Assessment of Anxiety in Youth With Autism Spectrum Disorder
Lauren J. Moskowitz1, Tamara Rosen2, Matthew D. Lerner2 and Karen Levine3, 1St. John’s University, Queens, NY, United States, 2Stony Brook University, Stony Brook, NY, United States, 3Harvard Medical School, Lexington, MA, United States
Abstract
Anxiety is one of the most common presenting problems for youth with autism spectrum disorder (ASD) and causes greater impairment than the symptoms of ASD alone. Despite the increased risk for anxiety disorders in ASD, there is limited research on the assessment of anxiety in this population, relative to the large body of literature on anxiety in typically developing children. This may be due, in part, to the difficulty of assessing anxiety in ASD. Anxiety is a multifaceted construct involving behaviors, cognitions, affect, and physiological arousal. These multiple components of anxiety are ideally assessed using a multimethod approach, including questionnaires, interviews, direct behavioral observation, and occasionally physiological responses elicited by feared stimuli. This chapter will review the variety of methods that can be used to assess anxiety in youth with ASD, discussing the advantages and disadvantages of each method as well as the psychometric properties and uses of various assessment tools.
Keywords
Anxiety; autism spectrum disorder; assessment; interviews; questionnaires; direct observation; physiological; multimethod
Assessment of Anxiety in Autism Spectrum Disorder
Background
Anxiety-related concerns are among the most common presenting problems for children and adolescents with autism spectrum disorders (ASDs) (White et al. 2009), with approximately 40% of youth with ASD meeting criteria for at least one anxiety disorder (van Steensel et al. 2011). This is likely an underestimate, given that anxiety is often overlooked in individuals with ASD. This may be due to a general clinical consensus that symptoms of anxiety are “better explained by the ASD itself” (White et al. 2009), a tendency known as diagnostic overshadowing. Another reason anxiety may be overlooked is because, historically, behaviorists have been reluctant to use the construct of “anxiety” to describe or explain behavior when discussing those with ASD (Groden et al. 1994). This may be because, unlike behaviors, the cognitions, affective/subjective state, and physiological arousal that are part of the construct of anxiety often cannot be directly observed in those individuals who cannot self-report (Groden et al., 1994). This perspective highlights the importance of attempting to address the multiple components of anxiety using a multimethod approach as well as the inherent difficulties of assessing anxiety in this population.
Difficulty of Assessing Anxiety in ASD
One reason anxiety is so difficult to assess is due to the inherent communication impairments in ASD. Given that up to one-half of individuals with ASD are functionally nonverbal and even those with verbal language often have difficulty describing their thoughts and feelings (Leyfer et al., 2006), traditional assessment of anxiety using self-report can be difficult or even impossible (Hagopian and Jennett, 2008). These communication deficits might also cause parents to be unaware of their child’s thoughts and feelings, limiting the usefulness of other-informant-reports, and underscoring the importance of a multimethod assessment approach. Second, the presence of co-occurring ID in over 50% of children with ASD (CDC, 2014) can further compound these difficulties with assessment, given that youth with ASD and ID often lack the ability to communicate their anxiety to an even greater extent than those with ASD who do not have ID (Davis et al., 2011). Third, symptoms of anxiety may present differently in ASD, making it more difficult for caretakers to identify the symptoms of anxiety. For example, those with ASD are more likely to express fear or anxiety through behaviors such as aggression, self-injury, and tantrums (White et al. 2009). Additionally, the content of the anxiety may differ for youth with ASD, such as fears of an unusual focus (e.g., fear of graffiti) in some youth with ASD (Kerns et al., 2014). Finally, given the symptom overlap in the current diagnostic classification system, clinicians often find it challenging to decide if symptoms such as social avoidance or compulsive checking should be conceptualized as part of the ASD or as a separate comorbid anxiety disorder (White et al. 2009). Ultimately, as Wood and Gadow (2010) suggested, the only way to distinguish between the overlapping symptoms of ASD and anxiety disorders may be to assess the function of the symptom—e.g., to assess the function of the social avoidance (i.e., disliking social interaction versus fear of being scrutinized or rejected). Still, it may be difficult or even impossible to assess these functions if a child cannot talk or report his own mental state. Faced with such difficulties, one might take a multistep approach, such as that advocated by Kerns and colleagues (2016) (i.e., consideration of development; assessment of discrete impairment; examination of overlap with candidate ASD symptoms) to identify whether a child may formally meet anxiety comorbidity criteria. Alternatively, one could prioritize a pragmatic approach, and conceptualize and treat patterns of behavior that are “phobia-like” as phobias, even if their derivation may not be fully understood or when it may be suspected that these are from a source such as hyperacusis or sensory over-reactivity (e.g., Koegel et al., 2004). In either case, a methodical and empirically-grounded assessment process, relying on a variety of different measures and informants to converge upon the presence of anxiety, is a vital first step.
Clinical Assessment of Anxiety in ASD
Given the prevalence and difficulties associated with anxiety in ASD, it is important to regularly screen and/or assess for anxiety during clinical evaluations. Toward this end, an array of measures are available that can be useful. We now review available questionnaires/rating scales, clinical interviews, clinician-rated symptom measures, direct observation measures, and physiological measures, and how they might be useful for the assessment of anxiety in ASD.
Tools for Assessing Anxiety in ASD
Questionnaires/Rating Scales
Self-report questionnaires
Self-report questionnaires are the most widely-used formal evaluative method for assessing anxiety in youth with and without ASD. However, virtually none of these instruments are specifically designed to assess anxiety in ASD (see Table 5.1 for a summary of questionnaire instruments). Among the most well-researched measures are the Multidimensional Anxiety Scale for Children (MASC-C; March et al., 1997), the Screen for Child Anxiety Related Emotional Disorders (SCARED-C; Birmaher et al., 1997), and the Spence Children’s Anxiety Scale (SCAS; Spence, 1998). However, these measures may be more appropriate for higher-functioning youth, given their emphasis on verbally-mediated symptoms. On the other hand, the Revised Child Anxiety and Depression Scale (RCADS; Chorpita et al., 2000) may be useful for lower-functioning youth (i.e., those with ID and/or minimal verbal ability).
