Spectrum or subtypes? A latent profile analysis of restricted and repetitive behaviours in autism

https://doi.org/10.1016/j.rasd.2018.10.003Get rights and content

Highlights

  • Restricted and repetitive behaviours and interests mainly differ in severity.

  • The clinical, behaviour and diagnostic features of autism form severity subgroups.

  • The phenotype of autism may be best defined as a spectrum of severity.

Abstract

Background

Autism Spectrum Disorder (ASD) is a heterogeneous condition. One way of understanding this heterogeneity is by investigating whether homogenous subgroups within the autism population exist. Some studies have attempted to do this by looking at social and communication skills. However, few studies have looked at subtyping using restricted and repetitive behaviours. While restricted and repetitive behaviours form part of the core features of autism, their presentation is diverse across different individuals on the spectrum. The aim of this study was to determine if restricted and repetitive behaviours could be used to identify potential subtypes of autism.

Method

This study used unsupervised clustering algorithms to differentiate subgroups of individuals on the autism spectrum based on their scores on the Repetitive Behaviour Scale-Revised (RBS-R).

Results

Three groups were found that reported low, medium and high levels of restricted and repetitive behaviours. These groups also differed on a range of clinical measures including problematic behaviours, autistic traits and adaptive behaviours.

Conclusions

Our findings indicate that subgroups of individuals with autism can be identified based on their level of restricted and repetitive behaviours. This highlights that restricted and repetitive behaviours may be best understood under a dimensional continuum of severity. This has implications for our understanding of the non-social characteristics of autism.

Introduction

Autism Spectrum Disorder (ASD) is a neurodevelopmentalcondition diagnosed based on social and communication impairments and restricted and repetitive interests and behaviours (American Psychiatric Association, 2013). However, autism is very heterogeneous in clinical presentation and trajectory (Fountain, Winter, & Bearman, 2012; Gotham, Pickles, & Lord, 2012; Lord & Bishop, 2015; Venker, Ray-Subramanian, Bolt, & Ellis Weismer, 2014). In addition, individuals on the spectrum can respond differently to the same intervention, even if they start the intervention at the same developmental level (Ben-Itzchak & Zachor, 2007, 2009). This is a pressing issue, as these diverse outcomes can limit the amount of benefit individuals on the spectrum receive from supports and interventions. In order to shed some light on this issue, the idea has been put forward that it may be time to think of autism as consisting of a number of subtypes rather than as one condition (Cholemkery, Medda, Lempp, & Freitag, 2016).

The Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) originally classified ASD under the umbrella of Pervasive Developmental Disorders, which included: Autistic Disorder, Asperger’s Disorder, Pervasive Developmental Disorder Not Otherwise Specified (PDD-NOS), Rett’s Disorder and Childhood Disintegrative Disorder (American Psychiatric Association, 1994). However, the DSM-5 (American Psychiatric Association, 2013) collapsed these into one category described as ‘Autism Spectrum Disorder’. This was based on the lack of research evidence distinguishing between the DSM-IV categories (Mayes & Calhoun, 2004; Mayes, Calhoun, & Crites, 2001; Verté et al., 2006). For example, Verté et al. (2006) compared three groups of children on the autism spectrum (High Functioning Autism, Asperger’s and PDD-NOS) based on their scores on the Children’s Communication Checklist. The authors found very few differences in scores between the three groups. Following this, they conducted hierarchical cluster analysis using the subscales of the ADI-R and found three clusters that were only very weakly related to the DSM-IV classifications. When comparing the three groups across the Autism Diagnosis Interview (ADI-R) subscales, it was found that these three clusters predominately differed in severity levels, rather than exhibiting unique profiles (Verté et al., 2006).

