Application of Sensory Subtypes: Understanding repetitive behaviors, adaptive behaviors, and motor skills in autistic children

Sensory processing differences are highly prevalent in autism (54–97%) and demonstrate high degree of individual variability (Dellapiazza et al., 2018; Jussila et al., 2020; Kirby et al., 2022). Researchers have attempted to capture the heterogeneity of sensory features in autism through sensory subtypes¹ (e.g. Ausderau et al., 2014a; Chen et al., 2022; Kadlaskar et al., 2022; Lane et al., 2014; Tillmann et al., 2020), given that sensory patterns (i.e., hypo- and hyper-responsiveness) are often co-occurring (Ausderau et al., 2014b; Kuiper et al., 2019; Scheerer et al., 2021). While different approaches to subtyping have emerged, sensory subtypes have produced distinct subgroups to better understand autistic children’s sensory profiles (Lane, 2021) with emerging research applying them to outcome measures. Sensory subtypes can be clinically meaningful as they increase understanding of the complexity of the heterogeneous relationship between sensory differences and autistic features (Kadlaskar et al., 2022). In addition, subtyping may help understand when co-occurring sensory features may be protective or interfere with meaningful daily activities for autistic individuals. Therefore, the purpose of the present study was to apply an existing sensory subtype methodology to a new group of autistic children to examine the relationship between sensory subtypes and multiple child outcome measures.

To date, multiple studies have used a variety of methods to examine sensory subtypes in autism and to see how these subtypes relate to core autism features, adaptive behavior, and motor skills. Based on a literature search of articles that used sensory subtyping in autistic populations alongside measures of autism features, adaptive behavior, or motor skills, a total of 11 studies were identified. These studies varied in participant demographics, sensory measures, and subtype generation. The participants in the studies ranged from small developmental age spans (e.g., 2–4 years old for Kadlaskar et al., 2022) to including participants into adulthood (e.g., 6–31 years-old for Tillmann et al., 2020). It is important to consider the age of the participants in the study, even though research has found overall high stability in autistic children’s sensory pattern over time (Baranek et al., 2019; Dwyer et al., 2020; Perez Repetto et al., 2017). Most of the studies required confirmation of an autism diagnosis, but there was varying additional exclusion criteria for co-occurring conditions of autism including ADHD (Scheerer et al., 2021, 2024). The most common sensory measurement was the Short Sensory Profile (SSP) (Dwyer et al., 2020, 2022; Kadlaskar et al., 2022; Lane et al., 2010; Scheerer et al., 2021, 2024; Tillmann et al., 2020; Uljarević et al., 2016). Additional sensory measurements were used to generate sensory subtypes including the Short Sensory Profile-2 (Simpson et al., 2019), Infant/Toddler Sensory Profile (Ben-Sasson et al., 2008), and the Sensory Experience Questionnaire (SEQ 3.0) (Ausderau et al., 2016). All of the sensory assessments were caregiver-reported measures to capture sensory features, but each measure used a unique classification of sensory patterns (Burns et al., 2017; Gunderson et al., 2023). Additionally, each paper used different statistical approaches to generate the subtypes. These included a latent profile analysis (Kadlaskar et al., 2022), a latent transition profile analysis (Ausderau et al., 2016), growth mixture model (Dwyer et al., 2020), factor mixture model (Dwyer et al., 2022; Tillmann et al., 2020), and cluster analysis (Ben-Sasson et al., 2008; Lane et al., 2010; Scheerer et al., 2021, 2024; Simpson et al., 2019; Uljarević et al., 2016). The measurement tool and grouping analysis play a critical role in creating a variety of sensory subtype classifications that have been used to relate to child outcomes (DeBoth & Reynolds, 2017).