Table 5.1
| Form | # of items; time | Age range | Possible raters | Subscales/domains | Reliability in ASD | Validity in ASD | Use in ASD |
| MASC (March et al., 1997) | 39 items; 15 min | 8–19 years | Child, Parent | Social anxiety, physical symptoms, harm avoidance, separation anxiety/panic | • Acceptable internal consistency (Wood et al., 2009) |
• Modest convergent validity (Storch et al., 2012); • Treatment sensitivity (Wood et al., 2009) |
Appropriate with conditions as outcome measure, but emphasis on language may make it limited to high-functioning youth (Lecavalier et al., 2014); self-report scale may be measuring the same anxiety constructs in ASD and TD youth, more so than parent-report (White et al., 2015) |
| SCARED (Birmaher et al., 1997, 1999) | 41 items; 10 min | 9–18 years | Child, Parent | Panic/ somatic, generalized anxiety, separation anxiety, social phobia, and school phobia | • Moderate internal consistency (Reaven et al., 2009) |
• Moderate convergent validity; • Mixed evidence for sensitivity and specificity (Kerns et al., 2015; Stern, Gadgil et al., 2014); • Some evidence for treatment sensitivity(Reaven et al., 2009) |
Potentially appropriate for as outcome measure, but emphasis on language may make it limited to higher-functioning youth (Lecavalier et al., 2014); evidence that the SCARED measures anxiety similarly in ASD and TD youth (Stern et al., 2014) |
| SCAS (Spence, 1998) | 44 items; 15 min | 7–14 years | Child, Parent | Separation anxiety, social phobia, obsessive-compulsive, panic-agoraphobia, generalized anxiety and physical injury fears | • Acceptable internal consistency, though not strong for all subscales (Kerns et al., 2016) |
• Acceptable sensitivity and specificity (Kerns et al., 2016) |
• Acceptable to moderate parent-child agreement (Kerns et al., 2016) • May be more appropriate for higher-functioning youth (Grondhuis and Aman, 2012) |
| RCMAS (Reynolds and Richmond, 1978) | 37 items; 10–15 min | 6–19 years | Child | Physiological anxiety, worry/oversensitivity, and social concerns | • Good internal consistency (Mazefsky et al., 2011) |
• Some evidence for specificity and sensitivity (Mazefsky et al., 2011) |
May be useful as screening measure, not appropriate as outcome measure (Lecavalier et al., 2014) |
| RCADS (Chorpita et al., 2000) | 47 items; 20 min | 9–18 years | Child, parent | Separation anxiety, social phobia, generalized anxiety, panic, obsessive-compulsive | • Acceptable test-retest reliability (Kaat and Lecavalier, 2015) • Poor inter-rater reliability (Kaat and Lecavalier, 2015), • Acceptable internal consistency (Hallett et al., 2013; Sterling et al., 2015) |
• Mixed findings regarding divergent validity (Kaat and Lecavalier, 2015; Sterling et al., 2015) • Modest convergent validity (Sterling et al., 2015) |
Potentially appropriate as outcome measure (Lecavalier et al., 2014); May be useful for ID, as 33% of individuals in Kaat and Lecavalier (2015) had IQ below 85 |
| ASC-ASD (Rodgers et al., 2016) | 24 items; 10 min | 8–15 years | Child, parent | Performance Anxiety, Uncertainty, AnxiousArousal, Separation Anxiety | • Good-to-excellent internal consistency and 1 month test-retest reliability and high parent–child agreement (Rodgers et al., 2016) |
• Good convergent, discriminant, and content validity (Rodgers et al., 2016) |
May be limited to verbally fluent youth with ASD (Rodgers et al., 2016) |
| STAIC (Spielberger, 1973) | 20 items; 20 min | 6–14 years | Child | State anxiety, trait anxiety | May be useful for distinguishing state and trait anxiety (Lanni et al., 2012) | ||
| SASC-R (La Greca and Stone, 1993); SAS-A (La Greca and Lopez, 1998) | 22 (SASC-R); 26 (SAS-A); 10–15 min | 8–18 years | Child | Fear of negative evaluation, generalized and specific social avoidance/distress | • Good internal consistency (Kaboski et al., 2015) |
• Treatment sensitivity (Kaboski et al., 2015) • Some evidence for convergent validity (Henderson et al., 2006) |
When items overlapping with ASD were removed, total scale had good internal consistency (Kuusikko et al., 2008) |
| CASI-4R (Gadow and Sprafkin, 2002) | 26 (anx items); 15–20 min | 5–18 years | Parent, teacher | Generalized anxiety disorder, social phobia, separation anxiety disorder, obsessive-compulsive disorder, specific phobia, panic attack | • 20-item version created to measure anxiety in ASD showed good internal consistency in children with and without cognitive impairment (Sukhodolsky et al., 2008; White et al., 2012) |
• Good convergent validity for 20-item version (White et al., 2012) |
20-item version is appropriate with conditions for use in ASD as outcome measure (Lecavalier et al., 2014) |
| SWQ (Spence, 1995) | 10; 10 min | 8–17 years | Parent, child | Social anxiety | • Acceptable internal consistency (Kerns et al., 2016) |
• Treatment sensitivity (Kerns et al., 2016) |
Weak relation of parent- and child-report; limited psychometric data (Kerns et al., 2016) |
| ASD-CC (Matson and Gonzalez, 2007) | 49 items | 2–16 years | Parent | Worry/Depressed and Avoidant subscales | • Worry/Depressed and Avoidant subscales of the ASD-CC have demonstrated moderately good internal consistency (Davis et al., 2011) |
• Worry/Depressed and Avoidant subscales of the ASD-CC have demonstrated good convergent & discriminant validity (Rieske et al., 2013) |
May be useful in assessing anxiety (Rieske et al., 2013) |
| ADAMS (Esbensen et al. 2003)a | 28 items; 10 min | 10–79 years | Informant-rating scale | Social avoidance, general anxiety, compulsive behavior | • Excellent internal consistency and test-retest reliability; interrater reliability good for Compulsive Behavior but poor for GAD in ID (Esbensen et al. 2003) |
• Valid for screening for OCD in individuals with ID (Esbensen et al. 2003)a |
Potentially appropriate as an outcome measure for anxiety in ASD and ID, but only examined in ID, not ASD (Lecavalier et al., 2014) |
| BISCUIT, Part II (Matson et al., 2009)a | 11 (anx items); 10 min | 17–37 months | Parent or caregiver | Anxiety/Repetitive Behavior Avoidance behavior | • Excellent internal consistency in youth with DD including ASD (Matson et al. 2009)a |