The DSM-5 reconceptualises ASD as one condition with three different levels of severity: Level 1 “Requiring support”; Level 2 “Requiring substantial support”; and Level 3 “Requiring very substantial support”. These levels are characterised through the two diagnostic dimensions: social communication and restricted interests and ritualistic and repetitive behaviours (see Table 1 for a summary of each level). However, the differentiating features between these categories still remains unclear, and little research has been done to establish the validity of these support levels. While some studies investigating subtypes outside of the DSM-IV classifications have found evidence to support the DSM-5 idea of a spectrum condition differentiated by severity, others have found subtypes with distinct features.

For example, using hierarchical clustering, Stevens et al. (2000) looked at expressive and receptive language skills, nonverbal IQ and social behaviour in children on the autism spectrum and identified a “high functioning” and “low functioning” group. Specifically, these two groups differed purely in severity on measures of non-verbal IQ, social interaction, communication, repetitive behaviours, language, and social behaviour. Conversely, Klopper, Testa, Pantelis, and Skafidas, (2017) used the items from the Autism Diagnosis Observation Schedule (ADOS-2; Lord et al., 2012) and the ADI-R (Rutter, Le Couteur, & Lord, 2003) with a group of children without intellectual disability, and found that these children could be separated into two subgroups with distinct social communication and repetitive behaviour profiles. Specifically, they found one group that had relatively more ADOS and ADI-R social communication difficulties, but slightly less ADI-R restricted and repetitive behaviours, and another that had relatively less social communication difficulties, but slightly more restricted and repetitive behaviours.

Another study conducted by Cholemkery et al. (2016) found support for both a severity and a subtype based model. Specifically, using K-means clustering with the item scores from the ADI-R questionnaire, the authors found a three cluster solution. This included one cluster that scored low across all the ADI-R domains (i.e. low in autistic traits), one cluster that scored high across all the ADI-R domains (i.e. high in autistic traits) and one cluster that showed more social and communication difficulties and less repetitive and restricted interests and behaviours.

In a larger scale study, Greaves-Lord et al. (2013) conducted latent profile analysis with a multisite sample of 949 children. The authors identified six groups that showed unique patterns of scores on the subscales of the Children’s Social Behaviour Questionnaire which measured social contact and interest, understanding social information, fear and resistance to changes, stereotyped behaviours, adaptation to social situations and executive function in daily life. However, when these groups were compared based on their scores on the Child Behaviour Checklist (CBCL), a different picture emerged. A comparison of average item scores showed that the six groups differed only in severity in emotional and problematic behaviours across all eight CBCL subscales rather than exhibiting unique profiles. The findings of these studies indicate that more research needs to be conducted in order to gain a better understanding of whether subtypes of autism are best characterised as having unique profiles or differences in severity levels.

The methodology used by Greaves-Lord et al. (2013), specifically coupling data-driven clustering techniques, such as latent class analysis and latent profile analysis, with data from large databases, is highly suitable for exploring the existence of subtypes within a population. One database that has yet to be used in this type of analysis is the Simons Simplex Collection (SSC). The SSC is an amalgamation of clinical and genetic data from approximately 2600 families of individuals on the spectrum collected from 12 different sites in the US. Some studies have used the clinical data in this database to explore subtypes of autism based on predefined criteria such as high IQ or higher autistic symptoms (Chaste et al., 2015). However, there is a lack of research examining this data using purely data-driven techniques. The clinical data within the SSC includes measures that assess autism diagnosis, autism severity, repetitive and restricted interests and behaviours, cognitive ability, problematic behaviours and adaptive behaviour. This range of clinical measures, coupled with the large sample size, makes the SSC a powerful resource waiting to be leveraged in the search for subtypes.