Sensory subtypes have been applied to explore core autism features which include differences in social communication skills and restricted or repetitive behaviors and interests (American Psychiatric Association, 2013). Social communication was evaluated in nine papers using the social communication questionnaire (Scheerer et al., 2021, 2024; Simpson et al., 2019; Uljarević et al., 2016), Social Responsiveness Scale-2nd edition (Kadlaskar et al., 2022; Tillmann et al., 2020), the Vineland Adaptive Behavior Scale socialization and communication domain (Ausderau et al., 2016; Kadlaskar et al., 2022; Lane et al., 2010; Scheerer et al., 2021; Tillmann et al., 2020), Autism Diagnostic Interview-Revised social and communication domains (Ben-Sasson et al., 2008), and the Autism Diagnostic Observation Schedule (ADOS) social and communication domains (Ben-Sasson et al., 2008; Tillmann et al., 2020). There were mixed findings with some studies finding subtypes with less sensory features had scores reflecting less challenges with socialization and communication compared to the subtypes with more sensory features (Ben-Sasson et al., 2008; Kadlaskar et al., 2022; Scheerer et al., 2021, 2024; Tillmann et al., 2020). Contrary to this finding, Lane and colleagues (2010) found the subtype with all sensory features elevated preformed better on the communication domain compared to the subtype with elevated sensory features except for movement sensitivity and low energy/weak which had the lowest communication score. Ausderau and colleagues (2016) found the sensory subtype with elevated hyporesponsiveness and SIRS (sensory interests, repetitions, and seeking behaviors) had lower socialization and communication scores. Two studies found no difference between subtypes and social communication scores (Simpson et al., 2019; Uljarević et al., 2016). Five papers investigated the relationship between sensory subtypes and repetitive behavior. Three studies used the Repetitive Behavior Scale- Revised (Scheerer et al., 2021, 2024; Tillmann et al., 2020), ADOS (Ben-Sasson et al., 2008; Tillmann et al., 2020), and one study used the Social Responsiveness Scale-2 Restricted and Repetitive Behaviors Score (Kadlaskar et al., 2022). All five studies found consistent results that the sensory subtype with more elevated sensory differences had more repetitive behaviors and the subtype with the least amount of sensory differences had the lowest reported repetitive behaviors (Ben-Sasson et al., 2008; Kadlaskar et al., 2022; Scheerer et al., 2021, 2024; Tillmann et al., 2020). While this was consistent, findings using the ADOS as the measurement did not find any difference in repetitive behaviors and social communication (Ben-Sasson et al., 2008; Tillmann et al., 2020). Therefore, the measurement tool for repetitive behavior is essential to consider when comparing the findings. Overall, sensory subtype research has found subtypes with more sensory differences report more autistic features.

There are mixed results of six sensory subtype papers being applied to adaptive behavior using the Vineland Adaptive Behavior Scale (VABS). Three studies found a differential relationship between adaptive skills and sensory subtypes with lower adaptive skills being more often associated with the subtype defined by an increase in sensory features (Scheerer et al., 2021; Tillmann et al., 2020). However, two studies found specific sensory features impacted adaptive behavior scores. Ausderau and colleagues (2016) found the sensory subtype with elevated hyporesponsiveness and SIRS (sensory interests, repetitions, and seeking behaviors) had lower adaptive behavior scores. Lane and colleagues (2010) found the subtype with differences in all the SSP domains except for movement sensitivity and low energy/weak had the lowest VABS total score. Two papers did not find any differences in sensory subtypes and adaptive behavior (Dwyer et al., 2020, 2022). The mixed findings of the relationship between sensory subtypes and adaptive behavior may be the result of different participant demographics and modeling methodology.

Sensory input is also essential for motor performance (Coll et al., 2020; King et al., 2009) and sensory subtypes may help to understand relationship between sensory and motor development. To date, only two studies reported on motor skills using caregiver rerport from the VABS motor skills domain (Lane et al., 2010; Scheerer et al., 2021). Lane and colleagues (2010) did not find a significant difference between subtypes, while Scheerer and colleagues (2021) found a subtype categorized by challenges on the SSP for movement and low energy/weakness had the lowest motor skills. This difference may be attributed to how Scheerer and colleagues emphasized motor difficulties in their subtypes. Limited research has investigated the relationship between sensory subtypes and motor skills, and using performance based motor assessments may be beneficial to parse out unique differences.

Additional research is still needed to understand if sensory subtypes adequately capture the heterogeneity of sensory features and provide clinically useful information related to child outcomes. Given the limited studies applying sensory subtypes and lack of clarity of how sensory subtypes may relate to other key features in autism, such as adaptive behavior, motor skills, social communication and repetitive behaviors, this study applied the sensory subtype methodology by Ausderau and colleagues (2014a) using the SEQ 3.0. The sensory subtype model was based on differential patterns of four sensory features described as hyperresponsiveness (increase response to sensory input); hyporesponsiveness (decrease response to sensory input); sensory interests, repetitions, and seeking behaviors (SIRS); and enhanced perception (increased acuity of sensory stimuli) (Ausderau et al., 2014b). The four applied sensory subtypes are mild (low expression of the four sensory response patterns), sensitive-distressed (high expression of hyperresponsiveness and enhanced perception), attenuated-preoccupied (high expression of hyporesponsiveness and SIRS), and extreme-mixed (high expression of the four sensory response patterns) (Ausderau et al., 2014a). The subtyping model controlled for select characteristics (autism symptom severity based on the Social Responsiveness Scale, IQ proxy, chronological age, gender, household income, and mother’s education) and demonstrated high stability (91%) of subtyping membership over one year (Ausderau et al., 2016). The current study aimed to apply this subtyping to a new sample of autistic children and test the difference among subtypes on motor skills, autism features, and adaptive behaviors. Based on the previous broader sensory subtyping research, we hypothesized the four sensory subtypes would have similar motor skills (Lane et al., 2010), extreme-mixed subtype would have more reported autistic features (Kadlaskar et al., 2022; Scheerer et al., 2021, 2024; Tillmann et al., 2020), and attenuated-preoccupied would have the lowest adaptive behavior and social communication scores (Ausderau et al., 2016). Finally, we explored the subcategories of repetitive and adaptative behavior measurements to determine if the four sensory subtypes revealed distinct behaviors.