Cutoffs and norms for each factor established for children with DD including ASD (Matson et al., 2009) | |
| DBC (Brereton et al., 2006)a | 9 (anx items); 15–20 min | 3–24 years | Parent or teacher | Anxiety subscale | • Excellent test-retest reliability, but moderate interrater reliability and internal consistency in youth with ID (Reardon et al., 2015)a |
• Satisfactory construct, criterion, convergent, divergent, and criterion validity in youth with ID without ASD (Reardon et al., 2015)a |
Used to compare psychopathology including anxiety in youth with ASD (73% of whom had ID) to ID alone (Brereton et al., 2006) |
| NCBRF (Aman et al., 1996)a | 15 (anx items); 7–8 min | 4–18 years | Parent or teacher | Insecure/Anxious subscale and Overly Sensitive subscale | • Good internal consistency and test retest reliability in ID (but not ASD; moderate to poor parent-teacher interrater reliability (Reardon et al., 2015)a |
• Some support for criterion validity; low convergent validity with DBC-Anxiety in ID, 37% with PDD (Reardon et al., 2015)a |
Used to characterize children with ASD, but has not been evaluated as a screen for anxiety disorders in ID or ASD (Reardon et al., 2015); not appropriate as anxiety outcome measure in ASD (Lecavalier et al., 2014) |
Note: ASD = autism spectrum disorder. ASC-ASD = Anxiety Scale for Children with ASD. ADAMS = Anxiety, Depression, and Mood Scale. BISCUIT = Baby and Infant Screen for Children with aUtism Traits. DBC = Developmental Behaviour Checklist. MASC = Multidimensional Anxiety Scale for Children. NCBRF = Nisonger Child Behavior Rating Form. SCARED = Screen for Child Anxiety Disorders. RCMAS = Revised Children’s Manifest Anxiety Scale. RCADS = Revised Child Anxiety and Depression Scale. STAIC = State-Trait Anxiety Inventory for Children. SASC-R = Social Anxiety Scale for Children revised. SAS-A = Social Anxiety Scale for Adolescents.
aSample includes Intellectual Disability (ID).
Potentially appropriate treatment outcome measures include the MASC, SCARED, and the RCADS, while the Revised Children’s Manifest Anxiety Scale (RCMAS; Reynolds and Richmond, 1978) may be more useful as a screening measure. The State-Trait Anxiety Inventory for Children (STAIC; Spielberger, 1973) may be useful for the specific purpose of distinguishing trait and state anxiety in ASD (Lanni et al., 2012), while the Social Anxiety Scale for Children revised (SASC-R; La Greca and Stone, 1993), the Social Anxiety Scale for Adolescents (SAS-A; La Greca and Lopez, 1998), or the Social Worries Questionnaire (SWQ; Spence, 1995) may be the measures of choice for assessing social anxiety. However, the SWQ is relatively under-researched. Although all of these self-reports were designed for typically developing (TD) children rather than those with ASD, the freely available Anxiety Scale for Children with ASD, Parent and Child versions (ASC-ASD), an adapted version of the RCADS which includes additional items related to sensory anxiety, intolerance of uncertainty and phobias, shows promising psychometric properties for youth with fluent speech (Rodgers et al., 2016).
Self-reports in ASD and ID
Although the RCMAS, SASC, and STAIC have each been used in one study with youth with ID (see Reardon et al., 2015 for a review), the majority of aforementioned self-report questionnaires have only been examined in youth with high-functioning ASD (HFA; i.e., those who have average or above average IQ and/or greater verbal abilities). Traditional self-report questionnaires, which typically require a second- or third-grade reading level, may be inappropriate for some children with ASD and ID due to limitations in cognition, communication, and comprehension. However, there is some evidence that self-reports that are modified can be reliable and valid for some individuals with ID (Hagopian and Jennett, 2008). For example, the Fear Survey Schedule for Children-Revised (FSSC-R) and the Fear Survey for Children With and Without Mental Retardation (FSCMR) have both been psychometrically evaluated with youth with ID (Reardon et al., 2015). Modifications to self-reports for those with ID include both verbal and visual presentation, simpler language, limiting the number of words, pictorial representations of response options (e.g., a visual scale of facial expressions of fear), illustrations/photos to provide visual representations of items, and neutral items to assess acquiescence or choosing the more positive response (Hagopian and Jennett, 2008).
Other Informant Rating Scales
Questionnaires completed by other informants (see Table 5.1), such as parents or teachers, can offer additional information beyond self-reports. First, other-informant-report versions are available for the MASC, SCARED, RCADS, SCAS, and SWQ. However, relations of parent and child report may be weaker for the MASC and SWQ. The Child and Adolescent Symptom Inventory-4th Edition Revised (CASI-4R; Gadow and Sprafkin, 2002) is an additional other-informant-report measure. A 20-item version of the anxiety scale has been created to assess anxiety specifically in ASD samples (Sukhodolsky et al., 2008), which may be appropriate as a treatment outcome measure in ASD (Lecavalier et al., 2014). Of note, in contrast to the previously mentioned self-reports, most of the youth with ASD in the study by Sukhodolsky et al. (2008) had ID, ranging from mild to severe/profound ID. Finally, the Autism Spectrum Disorders-Comorbidity for Children (ASD-CC; Matson and Gonzalez, 2007) was designed to assess co-occurring disorders in ASD, specifically. The Worry/Depressed and Avoidant subscales of the ASD-CC may be useful for assessing the presence of anxiety in ASD (Rieske et al., 2013). As youth with ASD may evince patterns of elevated self-perceptions when completing self-report measures compared to parent-ratings (Lerner et al., 2012), these other-informant measures may be useful in obtaining a more comprehensive picture of anxiety in ASD.
