Anxiety and Mood Disorder in Children With Autism Spectrum Disorder and ADHD

This study was approved by the Johns Hopkins University Institutional Review Board.

We performed a cross-sectional, network-based study of children with ASD who were enrolled in the Interactive Autism Network, referred to as IAN, between 2006 and 2013. IAN is a family-centered, online registry and research database that was created to accelerate ASD research by linking participants with studies and by sharing deidentified data for analysis.¹⁸ Children and adults with ASD may register for IAN along with parents and siblings. To register with IAN, participant probands must have a professional diagnosis of ASD. Approximately 60 000 people have consented to participate, including >18 500 children and 7500 adults with ASD. Children with ASD are 80% boys with an ethnic and racial profile of ∼80% white, 4% African American, 2% Asian American, and 10% Hispanic. Parents are primarily college educated (85%). IAN content is in English only. Every state in the United States is represented. IAN has provided recruitment and data services for >500 studies. The IAN registry has been clinically validated for children with a Social Communication Questionnaire-Lifetime (SCQ-Lifetime) total score cutoff of 12,^19,–21 and it has been verified by a review of parent- and professional-provided medical records.²² 

We included individuals in the IAN registry ages 6 to 17 years who had completed the IAN Child with Autism Spectrum Disorder Questionnaire (CAQ) (a baseline questionnaire with demographic and core clinical information), had a total score ≥12 on the SCQ-Lifetime,²⁰ and had a total T-score ≥60 on the Social Responsiveness Scale (SRS)-Parent Report²³ with no more than 6 missing responses. Children outside of the age range and/or with incomplete questionnaires and/or with reported diagnosis of schizophrenia were excluded.

The CAQ is a baseline questionnaire for children with ASD that asks parents questions about their children’s birth, ASD diagnosis, development, and additional medical history.

Parent report of additional diagnoses was obtained from the following questions on the CAQ: “Has [child name] ever been diagnosed with or received treatment for____?” Options included depression, bipolar disorder, ADHD, and anxiety disorder. In this study, parent-reported mood disorder was defined as a positive response to the above question for depression and/or a positive response to the above question for bipolar disorder. ID was defined as a positive response to the question, “Has [child name] ever been diagnosed with intellectual disability (also known as mental retardation)?” and/or an IQ score <70 on the question, “What was [child name]’s most recent IQ test score?”

For the purposes of this study, children with autism spectrum disorder with parent-reported attention-deficit/hyperactivity disorder are referred to as ASD (+) ADHD, and children with autism spectrum disorder without parent-reported attention-deficit/hyperactivity disorder are referred to as ASD (−) ADHD.

Age was calculated by using the date of birth and the date on which the CAQ was completed.

The SCQ-Lifetime is a 40-item, parent-report questionnaire that is designed as a screening test for ASD.²⁰ It is validated for ages 4 years and older. Scores range from 0 to 39 with a cutoff of 15 for ASD in a general population, and a cutoff as low as 11 is recommended for a high-risk population to optimize the area under the curve.²⁴ In this study, we used a cutoff of 12 as 1 of several inclusion criteria in the IAN registry per the manual’s recommendation to use a lower threshold if there are additional risk factors²⁰ because registrants of the IAN are considered high risk for ASD given that they have received a professional diagnosis of ASD per parent report. In the IAN registry, the SCQ-Lifetime total score cutoff of 12 has been validated against the Autism Diagnostic Interview with 99% accuracy.¹⁹ 

The SRS-Parent Report consists of 65 items and is designed to identify the presence and severity of social impairment in ASD.²³ The questionnaire is validated in ages 4 to 18 years. Scores range from 0 to 145, and T-scores are standardized for sex. A T-score on the SRS-Parent Report ≥60 is considered abnormal and associated with ASD. The SRS has strong psychometric properties, including an interrater reliability of 0.9 between parents, an internal consistency of >0.9, and discriminant validity between other developmental behavioral disorders, including ADHD, mood disorders, conduct disorder, and psychosis.^23,25 It has been validated against clinical evaluation and the Autism Diagnostic Interview with a sensitivity of 0.75 and a specificity of 0.96.^24,26 

Detailed methodology regarding data management and data analysis may be found in the Supplemental Information.