Methods

Participants

This study is a retrospective observational study that included 118 autistic children ages 6–18 years old (M = 10.6, SD = 3.3) who were initially from four studies. Three of the four studies (MRI studies n=104 and non-MRI n=14) examined brain-behavior relationships, requiring magnetic resonance imaging (MRI). Procedures for each study were in accordance with the institutional review board (Health Sciences Institutional Review Board, Protocol #2014–1248, #2014–1499, #2016–044, #2018–1067) and with the 1964 Helsinki declaration and its later amendments. Additional approval allowed for the multiple datasets to be combined (#2022–0413). In order for the data to be included in this study, the caregiver must have completed the Sensory Experience Questionnaire, version 3 (SEQ 3.0; Baranek, 2009) and the Social Responsiveness Scale, second edition (SRS-2; Constantino & Gruber, 2012) that were used for sensory subtyping (Ausderau et al., 2014b). Table 1 summarizes the overall child demographic characteristics and the four sensory subtypes. The subtypes contained unbalanced sample sizes with the mild and sensitive-distressed having the most participants, and the extreme-mixed and attenuated-preoccupied with the fewest. The four sensory subtypes had similar demographics, except the attenuated-preoccupied group captured the youngest participants.

Table 1.

Demographic characteristics

	All participants (n=118)	Mild (n= 53, 44.9%)	Sensitive-distressed (n=45, 38.1%)	Attenuated-preoccupied (n=9, 7.6%)	Extreme-mixed (n=11, 9.3%)
Sex
Male (%)	101 (85.6%)	45 (84.9%)	37(82.2%)	9 (100%)	10 (90.9%)
Race
White, Non-Hispanic	96 (81.4%)	46 (86.8%)	32 (71.1%)	9 (100%)	9 (81.8%)
White, Hispanic/Latinx	0 (0%)	0 (0%)	0 (0%)	0 (0%)	0 (0%)
Asian	6 (5.1%)	0 (0%)	4 (8.8%)	0 (0%)	2 (18.2%)
American Indian/Alaskan Native	1 (0.8%)	0 (0%)	1 (2.2%)	0 (0%)	0 (0%)
Black or African American	5 (4.2%)	2 (3.8%)	3 (6.7%)	0 (0%)	0 (0%)
More than one selected	10 (8.5%)	5 (9.4%)	5 (11.1%)	0 (0%)	0 (0%)
Age (years)*
Mean (SD)	10.6 (3.3)	11.2 (3.4)A	10.7 (3.3)A	7.5 (1.2)M,S	9.5 (3.0)
Range	6.1–17.8	6.2–17.5	6.1–17.8	6.2–9.43	6.4–16.1
IQ
Mean (SD)	99.6 (17.4)	101.6 (17.7)	98.9 (16.4)	102.9 (23.1)	89.5 (12.6)
Range	61–140	67–140	61–131	62–124	72–104
SEQ 3.0*
Total score
Mean (SD)	223.2 (47.8)	182.5 (26.0)S,A,E	247 (23.1)M,E	237.4 (19.0)M,E	310.6 (25.9)M,S,A
Range	124–348	124–237	187–290	210–272	258–348
Hypo score
Mean (SD)**	2.0 (0.5)	1.6 (0.4)S,A,E	2.0 (0.4)M,A,E	2.6 (0.3)M,S	2.8 (0.5)M,S
Range	1.1–3.7	1.1–2.7	1.2–2.9	2.4–3.4	2.2–3.7
Hyper score
Mean (SD)**	2.7 (0.6)	2.3 (0.4)S,E	3.1 (0.4)M,E	2.6 (0.4)E	3.7 (0.3)M,S,A
Range	1.4–4.2	1.4–3.3	2.1–4.2	1.9–3.3	3.4–4.2
SIRS score
Mean (SD)**	2.3 (0.7)	1.8 (0.4)S,A,E	2.6 (0.5)M,E	2.7 (0.4)M	3.3 (0.7)M,S
Range	1.1–4.1	1.1–2.8	1.1–4.1	2.3–3.7	2.2–4.1
EP score
Mean (SD)**	2.5 (0.7)	2.1 (0.5)S,E	2.8 (0.5)M,A,E	1.9 (0.3)S,E,	3.4 (0.6)M,S,A
Range	1.1–4.4	1.1–3.5	1.7–4.2	1.7–2.5	2.4–4.4
SRS-2 Total Raw Score*
Mean (SD)	99.3 (28.9)	81.1 (25.5)S,A,E	111.3 (22.1)M,E	103.4 (13.4)M,E	134.5 (17.3)M,S,A
Range	24–168	24–135	58–168	84–121	114–166

^*Measures used to derive the sensory subtypes

^**Mean, SD, and range were calculated using a 5-point Likert scale.