Other-Informant Rating Scales in ASD and ID
Although almost all of the aforementioned questionnaires (with the exception of the CASI) have only been examined in youth with HFA and there are none designed specifically to assess anxiety in children with ASD and comorbid ID, there are several global measures of emotional and behavioral problems (including anxiety) that have been developed for individuals with ID or DD. Among these broad-based informant questionnaires, the Developmental Behaviour Checklist (DBC; Brereton et al., 2006) and the Nisonger Child Behaviour Rating Form (NCBRF; Aman et al., 1996) have demonstrated the strongest psychometric properties of their subscales measuring anxiety symptoms, although their capacity to screen for anxiety disorders in youth with ID and/or ASD has not been evaluated and they may not be useful as an outcome measure for anxiety in ASD and/or ID. The Anxiety, Depression, and Mood Scale (ADAMS; Esbensen et al. 2003) appears to be a reliable and valid instrument for screening for OCD in individuals with ID, although its use with other anxiety disorders may be limited (Hagopian and Jennett, 2008) and it has not been examined in ASD specifically. The Baby and Infant Screen for Children with aUtism Traits (BISCUIT), Part II (Matson et al., 2009) was developed to screen for comorbid psychopathology—including Anxiety/Repetitive Behavior—in infants and toddlers who have a developmental delay including ASD. One potential limitation is that anxiety is combined with repetitive behaviors. Although there may be overlap, it would be important to differentiate symptoms of comorbid anxiety from repetitive behavior, which is a core ASD symptom, given that repetitive and/or restricted behaviors and interests (RRBIs) are associated with anxiety in ASD but remain distinct constructs (Kerns et al., 2016).
Clinical Interviews
While questionnaires are valuable for screening and assessing a wide variety of symptoms from multiple sources and are less time- and resource-intensive than interviews, diagnoses of anxiety disorders tend to be more accurately made using interviews, which allow the clinician to obtain more detailed information from the child and/or parent’s verbal reports as well as from observations of the child’s behavior during the interview. Semi-structured interviews can be tailored to the child and/or parent and allow the clinician to provide examples and clarify items. Although most interviews used to assess anxiety in youth with ASD have been designed for TD youth, two interviews have been modified for youth with ASD (Table 5.2).
Table 5.2
Interviews and clinician rating scales
| Measure | Time to complete | Age range | Possible interviewees | Format | Reliability in ASD | Validity in ASD | Use in ASD |
| ADIS-IV-C/P (Silverman and Albano, 1996) | 1–3 h | 7–18 | Parent and child | Semi-structured interview (interviewer-based) | Poor agreement between child and parent, but excellent parent and consensus agreement (Storch et al., 2012) and clinician-to-clinician agreement (Ung et al., 2014) | Discriminant validity between anxiety and ASD severity; convergent validity among differing reports of two of the anxiety subdomains (Renno and Wood, 2013) | Used for diagnosis and outcome in several RCTs (e.g., Wood et al., 2009); appropriate with conditions as an outcome measure (Lecavalier et al., 2014), although only examined in youth with HFA (IQ ≥ 70) |
| ADIS/ASA (Kerns et al., 2014)a | 15–30 min for ASA (+1–3 h for ADIS) | 7–18 | Parent and child | Semi-structured interview (addendum to ADIS) | Good-to-excellent interrater agreement and test-retest reliability (Kerns et al., 2014; Kerns et al., in press) | Adequate convergent and discriminant validity with other measures (Kerns et al., in press) | Can aid in differential diagnosis of ASD and anxiety; has not been used for diagnosis or outcome in treatment research (Kerns et al., 2014) |
| PARS (RUPP, 2002) | 20–30 min | 6–17 | Parent and child | Clinician rating scale | Excellent test-retest reliability and interrater reliability, but low internal consistency (Storch et al., 2012b); IQ > 70 | Convergent and divergent validity partially supported (moderately low correlations with other anxiety measures) (Storch et al., 2012b) | Sensitive to treatment (Johnco et al., 2015); Appropriate with conditions as outcome measure; child interview requires fluent language (may limit use to HFA) (Lecavalier et al., 2014) |
| CY-BOCS-PDD (Scahill et al., 2006) | 30 min | 6–17 | Parent and child | Clinician rating scale | Excellent interrater reliability and internal consistency (Scahill et al, 2006)b | Appears distinct from other measures of repetitive behavior (Scahill et al, 2006)b | Demonstrated sensitivity to change (McDougle et al., 2005) |
| K-SADS-PL (Kaufman et al., 1997) | 90 min to 2.5 h | 6–18 | Parent and child | Semi-structured interview (interviewer-based) | Excellent interrater reliability between interviewers in youth with HFA (Mattila et al., 2010; Zainal et al., 2014) | Preliminary evidence of convergent validity between SCAS-P and K-SADS-PL anxiety screen total score in youth with HFA (Zainal et al., 2014) | Used to assess prevalence of comorbid DSM-IV disorders including anxiety in youth with ASD (e.g., Gjevik et al., 2011) and used for diagnosis in treatment research (Reaven et al., 2009); has not been used as an outcome measure |
| ACI-PL (Leyfer et al., 2006)a | 1–3 h | 5–17 | Parent | Semi-structured interview (modified from K-SADS) | Good inter-rater reliability and test-retest reliability for OCD (Leyfer et al., 2006);b Mean IQ = 82 | Good concurrent validity for OCD, though validity for other anxiety disorders not examined (Leyfer et al., 2006)b | Used to assess prevalence of psychiatric disorders including anxiety in ASD (Mazefsky et al., 2011); has not been used as an outcome measure |
