Rachel Kent and Emily Simonoff, Institute of Psychiatry, Psychology & Neuroscience, London, United Kingdom
This chapter reviews the current knowledge about the prevalence of and risk factors associated with anxiety disorders in individuals with autism spectrum disorder (ASD). Prevalence rates for a diagnosis of an anxiety disorder are considered only from studies that use clinical diagnostic interviews. We report that anxiety disorders are common in individuals with ASD and overall prevalence rates for anxiety disorders as an aggregate group range between 42% and 79%. The prevalence rates for individual anxiety disorders are variable. Specific phobia, obsessive compulsive disorder, and social anxiety disorder are the most common but the rate and type of anxiety disorders may vary across age and ability level. Challenges in measuring anxiety in ASD and factors that may contribute to variation in overall rates and those for specific disorders are discussed. The objective of summarizing the research literature on anxiety in ASD to date is to set out clear guidelines for the future direction of research and also interpret the findings to highlight the key clinical implications.
Autism spectrum disorder; social anxiety; obsessive compulsive disorder; DSM-5; specific phobia
Autism spectrum disorder (ASD) is a developmental disorder characterized by qualitative impairments in social interaction and social communication such as difficulties in engaging in normal reciprocal conversation and difficulties understanding relationships in addition to a pattern of restricted interests and repetitive or stereotyped behavior such as an insistence on sameness. Recent estimates of the prevalence of ASD from the Centers for Disease Control and Prevention indicate that ASD occurs in as many as one in 68 people and is about 4.5 times more common among boys than girls (Christensen et al., 2016).
The most recent revision of the classification of psychiatric disorders, the Diagnostic and Statistical Manual, Fifth Edition (DSM-5; American Psychological Association, 2013) collapsed under the single term “autism spectrum disorder” a number of diagnostic entities that had been differentiated in the previous classifications (the Diagnostic and Statistical Manual, Fourth Edition (DSM-IV) and the International Classification of Disease, 10th Edition (ICD-10)). These included: autistic disorder, Asperger’s disorder (also known as Asperger syndrome), pervasive developmental disorder not otherwise specified (PDD-NOS), and atypical autism. Research involving any of these categorical disorders will be covered in this chapter and referred to as ASD unless results specifically address differences between previous diagnostic subgroups.
Individuals with ASD vary widely in their presentation of autistic symptoms, cognitive and language abilities, and the additional co-occurring disorders they experience. Other neurodevelopmental disorders including intellectual disability (e.g., Dykens and Lense, 2011), dyspraxia (e.g., Dziuk et al., 2007), and language impairment (e.g., Loucas et al., 2008) are increased in ASD, as is epilepsy. People with ASD also have elevated rates of a range of psychiatric disorders compared to the general population (de Bruin et al., 2007; Mukaddes and Fateh, 2010; Simonoff et al., 2008). One of these frequently co-occurring psychiatric disorders with ASD is anxiety.
Anxiety disorders embrace a constellation of conditions marked by subjective experiences of worry or fear. Such symptoms are part of normal experience and should only be considered indicative of an anxiety disorder when they are sufficiently frequent or severe to cause sustained and high levels of distress and/or impairment in everyday functioning. Classification systems subsume under anxiety a number of different disorders and, in general, the same criteria are used to classify anxiety disorders among people with ASD as in the general population. In this chapter, we will largely refer to disorders as described under DSM-IV, as much of the relevant research uses this system. This review summarizes the prevalence of anxiety disorders in aggregate and provides an overview of the prevalence rates for individual disorders.
The main changes between DSM-IV and DSM-5 are that obsessive compulsive disorder (OCD) and posttraumatic stress disorder (PTSD) are now included in domains other than anxiety. In addition, for agoraphobia, specific phobia, and social anxiety disorder (used interchangeably with social phobia) it is no longer necessary for the individual to recognize their anxiety as being excessive or unreasonable. Agoraphobia can also now be diagnosed as a distinct disorder without panic disorder. Finally, the diagnostic criteria for separation anxiety disorder no longer require an onset before 18 years of age.
Children and adolescents. A review of general population prevalence studies found rates in children and adolescents to vary widely depending on methodological differences, including the assessment period; prevalence rates for any anxiety disorder ranged from 2.2–8.6% in studies measuring the prevalence over three month periods, to 5.5–7.7% over a six-month period, 8.6–20.9% over a 12-month period, and 8.3–27% for lifetime prevalence (Costello et al., 2005). A systematic review (1980–2004) also found lifetime prevalence (16.6%) of any anxiety disorder to be higher than the 12-month prevalence (10.6%); in this review social anxiety disorder (4.5%), specific phobia (3%), and generalized anxiety disorder (GAD; 2.6%) were the most common anxiety disorders and OCD (0.54%), panic disorder (0.99%), and agoraphobia (1.6%) were less common (Somers et al., 2006).
Adults. In a U.S. population-representative study of adults the 12-month prevalence rate of DSM-IV anxiety disorders was 18.1% and the lifetime prevalence was 28.8% (Kessler et al., 2005; Kessler et al., 2005). In contrast, a population study in Europe found a 12-month prevalence of 6.4% and lifetime prevalence of 13.6% (Alonso et al., 2004). For the 12-month prevalence, across both samples of adults, specific phobia (8.7%; 3.5%) was the most prevalent, followed by social anxiety disorder (6.8%; 1.2%). GAD (3.1%; 1.0%) and panic disorder (2.7%) were also common whereas separation anxiety (0.9%), OCD (1.0%), and agoraphobia (0.8%; 0.4%) were the least frequent (Alonso et al., 2004; Kessler et al., 2005). A general finding, for both youth and adult populations, is the higher rate of anxiety disorders in females with a ratio of around 1.5:1 (Alonso et al., 2004; Costello et al., 2005).
Prevalence in individuals with intellectual disability (ID). In considering the rates of anxiety disorders in people with ASD, it is helpful to contrast these with rates reported for those with ID. About half the ASD population also have ID (Charman et al., 2011). ID may both constitute a risk factor for anxiety disorders and also a barrier to its detection as people with ID may have more difficulty communicating the subjective experiences and cognitions that underpin diagnosis.