Categorical dependent variables (i.e. sex, race) were analyzed using a Fisher-Freeman-Halton’s exact test. Continuous demographic variables were analyzed using a one-way ANOVA with post hoc pairwise comparison of means. Significant difference in demographic characteristic is noted with superscript of each subtype mild (M), sensitive-distressed (S), attenuated-preoccupied (A), and extreme-mixed (E). See supplemental table 1 for additional information.

IQ Intelligent Quotient was calculated based on the Wechsler Abbreviated Scale of Intelligence, 2nd Edition or on the Kaufman Brief Intelligence Test, Second Edition, based on the study; SEQ 3.0 Sensory Experience Questionnaire, third edition; hypo hyporesponsiveness; hyper hyperresponsiveness; SIRS sensory interests, repetitions, and seeking behaviors; EP enhanced perception; SRS-2 Social Responsiveness Scale, second edition

Across all four studies, all children had a prior diagnosis of autism spectrum disorder, which was also confirmed by the research team using the Autism Diagnostic Observation Scale, 2nd edition (ADOS-2; n=107) or the Autism Diagnostic Interview-Revised (ADI-R; n= 11). Participants were required to speak and understand English with an IQ score greater than 60 on the Wechsler Abbreviated Scale of Intelligence, 2nd Edition or on the Kaufman Brief Intelligence Test, Second Edition. Exclusion criteria included: 1) tuberous sclerosis, Down syndrome, fragile X, a history of severe head injury, or hypoxia-ischemia; 2) visual or hearing impairments; or 3) inability to walk or stand. An additional exclusion criterion for three of the four studies (n= 104 participants) was for contraindications to MRI or physical challenges that prevent lying comfortably in the MRI scan. For the original studies, informed consent was collected from the caregiver of the participant and the participant provided assent to participate in the study.

Measures

The Sensory Experience Questionnaire, Version 3 (SEQ 3.0; Baranek, 2009) is a caregiver questionnaire that evaluates a child’s sensory response patterns. The SEQ 3.0 contains 105 questions with the first 97 questions designed using a 5-point Likert scale with higher scores indicating increased sensory response and the last 8 questions are narrative responses to better understand the child’s sensory response. The questions are designed to understand the sensory features of autistic children ages 2–12 years old in different scenarios throughout daily activities and routines. The SEQ has been used on individuals between 3 and 22 years old (Walz & Baranek, 2006). The SEQ is a reliable and valid measurement (Baranek et al., 2006; Little et al., 2011). Scoring of the scale was based on past psychometric analysis to both score sensory subscales (Ausderau et al., 2014b) and to assign respondents to subtypes based on previous latent profiles (Ausderau et al., 2014a).

The Social Responsiveness Scale–Second Edition (SRS-2; Constantino & Gruber, 2012) is a caregiver report measuring autistic traits and severity. This study utilizes the school age form designed for children 4–18 years old. The SRS-2 has 65 questions scored on a 4-point ordinal scale with higher scores representing increased autism features. The SRS-2 is a reliable and valid assessment (Bruni, 2014). This study used the SRS-2 total raw scores as a covariate to create sensory subtypes.

Bruininks–Oseretsky Test of Motor Proficiency, Version 2 short form (BOT-2; Bruininks & Bruininks, 2005) short form was used to measure motor skills in children ages 4–21 years old. The short form takes about 20–25 minutes to test the child on 14 subtests designed to evaluate fine and gross motor skills. One composite score was utilized with higher scores representing better motor skills that were normed for the child’s age and sex. The BOT-2 is a valid and reliable measurement based on normative data in non-autistic children (Bruininks & Bruininks, 2005; Fransen et al., 2014; Gharaei et al., 2019).

Social Communication Questionnaire (SCQ; Rutter et al., 2003) is a caregiver questionnaire created to understand social communication differences associated with autism spectrum disorder for individuals 4 years and older. The SCQ contains 40 yes/no questions with higher scores indicating more challenges with social and communication skills. The lifetime version of the SCQ was used, which has stronger psychometric properties than the current form (Wei et al., 2015) The SCQ is a valid and reliable measurement tool (Chandler et al., 2007; Corsello et al., 2007).

