Abstract

Research has demonstrated high prevalence for anxiety disorders in Autism Spectrum Conditions (ASC), however there is limited research investigating if those with anxiety disorder diagnoses have elevated autistic traits. Indeed, it is argued that women with anxiety disorders may include some of the undiagnosed population of autistic women. Furthermore, sensory over-responsivity (SOR) and intolerance of uncertainty (IU) have been shown to mediate autistic traits and anxiety in autistic samples, but have not been investigated in anxious samples. 360 women with anxiety disorders completed online questionnaires to investigate whether SOR and IU mediated anxiety symptoms and autistic traits. Results were consistent with hypotheses. Women with anxiety disorders had elevated autistic traits compared to general population, with 21.1% meeting cut-off for ASC on the AQ. Autistic traits were predicted by all anxiety measures, with SOR and IU being significant partial mediators in this relationship. Clinical implications for understanding autism and anxiety in women are discussed.
Key Words: Autism, Anxiety, Women, Sensory, Intolerance of Uncertainty, Mediation
Corresponding author email: fiona.gullon-scott@ncl.ac.uk
Autism Spectrum Conditions (ASC, or autism) are developmental conditions characterised by persistent deficits in social communication, restricted and repetitive patterns of behaviour and atypical sensory processing (American Psychiatric Association, APA, 2013). There is debate surrounding the gender ratio in autism, with male:female ratios of 4:1, up to 8:1 in those without intellectual disability, commonly reported (APA, 2013), however a recent meta-analysis suggests a ratio closer to 3:1 due to a diagnostic gender-bias against females (Loomes, Hull & Mandy, 2017).
Oone proposed explanation is that females with ASC may present with different symptoms to males (Head, McGillivray and Stokes, 2014), with the Diagnostic and Statistical Manual of Mental Disorders (DSM-5;APA, 2013) criteria criticised for not reflecting female-specific symptoms (Mazurek et al., 2017). Consequences of this include delayed diagnosis for women (Navot, Jorgenson & Webb, 2017), resulting in less support and increased stress (Bargiela, Steward & Mandy, 2016). One way in which women with autism may differ from their male counterparts is through anxiety, with female children with ASC displaying higher anxiety than autistic males (May, Cornish & Rhinehart, 2014). Furthermore, in a sample of autistic adults 70% of women also met diagnostic criteria for Social Anxiety Disorder (SAD), compared to a third of males (Maddox & White, 2015).
Adults and children with ASC of both sexes show high comorbidity rates with a range of anxiety disorders, (Ezell et al., 2018; Mertens, Zane, Neumeyer and Grossman, 2017; Bitsika and Sharpley, 2015). A recent meta-analysis of 31 studies found at least 40% of young people with autism have at least one comorbid anxiety disorder (van Steensel, Bögels, & Perrin, 2011), with Specific Phobia being the most common (30%) followed by Obsessive Compulsive Disorder (OCD) and SAD (17%). Such rates are significantly higher than in the general population (Beesdo, Knappe & Pine, 2009; Walitza et al., 2011). Understanding the nature of anxiety in ASC is important, as increased anxiety is associated with increased functional impairments (van Steensel, Bögels, & Wood, 2013), and social difficulties in ASC may be worsened by comorbid anxiety (Factor, Ryan, Farley, Ollendick & Scarpa, 2017). Furthermore, anxiety has been shown to impact educational success, friendships and family members (Reaven, 2011) and to interfere with treatment for ASC (White, Oswald, Ollendick & Scahill, 2009).
While research on comorbidity in autistic adults is not as prevalent as in children, Nah, Brewer, Young and Flower (2018) reported 39-40% of 155 autistic adults scored within the moderate-to-severe range on the Depression Anxiety and Stress Scale (DASS-21; Lovibond & Lovibond 1995), similar to rates in children (van Steensel et al., 2011). Furthermore, 27% of autistic adults report moderate-to-severe anxiety symptoms (Murray, Kovshoff, Brown, Abbott & Hadwin, 2019), with lower age, female gender, and autism severity all independently contributing to individual differences in anxiety. Lever and Geurts (2016) report rates of anxiety as high as 53.6% in autistic adults, and a recent systematic review of 30 studies gave a pooled estimated for current anxiety disorder in autistic adults at 27%, and lifetime at 42%, with SAD and OCD being the most common current disorders (Hollocks, Lerh, Magiati, Meiser-Stedman and Brugha, 2019). Autistic traits in non-clinical samples have also been correlated with social anxiety, OCD and worry (Liew, Thevaraja, Hong & Magiati, 2015).