Children and adolescents. In a UK national sample, Emerson and Hatton (2007) defined ID by parent- and teacher-reported significant learning problems and found a point prevalence of any impairing anxiety disorder was 11.4% compared to 3.2% in those without ID. Consistent with this, Dekker and Koot (2003) assessed Dutch children attending schools for individuals with ID and reported that for 12-month prevalence rates 21.9% of the children met the criteria for an anxiety disorder with 10.5% having a significant level of impairment. Across studies, specific phobia (2.0%; 6.8%), separation anxiety (2.7%; 1.9%), GAD (1.6%), and social anxiety disorder (1.9%) were the most common (Dekker and Koot, 2003; Emerson and Hatton, 2007).
Adults. In a population sample of adults with ID, the point prevalence of any DSM-IV anxiety disorder was 2.4% (Reid et al., 2011); GAD was the most common (1.3%), and other disorders had very low rates (agoraphobia, 0.2%; panic disorder, 0.2%; social anxiety disorder, 0.1%). In a randomly selected sample of adults with moderate to profound ID the DSM-IV prevalence of any anxiety disorder was similar at 3.3%; 2.5% met criteria for specific phobia, and 0.8% for agoraphobia without panic (Bailey, 2007).
The co-occurrence of anxiety symptoms in ASD has been noted since the first descriptions of the condition by Leo Kanner (1943) and Hans Asperger (see Frith, 1991), both of whom observed that the children they described were fearful of both common and unusual or novel situations and objects as well as presenting with high levels of generalized worry, fear of social encounters, and obsessionality. Despite these early observations, there still remain a number of challenges, both conceptual and methodological, to the accurate diagnosis of anxiety disorders in people with ASD.
Conceptual considerations. First, there is a lack of consensus on what constitutes gold standard measures of anxiety in ASD and whether these measures should take account possible differences in the manifestation of symptoms in the ASD population. While there is general agreement that assessments need to distinguish the superficial similarities of core ASD symptoms where they overlap, e.g., social avoidance versus social anxiety, repetitive, stereotyped language versus reassurance-seeking, there is presently no agreement whether anxiety symptom definitions and/or criteria should be modified in ASD. Second, many people with ASD experience difficulties with emotional literacy and may find it difficult to express emotions. In assessing internal experiences, this leads to using informants, rather than self-reports, who are often inferring internal states based on contextual cues and other suppositions. Third, among those with more significant levels of ID, it is uncertain how to conceptualize the cognitive components of anxiety, including worry and anticipatory fear. Currently, as there are no reliable and valid biomarkers for anxiety, these issues remain unresolved.
Methodological considerations. A number of methodological considerations may influence the findings, and the most important ones are considered below.
Implementing diagnostic criteria and clinical thresholds. Diagnostic classification systems are designed primarily to assist clinicians in determining which condition(s) apply to patients presenting with psychiatric disorder. While these systems indicate which symptoms are subsumed under different disorders, they give much less guidance with respect to the operationalization into research criteria and how these should be applied in epidemiological studies. Diagnostic assessments ordinarily require the use of a psychiatric interview but these vary in mode administration (structured/semi-structured), the definitions of individual symptoms, and the application of diagnostic algorithms. Diagnostic assessments vary in whether they include a measure of functional impairment, and whether this is applied implicitly or later, more explicitly, in the diagnostic process. These methodological problems are pervasive in psychiatry research and not limited to studies in people with ASD but are likely an important contributor to variability across studies.
Sample ascertainment. The most accurate prevalence estimates will be obtained from studies that use epidemiological methods to derive samples that either include the entire population of interest (target population) or are randomly selected from this population. These methods of complete ascertainment require that the target population can be accurately identified and, for selected groups, this can be a time-consuming aspect of research. Many studies in ASD use other methods to identify the population of interest, leading to incomplete ascertainment. These may involve health registers, clinic or special school populations, and volunteer samples. All have the potential to deliver biased findings and it is often difficult to judge the degree of inaccuracy. As a general principle, ascertainment from more selected samples, particularly where the selection is associated with the area of interest, is more likely to be biased.
Control groups. Control or comparison groups refer to an additional sample of participants that may help to contextualize the findings reported for the target population. However the selection of control groups raises questions about whether to match or control for certain features when drawing comparisons. In ASD, such features might include the high proportion of affected males and the relatively lower IQ. There is no clear scientific consensus on how best to select control groups or account for sample differences in analysis, although “matching” eliminates the possibility of exploring in analysis the effect of that characteristic. Comparison groups should be assessed in a manner that is similar, if not identical, to the target population. Sometimes, comparisons are made with published normative data; in such cases, the applicability of these norms to the population from which ASD participants were drawn should be carefully considered.
Informants. In assessing psychiatric status among adults, self-reports are the most common source of information. Among youth and people with ASD and/or ID, however, a variety of different informants may be used, including self, parent/carer, and teacher. The agreement among these different informants is limited, however, with a meta-analysis reporting only a modest correlation between self-report and other informants (0.22 and 0.28) on questionnaire measures of behavioral and emotional problems (Achenbach et al., 1987; Rey et al., 1992).
Comer and Kendall (2004) found parent–child agreement for symptoms was better than agreement at the diagnostic level for three anxiety disorders measured using a diagnostic interview. Agreement between parent and child has also been found to be higher on reports of conduct or behavioral problems (0.51) than anxiety, fear, and obsessions (0.17; Edelbrock et al., 1986).
Inconsistency across informants may reflect actual variation in psychopathology across situations, failure of an informant to detect or endorse psychopathology, or other factors that affect the ability to respond accurately (Grills and Ollendick, 2003). Younger children are known to have comprehension difficulties with questions about psychopathology (Breton et al., 1995) and these may apply to older children and adults with ASD. The gold standard for research-level diagnoses is to use multiple informant data but it is not always possible in ASD as youth populations may not be able to complete a psychiatric interview.
Period of assessment. Prevalence estimates refer to the presence of a disorder during a particular time period, typically either “current” or lifetime. The duration of the current period may vary, usually between the last month and the last year, but this variation can have a considerable impact on prevalence rates especially for disorders that remit and recur.
Methods of assessment. Structured or semi-structured interviews, which vary in the degree of structure applied and extent to which the interviewer probes and makes judgments, are the gold standard for psychiatric diagnoses. Interviews allow clinicians or researchers to request additional information when symptoms are unclear. While questionnaire measures have been used to report psychopathology, even when validated cut-offs are available, these are considered inferior to interview measures and ordinarily would be used for screening rather than diagnostic purposes, especially when applied at the individual clinical level. In population studies, they have the advantage of ease of administration and may be useful in providing indicative prevalence rates, although these should always be viewed with caution.