Repetitive Behaviors Scale-Revised (RBS-R; Bodfish et al., 2000) is a caregiver questionnaire created to understand the frequency at which repetitive behaviors occurred in the last month for individuals ages 3–48 years old. The RBS-R contains 43 questions that use a 4-point Likert scale with higher scores indicating increased repetitive behaviors. The 43 questions create one total score and 6 distinct subscales of stereotyped behavior, self-injurious behaviors, compulsive behaviors, ritualistic behaviors, sameness behaviors, and restricted behaviors. The RBS-R was reported to have medium reliability and high internal consistency (Lam & Aman, 2007). The RBS-R is not norm referenced requiring age to be controlled in the analysis.

Vineland Adaptive Behavior Scales, Version 2 (VABS-2; Sparrow et al., 2005) measures adaptive behavior from birth through adulthood. This study used the parent/caregiver rating form which is normed for ages 3–21 years old. The VABS provides a composite score (ABC) and 3 subdomains (communication, daily living skills and socialization). The VABS is a valid and reliable measurement of adaptive behavior with norm referenced scores (Pepperdine & McCrimmon, 2018).

Data analysis

Factor scores for the SEQ, 3.0 (hyperresponsiveness, hyporesponsiveness, SIRS, and enhanced perception) were estimated via confirmatory factor analysis, and subtypes were determined with a confirmatory latent profile analysis. Both factor score and latent profile analysis were done using Mplus version 7.4 (Muthén & Muthén, 2015) with code adapted from past research (Ausderau et al., 2014a, 2016). While the original model included multiple covariates, we adapted the model to include only age and SRS-2 scores (Ausderau et al., 2014a, 2016). Supplemental Table 1 demonstrates how the subtypes varied according to the subtype-formation variables (age, SEQ 3.0, and SRS-2 scores).

Data analysis was completed using R version 3.6.2 (R Core Team, 2019). Primary child outcome measures were analyzed using a one-way ANOVA with post hoc pairwise comparison of means. As the RBS-R is not age normed like the other measures examined, age was included as a covariate in the ANOVA and used adjusted means for the pairwise comparison. Where ANOVA model assumptions were not met, we applied Kruskal-Wallis followed by Dunn’s test for pairwise differences to analyze differences among the four sensory subtypes. P-values were adjusted using Benjamini-Hochberg correction (Benjamini & Hochberg, 1995). P-values were not adjusted for pairwise analysis due to small sample size, which increases the risk of type II error. Absolute value of Cohen’s d (Cohen, 1988) was used for effect size for all pairwise comparisons. In the case of missing data for the motor (n=6), social communication (n=1), or adaptive behavior scores (n=29), participants with missing data were omitted from the analysis, and only complete cases were utilized.

Additional follow-up statistical tests were completed to analyze the RBS-R subcategories and VABS-2 subdomains using a mixed model ANOVA. Post hoc pairwise differences were tested using estimated marginal means. RBS-R subcategories were scaled based on the number of questions. The RBS-R mixed model ANOVA included age as a covariate.

Results

Motor skills

Six participants were missing BOT-2 SF scores (mild [n=2], sensitive-distressed [n=3], and attenuated-preoccupied [n=1]). Figure 1 shows the standardized motor score for each subtype with Cohen’s d effect size reported between subtypes (Table 2). Motor skills were significantly different among the four subtypes. Pairwise comparison determined multiple significant differences between subtypes. Attenuated-preoccupied subtype had a large-sized significant difference with increased motor skills compared to the sensitive-distressed and extreme-mixed subtype. The mild subtype had a medium-sized significant difference with increased motor skills compared to the sensitive-distressed subtype. No other significant differences were observed.

Figure 1:

Motor skills was measured using the BOT-2 short form reporting the motor standard scores. Standard scores were reported with each subtype represented by a different color. Absolute value of Cohen’s d is reported for pairwise comparison.

Table 2.

Statistical analysis for primary child outcomes

	BOT-2 SF	SCQ Total Raw Score	RBS-R Total Score	VABS-2 ABC
F/H statistic	8.1 *	19.3 *	21.2 *	11.6 *
Pairwise comparison- p-values
Mild vs Sensitive-distressed	.046	.034	<.001	.066
Mild vs Attenuated-preoccupied	.343	.069	.002	.284
Mild vs Extreme-mixed	.082	<.001	<.001	.001
Sensitive-distressed vs Attenuated-preoccupied	.044	.543	.679	.990
Sensitive-distressed vs Extreme-mixed	.628	.024	.071	.044
Attenuated-preoccupied vs Extreme-mixed	.043	.231	.091	.134

^*p-value less than .05 with BH adjustment

Pairwise comparison reported the p-value between sensory subtypes for each primary child outcome measure. Bold values represent p<.05

Core Autism features

One participant was missing an SCQ score in the sensitive-distressed group. Social communication was significantly different among the four subtypes (Figure 2A and Table 2). The extreme-mixed subtype had a significantly large-sized difference scoring higher on the SCQ compared to the mild and sensitive-distressed group. A medium-sized significant difference was found with the mild group scoring lower on the SCQ compared to the sensitive distressed subtype. There were non-significant effects when comparing the attenuated-preoccupied to the extreme-mixed and to the mild group.