While there is a clear relationship between autism and anxiety, there is also evidence to suggest that anxiety reported in individuals with ASC differs in presentation from those in non-ASC populations (Halim, Richdale & Uljarević, 2018; Kerns and Kendall, 2014). Consequently, atypical presentations of anxiety may complicate diagnostic processes and measures, with Wigham and McConachie (2014) suggesting caution should be exercised when using anxiety measures in autistic samples.
Research has also investigated the relationship between autism and specific anxiety disorders such as OCD, Specific Phobia, Generalised Anxiety Disorder (GAD), and SAD due to the high comorbidity rates in ASC for each of these disorders (Leyfer et al., 2006). The relationship may be bidirectional, with research suggesting around 32-36% of children or young people with OCD showing clinically significant ASC traits (Stewart et al., 2016; Griffiths, Farrell, Waters & White, 2017), and anxiety may increase RRBs (Stewart et al., 2016; Russell, Frost and Ingersoll, 2019). Within samples of children with autism those with anxiety displayed more RRBs than those without, and higher levels of anxiety were associated with higher levels of insistence on sameness in the anxious group, whereas anxiety was associated with sensory-RRBs in both groups (Rodgers, Glod, Connolly & McConachie, 2012). However, the causal relationship between RRBs and anxiety is not clear, with Ooi et al., (2018) suggesting such behaviours may serve to reduce anxiety, while Rodgers et al., (2012) argue anxiety may be a consequence of RRBs, and that other cognitive factors such as IU are likely to play a role.
Another specific anxiety disorder with high comorbidity rates with autism is SAD. Joshi et al., (2013) compared prevalence rates of SAD with age- and gender-matched adults with and without ASC, finding that the ASC group displayed significantly higher rates of lifetime (56%) and current (40%) SAD relative to the non-ASC group (19% and 16% respectively). Maddox and White (2015) found that 50% of adults with ASC met diagnostic criteria for SAD, and Tonge, Rodebaugh, Fernandez and Lim (2016) reported 70.8% of adults with Social Anxiety Disorder scored above cut-off on a measure of autistic traits compared to 10.74% in the control group. Research also supports an association using continuous measures of SAD, with autistic adults displaying higher levels of SAD symptoms (Kleinhans et al., 2010). Finally, Bejerot, Eriksson and Mörtberg (2014) found that adults with ASC had significantly higher SAD and social avoidance scores compared to non-ASC samples, with SAD scores correlating with autistic traits, and SOR recognised as a possible contributing factor in this relationship. Further possible explanations for the link between SAD and ASC include how awareness of social difficulties and a sense of loneliness may contribute to the development of social anxiety (Kuusikko et al., 2008). White, Bray and Ollendick (2012) claim social anxiety and the social deficits of ASC likely have a bidirectional relationship, where anxiety is heightened by awareness of social difficulties, yet increased anxiety may impair the ability to make accurate interpretations in social situations, or lead to avoidance and consequently fewer opportunities to learn and practise social skills.
Only a handful of studies have investigated whether individuals with anxiety have increased autistic traits. Van Steensel et al. (2013) found children with anxiety disorders had higher autistic traits than those without an anxiety disorder, with 31% scoring within the ASC range. Pine, Guyer, Goldwin, Towbin and Leibenluft (2008) found youths with anxiety or depressive disorders had higher scores on ASC symptom scales than those without, and Stewart et al., (2016) demonstrated that 36.3% of 5-8 year olds with OCD met the clinically significant threshold for ASC. Similarly, Towbin, Pradella, Gorrindo, Pine and Leibenluft (2005) found that 62% of children with mood and anxiety disorders scored in the ASC-likely range in at least one measure. Finally, Puleo and Kendell (2010) found elevated autistic traits in anxious youth, on average 13 points higher on the Social Responsiveness Scale-Parent (Constantino & Gruber, 2005) than typically developing children. Understanding autistic traits in anxious individuals is significant for determining effective treatment, and while studies have demonstrated elevated autistic traits in anxious children, research with anxious adults is limited.
