Introduction
To respond with anxiety to a threatening situation or stimuli is normal and healthy; however, when such a reaction becomes excessive and impairs daily functioning, one is classified as having an anxiety disorder (APA
2000). Anxiety disorders are among the most common disorders in childhood, with prevalence rates between 8.3 and 20.9 % (Costello et al.
2005). In addition, comorbidity among anxiety disorders as well as with other DSM-IV disorders is substantial (e.g., Kendall et al.
2010). Symptoms of anxiety disorders may include preoccupations and repetitive behaviors (e.g., obsessions and compulsions), avoidance of (social) situations, and speech problems (e.g., dysfluency), which are also commonly seen in children with autism spectrum disorders (ASD) (Hartley and Sikora
2009; Wood and Gadow
2010). ASD is characterized by impairments in three domains; (a) the social domain (e.g., impairments in the use of nonverbal behaviors, lack of sharing, lack of social or emotional reciprocity), (b) the communication domain (e.g., impairments in the ability to initiate or sustain a conversation, stereotyped and repetitive language), and (c) the domain of repetitive, stereotyped interests and behaviors (e.g., preoccupations, nonfunctional routines or rituals) (APA
2000). In addition, prevalence rates of anxiety disorders in youth with ASD are much higher (nearly 40 % estimated by a recent meta-analysis; van Steensel, Bögels and Perrin 2011) compared to those found in typically developing children, and the differentiation between ASD and anxiety disorder symptoms, particular in the case of obsessive compulsive disorder and social anxiety disorder, can be difficult.
A study of Hartley and Sikora (
2009) explored which criteria effectively discriminate ASD from anxiety. They administered a semi-structured interview of DSM-IV-TR criteria for autistic disorder to the parents of children with ASD, children with anxiety disorders, and children with ADHD. In this study it was found that the domain of restricted/repetitive/stereotyped patterns could not discriminate between children with ASD and children with anxiety disorders, while the best differentiation between children with anxiety and children with ASD could be made through the communication domain. Within the social relatedness domain it was found that the children with ASD had higher endorsements of impaired non-verbal behavior and lack of seeking to share compared to the children with anxiety disorders, however, social/emotional reciprocity was a poor indicator for ASD (Hartley and Sikora
2009). In addition, recent studies have found elevated scores of ASD in children with anxiety and/or mood disorders (Towbin et al.
2005; Pine et al.
2008). The study of Towbin et al. (
2005) examined ASD traits in a sample of children with mood and anxiety disorders with three ASD measures, namely the Social Communication Questionnaire (SCQ; Berument et al.
1999), the Social Responsiveness Scale (SRS; Constantino et al.
2000), and the Children’s Communication Checklist (CCC-2; Bishop
1998). In this study, 48.0 % of the sample scored in the ASD range on at least one measure. A subsequent study of Pine et al. (
2008) examined ASD scores in youths with anxiety and/or mood disorders. The same three instruments were used to assess ASD; however, in this study a comparison group of healthy participants was added. Results revealed that 4.0–24.6 % of the clinically anxious children scored above ASD thresholds depending on which measurement was used (note that of the children with mood disorders these percentages ranged between 7.4 and 75.0 %). In addition, it was found that children with anxiety and/or mood disorders had significantly higher ASD scores compared to the children in the healthy sample (Pine et al.
2008).
In sum, research so far has found (a) high prevalence rates of anxiety disorders in children with ASD (e.g., van Steensel et al.
2011), (b) several similarities (and differences) between children with ASD and children with anxiety disorders with respect to the DSM-IV-TR-criteria of autistic disorder (Hartley and Sikora
2009), and (c) elevated scores on ASD measures in children with anxiety and/or mood disorders (Towbin et al.
2005; Pine et al.
2008). What is not yet known is whether these ASD-like behaviors in children with anxiety disorders (and mood disorders) are precursors of the anxiety disorders or not. Do current anxiety disorders manifest, in part, as ASD-like behavior (but possibly representing methodological artifact or phenocopy rather than true ASD), or, alternatively (or additionally) do ASD symptoms in some children with anxiety disorders begin early in life, consistent with the clinical course of true ASD, and perhaps cause anxiety and mood disorders later in childhood via stress generation (see Wood and Gadow
2010). Of note, the study of Hallett et al. (
2010) examined the relationship between autistic-like and internalizing traits in children from the normal population using a longitudinal design. The authors found evidence for an asymmetric bi-directional relation between the two; autistic-like traits measured at age 7 contributed to internalizing traits measured at age 12 and vice versa (although the latter relation—early internalizing traits contributing to latter autistic-traits—was found to be somewhat smaller compared to the first; Hallett et al.
2010).
The aim of this study was to compare children with anxiety disorders to typically developing children with respect to current ASD-like behaviors as well as ASD symptoms in early development (rated retrospectively). We studied a group of children with mixed anxiety disorders. Although a ‘pure’ group of, for example, children with social anxiety disorders only may also be of interest, comorbidity among anxiety disorders is high and therefore ‘pure’ cases hardly exist in the real world of children seeking treatment for anxiety (e.g., Kendall et al.