Figure 2:

Autism features were evaluated investigating social communication using the SCQ and repetitive behavior was measured using the RBS-R. A) SCQ total raw score, and B) Composite score of the RBS-R. The RBS-R has 6 subscale scores of C) compulsive behavior, D) restricted behavior, E) ritualistic behavior, F) sameness behavior, G) self-injurious behavior, and H) stereotyped behavior. Absolute value of Cohen’s d is reported for pairwise comparison. * One participant in the sensitive-distressed group is missing an SCQ score (n=344), RBS-R total and subscales have all participants (n=45).

All participants had complete RBS-R data. Repetitive behavior was significantly different among the four subtypes (Figure 2B and Table 2). The mild subtype had large-sized significant difference with lowest repetitive behavior total score compared to the sensitive-distressed, attenuated-preoccupied, and extreme-mixed subtypes. There was a non-significant, but with medium effect sized difference of the extreme-mixed reporting more repetitive behavior compared to sensitive-distressed and attenuated-preoccupied subtype.

To understand differences in sensory subtypes for repetitive behavior the six subscale scores were analyzed. Figure 2B–G shows the differences in repetitive behavior subscales for the four sensory subtypes (Supplemental Table 2). There was a statistically significant two-way interaction between sensory subtype and RBS-R subgroup in explaining repetitive behavior scores. The mild subtype reported significantly less prominent repetitive behaviors compared to the sensitive-distressed subtype with a large sized effect for all RBS-R subgroups except self-injurious (medium effect size). The mild subtype reported large-sized significant difference with less prominent repetitive behaviors for all RBS-R subgroups except self-injurious compared to the attenuated-preoccupied subtype. The extreme-mixed subtype consistently reported large effect sized and significantly more repetitive behaviors on all RBS-R subgroups compared to the mild subtype. Extreme-mixed also reported more repetitive behaviors compared to the sensitive-distressed with significantly large effect size for all RBS-R subgroups except ritualistic (small effect size) and restricted (not significant large effect size). In addition, the extreme-mixed reported more repetitive behaviors compared to the attenuated-preoccupied subtype with significantly large effect size for all RBS-R subgroups except ritualistic (small effect size). Comparing sensitive-distressed to the attenuated-preoccupied group, there was no significant differences between the two subtypes of the six RBS-R subgroups. However, restricted behavior had a medium effect size between sensitive-distressed and attenuated-preoccupied.

Adaptive behaviors

Twenty-nine participants were missing VABS-2 scores (mild [n=12], sensitive-distressed [n=13], attenuated-preoccupied [n=2], and extreme-mixed [n=2]). To determine differences in adaptive behavior the VABS-2 adaptive behavior composite score was used (Figure 3A & Table 2). The extreme-mixed subtype had a large-sized significantly lower adaptive behavior total score compared to the mild and sensitive-distressed subtype. The extreme-mixed subtype had a large-sized, but not significantly different VABS-2 ABC score compared to the attenuated-preoccupied subtype. The mild subtype had a medium effect sized difference of higher score on the VABS-2 compared to the attenuated-preoccupied group. The mild subtype had higher adaptive behavior total score on the VABS-2 compared to the extreme-mixed subtype.

Figure 3:

Adaptive behavior was measured using the VABS-2. A) Adaptive behavior composite score. The VABS-2 has 3 subdomains of C) communication D) daily living skills, and E) social skills. Absolute value of Cohen’s d is reported for pairwise comparison.

To understand differences within adaptive behavior the subdomains of daily living skills, communication, and social skills of the VABS-2 (Figure 3B–D & Supplemental Table 3). The two-way interactions between sensory subtype and VABS-2 subdomains were not significant, but due to small sample sizes the pairwise comparison was completed. The mild subtype had large-sized significant difference with higher score in all 3 subdomains of the VABS-2 compared to the extreme-mixed subtype. Mild also had significantly higher communication score compared to sensitive-distressed subtype (small effect size). The extreme-mixed had large-sized significant difference with lower scores on the VABS-2 social skills and daily living skills subgroup compared to the sensitive-distressed subtype. The extreme-mixed was also found to have a large-sized significantly lower VABS-2 daily living skills compared to the attenuated-preoccupied group. The extreme-mixed reported a large effect size, but not significant different for communication and social skills comparing to the attenuated-preoccupied subtype. There were no significant differences when comparing mild to the attenuated-preoccupied subtype, but social skills did report a medium effect size between the two subtypes. Similarly sensitive-distress and attenuated-preoccupied subtypes reported no significant difference but had a medium effect size for communication.