| P-ChIPS (Fristad et al., 1998) | 60 min | 6–17 | Parent | Structured interview (respondent-based) | Good-to-excellent interrater reliability (except for GAD in youth with IQ < 70) | Fair concordance between P-ChIPS and CASI (Witwer et al., 2012) in ASD; some challenges related to face and content validity (Witwer and Lecavalier, 2010)b; Mean IQ = 68 (range 42–150) | Used to assess prevalence of comorbid disorders including anxiety in ASD, both with and without ID (Witwer and Lecavalier, 2010; Witwer et al., 2012); has not been used as an outcome measure |
| Internal consistency good for SoP, fair for SP & SAD, poor for GAD & OCD (Witwer et al., 2012)b | |||||||
| DISC-IV (Shaffer et al., 2000) | 70–120 min per informant | 2–18 | Parent and child | Structured interview (respondent-based) | Reliability not examined in youth with ASD | Validity not examined in youth with ASD | Used to assess prevalence of psychiatric disorders including anxiety in youth with ASD (deBruin et al, 2007); has not been used as an outcome measure |
| CAPA (Angold et al., 1995) | 60 minutes per informant | 9–17 | Parent and Child | Combines interviewer-based & respondent-based format | Reliability not examined in youth with ASD | Validity not examined in youth with ASD | Used to assess prevalence of psychiatric disorders including anxiety in ASD (Simonoff et al., 2008); has not been used as an outcome measure |
| PAPA (Egger et al., 1999) | 1.5–2 h | 2–5-year old | Parent | Combines interviewer and respondent-based format | Reliability not examined in youth with ASD | Validity not examined in youth with ASD | Used to assess prevalence of psychiatric disorders including anxiety in ASD (Salazar et al., 2015)b |
Note: ADIS-IV-C/P = Anxiety Disorders Interview Schedule for DSM-IV, Child and Parent Versions. ACI-PL = Autism Comorbidity Interview – Present and Lifetime Version. ADIS/ASA=Autism Spectrum Addendum to the ADIS-P. CAPA = Child and Adolescent Psychiatric Assessment. CY-BOCS-PDD=Children’s Yale-Brown Obsessive Compulsive Scale for Pervasive Developmental Disorders. DISC = Diagnostic Interview Schedule for Children. K-SADS = Schedule for Affective Disorders and Schizophrenia in School-Aged Children. PAPA = Preschool Age Psychiatric Assessment. PARS=Pediatric Anxiety Rating Scale. P-ChIPS=Children’s Interview for Psychiatric Syndromes – Parent Version.
aDesigned for ASD.
bIncludes ID.
Of the interviews designed for TD youth, the Anxiety Disorders Interview Schedule for DSM-IV, Child and Parent Versions (ADIS-IV-C/P; Silverman and Albano, 1996) is the only interview specifically developed for assessing anxiety disorders, covering multiple dimensions of anxiety, and is generally considered the “gold standard.” The ADIS-IV-CP is the interview with the most research support in ASD; it has been validated in youth with HFA, used for both diagnosis and treatment outcome, and uses visual prompts to supplement a child’s language. Another semi-structured interview, the Schedule for Affective Disorders and Schizophrenia in School-Aged Children (K-SADS-PL; Kaufman et al., 1997), has been used less frequently in ASD and is lighter in its coverage of anxiety disorders than the ADIS, but has demonstrated reliability in youth with HFA, is free-of-charge, and requires less training than the ADIS. One limitation of both the ADIS and K-SADS is that they do not have a systematic approach for distinguishing symptoms of anxiety from the core features of ASD and may not capture the unique manifestations of anxiety in ASD; as such, they have been modified for ASD.
The Autism Comorbidity Interview—Present and Lifetime Version (ACI-PL; Leyfer et al., 2006) was modified from the K-SADS to differentiate between the core symptoms of autism and symptoms of other mental disorders including anxiety, which may make it more stringent than other interviews that do not differentiate (see Mazefsky et al., 2012); however, OCD was the only anxiety disorder to be examined psychometrically. More recently, Kerns and colleagues (2014) developed the Autism Spectrum Addendum to the ADIS-P (the ADIS/ASA), a set of guidelines and supplementary items designed to differentiate traditional DSM anxiety disorders in ASD from the more ambiguous symptoms (e.g., worries regarding schedule or environmental changes) often present in ASD. Recent findings show strong psychometric properties of the ADIS/ASA (Kerns, in press), which is the only measure designed to probe for more varied symptoms of anxiety in ASD.
Structured, or respondent-based, interviews—such as the Children’s Interview for Psychiatric Syndromes – Parent Version (P-ChIPS; Fristad et al., 1998) and Diagnostic Interview Schedule for Children, Version IV (DISC-IV; Shaffer et al., 2000)—and interviews that combine respondent-based and interviewer-based approaches—such as the Child and Adolescent Psychiatric Assessment (CAPA; Angold et al., 1995) and Preschool Age Psychiatric Assessment (PAPA; Egger et al. 1999)—have also been used to assess comorbid psychiatric disorders including anxiety in youth with ASD. The only one of these that has reported reliability and validity in ASD is the P-ChIPS (Witwer et al., 2012). Although these four interviews allow for direct comparison with TD populations and are briefer and require less training than the ADIS and ACI, these interviews may inflate the prevalence of anxiety in ASD by not discriminating symptoms of anxiety from symptoms of ASD (Mazefsky et al., 2012), or underestimate anxiety in ASD by capturing only those symptoms that present in a similar manner as in typically developing youth (Kerns and Kendal, 2012). In sum, although the ADIS/ASA, ACI-PL, and P-ChIPS are promising interviews, we need further examination of psychometric properties of these measures in youth with ASD (Kerns et al., 2016).