There were 3319 children who met inclusion criteria for this study, of whom 1503 (45.3%) reported a diagnosis of or treatment for ADHD. Demographics, the prevalence of parent-reported ID, mean SRS-Parent Report total raw scores, and the presence of comorbid parent-reported anxiety disorder and mood disorder for the entire cohort are presented in Table 1 in addition to a comparison between the ASD (+) ADHD and ASD (−) ADHD groups. Survey completion was near contemporaneous, with 92.2% completing both the CAQ and the SRS within 1 calendar year and 96.5% within 2 calendar years. The cohort was primarily male (82.9%), white (87.2%), and non-Hispanic (92.4%), with a mean age of 10.3 years. Of the children, 649 (19.6%) were reported to have ID, 1025 (30.8%) were reported to have a diagnosis of or treatment for an anxiety disorder, and 532 (16.0%) were reported to have a diagnosis of or treatment for a mood disorder. A statistically significant difference in the sex proportion and prevalence of parent-reported ID was found when comparing the ASD (+) ADHD and ASD (−) ADHD groups. The ASD (+) ADHD group was older than the ASD (−) ADHD group and had higher ASD severity per the SRS-Parent Report total raw score. We found no significant difference in either race or ethnicity between the groups.

In Table 2, we provide the results of generalized linear model (GLM)²⁷ analyses in which we compare the presence of anxiety or mood disorders with the presence or absence of ADHD. The ASD (+) ADHD group had an increased risk of reported anxiety disorder (adjusted relative risk 2.20; 95% confidence interval [CI] 1.97–2.46) and mood disorder (adjusted relative risk 2.72; 95% CI 2.28–3.24) compared with the ASD (−) ADHD group. Increasing age was the most significant contributor for both anxiety disorder and mood disorder (both P < .001), and the absence of report of ID was a significant contributor for mood disorder only (P < .001). Given the association between increasing age and parent-reported ADHD, we also analyzed relative risks by age subgroups (school-aged and adolescent) to better appreciate a clinical practice perspective. As expected, we found an increased prevalence of both anxiety disorder and mood disorder in the adolescent group compared with the school-aged group for both the ASD (+) ADHD and ASD (−) ADHD groups; however, there were higher relative risk ratios for the school-aged group compared with the adolescent group for both anxiety disorder and mood disorder. Within the age subgroups, we also found the same pattern as in the full data set that increasing age was the most significant contributor to the presence of both anxiety and mood disorders (for both age groups and both conditions: P < .001), and absence of report of ID was a significant contributor for mood disorder only (school-aged: P = .041; adolescent: P = .001). Neither sex, nor race, nor ethnicity were significant in any of the GLM analyses.

To our knowledge, this is the largest study in which researchers compare comorbidities in individuals with ASD alone and ASD with ADHD. It is also 1 of the largest in which researchers compare the clinical phenotypes of these populations. We found an extremely high prevalence of parent-reported ADHD among children with ASD, with ADHD affecting 45.2% of the children, which is commensurate with previous studies that reveal a 31% to 95% co-occurrence.^28,–31 Previous studies reveal that there may be a genetic or symptom overlap of these disorders.^3,32 Nonetheless, this should not invalidate either diagnosis, especially when diagnosis-specific treatments are available.

Our primary study findings were that children with both ASD and ADHD are at an increased risk for being diagnosed with or treated for anxiety and mood disorders when compared with those with ASD alone. These are supported by a 2011 study of adolescents in special education that revealed increased rates of antidepressant and/or antianxiety medication use among children with ASD and ADHD in comparison with ASD only.³³ Furthermore, the prevalence of reported anxiety and mood disorders increases with age, independent of ADHD, which is unsurprising given that the CAQ asks if a child has ever been diagnosed with these conditions, leading to an inevitable cumulative diagnosis with time. Additionally, both groups follow the same trajectory as typically developing peers in that the onset of symptoms consistent with mood and anxiety disorders is most often seen in adolescence, which may explain the higher prevalence of these disorders in the older cohort. In contrast, the relative risks of anxiety and mood disorders are greater in the younger, school-aged children than in the older adolescents for those with ADHD compared with those without ADHD. This suggests that ADHD may make children with ASD more vulnerable to an earlier onset of the symptoms of anxiety or mood disorders or more likely to exhibit detectable symptoms at an earlier age.