Discussion

The goal of this study was to apply an existing sensory subtype method (Ausderau et al. 2014a, 2016) to a novel cohort to understand how sensory subtypes differ in child outcome measures. This study identified distinctions among the four sensory subtypes (mild, sensitive-distressed, attenuated-preoccupied, and extreme-mixed) on measures of motor skills, core autism features and adaptive behaviors. The study ultimately found the extreme-mixed subtype consistently performed differently than the mild subtype. The mild subtype also found to have unique repetitive behavior findings, demonstrating lower repetitive behaviors in almost all subgroups compared to the sensitive-distressed, attenuated-preoccupied, and extreme-mixed subtypes. Additionally, the study differentiated the attenuated-preoccupied subtype having increased motor skills compared to the sensitive-distressed and extreme-mixed subtype when using a performance-based motor assessment. Application of sensory subtypes is one way to better capture how co-occurring sensory patterns impact child outcomes and can be used to tailor evaluations and interventions.

This study used an existing subtype methodology with a different sample to further understand the relationship of co-occurring sensory features with multiple child outcome measures. Overall, this study had a small-to-moderate sample size of 118 autistic participants. Most studies included approximately 200 participants (Ben-Sasson et al., 2008; Dwyer et al., 2020, 2022; Kadlaskar et al., 2022; Simpson et al., 2019) but varied from approximately 50 autistic participants (Lane et al., 2010; Uljarević et al., 2016) to greater than 300 autistic participants (Ausderau et al., 2016 [n=960]; Scheerer et al., 2021 [n=599], 2024 [n=495]; Tillmann et al., 2020 [n=332]). Based on previous sensory subtyping literature and the inclusion/exclusion criteria and age range of the current study, it was expected that there would be unequal distribution of the participants captured into the four subtypes, which was reflected in the current findings. Sensory subtypes typically found that more children were distributed into the sensory subtype with mild sensory differences and less children are distributed into the sensory subtype with more differences (Ausderau et al., 2016; Dwyer et al., 2020, 2022; Kadlaskar et al., 2022; Scheerer et al., 2024; Tillmann et al., 2020). In comparison to Ausderau (2014a, 2016), this study captured more autistic children in the mild and sensitive-distressed subtypes and less in the attenuated-preoccupied and extreme-mixed subtypes. Having more children categorized as such may be due to the in-person nature of the current study (compared to survey research), the IQ cutoff (IQs>60), or the demands of the MRI environment in three of four of the studies examined for this analysis. Additionally, we found that our attenuated-preoccupied subtype captured a narrower age range of younger participants. The majority of the sensory subtyping did not find a significant difference between age and distribution to sensory subtypes, but two papers did identify a significant relationship between age and sensory subtype (Ausderau et al., 2016; Scheerer et al., 2021). This study reflected the original findings from the Ausderau (2014a, 2016) which identified the attenuated-preoccupied captured younger autistic participants. Within the context of previous subtyping methodology, this sample reflects previous findings in terms of the subtype demographics without regenerating a new model to fit participants and supporting the clinical utility for practitioners.

This study further expanded into how the sensory subtypes differ in motor skills. We hypothesized the sensory subtypes would have similar motor skills (Lane et al., 2010). However, this study found the mild and attenuated-preoccupied subtypes had higher motor skills compared to the extreme-mixed and sensitive-distressed subtypes. In contrast to the previous subtyping studies that have investigated motor skills using the VABS motor domain which relies on caregiver-report (Lane et al., 2010; Scheerer et al., 2021), this study used the BOT-2 which is a performance-based assessment of motor skills (Bruininks & Bruininks, 2005). The VABS-2 motor domain is only designed for children younger than 7 years old (Sparrow et al., 2005), while the BOT-2 has larger age range up to 21 years old (Bruininks & Bruininks, 2005), which better represents this study sample. The use of a performance-based assessment may ultimately impact why this study found differences between sensory subtypes. Additionally, the extreme-mixed and sensitive-distressed subtype have elevated hyperresponsiveness and enhanced perception, which might be contributing to lower motor skills. Elevated sensory features have been associated with lower motor scores (Coll et al., 2020; Liu, 2013; Surgent et al., 2021). Chen and colleagues (2024) also found increasing hyperresponsiveness was associated with lower motor skills and lower participation outcomes. Hyperresponsive scores may be influencing both the extreme-mixed and sensitive-distressed subtype to have lower motor scores. However, additional research is still needed to better understand if increased hyperresponsiveness is solely driving motor skills or if co-occurring sensory patterns (e.g. role of elevated enhanced perception and hyperresponsiveness) further impacts motor skills.