Interviews in ASD and ID
Most of the interviews designed to assess anxiety in TD youth have only been examined in research studies with youth with ASD whose IQ > 70. However, Witwer et al. (2012) examined the reliability and validity of the P-ChIPS in parents of youth with ASD whose IQs ranged from 42 to 150. Although overall interrater agreement for GAD was excellent, IQ < 70 impacted interrater agreement for GAD, suggesting that some modifications to the P-ChIPS may be necessary in youth with ASD and ID (Witwer et al., 2012). For individuals with ASD and ID, informant reports that were designed for TD children may need to be modified to take into account that caregivers may not know what their children are thinking or feeling. For example, Cordeiro, Ballinger, Hagerman, and Hessl (2011) modified the ADIS-P for the parents of individuals with fragile X syndrome by eliminating the screening question criteria, which allowed for a diagnosis of social phobia in individuals who exhibited clinically significant impairment as a result of social phobia symptoms but were unable to verbalize “a worry that they might do something embarrassing.”
Clinician-Rated Symptom Measures
The Pediatric Anxiety Rating Scale (PARS; RUPP, 2002) is a clinician-rated measure of anxiety symptoms severity that was designed to rate the current frequency, severity, and associated impairment of anxiety symptoms across a range of anxiety disorders. The ADIS and PARS, which both have strong psychometric properties in youth with HFA, were the only clinician interviews that were classified as “appropriate with conditions” for use as an outcome measure in ASD, although their use may be limited to those who are more verbal and/or have a higher IQ (Lecavalier et al., 2014). Another semi-structured clinician-rated instrument is the Children’s Yale-Brown Obsessive Compulsive Scale for Pervasive Developmental Disorders (CY-BOCS-PDD; Scahill et al., 2006) which was designed to assess the symptoms and severity of OCD in youth with ASD, 61% of whom had ID. The CY-BOCS-PDD only includes the Compulsions checklist from the original CY-BOCS (obsessions were dropped because of cognitive and communication limitations in ASD) and added repetitive behaviors common to youth with ASD. Although the CY-BOCS-PDD demonstrates strong psychometric properties, it can be difficult to truly ascertain the presence of OCD by only relying on compulsions, without assessing obsessions, given the functional relationship between obsessions and compulsions. Of note, although the CY-BOCS-PDD performed somewhat differently for children in the normal IQ range (IQ ≥ 70) versus those with ID (IQ < 70), most differences were not significant.
Direct Observation
As a result of difficulties with self-report and other-informant-report, anxiety in individuals with ASD, particularly those with comorbid ID and/or minimal verbal abilities, must often be inferred from the individual’s overt behavior or “fear responses” via direct observation (Rosen et al., 2016). Although information collected from interviews and questionnaires can narrow the focus and help clinicians to formulate hypotheses regarding the controlling variables of anxiety, direct behavioral observation is often needed to clarify and validate these findings in children with ASD, particularly those with both ASD and ID (Hagopian and Jennett, 2008). Direct observations may generate more detailed and likely more accurate information than questionnaires/interviews about the situations that evoke anxiety in youth with ASD, the behaviors they display when they are anxious, and the antecedents and consequences related to their anxiety. Important variables that are overlooked in questionnaires and interviews are sometimes uncovered during direct observation. However, observation requires more time and financial commitment, and may restrict the opportunity to observe a range of anxious behaviors and anxiety-provoking situations.
One approach to direct observation is the Behavioral Avoidance Test (BAT), which involves progressively exposing the child to the feared stimulus while assessing the child’s avoidance response, subjective level of anxiety, physiological reactions, and/or behavioral responses (Hagopian and Jennett, 2008). BATs can be used to observe levels of anxiety during assessment as well as during and after intervention to evaluate treatment outcomes. While certain anxiety disorders lend themselves to this design (e.g., phobias, OCD), it may be more challenging to identify or control stimuli that evoke anxiety in children with more generalized anxiety (Hagopian and Jennett, 2008). Although the BAT has been widely used for assessing anxiety disorders, it may be particularly important to include a BAT in the assessment of anxiety in youth with ASD and ID, given the limits of self-report and interview in this population (Hagopian and Jennett, 2008). When a BAT is not feasible, naturalistic observations can still be used to assess anxiety during the interview and/or in the child’s home, school, or community. Practically, having caretakers record children’s anxious behavior may be a better option (Hagopian and Jennett, 2008). However, due to the limited verbal abilities of many individuals with ASD and ID, researchers often use indirect measures of behavior such as “distance to the avoided stimulus” or “number of steps completed within a hierarchy” to measure fear or avoidance responses rather than measuring direct behaviors that indicate anxiety (Rapp, Vollmer, and Hovanetz, 2005).
Physiological Measures
The use of informant rating scales, clinical interviews, and behavioral observations is standard practice in clinical assessment. However, there are advancing efforts to translate physiological measures, which represent a variety of indices that reflect bodily and neural responses related to anxiety, in clinical research (De Los Reyes and Aldao, 2015). Though such measures are not yet diagnostic or in wide use in clinical practice, they can add valuable information as well as metrics of change inaccessible to self-report due to concerns about insight or verbal abilities.
Cortisol is a hormone that is responsible for regulating the human stress response (e.g., Stansbury and Gunnar, 2011). Cortisol is often assessed via salivary samples, and has been used to index the stress response in ASD, wherein individuals are exposed to situations thought to elicit stress or anxiety (e.g., Corbett et al., 2008). However, there is evidence to suggest that cortisol may be measuring generalized stress, rather than anxiety per se, in ASD (Lanni et al., 2012).
Heart rate, often measured by number of beats per minute via an electrocardiograph (ECG), is thought to index an individual’s general state of arousal (Berntson et al., 1997). One examination found that individuals with ASD, compared to a TD group, had a higher heart rate in both low-anxiety and high-anxiety conditions (Kushki et al., 2013). However, a subsequent study found a blunted heart rate response following a stress test in anxious ASD youth, compared to the TD and non-anxious ASD groups, while reduced heart rate was related to higher levels of anxiety symptoms in the anxious ASD youth (Hollocks et al., 2014). Taken together, while ASD youth show increased heart rate to anxiety triggers, the relationship of heart rate to trait levels of anxiety in ASD is unclear.