In order to review the most reliable and valid data on the prevalence of anxiety disorders in ASD, only studies using diagnostic interviews that apply specific DSM or ICD criteria will be used. The studies’ methodological characteristics are summarized in Table 2.1 and the prevalence rates of both aggregate and specific anxiety disorders are shown in Table 2.2. In total 17 studies used clinical interviews to assess anxiety disorders in ASD. The majority (N=12) focused on children and adolescents and only four studies examined prevalence in adults. Maddox and White (2015) looked only at social anxiety in a sample of adolescents and adults.
Table 2.1
Table presenting the sample characteristics and measurement instruments used for studies assessing the prevalence of anxiety in ASD using diagnostic interviews
Study | # ASD | Control group | Age | Ability levela | Gender; % males | Diagnosis of ASD | Sampling technique (ASD) | Anxiety measure |
Children and adolescents | ||||||||
Population studies | ||||||||
Simonoff et al. (2008) | 112 | – | M=11.5 | M=72.7 | 88% | Clinical consensus diagnosis (ICD-10 criteria) informed by ADOS & ADI-R | Population cohort (56,946) of children with PDD diagnosis or Statement of Special Educational Needs | CAPA; Parent report |
R=10–13.9 | SD=21.6 | |||||||
R=19–124 | ||||||||
Community samples | ||||||||
Mattila et al. (2010) | 50 | – | M=12.7 | FSIQ>75 | 76% | ADI-R, ADOS & ASSQ | Combination of community sample (n=18) and a clinic study (n=32) | K-SADS; Parent and self-report |
SD=1.5 | ||||||||
R=9.8–16.3 | ||||||||
Leyfer et al. (2006) | 109 | – | M=9.2 | M=82.6 | 94% | ADI-R, ADOS & DSM-IV-TR criteria | Two previous samples. Not clinically referred. | ACI; Parent report |
SD=2.7 | SD=23.4 | |||||||
R=5.1–17 | R=42–141 | |||||||
Salazar et al. (2015) | 101 | – | M=6.7 | M=66.4 | 56% | ADI-R, 3di, DISCO, ADOS | Population sample living in either of two London boroughs with an ASD diagnosis | PAPA; Parent report |
SD=1.1 | SD=28 | |||||||
R=4.5–9.3 | R=19–120 | |||||||
Clinic samples | ||||||||
de Bruin et al. (2007) | 94 | – | M=8.5 | M=91.2 | 88% | PDD-NOS research criteria. ADOS on 93.6% of the children. | Consecutive referrals (2 years); outpatients of child and adolescent psychiatry department. | DISC-IV; Parent report |
SD=1.9 | SD=17.4 | |||||||
R=6–12 | R=55–120 | |||||||
Gjevik et al. (2011) | 71 | – | M=11.8 | NVIQ: | 82% | Research ADI-R | Attending selected special education needs school | K-SADS; parent report |
SD=3.3 | M=65.2 | |||||||
R=6.2–17.9 | SD=29.6 | |||||||
R=30–129 | ||||||||
Green et al. (2000) | 20 | 20 CD | AS: | AS: | 100% | ICD-10 clinical criteria for AS | Clinical referral | Modified Isle of Wight Semi-structured Informant and Child Interviews. Parent- and self-report |
M=13.8 | M=92.2 | |||||||
R=11–19 | SD=17.7 | |||||||
R=71–141 | ||||||||
CD: | CD: | |||||||
M=14.5 | M=91.2 | |||||||
R=11–18 | SD=9.1 | |||||||
R=74–107 | ||||||||
Joshi et al. (2010) | 217 | 217 | ASD: | – | ASD: 87% | DSM-III checklist. AD: n=25; PDD-NOS: n=192 | Consecutive referrals to pediatric clinic (N=2323); 217 meeting ASD criteria and 217 not meeting criteria. | K-SADS; parent report |
M=9.7 | Non-ASD: 72% | |||||||
SD=3.6 | ||||||||
Non-ASD: | ||||||||
M=10.9 | ||||||||
SD=3.5 | ||||||||
Mukaddes et al. (2010) | 60 | – | HFA: | HFA: | – | DSM-IV criteria; 30 children and adolescents with diagnosis of HFA and 30 with diagnosis of AS. | Selected from 454 children and adolescents referred to ASD clinic | K-SADS; unclear informant |
M=10.3 | M=90.5 | |||||||
R=6.2–14.4 | R=70–127 | |||||||
AS: | AS: | |||||||
M=11.0 | M=106.5 | |||||||
R=7.0–15.5 | R=82–138 | |||||||
Mukaddes and Fateh (2010) | 37 | – | M=10.9 | M=116 | 87% | AS according to DSM-IV criteria | Recruited from private clinic for general psychiatry | K-SADS; unclear informant |
R=6–20 | R=90–139 | |||||||
Muris et al. (1998) | 44 | – | M=9.7 | M=79.5 | – | Clinical diagnosis of AD (n=15) or PDD-NOS (n=29) | Clinical sample | DISC (version 2.3); parent report |
SD=4.8 | SD=14.0 | |||||||
R=2–18 | R=59–116 | |||||||
Witwer and Lecavalier (2010) | 61 | – | M=11.2 | M=68.4 | 82% | ADI-R: 16 AS, 17 AD, 26 PDD-NOS. | Advertisements to those receiving services | P-ChIPS; parent report |
SD=3.8 | SD=23.3 | |||||||
R=6–17 | R=42–150 | |||||||
Subclinical symptoms | ||||||||
Caamaño et al. (2013) | 25 | 25 | ASD: | ASD: | ASD: 96% | Developmental history and medical reports were scored according to DSM-IV or Gillberg’s criteria. The ADOS was used if the two sets of criteria did not match (35%). | Volunteer sample | K-SADS-PL: Parent- and self-report |
M=12.8 | M=97.9 | Non-ASD: 92% | ||||||
SD=2.9 | SD=27.6 | |||||||
Non-ASD: | Non-ASD: | |||||||
M=12.5 | M=114.4 | |||||||
SD=2.9 | SD=16.1 | |||||||
Adults | ||||||||
Hofvander et al. (2009) | 122 | – | Med=29 | – | 67% | DSM-IV AD criteria and the Gillberg & Gillberg Asperger criteria. Used ASDI: 5 AD; 67 AS; 50 PDD-NOS | Referrals to two ASD clinics | SCID-I; unclear informant |
R=16–60 | ||||||||
Joshi et al. (2013) | 63 | 63 | ASD: | ASD: | ASD: 65% | Clinical structured diagnostic interview: 41 AD; 16 AS; 6 PDD-NOS | Clinical referrals to ASD clinic (ASD group) or general psychopharmacology program (control group; age and sex matched) | SCID-I; self-report |