In the current literature, the majority of studies focus mainly on social skills and communication abilities when investigating subtypes (Greaves-Lord et al., 2013; Stevens et al., 2000; Volkmar, Cohen, Bregman, Hooks, & Stevenson, 1989; Wing & Gould, 1979). This means that subtypes defined based on restricted and repetitive behaviours and interests (RRBIs) is an area that is under-researched. RRBIs form a core aspect in the diagnosis of autism. However, the severity and presentation of RRBIs can vary substantially across individuals on the autism spectrum. For example, RRBIs can range from repetitive motor movements, to obsessions with parts of objects, to insistence in maintaining the same routine every day. In some individuals, these behaviours may only be observed in specific circumstances, while in others they may impact day to day activities (American Psychiatric Association, 2013). In addition, there is some evidence that RRIBs have different aetiology to social skills and communication challenges experienced by individuals on the spectrum (Happe & Ronald, 2008). A series of studies conducted by Happe et al. (Brunsdon & Happe, 2014; Happe & Ronald, 2008; Happe, Ronald, & Plomin, 2006) has found that the social and non-social aspects of autism cannot be explained by a single underlying cause at the genetic, neurobiological or cognitive level. These findings provide support for the need for more detailed research that explores the social and non-social domains of autism separately.

While some subtyping studies have examined RRBIs using the ADOS or ADI-R subscales (Ingram, Takahashi, & Miles, 2008; Klopper et al., 2017), these measures only provide a single score assessing repetitive behaviours, and are therefore limited in their capacity to characterise the true heterogeneity of RRBIs in autism. It is important to evaluate more comprehensive measures of RRBIs in order to understand whether they may provide an indication of any subtypes of autism that differ in presentation and outcomes. The Repetitive Behaviour Scale-Revised (RBS-R; Bodfish, Symons, Parker, & Lewis, 2000) is a specialised questionnaire that offers a more comprehensive assessment of the different types of RRBIs and their severity. The RBS-R consists of 43 items separated into different subscales that assess stereotyped (movement or actions that are repeated in a similar manner), self-injurious (movements or actions that have the potential to cause redness, bruising or other injury to the body, and that are repeated in a similar manner), compulsive (behaviours that are repeatedly performed according to a rule, or involves things being done ‘just so’), ritualistic (performing activities of daily living in a similar manner), sameness (resistance to change, insisting that things stay the same) and restricted (limited in range of focus, interest or activity) behaviours. It therefore offers a more comprehensive assessment of RRBIs and the different ways they can manifest than the ADOS or ADI-R.

The aim of this study was to determine if differing presentations of repetitive and restricted interests and behaviours could be used to identify and distinguish subtypes of autism. The first aim was to use LPA to extract subgroups of individuals on the autism spectrum based on their scores on the RBS-R. The second aim was to determine whether these subgroups had any unique profiles by comparing the groups across the range of measures present within the SSC. This included measures of cognitive ability, autism severity, problematic behaviours and adaptive behaviours.

Section snippets

Participants

Data was obtained from the Simons Simplex Collection (SSC). 2759 participants aged 4–18 years (M = 9.02, SD = 3.57) had completed the RBS-R. This included 2384 males and 375 females. Informed consent was obtained at each data collection site. Please see Fischbach and Lord (2010) for a complete report of the data collection process. The study was approved by the Human Research Ethics Committees of the University of New South Wales and work was carried out in accordance with the ethical standards

Results

The results of the LPA are provided in Table 2. The LPA identified 3 classes as the best fitting model to the data. Table 3 shows descriptive statistics of the three classes. These three classes did not differ in terms of age (F(2, 2758) = 0.162, p =  0.851) or sex (χ2 = .502; p =  0.778).

The three classes were compared across RBS-R subscales, cognitive ability, SCQ scores, CBCL internalising, externalising and total problematic behaviour scores and Vineland Communication, Daily Living Skills,

Discussion

The aim of this study was to determine if subtypes of autism could be distinguished based on restricted and repetitive interests and behaviours. Using LPA, subgroups of individuals were extracted based on their scores on the RBS-R subscales. The results indicated that a three class solution best fit the data. The three classes were then compared across a range of clinical measures. The results indicated that the three classes differed on RRBIs, cognitive ability, autistic traits, problematic

