A Comparison of Sensory Subtyping Models in Children with Autism Spectrum Disorder

Background: Heterogeneity in sensory features in children with ASD is a barrier to the provision of customized and effective therapies (IACC, 2011). To date, two research groups have tested various subtyping models based on different parent-report measures and yielding different numbers of subtypes (range = 3-5 subtypes) across different cohorts (Ausderau et al, 2014; Lane et al, 2014; Lane et al, 2010; Lane et al, 2011).

Objectives: The aims of this study were: (1) test a 3-, 4- and 5-subtype model using the Short Sensory Profile (SSP) questionnaire with an independent and large national sample, and (2) compare the resulting classifications of children on the best-fitting SSP model to the established 4-subtype solution on the Sensory Experiences Questionnaire, version 3.0 (SEQ-3) (Ausderau et al., 2014) using a single cohort. From here, recommendations will be made for a unified model of sensory subtypes.

Methods: Extant data from the SEQ-3 and SSP were analyzed from participants with ASD (n=550), ages 2-12, as part of a national online survey. The SEQ-3.0 has 4 dimensional factors (i.e., hyporesponsiveness; HYPO, hyperresponsiveness; HYPER, sensory interests, repetitions, and seeking behavior; SIRS, and enhanced perception; EP); these factor scores were used in the subtyping analyses (Ausderau et al., 2014). For the SSP, we generated domain z scores (tactile sensitivity, taste/smell sensitivity, movement sensitivity, underresponsive/seeks sensation, auditory filtering, low energy/weak, visual/auditory sensitivity). Then, a latent profile analysis was conducted in MPLUS using the SSP z scores and relevant covariates; statistical fit indices (AIC, BIC, LMR, entropy, BLRT) were analyzed. The similarities and differences between the various subtype models and their theoretical underpinnings were qualitatively evaluated.

Results: Fit indices were mixed for 3-, 4-, and 5-subtype solutions, and we did not find conclusive support for a single subtype solution on the SSP, as was clear with the SEQ-3. The SEQ-3 subtypes included both high and low sensory severity subtypes, as well as two qualitatively different subtypes (one more hypo- and one more hyper-responsive). The SSP subtypes were strongly influenced by two domains: low energy/weak and movement, which are not well represented by items on the SEQ-3. A 3-subtype solution on the SSP was considered, but found to be theoretically less informative (i.e., followed a low, medium, high severity pattern, but no qualitative differences in patterns). The SSP 5-subtype solution showed good fit indices and was more consistent with Lane et al. (2011). Comparison of children’s classifications across the two measures within the same cohort of 550 children is in progress.

Conclusions: This study is the first to subtype using two sensory measures within the same cohort of children. A unified model of sensory subtypes in ASD would have significant benefits to the field by clarifying latent constructs and providing guidelines for assessment and targeted interventions; however, theoretical models driving the development of these assessments impact how sensory features are characterized in ASD, and may contribute to inconsistencies in findings across studies. Clinical consensus on the role of various constructs (e.g., sensory modulation vs. motor/praxis) in characterizing sensory subtypes is recommended.