M=29.2 | M=104.4 | Non-ASD: 65% | ||||||
R=18–63 | R=55–136 | |||||||
Non-ASD: | Non-ASD: | |||||||
M=29.3 | M=106.8 | |||||||
R=18–65 | R=77.5–133 | |||||||
Lever and Geurts (2016) | 138 | 170 | ASD: M=46.5 Non-ASD: M=45.9 |
ASD: M=113.8 Non-ASD: M=113.3 |
ASD: 70% Non-ASD: 57% |
Clinical diagnoses: 21 Autistic Disorder; 69 AS; 43 PDD-NOS; 5 ASD. ADOS module 4 | ASD group: Advertisements to those accessing clinical services. Comparison group advertisements at university and on social media | MINIPlus; self-report |
Lugnegård, Hallerbäck, and Gillberg (2011) | 54 | – | M=6 | M=102 | 48% | AS diagnosis confirmed using the DISCO and clinical judgment | Current or previous patients of ASD clinics | SCID-I; self-report |
SD=3.9 | SD=12 | |||||||
Social anxiety disorder (SAD) only | ||||||||
Maddox and White (2015) | 28 | 26 SAD | ASD: | ASD: | ASD: 54% | ASD: ADOS-2 and clinical interview | Volunteer sample | ADIS-IV-C/P; self-report |
25 TD | M=23.9 | M=107 | SAD: 50% | SAD: ADIS-IV-C/P criteria for SAD | ||||
SD=6.9 | SD=16.5 | TD: 48% | ||||||
R=16–42 | R=80–141 | |||||||
SAD: | SAD: | |||||||
M=25.9 | M=109 | |||||||
SD=7.1 | SD=10.5 | |||||||
R=16–42 | R=88–127 | |||||||
TD: | TD: | |||||||
M=24.8 | M=114 | |||||||
SD=7.3 | SD=10.8 | |||||||
R=17–44 | R=91–133 |
aFull Scale IQ unless otherwise stated. NVIQ=Non-verbal IQ, M=mean, Med=median, SD=standard deviation, R=range, AD=autistic disorder, PDD-NOS=pervasive developmental disorder not otherwise specified, AS=Asperger syndrome, CD=conduct disorder, HFA=high-functioning autism. CAPA: Child and Adolescent Psychiatric Assessment (Angold and Costello, 2000); K-SADS: Kiddie – Schedule for Affective Disorders and Schizophrenia for School-Age Children (Puig-Antich and Chambers, 1978); ACI: Autism Comorbidity Interview (Leyfer et al., 2006); PAPA: Preschool Age Psychiatric Assessment (Egger and Angold, 2004); DISC: Diagnostic Interview Schedule for Children (Shaffer et al., 1996); P-ChIPS: Children’s Interview for Psychiatric Syndromes-Parent Version (Weller et al., 1999); SCID: Structured Clinical Interview for DSM-IV Axis I Disorders (First et al., 2002); MINIPlus: the mini International neuropsychiatric interview (Sheehan et al., 1998); ADIS-IV-C/P: anxiety disorder interview schedule for DSM-IV (Silverman and Albano, 1996); ADI-R: Autism Diagnostic Interview – Revised (Rutter et al., 2003); ADOS: Autism Diagnostic Observation Schedule (Lord et al., 2012); ASSQ: Autism Spectrum Screening Questionnaire (Ehlers et al., 1999); 3di: developmental, dimensional and diagnostic interview (Skuse, 2013); DISCO: diagnostic interview for social and communication disorders (Wing et al., 2002); ASDI: the Asperger syndrome diagnostic interview (Gillberg et al., 2001).
Table 2.2
Table showing the prevalence rates (%) for DSM-IV specific anxiety disorders in ASD across the reviewed studies using diagnostic interviews
Prevalence period | GAD | Social anxiety disorder | Specific phobia | Panic disorder | Agoraphobia | Separation anxiety | OCD | Any anxiety disorder | |
Children & adolescents | |||||||||
Population samples | |||||||||
Simonoff et al. (2008) | 3 months | 13.4 | 29.2 | 8.5 | 10.1 | 7.9 | 0.5 | 8.2 | 41.9 |
Community samples | |||||||||
Leyfer et al. (2006) | Lifetime | 2.4 | 7.5 | 44.3 | 0 | – | 11.9 | 37.2 | – |
Mattila et al. (2010) Combined sample | Current/lifetime | – | 4/6 | 28/34 | 2/2 | 2/2 | 2/8 | 22/28 | 42/56 |
Community sample | Current/lifetime | – | 6/6 | 33/39 | 6/6 | 0/0 | 6/17 | 11/11 | 39/50 |
Clinic sample | Current/lifetime | – | 2.5/5 | 28/33 | 2.5/2.5 | 2.5/2.5 | 2.5/5 | 25/33 | 45/58 |
Salazar et al. (2015) | 3 months | 66.5 | 15.1 | 52.7 | 3.1 | 18.0 | 18.6 | – | 78.9 |
Clinic samples | |||||||||
de Bruin et al. (2007) | 1 year | 5.3 | 11.7 | 38.3 | 1.1 | 6.4 | 8.5 | 6.4 | 55.3 |
Green et al. (2000) | 3 months | 35 | – | 10 | – | – | – | 25 | – |
Mukaddes et al. (2010) Combined | Lifetime | 10 | 13.3 | 53.4 | 1.7 | – | 13.4 | 37.2 | 78.3 |
AS | Lifetime | 16.7 | 13.3 | 46.7 | 0 | – | 10 | 36.7 | 73.3 |
HFA | Lifetime | 3.3 | 13.3 | 60 | 3.3 | – | 16.7 | 37.7 | 83.3 |
Mukaddes and Fateh (2010) Combined | Lifetime | 5.4 | 5.4 | 13 | 5.4 | – | 2.7 | 32 | 59.5 |
Children | Lifetime | 0 | 0 | 21 | 0 | – | 4 | 17 | 43 |
Adolescents | Lifetime | 14 | 14 | 0 | 14 | – | 0 | 57 | 85 |
Muris et al. (1998) | 6 months | 22.7 | 20.5 | 63.6 | 9.1 | 45.5 | 27.3 | 11.4 | 84.1 |
Witwer and Lecavalier (2010) | 1 month | 24.6 | 16.4 | 67.2 | – | – | 14.8 | 4.9 | – |
IQ>70 | 1 month | 50 | 22.7 | 68.2 | – | – | 13.6 | 13.6 | – |
IQ<70 | 1 month | 8.3 | 13.9 | 67.2 | – | – | 13.9 | 0 | – |
Gjevik et al. (2011) | Lifetime | 0 | 7 | 31 | – | – | 0 | 10 | 42 |
Joshi et al. (2010) | Lifetime | 35 | 28 | 37 | 6 | 35 | 37 | 25 | 61a |
Subclinical symptoms | |||||||||
Caamaño et al. (2013) | Subclinical | 32 | 40 | 36~ | 20 | 36~ | 28 | 48 | – |
Adults | |||||||||
Lugnegård et al. (2011) | Lifetime | 22 | 22 | – | 13 | 15 | – | 7 | 56 |
Women | Lifetime | 25 | 18 | – | 18 | 14 | – | 4 | 57 |
