Repetitive and restricted behaviours and anxiety in autism spectrum disorder: protocol for a systematic review and meta-analysis

There is a body of evidence highlighting the association between RRBs and anxiety symptoms in individuals with ASD [12]. Specifically, cross-sectional studies have found moderate to large correlations between the two constructs (r = .41–.69) with individuals experiencing higher anxiety engaging in more RRBs [13‐15]. Several longitudinal studies exploring the nature of this relationship have found that the severity of RRBs predicts later anxiety, suggesting that a unidirectional relationship exists between the two constructs (p < .001–p = 0.026) [12, 16, 17]. The strong and consistent associations found across studies suggest the RRBs and anxiety are interrelated and that the presence of RRB’s precedes the onset of anxiety [12, 16, 17]. Subsequently, RRBs have been conceptualised as an early manifestation of anxiety [12, 16] or a maladaptive response to negative affect which subsequently increases one’s risk of anxiety over time [10, 12, 18]. Although engaging in RRBs may temporarily reduce anxiety symptoms, their continued use may impede one’s ability to adopt more adaptative coping mechanisms, reduce their ability to engage in their environment, and therefore ultimately perpetuate feelings of anxiety [10, 12, 18]. This parallels the trajectory observed in OCD whereby individuals engage in compulsions to neutralise anxiety associated with obsessions [4, 18]. However, in reality, these compulsions add considerable distress to individuals with OCD as they do not address the source of anxiety and interfere with daily functioning [4, 18]. A similar association is seen in posttraumatic stress disorder, where efforts to suppress or avoid memories of an index trauma are helpful in the short term, but lead to long-term increases in intrusive re-experiencing and hyperarousal [19]. As the hypothesised relationship between anxiety and RRBs in ASD is thought to mirror the relationship between obsessions and compulsions in the OCD literature, it is plausible to suggest that RRBs could identify individuals who are vulnerable to experiencing future anxiety. This could help clinicians identify and treat anxiety in young children, preventing the manifestation of significant distress in later years.

Although the literature consistently reports that RRBs are associated with anxiety, it is important to clarify if this relationship exists for all or specific RRB subtypes. There is a consistent body of evidence to suggest that insistence on sameness and restricted interests are linked to anxiety [3]. However, the relationship between the remaining subtypes and anxiety remains unclear, with conflicting evidence emerging from the literature. This lack of clarity may be associated with the varying definitions of RRBs. Whilst some studies have explored subtypes individually [3, 20], others have combined the subtypes into two broad domains known as low-order and high-order RRBs [10, 21]. Low-order RRBs refer to repetitive movements that may be motivated by sensory feedback, whereas high-order RRBs include restricted interests and insistence on sameness behaviours [22]. These domains have been criticised for being too broad and therefore disregarding important differences in the functions served by RRB subtypes [3]. The grouping of low-order RRBs may limit our understanding of the relationship between these RRB subtypes and anxiety as they are thought to have contradictory functions. For instance, sensory processing abnormalities may increase one’s risk of experiencing anxiety as having a low threshold for sensory input is thought to exacerbate environmental stressors [17]. This is consistent with the finding that hyperactivity to sensory input in toddlers predicts later anxiety [17]. 

Contrastingly, repetitive movements are thought to enable one to manage anxiety that stems from sensory processing abnormalities [23]. A discrepancy is therefore highlighted between the two low-order RRBs, with one thought to contribute to anxiety, whilst the other a response. 

Subtypes of high-order RRBs are also understood to uniquely moderate anxiety symptoms. For instance, restricted interests are thought to be a coping response to anxiety [18], whilst insistence on sameness behaviours may enable one to control their environment and thus buffer the effect of anxiety [23]. These distinctions are consistent with the results of factor analyses which have found restricted interests and insistence on sameness behaviours to be distinct subtypes [24‐26]. Thus, it is critical to examine these subtypes individually in order to clarify their differential relationship with anxiety. This knowledge could help clinicians tailor diagnostic tools and treatment interventions as specific RRBs may be utilised as a marker for anxiety, whilst others could be targeted for interventions.

To effectively conceptualise the associations between RRBs and anxiety, it is important to also understand the inter-relationships amongst RRB subtypes. There is emerging evidence to suggest that RRB subtypes are interrelated [23, 27]. Specifically, sensory processing abnormalities have been shown to negatively correlate with repetitive movements and high-order RRBs, whilst a positive association has been observed between repetitive movements and high-order RRBs [23, 27]. However, not all studies yield consistent results with Lidstone et al. [13] finding that low registration (a measure of hyporeactivity) and sensory sensitivity (a measure of hyperreactivity) were not associated with repetitive movements.