One theory that may explain increased anxiety in autism, particularly in women, is that of camouflaging, which refers to behaviours that mask aspects of oneself from others in order to “pass” in everyday social interactions (Hull et al., 2017), and may be more common in autistic women than their male counterparts (Schuk, Flores & Fung, 2019). Such behaviours include making eye-contact, using learned phrases and following social scripts (Lai & Baron-Cohen, 2015), which autistic women describe as mentally exhausting and stressful (Hull et al., 2017). Indeed, those identified as being high camouflagers had higher anxiety symptoms than those who were low in camouflaging behaviours (Cage & Troxell-Whitman, 2019). Camouflaging may also explain the late diagnosis of ASC often seen in women (Lai et al., 2017) due to masking the external symptoms of ASC, with women scoring lower on measures of external symptoms such as the Autism Diagnostic Observation Schedule (Lord et al., 2000) than males, but showing similar scores on measures of internal symptoms (Lai et al., 2017).
South and Rodgers (2017) propose a further model that may explain this relationship, suggesting atypical sensory function, alexithymia and IU strongly predict anxiety in autistic samples. Addressing sensory function first, sensory processing abnormalities are common in ASC (Rogers & Ozonoff, 2005) and can include both sensory over- and under-responsivity in the same individual (Ben-Sasson, Cermak, Orsmond, & Tager-Flusberg, 2007), though the former has received more attention regarding anxiety in ASC. Sensory over-responsivity (SOR) is defined as an extreme negative response to (or avoidance of) sensory stimuli such as loud noises or unexpected touch (Liss, Saulnier, Fein & Kinsbourne, 2006). A meta-analysis of 14 studies demonstrated significant differences in both the presence and frequency of sensory symptoms between those with and without ASC (Ben-Sasson et al., 2009), with 95% of autistic samples demonstrating some degree of sensory processing dysfunction (Tomchek & Dunn, 2017).
SOR is associated with anxiety as well as increased functional impairment (Ben-Sasson et al., 2008), with significant correlations between sensory features and a number of anxiety disorders (Bitsika, Sharpley and Mills, 2016). Furthermore, in ASC samples, more severe anxiety was associated with more severe sensory dysfunction (Uljarević, Lane, Kelley & Leekam, 2016), with a positive correlation between sensory-defensiveness and anxiety in children and teens with Asperger’s Syndrome. Green and Ben-Sasson (2010) propose a number of models which could explain the relationship between SOR and anxiety: the primary SOR model proposes that anxiety is caused by, or is a symptom of SOR, whereas the primary anxiety model posits that SOR is caused by anxiety. Additionally, they argue the possibility of a third, noncausal model where both anxiety and SOR are caused by another factor. Green et al., (2013) found a uni-directional relationship in autistic toddlers, where SOR predicted anxiety, but anxiety was not predicted by SOR. Similarly, in a non-clinical sample of children, auditory and tactile SOR was associated with higher fear and anxiety (Goldsmith, Van Huelle, Arneson & Schreiber, 2006). In an adult general population sample, SOR and stress were mediators in the relationship between autistic traits and anxiety (Amos, Byrne, Chouinard & Godber, 2018), supporting Green and Ben-Sasson’s primary SOR model. This was further supported by both Lane, Reynolds, Dumenci (2012) who found that in children with ASC diagnoses the magnitude of physiological responses to sensory challenges was a mediator between arousal and anxiety, and Lidstone et al., (2013) who found that while the autistic trait of insistence on sameness was significantly correlated with anxiety, this relationship was mediated by sensory avoiding and sensitivity.
In support of the primary anxiety model, Horder, Wilson, Mendez and Murphy (2014) found that anxiety partially explains the relationship between ASC and sensory abnormalities in clinical and non-ASC adult samples. Scores on three different sensory scales were significantly correlated with autistic traits and remained significant after controlling for anxiety scores, indicating that although correlated, anxiety and sensory abnormalities are still distinct phenomena within ASC. Additionally, self-ratings of sensory processing in the visual domain, sensory sensitivity, or both predicted a range of anxiety disorders, including GAD, SAD, phobia and OCD (Bitsika et al., 2016). In non-ASC female students’ higher social anxiety was associated with more abnormal sensory responses, a relationship that was mediated by the level of autistic traits (Ludlow, Roberts & Gutierrez, 2015). Understanding the role SOR has in anxiety and ASC has clinical relevance regarding the need for sensory neutral environments to manage anxiety, and also suggests that SOR is a possible target for treatment of anxiety in ASC.