The third measure is respiratory sinus arrhythmia (RSA), or vagal regulation of one’s heart rate and sympathetic influences (Berntson et al., 1997). Like heart rate, RSA is measured using an ECG, as it is indexed via the variability in heart rate. Studies have found evidence for a decreased RSA in ASD, which is related to symptoms of anxiety (Bal et al., 2010; Guy et al., 2014), including during threatening situations (Van Hecke et al., 2009). These studies suggest increased levels of generalized arousal in ASD.
The fourth measure is the error-related negativity (ERN), an error-related potential (ERP) component, which is measured using an electroencephalogram—an electrode array affixed to the head (see Olvet and Hajcak, 2009). The ERN is thought to represent monitoring of one’s own performance; increased monitoring and corresponding ERN are associated with higher anxiety levels (see Olvet and Hajcak, 2009), a relationship which has been found in ASD (Henderson et al., 2015; for countervailing findings see Henderson et al., 2006). However, the inconsistent findings in this population suggest more research is needed to uncover the potentially unique patterns of self-monitoring seen in youth with ASD.
Presently, additional research is needed to more firmly establish how the aforementioned measures index the heterogeneous presentation of anxiety in ASD. Specifically, questions regarding clinical and incremental utility, feasibility, cost, and training will need to be explored. In the future, though, these measures may be integrated into clinical practice to improve assessment and diagnosis. However, it is important to note that measures of physiological arousal in individuals with ASD are not on their own diagnostic or indicative of anxiety, and could also be reflective of other physical or emotional states (e.g., anger). This is because the process of labeling one’s state of affective arousal as “anxiety” or any other emotion is highly influenced by the situational context in which the arousal occurs (Bandura, 1988). For example, if one’s heart were racing while exercising, the arousal would not likely be interpreted as anxiety, whereas if one’s heart were racing while taking an exam, the arousal might be interpreted as anxiety because of the context in which the arousal occurs. Thus, information gained from physiological measures is most useful when interpreted in the context of a more comprehensive evaluation that incorporates multiple assessment methods from different informants (Moskowitz et al., 2013).
Multimethod Assessment of Anxiety in Youth With ASD
Just as increased heart rate might indicate anxiety in one context but excitement or anger in another context, behaviors such as running away or crying might indicate anxiety in a child with ASD at certain times in certain contexts, but that same child might also run away or cry because he is tired, in pain, angry, sad, or because he dislikes an object, person, or situation (or simply prefers another situation). Indeed, Hagopian and Jennett (2014) differentiated between “simple avoidance” in which the individual with ASD avoids non-preferred stimuli or situations (e.g., wearing shoes, academic task) versus “anxious avoidance” in which the individual exhibits avoidant behavior accompanied by traditional symptoms of anxiety including facial expressions indicative of fear, increased physiological arousal and, if possible, self-reported anxiety. One way to differentiate between these two scenarios is to assess the context in which avoidant behavior and physiological arousal occur, as part of a multimethod assessment.
Given the difficulty of assessing anxiety in youth with ASD, particularly those who have ID or are minimally verbal, a multimethod assessment of the behavioral, physiological, and cognitive/affective/contextual components of anxiety is often warranted. To this end, Moskowitz et al. (2013) developed a multimethod assessment strategy to operationally define the construct of “anxiety” in children with ASD and ID. We measured the affective/cognitive/contextual component of anxiety using parent-report questionnaires and blind observers’ Likert-type ratings, the physiological component of anxiety using heart rate (HR) and RSA, and the behavioral component of anxiety via direct observation (using idiosyncratic behavioral indicators of anxiety for each child). Regarding the affective/cognitive/contextual component, given the importance of identifying the context in which behaviors and physiological arousal occur, other-informant-report questionnaires such as the Stress Survey Schedule (SSS; Groden et al., 2011) can be used to identify situations that evoke anxiety in children with ASD, or one can simply ask parents and/or teachers (and even the child himself, if possible) which situations appear to evoke anxiety. In addition to relying on parent report, Moskowitz et al. (2013) also indexed the contextual or cognitive/affective component of anxiety (i.e., subjective fear or apprehension) by having blind observers rate the child’s appearance of anxiety on a Likert-type scale (similar to Koegel et al., 2004). Regarding the physiological component, although increased heart rate or lower RSA in certain contexts can point to the presence of anxiety in those contexts, physiological indices are usually not realistic for parents, teachers, or providers to measure. In lieu of a device to measure physiology, it is still possible to assess observable symptoms that indicate physiological arousal, such as visible muscle tension, rapid breathing, sweating, flushed face, or trembling (Cautela, 1977). Although signs of physiological arousal appear difficult for caretakers to recognize in individuals with ASD and ID, increasing informants’ awareness can help them to recognize signs such as tenseness and restlessness (Helverschou and Martinsen, 2011). Finally, regarding the behavioral component, to generate possible anxious behaviors, we initially created a comprehensive list of behavioral indicators of anxiety derived from a variety of sources, such as the Cues for Tension and Anxiety Survey Schedule (CTASS; Cautela, 1977), as well as from clinical observations of each child. Parents, teachers, staff members, or other caregivers identified behaviors the child typically displayed when anxious from this list of behavioral descriptors as well as identified other idiosyncratic behaviors that were not on the list but which the child displayed when anxious. Using this multimethod approach, we found that substantially more problem behaviors (as well as higher HR and lower RSA) occurred in High-Anxiety than in Low-Anxiety contexts (Moskowitz et al., 2013), suggesting that children with ASD may engage in problem behaviors to escape or reduce their anxiety. In sum, it is important to conduct a multimethod assessment incorporating multiple informants and direct observation when assessing for the presence of anxiety in children with ASD and ID because, although any behavior on its own does not necessarily indicate anxiety (e.g., someone may cry because he is feeling afraid, sad, angry, in pain, or ill), multiple sources of converging data may support that the behavior is a sign of anxiety.