Men | Lifetime | 19 | 27 | – | 8 | 15 | – | 12 | 54 |
Hofvander et al. (2009) | Lifetime | 15 | 13 | 6 | 11 | 11 | – | 24 | 50 |
Lever and Geurts (2016) | Lifetime | 15.9 | 15.2 | 11.6 | 15.2 | 21 | – | 21.7 | 53.6 |
Young (19–38) | Lifetime | 17.4 | 21.7 | 10.9 | 23.9 | 21.7 | – | 28.3 | 65.2 |
Middle (39–54) | Lifetime | 19.1 | 21.3 | 14.9 | 12.8 | 19.1 | – | 21.3 | 53.2 |
Older (55–79) | Lifetime | 11.1 | 2.2 | 8.9 | 8.9 | 22.2 | – | 15.6 | 42.2 |
Joshi et al. (2013) | Current | 29 | 40 | 18 | 3 | 24 | 3 | 16 | 38a |
Lifetime | 35 | 56 | 32 | 15 | 35 | 21 | 24 | 59a | |
Social anxiety disorder only | |||||||||
Maddox and White (2015)– | – | 50 | – | – | – | – | – | – |
a2 or more anxiety disorders; ~agoraphobia and specific phobia combined.
Prevalence of any anxiety disorder. In a population cohort of children with PDD or a statement of special educational needs, Simonoff et al. (2008) found a 3-month prevalence of 42% for any anxiety disorder. In contrast, the community sample used by Salazar et al. (2015) reported a 3-month prevalence of 78.9%; however, the interview used in this study is known to produce relatively high rates of disorders (see Egger and Angold, 2006). Mattila et al. (2010) directly compared recruitment strategies and period prevalence; using a sample of children with Asperger Syndrome (AS) or high-functioning autism (HFA), they found rates of 45% and 58%, respectively for current and lifetime prevalence among those who were clinically referred compared to rates of 39% and 50% among those recruited from the community.
There is consensus among the 12 studies that the proportion of children and adolescents with ASD and one or more anxiety disorders is high. The majority of studies report a prevalence rate of around 50%, although some, such as Mukaddes et al. (2010) and Salazar et al. (2015), report levels as high as 78% and 79%, respectively. The variability in rates is likely due to the methodological differences highlighted. Overall prevalence estimates range from 42% to 79%, which is considerably higher than rates reported in the general or ID population of children and adolescents (e.g., Costello et al., 2005; Dekker and Koot, 2003).
Tables 2.1 and 2.2 categorize the reviewed studies into child/adolescent and adult samples, although some overlap does occur in the younger samples extending up to 20 years old (Green et al., 2000; Mukaddes and Fateh, 2010) whereas the adult sample used by Hofvander et al. (2009) included participants from 16 years old. The lifetime prevalence rates of any anxiety disorder among adults with ASD who had been recruited from clinical samples ranged from 50% to 59%. These rates are substantially higher than the lifetime prevalence in typically developing adults (28.8%; Kessler et al., 2005). Lever and Geurts (2016) collected questionnaire and interview data on different age subsets of adults with ASD. The adults were split into three age groups: young adults (19–38); middle-aged adults (39–54), and older adults (55–79). In the ASD sample as a whole the prevalence of any anxiety disorder was 54% which was significantly higher than 14.7% of the age and IQ matched comparison group. The young and middle-aged adults had the most anxiety (65.2% and 53.2%, respectively), which is in line with the other studies in adults with ASD, whereas the rate among older adults was lower at 42.2%.
Specific anxiety disorders. Specific phobia was the most prevalent diagnosis across 8 of the 12 reviewed studies on children and adolescents with ASD (31–67%). Some studies report lower rates, such as 8.5% in Simonoff et al. (2008); however, in this study evidence of functional impairment was additionally required to receive a diagnosis. In the four studies of adults with ASD specific phobia was still common but not consistently as high; prevalence rates ranged from 6% to 32%. Tentative findings indicate the prevalence of specific phobias may decrease as age increases: Mukaddes and Fateh (2010) found that while the majority of children in their sample met criteria none of the adolescents did; Salazar et al. (2015) also found a trend that specific phobia decreases from 62% in those 7.5 years old and under to 44.6% in those older than 7.5 years.
The rates of OCD across the reviewed studies are variable (4.9–37.2% in children and adolescents and 7–24% in adults) but consistently reported at moderate rates. The highest prevalence was reported by Leyfer et al. (2006) who applied modified diagnostic criteria such that a diagnosis of OCD could be made on observable signs and symptoms only rather than caregivers being asked to infer children’s subjective mental experiences. They argued without this adjustment only a minority of children in their sample would have met criteria. The difference between repetitive behaviors in OCD and ASD is typically considered a qualitative one; in ASD, repetitive behaviors are a source of pleasure whereas in OCD they are associated with distress and anxiety. However, eliciting the associated feelings is often extremely difficult. The anxiety related to repetitive behaviors could be overlooked in ASD and this may explain the variable rates across studies.