2010 reported that almost 80 % of the anxiety disordered youth with a principal diagnosis of generalized anxiety disorder, separation anxiety disorder or social anxiety disorder, was also diagnosed with at least one of these three disorders).
Discussion
This study explored autistic symptoms and ASD-like behaviors in children with anxiety disorders. The main results were: (a) Parents reported that their school-aged children with anxiety disorders had significantly more ASD symptoms in early childhood than typically developing children; (b) over one-third of children with an anxiety disorder, but no known history of ASD, exceeded at least one of the three ADI-R thresholds for early childhood clinically significant ASD symptoms; and (c) early ASD symptoms were found to be related to current ASD-like behaviors as well as current symptoms of social anxiety disorder.
It was found that ASD symptoms seem to be present early in life (based on retrospective reports of parents) in some children who have developed clinical anxiety but who are not recognized by professionals as having current ASD. Cluster analysis revealed that a small number of children (
n = 4) showed a classic, if moderate, ASD-like profile (high scores on both the social and communication domains, as well as moderate repetitive behaviors), while the majority of the children with anxiety disorders (
n = 25) were characterized by a milder ASD-like early childhood phenotype. Commonly endorsed items for the latter group included items such as early childhood impairments in spontaneous imitation, imaginative play (with peers), and offering to share, but at a low-severity level. Interestingly, the most commonly endorsed symptoms, and significant differences between clinically anxious and controls, were found for items in the communication domain. Similarly, the study of Hallett et al. (
2010) found that early communication difficulties in children from the general population contributed more strongly to internalizing traits in later life than social difficulties and repetitive behaviors.
With the exception of symptoms of obsessive compulsive disorder and post traumatic stress disorder, all other anxiety symptoms were found to be associated with current ASD-like behaviors. In contrast, only one (symptoms of social anxiety disorder) was found to be associated with early ASD symptoms (as measured with the ADI-R). It should be noted that the instruments to assess anxiety symptoms (SCARED-71) and current ASD-like behaviors (CSBQ) are more similar (likely resulting in higher correlations between these two instruments) compared to the instrument used to assess early ASD symptoms (ADI-R). However, it may also be that children with a moderate or high degree of ASD symptoms early in life, may be more prone to develop (symptoms of) some (social anxiety) rather than other anxiety disorders. In addition, although different instruments were used to assess early ASD symptoms and current ASD-like behaviors, evidence for moderate ASD-symptom stability was found in the current study (i.e., the correlation between early ASD symptoms and current ASD-like behaviors was found to be medium and significant,
r = .36), however, it was not as large as in other studies that used the same instrument to asses ASD traits over time (e.g., the correlation between ASD traits over a 5-year period of time was found to be .59 for boys and .55 for girls in the study of Hallett et al.
2010).
With respect to current ASD-like behaviors, this study found almost one-third (31.0 %) of the clinically anxious sample falling in the ASD range on the current ASD-like behavior instrument (CSBQ). This finding is in accordance with previous studies (Towbin et al.
2005; Pine et al.
2008), although Pine and colleagues reported a somewhat lower percentage for the children with anxiety disorders (i.e., 4.0–24.6 %). The relatively high percentage may partly be explained by symptom overlap in the instrument used to assess current ASD-like behaviors (e.g., items of the CSBQ such as ‘makes little eye-contact’ or ‘does not initiate to play with others’ may very well be present in socially anxious children, while those behaviors may not necessarily be attributable to ASD). Nevertheless, it may be important in clinical practice to explore current ASD-like behaviors in children with anxiety disorders as the study of Puleo and Kendall (
2011) found that the presence of ASD traits in anxious children not identified to have ASD may have consequences for treatment choice. That is, children with moderate ASD traits were found to profit more from family oriented cognitive behavioral therapy (CBT) compared to individual CBT (Puleo and Kendall
2011). Note also that Sofronoff et al. (
2005) found that for the treatment of anxiety in children with Asperger, active parental involvement enhanced the effects of this intervention. Contradictory, in typically developing children with anxiety disorders an additional effect of a family component is not always found (e.g., In-Albon and Schneider
2007; Bodden et al.
2008); however, those studies did not explore the possible role of ASD traits.
Finally, limitations of the study have to be addressed. The first limitation is that ASD symptoms in early development were assessed retrospectively, and although the ADI-R has good psychometric properties (e.g., Lord et al.