The specific etiology behind the relationships among these conditions is unclear at this time. It is possible that there is a genetic basis for an increased risk of multiple psychiatric disorders, as has been found with ASD and ADHD.³² Alternatively, it is possible that 1 syndrome is an early manifestation of the other, or the development of 1 syndrome increases the risk for the other. One may also consider that children with ADHD and ASD are at an increased risk for behavioral problems,^8,10 and these behaviors may contribute to anxiety or mood symptoms. This may also contribute to the differences in SRS scores between the groups, which is discussed below.

Referral bias may explain an increased risk for reported anxiety and mood disorders in children with ASD and ADHD in comparison with ASD alone because practitioners who diagnose ADHD may be more likely to also diagnose anxiety or mood disorders. However, this question was addressed in a previous study of the IAN registry,¹² in which children with both ASD and ADHD were less likely to have a third diagnosis than to not (odds ratio 0.1, 95% CI 0.1–0.2), implying that referral bias is unlikely in this sample. Registration bias may also influence the findings if parents of children with multiple comorbidities are more likely to participate in IAN.

Evolving diagnostic criteria may also influence population-based studies. The Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition, has broadened the construct of autism toward a spectrum and narrowed the diagnostic criteria for ASD, although the definitions of ADHD and mood disorders are similar to those outlined in the previous edition. The evolution of definitions and allowing the coexistence of multiple psychiatric diagnoses acknowledges and may affect medical recognition and treatment. The high rates of comorbidity in this study may thus reflect changing practice with the evolution of the Diagnostic and Statistical Manual of Mental Disorders.

Pharmacotherapy may also contribute to our findings because ADHD, anxiety, and mood disorders all have treatments that are widely available and increasingly used in practice.³⁴ Notably, IAN asks whether a child has ever been diagnosed with or treated for these comorbidities, acknowledging that with the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, many children with ASD were not assigned a diagnosis because symptoms may overlap but were prescribed medication for hyperactivity, anxiety, or mood symptoms in the absence of a formal diagnosis. With this in mind, our rates of ADHD, anxiety, and mood disorders may reflect the rate of symptoms that are consistent with these disorders rather than formal diagnosis. Frequently still, diagnoses are not used until intervention is needed, which suggests that our sample may be underidentifying these comorbidities if the children are not being medically treated.

Recognizing the increased risk for psychiatric disorders in this population has implications for clinical practice. This may be challenging in ASD because symptoms of anxiety and mood disorders may present differently in these children than in typically developing children. Unfortunately, information regarding how anxiety and mood disorders were diagnosed and/or treated was not available for this study. Further research is needed to better understand how mood and anxiety disorders present in both ASD and ADHD populations to optimally assess and diagnose these disorders. Importantly, both anxiety and mood disorder symptoms are treatable medical conditions through psychotherapy³⁵ and medication.³⁶ Recognizing and treating the symptoms can impact quality of life³⁷ and improve other short- and long-term outcomes, with further knowledge also being needed about effective, evidence-based treatments for these comorbidities in ASD.

We found that the presence of ADHD has a small association with greater ASD symptom severity, as reflected in the SRS score, suggesting that children with increased ASD severity are either more likely to be diagnosed with ADHD, or a dual diagnosis of ASD and ADHD impacts ASD symptoms. Researchers in another study found similarly that children with ASD and ADHD have SRS scores ∼3 points higher than children with ASD who do not have ADHD.³⁸ There is also evidence suggesting that children with ADHD alone may have higher SRS scores than the normative population,³⁹ suggesting that a behavioral overlap between ASD and other psychiatric disorders exists. The clinical implication of a small increase in ASD symptom severity in children with both ASD and ADHD is unclear. Six points on the SRS may not translate to appreciable differences in an individual child’s outcome, but such a difference may have a broader social or economic impact among this population. It is possible that the SRS is not an adequately sensitive or specific tool to assess ASD function in this setting, and additional studies of ASD symptomatology in the context of ADHD are needed.