The prevalence of social anxiety disorder across children and adolescents was also common, although rates varied from 4% to 29.2%. Interpretation of the symptom descriptions, such as social avoidance being coded as either core to ASD or as part of distinct social anxiety, may again help to explain the discrepancies in diagnostic rates. The description adopted by Gjevik et al. (2011) required observable and expressed symptoms such that social avoidance would not be sufficient in and of itself to meet diagnostic criteria. Such descriptions could pose a particular problem in ASD because of reduced emotional literacy and poorer communication. Leyfer et al. (2006) argue that individuals with ASD may try to avoid the non-social aspects of interactions such as noise and that social anxiety disorder per se is less prevalent. The lower prevalence rate reported by Leyfer et al. (2006) may, therefore, reflect the modifications to the criteria applied, which specifically distinguish between fear and avoidance of either the social or the non-social aspects of social situations.
Generalized anxiety disorder (GAD) was found to be prevalent across the four studies of adults (6–32%) and although the rates were more variable in children and adolescents, GAD was also a common anxiety disorder (0–66.5%) in most studies. In contrast to the high rates (e.g., Joshi et al., 2010; Green et al., 2000; Salazer et al., 2015), none of the parents in Gjevik et al. (2011) reported subjective anxiety symptoms required for a diagnosis of GAD. One explanation may be the substantial number of lower cognitive ability participants in the sample. This is supported by the findings of Witwer and Lecavalier (2010) who found GAD was significantly less prevalent in the low-functioning group (8.3%; IQ<70) than the high-functioning ASD group (50%; IQ>70). Panic disorder was the least common anxiety disorder in children and adolescents with rates ranging from 0% to 10.1%, but appears to be more prevalent in adults (11–15%). Overall, the findings of the current review are in line with a meta-analysis of children and adolescents with ASD conducted by van Steensel et al. (2011), which included both standardized interviews and questionnaires (N=2,121). The meta-analysis found specific phobia to be the most common anxiety disorder in ASD (30%), then OCD (17%), social anxiety disorder and agoraphobia (17%) followed by GAD (15%), separation anxiety (9%), and panic disorder (2%).
The diagnostic interview studies reviewed above are limited in their inclusion of comparison groups and range of age or ability level within any one sample. The addition of questionnaire studies in the following section is used to be able to draw conclusions about risk factors and correlates of anxiety in ASD.
Anxiety symptomatology in ASD and comparison groups. Of the reviewed diagnostic interview studies that did include comparison groups, individuals with ASD were found to be significantly more likely to meet diagnostic criteria for anxiety: children with AS had significantly more anxiety symptoms than children with conduct disorder (Green et al., 2000) and adults with ASD had significantly higher rates of multiple anxiety disorders (two or more), agoraphobia, OCD, and social anxiety disorder than non-ASD referrals to a general psychiatric clinic (Joshi et al., 2013).
A substantial proportion of the studies using questionnaires to measure anxiety symptoms, rather than diagnostic rates, compare groups of individuals with ASD to groups of typically developing individuals and those with other clinical conditions. These studies have found individuals with ASD to score significantly higher than individuals with typical development (Bellini, 2004; Kim et al., 2000; Lopata et al., 2010; Thede and Coolidge, 2007; Weisbrot et al., 2005), ID without ASD (e.g., Brereton et al., 2006; Gillott and Standen, 2007), Down’s syndrome (Evans et al., 2005), Williams syndrome (Rodgers et al., 2012), and specific language impairment (Gillott et al., 2001). Individuals with ASD have also been found to have comparable levels of anxiety to non-ASD individuals with a clinical anxiety disorder (Farrugia and Hudson, 2006) and parents of children with ASD reported higher total levels of anxiety, OCD symptoms, and fear of injury anxiety than parents of clinically anxious children (Russell and Sofronoff, 2005).
Anxiety in ASD subgroups. The main ASD diagnostic subgroup comparisons have been conducted between individuals with a clinical diagnosis of AS and HFA. Using a clinical diagnostic interview no significant differences in anxiety diagnoses were found between groups of adolescents with AS or HFA, who were diagnosed according to the DSM-IV criteria (Mukaddes et al., 2010). In addition, no significant differences were found on a questionnaire measure of anxiety symptoms in 9- to 14-year-old children with AS or HFA (Kim et al., 2000) who were categorized according to the presence (AS) or absence (HFA) of spontaneous phrase speech by 36 months of age and presence (HFA) or absence (AS) of persistent deviant language development. In contrast, Thede and Coolidge (2007) found children with AS (no parent reported language delays) had significantly higher scores on an anxiety questionnaire than the HFA group and when Tonge et al. (1999) controlled the effects of age and cognitive level across children and adolescents with HFA and AS, diagnosed according to DSM-IV criteria, they found the AS group to be significantly more anxious.
When examining all ASD subgroups Weisbrot et al. (2005) found school age children with autistic disorder (AD) were reported by parents and teachers to have the lowest levels of anxiety, whereas children with a diagnosis of AS or PDD-NOS were rated to have significantly more obsessional symptoms than children with AD and those with AS had more severe generalized anxiety than both other groups. It is important to note, however, that the assignment of diagnostic labels under the DSM-IV was not consistent across professionals or clinical sites (e.g., Lord et al., 2012). It may, therefore, be more beneficial to look for additional explanatory factors such as IQ level rather than diagnostic subsets. For example, Muris et al. (1998) found some anxiety disorders were more frequently seen in children with PDD-NOS than in those with AD (e.g., simple phobia, separation anxiety disorder). However, the children who were labeled as PDD-NOS were both older and had significantly higher IQ scores than the AD group.
Anxiety and ASD severity. It has been suggested that increased ASD symptoms may increase individual’s risk of developing anxiety (Wood and Gadow, 2010) but the evidence has been inconsistent. Whereas Mayes et al. (2011) found ASD severity as rated by parents was the best predictor of level of anxiety in children, Eussen et al. (2013) found lower ASD symptom severity was related to higher anxiety symptoms. Furthermore, Rieske et al. (2012) investigated the role of social impairments in predicting co-occurring anxiety and found the addition of this information to a regression model only explained an additional 1.3% of the variance (the full model explained 56%).
