utilizes multiple sources for ASD case ascertainment; however, it is limited to information that has been previously documented in records. Large, population-based cohort studies that are both generalizable and have strong ascertainment capabilities will provide the best insight into the relationship between ASD and GI problems.
Certain behaviors, including abnormalities in sleep patterns, abnormalities in mood or affect, and argumentative, oppositional, defiant or destructive behavior were described significantly more often in children with ASD who also had GI problems than in those with ASD and no history of GI problems. These features (often described as characteristics of autism) may be more common among children with autism who also have GI problems. However, because these behaviors are also frequent in children with ASD and no GI problems (nearly all children had 1 or more behaviors), they are unlikely to efficiently predict GI problems in children with ASD. Consideration of medical, biological, or physiological co-occurring conditions, genetic susceptibility, diet and nutrition, and medication use are necessary to determine whether in children with ASD both behavioral presentation and GI problems might be associated with other underlying factors.
This work was supported by a grant from the Autism Science Foundation and by the Centers for Disease Control and Prevention through Cooperative Agreements UR3/CCU523235 and UR3/DD000078 as part of the Autism and Developmental Disabilities Monitoring (ADDM) Network. We gratefully acknowledge ADDM project coordinators, clinician reviewers, abstractors, ADDM investigators who contributed to the surveillance project and data collection. We also thank Dr. Lisa Miller for her helpful comments on an earlier version of this manuscript.