Case Example
Jon was a 6-year-old boy of Jamaican descent, diagnosed with ASD and ID, who lived at home with his mother, father, and two younger brothers. Jon’s level of adaptive functioning was in the low range, and he communicated through the use of 1–2-word phrases, primarily using language for making requests, negating, and scripted speech. Jon met criteria for Specific Phobia of birthday parties (classified as “Other Type”) when his mother was interviewed using the ADIS-IV-C/P; she endorsed that he was more afraid of birthday parties than other kids his age, that he tries his hardest to avoid them or became extremely upset if he was forced to stay in the situation, and that his level of distress and avoidance interfered with both family and classroom activities. His fear of birthday parties had persisted for several years and there were never times when he was able to be around people singing “happy birthday!” in the presence of a birthday cake and remain in the situation calmly.
Affective/Cognitive/Contextual Component of Anxiety
To identify the context that was most likely to be associated with anxiety, Jon’s parents were given the Stress Survey Schedule (SSS); they wrote in “birthday parties” when asked if there were any other stressors that were not listed, rating it as a “5” (with 1 being no anxiety to 5 being severe anxiety). Interviews based on the Functional Assessment Interview (FAI; O’Neill et al., 1997) were conducted with Jon’s parents to identify, in greater detail, the specific events or situations that predicted the occurrence of anxiety. During the interview, Jon’s parents identified birthday parties as the context that was most likely to evoke anxiety for Jon. As soon as children in class or family members at home or people in community settings starting sitting happy birthday while presenting a cake with lit candles, Jon immediately bolted out of the room or cowered in a corner if he could not leave the room, while crying and exhibiting other anxious behaviors. This context was confirmed as the most likely to evoke anxiety by Jon’s teacher as well as by the clinician’s direct observations. To further support the affective component of anxiety (as Jon himself could not self-report cognitions or emotions), on a scale of 0 (no anxiety) to 3 (high anxiety), Jon’s appearance of anxiety was rated by blind observers as an average of 2.75 for each of the happy birthday contexts compared to a 0 for each of the low-anxiety contexts (Moskowitz et al., 2013).
Physiological Component of Anxiety
Jon wore the Alive heart rate monitor, a portable, wireless device with electrode transmitters that adhered to his chest and a receiver that was placed in a small fanny pack worn around his waist. We found that Jon exhibited significantly higher heart rate in the high-anxiety context (happy birthday) than in the low-anxiety context, and lower RSA in the high- than low-anxiety context (with the difference in RSA between the high- and low-anxiety context approaching significance) (Moskowitz et al., 2013).
Behavioral Component of Anxiety
Jon behaved differently in response to situations he simply disliked versus anxiety-provoking situations. He often verbally objected to things he did not like (such as if he were served pasta with tomato sauce instead of plain pasta, he would say “No!” and push it away) and he yelled or tantrummed when he was denied access to something such as his favorite video, but this was very different from the crying and fearful facial expression that occurred in response to birthday parties. The main way in which we discriminated between fear/anxiety versus dislike was by defining anxious behaviors and problem behaviors separately. Using a comprehensive list of behavioral descriptors compiled from existing measures (e.g., CTASS) as well as novel, idiosyncratic behaviors that were identified by Jon’s mother, father, and teacher, we identified several “anxious behaviors” for Jon: clinging onto his mother, crying/tearfulness, freezing (lack of movement except for respiration), cowering (e.g., turning into corner), anxious vocalizations (e.g., whimpering, moaning, or idiosyncratic throat noises), and a fearful/anxious facial expression consisting of eyes wide open or eyes rapidly darting back and forth, eyebrows sloping down in an inverted V-shape, and frowning (turning down of the mouth). Problem behaviors identified for Jon were yelling or screaming, elopement (running away; leaving the room or attempting to leave the room), pushing another person, and pulling his mother’s hair. Although Jon exhibited both anxious behavior and problem behavior in the context of happy birthday (given that his anxious behavior often escalated into problem behavior), he exhibited only problem behaviors such as yelling—not anxious behaviors—in contexts that were merely disliked versus anxiety-provoking. This made it easier to recognize that his anxious behaviors in response to “happy birthday” were distinctly different than his other problematic behaviors, and to conceptualize this as anxiety-based.
Conclusion
Despite longstanding clinical accounts of anxiety in children with ASD (e.g., Kanner, 1943), affective states such as fear and anxiety were rarely discussed or acknowledged in this population until relatively recently. Although fear and anxiety are reported to be more common in youth with ASD than in TD youth (e.g., Kuusikko et al., 2008) and those with other DDs (e.g., Brereton et al., 2006), it is worth remembering that almost all of the measures used to assess anxiety in ASD—with the exception of the ADIS/ASA, ACI-PL, and ASC-ASD—were designed for individuals without ASD. The prevalence of anxiety in ASD may be misrepresented given these measurement limitations, the varied and overlapping presentation of anxiety and ASD symptoms, and an over-reliance in anxiety assessment on self-reported or verbally-mediated symptoms (e.g., worries) that are at odds with deficits in communication and socio-emotional insight characteristic of ASD. There is a need for self-report and other-informant-report questionnaires and interviews designed to address these challenges, particularly in youth with ASD who have comorbid ID and/or are minimally verbal. We advocate for a multimethod approach using a variety of tools (e.g., interviews, questionnaires, direct observation) as well as multiple informants—and interpreting behaviors and physiological arousal within a given context—to assess anxiety in individuals with ASD. We must continue to further examine the best ways to discriminate between anxiety and other affective states, particularly in those with ID or who are minimally verbal, so that parents, teachers and clinicians do not misattribute anxiety to anger, frustration, sadness, boredom, or excitement, or vice versa.