We also found a difference in the rates of ID among those children with ASD with and without ADHD. In our cohort, those with ADHD had slightly lower rates of ID. It may be that ADHD symptoms are more easily or frequently detected in children with normal intellect or that the genetic phenotype associated with ASD and ADHD is also associated with normal intellect. Differential rates of ID among those children with ASD with and without ADHD may also be a function of diagnostic overshadowing (eg, ascribing inattention and/or impulsivity to ID rather than ADHD). Researchers in future studies examining this question may help clarify whether this association is replicable and what its clinical implications may be.

The diagnosis of ASD has been validated in the IAN database with 98% accuracy,^19,21 but similar data are not available for the other diagnoses in this study. Although performing standardized, comprehensive psychiatric assessment is the gold standard for diagnosis, participant report is efficient in sampling a large population, with data for other diagnoses supporting that such report is valid with equal accuracy (S. Terry, MA, personal communication, 2017).⁴⁰ Furthermore, the parent-reported diagnoses in this study are supported by similar rates of comorbidity with ASD in large epidemiologic studies.^{28,30,31,33,41,–43} However, we do acknowledge that participant-reported data may be susceptible to recall or reporting bias. Self- and parent-report data have also demonstrated statistical validity in the social sciences⁴⁴ and is frequently relied on for the diagnosis of ADHD⁴⁵ and anxiety disorders.⁴⁶ Furthermore, there is similar precedent for the use of parent-reported diagnoses in other large epidemiologic studies of children, such as the National Health Interview Survey,^47,–49 in which the language is identical to that used in the IAN questionnaire. Incorporating psychiatric diagnostic questionnaires may help validate this report in the future.

Both ADHD and ID may be underreported in this cohort, as is seen with chronic health conditions.⁴⁸ For ID specifically, parents may be underinformed or misinformed of their children’s intellectual skills. We further acknowledge that our definition of ID (parent report or IQ <70) does not conform to the current Diagnostic and Statistical Manual of Mental Disorders definition because it does not incorporate adaptive functioning.

This study represents a cross-sectional sample of lifetime information, and we did not assess individuals longitudinally. Therefore, our trends in age groups are based on prevalence rather than incidence. Longitudinal data may help clarify the relationships between these conditions and age.

Because computer and Internet access are required to complete the IAN questionnaires, there is bias toward participants of higher socioeconomic status.⁵⁰ We have assumed that this bias is constant throughout the sample, although this sample is not precisely representative of the general population.

ADHD affects nearly half of the children with ASD. This subgroup of individuals with ASD may represent a distinct clinical phenotype, with different diagnostic and therapeutic implications. Better understanding the differences between children with ASD with and without ADHD is crucial to designing effective interventions.

Our study supports that anxiety and mood disorders, although highly prevalent in those with ASD alone, are even more prevalent in individuals who have ADHD. They are also more prevalent with increasing age. The identification of psychiatric conditions in children with ASD is important because these disorders are treatable and affect quality of life. Physicians who treat children with ASD should be vigilant about screening for anxiety and mood symptoms, particularly in those with ADHD.

ADHD

attention-deficit/hyperactivity disorder

ASD

autism spectrum disorder

ASD (−) ADHD

children with autism spectrum disorder without parent-reported attention-deficit/hyperactivity disorder

ASD (+) ADHD

children with autism spectrum disorder with parent-reported attention-deficit/hyperactivity disorder

CAQ

Child with Autism Spectrum Disorder Questionnaire

CI

confidence interval

GLM

generalized linear model

IAN

Interactive Autism Network

ID

intellectual disability

SCQ-Lifetime

Social Communication Questionnaire-Lifetime

SRS

Social Responsiveness Scale

We acknowledge the individuals with ASD, their families, the researchers, and the health care professionals who make IAN possible through the generous contribution of their time and effort.