The relationship between specific RRBs and anxiety may be influenced by the presence of additional RRBs. Black et al. [28] found that hypersensitivity mediated the relationship between anxiety subtypes and high-order RRBs in individuals with ASD. There is additional evidence to suggest that anxiety mediates the relationship between sensory processing abnormalities and high-order RRBs [27]. Drawing from these findings, it is plausible to suggest that restricted interests and insistence on sameness may enable individuals to reduce anxiety associated with sensory input [13, 23]. Additionally, anxiety and intolerance of uncertainty, a trait associated with high worry and avoidance, were found to partially mediate the relationship between sensory processing abnormalities and repetitive movements [23, 27]. These findings suggest that repetitive movements might function to regulate arousal; however, this process may also reduce anxiety [23]. Collating this information may help clinicians understand if these behaviours are interrelated and whether clusters or specific RRBs are associated with anxiety.

Other individual traits or characteristics may also influence the relationship between RRBs and anxiety. For instance, intolerance of uncertainty has been linked to both anxiety and RRBs [22, 23, 29]. Hwang et al. [29] found that intolerance of uncertainty significantly mediated the relationship between anxiety and hyposensitivity, hypersensitivity, and high-order RRBs. Age and intelligence quotient (IQ) have also been identified as factors relating to both anxiety and RRB subtypes [3, 30]; however, the role they play within the RRB-anxiety relationship remains unclear. Findings to date are mixed with some studies finding that anxiety increases with age but not with RRBs [17, 31], whilst others have found the reverse [29]. As all individuals with ASD present with at least two RRBs as per the Diagnostic and Statistical Manual of Mental Disorders (DSM: 1), identifying and collating the factors associated with the RRB-anxiety relationships could help clinicians identify who are most vulnerable to developing anxiety symptoms.

Our systematic review aims to (1) meta-analytically assess the association between RRB subtypes and anxiety symptoms in individuals with ASD, (2) meta-analytically assess the association between the RRB subtypes in the context of the anxiety-RRB relationship, and (3) collate factors associated with both RRBs and anxiety symptoms. These aims will be addressed via the following research questions:

This review is registered on the PROSPERO database (CRD42020185434).

Observational studies will be included in the review if they meet the following eligibility criteria:

Articles will be excluded if they include less than five participants or if they are an opinion piece, editorial, or review. The reference list of relevant reviews will be checked to identify additional research.

The literature search will be conducted across five databases: CINAHL Plus, Cochrane Central Register of Controlled Trials, Ovid MEDLINE, PsycINFO, and Scopus. The search strategy was developed alongside a senior tertiary librarian to identify studies which meet the above eligibility criteria. The reference list of all eligible articles will also be searched to ensure literature saturation. The search strategy for Ovid MEDLINE is included as an additional file (Additional file 1).

Unpublished materials will also be identified by searching clinical trial registries, ClinicalTrials.​gov and Australian New Zealand Clinical Trials Registry (ANZCTR), and thesis databases, Trove and ProQuest Dissertations and Theses Global. The first 50 results of an advanced Google search will also be reviewed to identify papers that were not included in the databases.

The search will be limited to humans.

Articles identified from the literature search will be exported into Endnote X9 and duplicates will be removed.

The titles and abstracts of remaining articles will be imported into Covidence, and their eligibility will be assessed by two independent reviewers. Eligible articles will have this process repeated for the full article. If an article is missing information, the author will be asked to provide this information.

Reviewers will solve disagreements by discussing the matter in relation to the inclusion criteria. If consensus is not reached, a third reviewer will be engaged to resolve this.

A PRISMA flow diagram will highlight the studies remaining at each stage.

A standardised form will be utilised by two reviewers to extract the following data for each study: (1) study details including the study design; (2) sample size; (3) participant demographics including gender, age, IQ, level of functioning, and severity of ASD symptoms; (4) how ASD diagnoses were confirmed; (5) anxiety type/s investigated; (6) anxiety measure/s used; (7) RRB subtype/s investigated; (8) RRB measure/s used; (9) factors associated with RRBs and anxiety; (10) reported outcomes, statistical analyses, adjustments, and significance; and (11) information for assessing methodological quality (see Additional file 2).

If overlapping studies are identified, we will include the study with the largest sample size to avoid including duplicate samples or studies as this may bias the results.

When data is missing, we will contact the first author of the study via email and ask them to provide this information. If this information is not provided, we will aim to retain the study in the review. However, a sensitivity analysis will be run to assess the impact of missing data on the results.

One reviewer will utilise a modified version of the Newcastle-Ottawa Scale to assess the methodological quality of each included study [32]. This scale has been adapted for use with cross-sectional studies to assess selection, comparability, and outcomes (see Additional file 3).