1994), recall bias cannot be excluded. In addition, it cannot be ruled out that recall-bias is different in the two groups. That is, it might be that parents of a child with an anxiety disorder report more negative aspects of their child’s development (e.g., due to the stress associated with having a child with an anxiety disorder) compared to parents of control children. However, it was also found that the early ASD scores from parents who were interviewed after their child received treatment (perhaps being more in a positive state of mind because of successful treatment outcome) did not differ from the scores of parents interviewed before their child received treatment. Furthermore, the distinct clusters of early ASD symptoms revealed in the cluster analysis suggest that only a minority of children were reported to have a pattern of significant ASD symptomatology in early childhood, suggesting at the least that such a recall bias could not have affected the parents of children with anxiety disorders uniformly. Nonetheless, as recall bias cannot be ruled out when using a retrospective measure, a potential supplemental research strategy for future studies would be to ask the parents to provide home videotapes about their child’s early development, and then rate the ASD-related behaviors (note that ASD-research already made use of such an approach; e.g., Osterling et al.
2002; Werner and Dawson
2005).
A second limitation of the study is that we could not establish temporal precedence for we did not assess anxiety early in life. It may well be that anxiety disorders were already present early in life and may have not been noticed or treated until later in life. For example, social anxiety disorder may have its onset early in development and behavioral inhibition is thought to be a precursor (e.g., Bögels et al.
2010; Rapee and Spence
2004). Behavioral inhibition is viewed as a temperament style that consists of a pattern of behaviors such as avoidance, withdrawal, shyness, and reticence that manifest in response to novelty or unfamiliarity (Hirshfeld-Becker et al.
2007). Although there may be some overlap between items of the ADI-R and behavioral inhibition under select novel conditions (e.g., items of the ADI-R asking about ‘seeking to share one’s enjoyment with others’ or ‘initiating social talk’), the ADI-R primarily includes items that share (very) little overlap with behavioral inhibition (e.g., ‘imaginative play’ or ‘facial expressions’). In addition, post hoc analyses revealed significant differences between the clinically anxious children and the typically developing children for such non-overlapping items, suggesting that the early life ASD symptoms found in clinically anxious children are not solely attributed to the diagnostic overlap with other constructs like behavioral inhibition. However, considering the cross-sectional nature of this study, results about the relation between elevated ASD symptoms in early development and anxiety disorders later in life should be interpreted with caution and viewed with the perspective that these are preliminary findings that should be further explored.
Third, relatively few children (
n < 10) were diagnosed with obsessive compulsive disorder, panic disorder, agoraphobia and post traumatic stress disorder. It is possible that this confounded the results. For example, certain autistic-like behaviors are common in children with obsessive compulsive disorder (e.g., Ivarsson and Melin
2008), and autistic-like behaviors in adults were found to be more common in those with obsessive compulsive disorder than in those with social anxiety disorder (Bejerot and Mörtberg
2009). It could be that differences between the clinical group and controls would have been even larger if more cases of obsessive compulsive disorder were included. Larger sample sizes of ‘pure’ cases (e.g., children with just social anxiety disorder or obsessive compulsive disorder) would in some ways have been preferable to explore the specific associations between anxiety disorders and ASD symptoms early in life. However, as noted in the introduction, clinically anxious children are often diagnosed with multiple anxiety disorders (Kendall et al.
2010), making such comparisons complicated and unrepresentative of treatment-seeking children with clinical anxiety.
A final limitation is that we did not have IQ data available for the two groups and that the clinically anxious children consisted of a very heterogeneous sample. That is, our sample of children with anxiety disorders consisted of a variety of anxiety disorders and most of the children had multiple anxiety disorders. On the other hand, all were highly verbal and able to respond appropriately to our anxiety disorders diagnostic interview, denoting roughly age-appropriate levels of verbal communication.
Despite the addressed limitations, the relationship between anxiety and ASD is interesting and warrants further investigation. Not only are anxiety disorders highly prevalent and perhaps somewhat (phenomenologically) endemic in individuals with ASD, but also—as found in this and other studies—early ASD symptoms and current ASD-like behaviors are significantly more prominent in children with anxiety disorders. Interesting, the two disorders may share some deficits in the same brain region. For example, for both children with ASD as well as clinically anxious children, anxiety symptoms were found to be associated with abnormalities in the functioning of the amygdala (Thomas et al.
2001; Juranek et al.
2006). In addition, similar genetic markers are found for anxiety in children with autism and typically developing children (e.g., Gadow et al.
2008). However, results concerning the amygdala are inconclusive, other brain areas (as well as other biomarkers) are found to be abnormal in ASD (e.g., Brambilla et al.
2003; Hughes
2009), and apparent similarity may disguise underlying differences. Gregory and Eley (
2007) concluded that environmental factors are at least equally important as genetic factors for the differentiation of individual anxiety levels in children. Furthermore, Hallett et al. (
2009) found little evidence for genetic influences with respect to the phenotypic correlation between autistic traits and internalizing traits in the general population, and propose that anxiety may be a response to autistic-like difficulties. Likely, the stress experienced by many children with moderate to high ASD symptoms promotes anxiety and mood disorders (Wood and Gadow
2010), however, more empirical research is needed. Although part of the results (found here and elsewhere) may be conflicted by diagnostic overlap between ASD and anxiety, the present findings do warrant further investigation of the relationship between the two disorders, and more specifically of the role of ASD symptoms in the development, maintenance, and treatment of anxiety disorders.