One methodological factor may have a large impact on interpreting the results regarding ASD severity. Most studies have looked at an overall rating of ASD severity when it is possible that the social and communication symptoms and the restricted and repetitive behavior symptoms may have differential effects on anxiety. Magiati et al. (2016) found that in contrast to the social and communication score, the repetitive speech and stereotyped behavior score was found to predict overall anxiety scores. Hallett et al. (2013) found children with higher restricted and repetitive behavior scores had more panic and OCD symptoms whereas higher social and communication symptoms were associated with higher separation anxiety symptoms and lower social anxiety symptoms. Such studies are not designed, however, to disentangle different manifestations of the same underlying phenomenology, e.g., repetitive speech from seeking reassurance, lack of social interest from social avoidance, and more sensitive measurement strategies are needed to further explore these findings.
Anxiety across age. The studies that have attempted to explore anxiety in ASD across ages have been inconsistent, with some studies finding no association between chronological age and anxiety (Farrugia and Hudson, 2006; Hallett et al., 2013; Strang et al., 2012; Sukhodolsky et al., 2008) and other studies suggesting that anxiety severity increases with age across children and adolescents (e.g., Green et al., 2012; Kuusikko et al., 2008; Vasa et al., 2013). Davis et al. (2011) used a cross-sectional approach to compare anxiety across ages in 131 toddlers (aged 17–36 months), children (aged 3–16 years), and adults (aged 20–65 years) with ASD. They found anxiety levels increased from toddlerhood to childhood, decreased from childhood to young adulthood, and increased again from young adulthood into older adulthood. However, different anxiety measures were used in the different age groups. Lever and Geurts (2016) also used a cross-sectional design in adults and found a slight decrease in older adults (aged 55–79) but these adults were older again than those in Davis et al. (2011).
Some research has indicated that the trajectory of anxiety across ages may depend on the specific anxiety disorder being measured. The findings from a meta-analysis found that OCD and separation anxiety were more common in younger children with ASD and that GAD and overall levels of anxiety were higher in older children and adolescents with ASD (van Steensel et al., 2011). More recent findings may support the idea of anxiety specific age-effects as Magiati et al. (2016) found age was positively (although weakly) associated only with social anxiety and OCD. Lever and Geurts (2016) found specific phobia to significantly decrease (22% to 2%) in a sample of young, middle-aged, and older adults. Using a diagnostic interview, Salazar et al. (2015) found agoraphobia, GAD, and separation anxiety all had significantly higher prevalence in the older children (over 7.5 years old) than the younger children (below 7.5 years old). Mukaddes and Fateh (2010) also found the prevalence of specific anxiety disorders to differ between children and adolescents; e.g., 57% of adolescents met criteria for OCD but only 17% of children, and 14% of adolescents met criteria for social anxiety which was not present in any children. In contrast 21% of children met criteria for specific phobia but no adolescents met the criteria. It may be misleading, therefore, to look at prevalence rates in anxiety disorders, especially for each specific disorder across a large age range.
In their review Kerns and Kendall (2012) suggested anxiety patterns may follow two trajectories depending on the type of anxiety: anxiety symptoms strongly related to ASD such as specific phobia, social anxiety disorder, and compulsions may remain constant across the lifespan whereas anxiety symptoms that are similar to those seen in the general population like GAD may follow the same developmental course of being variable across time and more prevalent in older children and adolescents. This requires further exploration.
Anxiety and ability level. A large majority of the work on anxiety in ASD has been conducted with high-functioning individuals (IQ>70) as assessing anxiety in lower functioning individuals is more complex due to the limited ability of individuals with ASD to express their emotions, and parents and caregivers’ difficulty in distinguishing anxiety behaviors from other negative emotions. Studies that have included participants with ASD and ID have found a range of rates of anxiety using parental questionnaires (11–48%; Bakken et al., 2010; Lecavalier, 2006; Sukhodolsky et al., 2008); Bradley et al. (2004) compared samples of individuals with ID and found anxiety to be present more often in individuals with ASD and ID than in individuals with just ID alone.
Studies comparing IQ groups have found inconsistent results. Two reviewed studies utilizing diagnostic interviews found no significant IQ differences in those who did or did not have a co-occurring anxiety disorder across samples with a wide range of IQ scores (de Bruin et al., 2007; Simonoff et al., 2008). However, questionnaire measures of anxiety symptoms have found more variation in findings across IQ. Higher anxiety symptoms have been shown to be more frequent in children with an IQ above 70 and functional language (Gadow et al., 2005; Hallett et al., 2013; Weisbrot et al., 2005) leading Gadow et al. (2005) to hypothesize that high-functioning individuals with ASD have an increased awareness of their differences and this in turn will lead to more anxiety. It may also be that questionnaire measures are better at capturing the presentation of anxiety in higher rather than lower functioning individuals. In contrast, Brereton et al. (2006) found overall psychopathology including anxiety, as measured using questionnaires, was not affected by age, sex or IQ in 381 young people with ASD who ranged from normal IQ to severe intellectual disability.
Witwer and Lecavalier (2010) found diagnostic rates of specific anxiety disorders, based on diagnostic interviews, did vary according to IQ. GAD was significantly more common in the high-functioning (IQ>70) than low-functioning ASD group (IQ<70). Although not significant 13.6% of the high-functioning group also met criteria for OCD but none of the low-functioning group did. Whereas prevalence rates for specific phobia and separation anxiety were nearly identical across the two IQ groups. Using questionnaire measures Sukhodolsky et al. (2008) also reported anxiety symptoms related to specific phobia and social anxiety disorder were prevalent across the whole IQ range but symptoms related to, generalized anxiety were significantly more prevalent in higher functioning children and adolescents with ASD (IQ>70) than the lower functioning group with ASD (IQ<70).
Clinical implications. The current review indicates that a large proportion of individuals with ASD also meet diagnostic criteria for at least one anxiety disorder. Existing research suggests that anxiety is the most common reason for clinical referral (Ghaziuddin, 2002). It has also been shown that having anxiety as well as ASD may cause additional burden to the individual and lead to worse functional outcomes (Wood and Gadow, 2010). In combination, these findings indicate that the assessment and diagnosis of anxiety disorders should be a priority for clinicians caring for those with ASD as the diagnosis of an additional co-occurring disorder such as anxiety may lead to a more informed management plan and specific intervention programs.