Limitations

While the large sample size of the SSC was beneficial in helping to avoid some of the problems commonly faced by previous studies in this area (Beglinger & Smith, 2005), this study was subject to some limitations. For example, the SSC only collected data from individuals aged 4 to 18, and thus does not fully represent the population of individuals on the autism spectrum. The results showed that there were no differences in age between the three severity groups. However, future research should

Conclusion

This study identified three groups of individuals with autism that had low, medium and high levels of restricted and repetitive behaviours. This highlights that the presentation of restricted and repetitive behaviours and interests among individuals with autism may be best understood from a dimensional perspective. The subgroups could also be differentiated by their difficulties with adaptive behaviours, autistic traits and problematic behaviours. This shows the importance of supplementing

Acknowledgments

We would like to acknowledge the financial support of the Cooperative Research Centre for Living with Autism (Autism CRC), established and supported under the Australian Government's Cooperative Research Centres program. LZ was supported by an Australian Postgraduate Award and acknowledges the financial support of the Cooperative Research Centre for Living with Autism (Autism CRC). We are grateful to all of the families at the participating Simons Simplex Collection (SSC) sites, as well as the

References (59)

  • H. Akaike

    Factor analysis and AIC

  • American Psychiatric Association

    DSM-IV: Diagnostic and statistic manual of mental disorders

    (1994)
  • American Psychiatric Association

    Diagnostic and statistical manual of mental disorders

    (2013)
  • K.K. Ausderau et al.

    Sensory subtypes in children with autism spectrum disorder: Latent profile transition analysis using a national survey of sensory features

    Journal of Child Psychology and Psychiatry, and Allied Disciplines

    (2014)
  • G.T. Baranek et al.

    Hyperresponsive sensory patterns in young children with autism, developmental delay, and typical development

    American Journal of Mental Retardation: AJMR

    (2007)
  • L. Beglinger et al.

    Concurrent validity of social subtype and IQ after early intensive behavioral intervention in children with autism: A preliminary investigation

    Journal of Autism and Developmental Disorders

    (2005)
  • A. Ben-Sasson et al.

    Sensory clusters of toddlers with autism spectrum disorders: Differences in affective symptoms

    Journal of Child Psychology and Psychiatry

    (2008)
  • J.W. Bodfish et al.

    Varieties of repetitive behavior in autism: Comparisons to mental retardation

    Journal of Autism and Developmental Disorders

    (2000)
  • B.A. Boyd et al.

    Sensory features and repetitive behaviors in children with autism and developmental delays

    Autism Research

    (2010)
  • V.E. Brunsdon et al.

    Exploring the ‘fractionation’ of autism at the cognitive level

    Autism

    (2014)
  • H. Cholemkery et al.

    Classifying autism spectrum disorders by ADI-R: Subtypes or severity gradient?

    Journal of Autism and Developmental Disorders

    (2016)
  • C.D. Elliott

    Differential ability scales–second edition (DAS-II)

    (2007)
  • A.J. Esbensen et al.

    Age-related differences in restricted repetitive behaviors in autism spectrum disorders

    Journal of Autism and Developmental Disorders

    (2009)
  • C. Fountain et al.

    Six developmental trajectories characterize children with autism

    Pediatrics

    (2012)
  • T.W. Frazier et al.

    Behavioral and cognitive characteristics of females and males with autism in the Simons Simplex Collection

    Journal of the American Academy of Child and Adolescent Psychiatry

    (2014)
  • K. Gotham et al.

    Trajectories of autism severity in children using standardized ADOS scores

    Pediatrics

    (2012)
  • K. Greaves-Lord et al.

    Empirically based phenotypic profiles of children with pervasive developmental disorders: Interpretation in the light of the DSM-5

    Journal of Autism and Developmental Disorders

    (2013)
  • F. Happe et al.

    The ‘Fractionable autism triad’: A review of evidence from behavioural, genetic, cognitive and neural research

    Neuropsychology Review

    (2008)
  • F. Happe et al.

    Time to give up on a single explanation for autism

    Nature Neuroscience

    (2006)
  • View full text