The extreme-mixed subtype consistently had lower adaptive behavior, whereas the mild subtype consistently had higher adaptive behavior. This study originally hypothesized that the attenuated-preoccupied subtype would have the lowest adaptive behavior composite score, since this study applied the sensory subtype methodology by Ausderau and colleagues (2014a). Elevated sensory features have been associated with lower adaptive behavior (Neufeld et al., 2021), which impacts the child’s leisure participation (Askari et al., 2015), learning (Butera et al., 2020), social communication (Dakopolos & Jahromi, 2019; Dunn et al., 2016), emotional regulation (Erfanian et al., 2018), and executive functioning (Pastor-Cerezuela et al., 2020). This study, within the context of the broader sensory subtyping literature, further strengthens the relationship that lower adaptive skills are associated with an increase in sensory features (Scheerer et al., 2021; Tillmann et al., 2020).

Similar to the adaptive behavior finding, the extreme-mixed subtype had more reported autistic features and the mild subtype had lower reported autistic features, which was consistent with the hypothesis that the extreme-mixed subtype would have more reported autistic features (Kadlaskar et al., 2022; Scheerer et al., 2021, 2024; Tillmann et al., 2020). Sensory subtyping literature has found increased sensory features reflected more challenges with socialization and communication (Ben-Sasson et al., 2008; Kadlaskar et al., 2022; Scheerer et al., 2021, 2024; Tillmann et al., 2020) and restricted and repetitive behaviors (Ben-Sasson et al., 2008; Kadlaskar et al., 2022; Scheerer et al., 2021, 2024; Tillmann et al., 2020). The broader literature has found hyperresponsiveness is associated with restricted and repetitive behavior (Chen et al., 2009; Feldman et al., 2020; Zetler et al., 2022), while hyporesponsiveness has been reported associated with social communication challenges (Feldman et al., 2020, 2021). This study found the extreme-mixed subtype was distinct from the sensitive-distressed and attenuated-distressed subtypes, highlighting that the co-occurrence of elevated sensory features may be more likely to impact the child outcome than just one elevated sensory feature. Additionally, emerging neurodiversity affirming research found that non-harmful repetitive behaviors are useful and provide a way for autistic individuals to self-regulate (Kapp et al., 2019; Manor-Binyamini & Schreiber-Divon, 2019). This study highlights that autistic individuals with increased co-occurring sensory features may utilize repetitive behaviors more as an adaptive strategy for regulation within their environments (Gal et al., 2002; Zetler et al., 2022). The extreme-mixed subtype might benefit more from interventions supporting regulation and environmental modifications to support participation.

This study had multiple limitations that should be considered when interpreting the findings. The inclusion criteria limited the sample based on the child’s IQ score (>60) and, for three of the four studies, the ability to attempt an MRI scan. Therefore, sensory, motor, and adaptive behavior scores may differ in this sample compared to other study samples that used survey methodologies or did not include an MRI scan. Moreover, the IQ cutoff and MRI demands likely limited the representation of autistic children across the whole of the autism spectrum, which will be a key area for future research. Another limitation was the demographics of the study being predominantly Caucasian. Studies with demographics that more accurately mirror the demographics of the country are needed. In addition, this study was the compilation of data from multiple studies which led to missing VABS-2 data for 29 participants. Another limitation was the use of the SRS-2 as a variable in the model to generate subtypes, which may have impacted the evaluation of how the sensory subtypes differed in their SCQ, RBS-R, and VABS-2 scores, as these latter variables are known to be intercorrelated with SRS-2 scores (Bölte et al., 2008; Hooker et al., 2019; Pine et al., 2006; Uljarević et al., 2021). Further interpreting this study within the sensory subtype literature is challenging due to differences in participant demographics, sensory measures, and subtype generation (DeBoth & Reynolds, 2017; Simpson et al., 2019). Future research is needed in more representative samples and in determining subtypes both with and without measures of core autism features to better understand how sensory subtypes may differ on key outcome variables.

Conclusion

Understanding the relationship between sensory subtypes with child outcome measures is essential to explore the heterogeneity of sensory features, further adding to the subtype literature. This study applied an existing sensory subtyping methodology to a new cohort of children. The four sensory subtypes exhibited group differences in autistic children ages 6–18 years old with an IQ above 60. This study was the first to use a performance-based outcome of motor skills which found the attenuated-preoccupied subtype had increased motor skills compared to the sensitive-distressed and extreme-mixed subtype. The extreme-mixed subtype demonstrated lower adaptive behavior and more autistic features compared to the mild subtype. Application of sensory subtypes to the evaluation of autistic children could better target interventions to support autistic children’s daily participation. Specifically identifying children in the extreme-mixed subtype would help therapists personalize interventions to support motor skill development, self-regulation, and adaptive behavior.