The importance of assessment and diagnosis is even greater given that interventions used to treat anxiety in the general population have also been shown to improve anxiety in ASD, with some modification to the intervention (Sukhodolsky et al., 2013). Clinicians may also need to be aware of the high level of anxiety symptoms in individuals with ASD that may not meet diagnostic cut-offs. Many children and adolescents with ASD present with subclinical anxiety; Caamaño et al. (2013) found 76% of children and adolescents with ASD were at threshold or subthreshold level for diagnosis. In addition, a considerable percentage of children show impairment in relation to a specific anxiety disorder but do not always meet diagnostic cut-offs (Witwer and Lecavalier, 2010).
Research implications. To gain the most accurate understanding of the prevalence of anxiety disorders in ASD, consensus on a number of conceptual and methodological questions needs to be reached. These questions and avenues for future research are presented below.
Firstly, in contrast to the general population in which prevalence rates are based on population samples, only one study has used a diagnostic interview to measure prevalence rates in a population based study of individuals with ASD (Simonoff et al., 2008). The majority of studies used clinically referred samples, which could inflate prevalence rates due to these individuals being those that require clinical services. Estimates of prevalence rates also need to consider which informants are involved and how to combine responses across different informants, as well as the assessment period.
Anxiety disorders as a whole were prevalent across the age ranges sampled but investigation into specific anxiety disorders highlights that the prevalence of these disorders may vary across the lifespan (e.g., Davis et al., 2011; Lever and Geurts, 2016). Currently, these findings are limited to cross-sectional studies. Longitudinal studies will allow a better understanding of the developmental changes in anxiety disorders in ASD and the bidirectional effects of risk factors. These studies should also measure specific anxiety diagnoses across time as differences in trajectories for different disorders may obscure overall developmental effects.
Diagnostic interviews that apply DSM or ICD criteria are currently the favored method for estimating the prevalence of anxiety disorders; however, the measurement of anxiety symptoms and clinical cut-offs for diagnoses vary across studies with different research groups using different ways to describe anxiety symptoms. One study made modifications to an existing interview to better capture the presentation of anxiety in ASD (Leyfer et al., 2006), and Witwer and Lecavalier (2010) excluded questions that asked about functioning communication, e.g., worry in separation anxiety and GAD as well as thoughts in OCD for participants with lower IQ or language ability. These interviews are limited in providing information about anxiety in ASD. None of the diagnostic interviews have been validated in an ASD sample and this is a necessary first step in obtaining accurate prevalence data. Validation is challenging as there is no single set of biomarkers that accurately index anxiety disorders. Hence, a multimodal approach is required and should include discriminant and convergent validity, as well as longitudinal studies and evaluation of treatment response. It will also be beneficial to distinguish between the symptoms of ASD and anxiety which are similar, e.g., social avoidance and social anxiety; and also to question whether diagnostic criteria or symptom descriptions of anxiety should be modified in ASD.
The triggers or causes and presentation of anxiety in individuals with ASD can be different from the general population; anxieties centered around changes in routines, for example, do not currently fit easily into the diagnostic criteria (Ozsivadjian et al., 2012). Clinicians also face challenges in differentiating anxiety from other syndromes involving the regulation of emotion such as anger, low mood, or oppositionality (Muris et al., 1998). Ozsivadjian et al. (2012) highlight the main differences in presentation of anxiety in ASD. Parents frequently reported that changes to routines, social difficulties such as difficulties with perspective taking and social expectations as well as sensory sensitivities were common triggers in their children and that the presentation of anxiety was usually through challenging behavior or avoidance rather than verbally. These observational reports require confirmation with more experimental studies.
The introduction detailed the main changes in diagnostic criteria between DSM-IV-TR and DSM-5 for anxiety disorders. Little work has been published using the new criteria and therefore it is unknown what impact these changes may have on the prevalence rates in the general population but especially in ASD. For example, the onset of separation anxiety disorder no longer needs to be before 18 years of age, and the awareness of the individual in viewing their anxiety as excessive to receive a diagnosis of phobias may lead to further individuals with ASD meeting criteria as according to DSM-IV they would need to provide some insight into their anxious feelings which they may find difficult.
Future directions for research should include gaining a better understanding of how these two disorders overlap. This will require comparison of genetic markers, neurophysiological and brain-based measures to identify unique and shared mechanisms. We can also ask research questions about whether having both anxiety and ASD impacts on the individual beyond having one disorder of either ASD or anxiety. It has been suggested that anxiety may mediate difficulties already found in ASD such as social avoidance when individuals with ASD have difficulties initiating social contact which encourages social anxiety and prevents any attempts or improvements in social interaction (e.g., Bellini, 2004). The pathways and causal nature of these relationships need further exploration.
Individuals with ASD may find it difficult to recognize and describe their emotions. One avenue for future research should be on methodology which aids individuals with ASD and especially children or lower functioning individuals to report feelings of anxiety. Instead of questionnaire methods, picture-based rating scales could be an alternative. However, May et al. (2015) found that neither parent nor child questionnaire anxiety ratings were correlated to children’s “worry thermometer” ratings in which the children rated their anxiety on a picture scale. In addition to modifying methods of self-report, another way to overcome the reporting difficulties would be to use objective measures such as physiological or cognitive markers of anxiety. Indices such as heart rate variability (e.g., Appelhans and Luecken, 2006; Schmitz et al., 2011), galvanic skin response, or cortisol levels (e.g., van West, Claes, Sulon, and Deboutte, 2008) have been used to measure anxiety levels. For example, in a sample of children with ASD, anxiety, or both, the ASD plus anxiety group showed blunted cortisol and heart rate response to psychosocial stress compared to the other groups and these responses were also related to increased anxiety symptoms (Hollocks et al., 2014). Such physiological approaches may improve our understanding of anxiety presentation with and without ID, although measurement biases such as state anxiety due to research participation would need to be controlled for.
In conclusion, the best estimate prevalence rates of DSM-IV anxiety diagnoses in ASD are high. The variability in prevalence rates is likely explained by the different methodologies adopted across studies. Future work on validating measures for individuals with ASD and an understanding of the ascertainment and measurement of samples will further the understanding on the co-occurrence of ASD and anxiety.