South and Rodgers (2017) suggest alexithymia may play a role in the relationship between autism and anxiety, as it is common in both ASC (Poquereusse, Pastore, Dellantonio, & Esposito, 2018) and anxiety (Karukivi et al., 2010). Milosavljevic et al., (2015) found 55% of autistic adolescents scored above cut-off for alexithymia, a significantly elevated rate compared to 16% in the non-ASD group. Furthermore, in the ASC group alexithymia was associated with increased anxiety and sensory processing atypicalities. Indeed, alexithymia has previously been associated with SOR, with individuals being hypersensitive to touch and pain (Nyklı́ček & Vingerhoets, 2000). In young adults with autism, alexithymia and emotional regulation predicted anxiety and mediated associations between autistic features and anxiety (Morie, Jackson, Zhai, Potenza & Dritschel, 2019). Therefore, alexithymia may partially explain the relationship between anxiety and autism. Though there is some overlap between alexithymia and ASC, evidence suggests they are unique (Harms, Martin & Wallace, 2010; Bird & Cook, 2013).
The final variable in the South and Rodgers (2017) model is intolerance of uncertainty (IU) - the belief that uncertainty is upsetting and negative (Hodgson, Freeston, Honey & Rodgers, 2017), a shared mechanism in both anxiety and ASC (Chamberlain et al., 2013). IU has been identified as a critical construct underpinning a range of anxiety disorders, even after controlling for trait anxiety and negative affect (Norr et al., 2013), and is viewed as a dispositional risk factor for the development and maintenance of anxiety (Carleton, 2012). IU has also been associated with ASC, with higher IU in autistic children and adolescents compared to controls, and IU being a significant mediator in the relationship between autism and anxiety (Neil, Olsson & Pellicano, 2016). Boulter et al., (2014) identified a causal pathway from IU to anxiety in ASC, in that IU accounted for significantly higher levels of anxiety in autistic children and adolescents compared to TD groups. IU may also explain the relationship between RRBs, and the atypical sensory reactions seen in autistic children (Wigham et al., 2015), with evidence for an indirect path from sensory atypicalities to RRBs, mediated by IU and anxiety. Additionally, Glod, Riby and Rodgers (2019) found evidence for a direct relationship between sensory features and repetitive behaviours in autistic children and suggest anxiety and/or IU mediated the relationships between both hyper- and hypo-sensory responsiveness and sensory motor behaviours. Joyce, Honey, Leekam, Barrett and Rodgers (2019) also report a positive relationship between IU and RRBs in autistic adolescents, as well as a positive association between anxiety and RRBs.
Finally, research has also looked at combinations of SOR, alexithymia and IU. Correlations between ASC symptoms and anxiety were almost completely accounted for by a combination of alexithymia, emotional acceptance and IU, explaining 64% of the relationship between ASC symptom severity and anxiety (Maisel et al., 2016). Furthermore, IU has been linked to SOR, because after controlling for anxiety IU remained a significant predictor of sensory sensitivities (Neil et al., 2016).
The majority of research on anxiety in ASC has focused on children and adolescents, with autistic adults being underrepresented. Moreover, most research looks at the relationship between autism and anxiety in those with a primary diagnosis of autism, with little research investigating autistic traits in those with anxiety disorders. Additionally, research on autism in women is relatively limited. Females are at higher risk of their ASC being undiagnosed, or being missed arguably due in part to diagnoses of other presentations, such as anxiety (Lai & Baron-Cohen, 2015; Pellicano, Dinsmore & Charman, 2014).
The present study therefore investigated the autistic traits in women with a diagnosis of an anxiety disorder, including SAD, GAD, Specific Phobia or OCD, due to the emerging theory that undiagnosed autistic women may be over-represented in women with anxiety disorder diagnoses (Lai et al., 2017), and explored the influence of IU and SOR as mediating variables between ASC traits and anxiety in this population. Additionally, since research suggests there are specific associations with ASC and disorders such as OCD and SAD the present study utilised a number of different anxiety scales in order to investigate the relationship of subgroups of anxiety with autistic traits.
Hypotheses were:
1. Women with anxiety disorder diagnoses will have elevated autistic traits, being more likely to reach clinical thresholds than those without anxiety diagnoses.
2. Autistic traits will be predicted by scores on OCD, GAD, social phobia and SAD scales.
3. The relationship between ASC and each anxiety disorder will be partially mediated by IU and SOR.