An Intervention for Sensory Difficulties in Children with Autism: A Randomized Trial

Difficulty processing, integrating and responding to sensory stimuli has been described as a feature of autism spectrum disorders (ASD) since the disorder was first identified. Current estimates show that between 45 and 96 % of children with ASD demonstrate these sensory difficulties (Ben-Sasson et al. 2009; Lane et al. 2010) and sensory features (i.e.: hyper- or hypo reactivity to sensory input or unusual interest in the sensory aspects of the environment) are now included as one of four possible manifestations of ‘Restricted, Repetitive Patterns of Behavior, Interests, or Activities’ (American Psychiatric Association 2013). Families report that behaviors associated with difficulty processing and integrating sensory information create social isolation for them and their child, restrict participation in daily living activities (Schaaf et al. 2011) and impact social engagement (Hilton et al. 2007, 2010; Baker et al. 2008; Ashburner et al. 2008; Reynolds et al. 2011; Watson et al. 2011; Hochhauser and Engel-Yeger 2010). Consequently, interventions to address problems associated with difficulty processing sensory information, such as occupational therapy using sensory integration (OT/SI), (Ayres 1972, 1979, 1989) are among the most often requested services by parents of children with ASD (Mandell et al. 2005; Green et al. 2006; Goin-Kochel et al. 2009). There is emerging evidence regarding positive outcomes of OT/SI for children with ASD (Pfeiffer et al. 2011; Fazlioglu and Baran 2008; and see Schaaf 2011 for a review), however, methodological limitations preclude definitive conclusions. Hence, there is the need for a rigorous study of OT/SI that includes a manualized protocol and measurement of treatment adherence (Case-Smith and Arbesman 2008; Watling et al. 2011). Fortunately, a validated measure of treatment fidelity that describes the key principles of the sensory integrative approach and provides guidelines for best practice is now available (Parham et al. 2011, 2007; May-Benson et al., in press). Importantly, this measure provides a means to evaluate the fidelity of OT/SI in a clinical trial while assuring internal and external validity; a standard that is followed in the current study.

A second advancement that enhances the testing of this intervention is data showing that Goal Attainment Scaling (GAS) is a useful outcome measure for studies of interventions for ASD (Ruble et al. 2012). GAS is used to measure functional and meaningful aspects of an individual’s progress (Mailloux et al. 2007; Kiresuk et al. 1994). In autism, inherent heterogeneity often confounds findings, and thus, it is important to utilize outcome measures that are sensitive to individual outcomes. GAS has been shown to be a substantive and sensitive approach to evaluate progress on individualized goals in randomized controlled trials of psychosocial interventions for children with autism provided that specific quality indicators are present. These include that goals are independently rated, evaluated for equivalence between groups (comparability), scaled with equidistance, have measurable criteria, and clear, identifiable benchmarks (Ruble et al. 2012), recommendations that we followed in this study. A further strength of using GAS is that it provides a means to identify and measure outcomes that are parent-chosen and thus, meaningful to family. Given the increased emphasis on measurement of outcomes that are meaningful to the client or family (PCORI, http://www.pcori.org), the use of GAS provides a model for best practice.

Given the need for a rigorous randomized trial of OT/SI for individuals with ASD, the primary purpose of this study is to evaluate the efficacy of OT/SI following a manualized protocol on individual goal attainment (primary outcome) in comparison to usual care (UC). The secondary purpose was to evaluate the impact of this approach on the child’s sensory behaviors, adaptive behaviors and functional skills.

Our main goal was to evaluate the effects of the intervention on parent-reported, individual goal attainment using GAS (primary outcome). Secondarily, we evaluated the effects of the intervention on sensory behaviors, functional and adaptive behaviors using the PDDBI, PEDI, and the VABS II. Normality of primary and secondary outcome measures was evaluated prior to evaluating group differences. Scores on these secondary outcomes were not normally distributed and thus, non-parametric statistical tests were used to account for violation of the assumption of normality. In addition, although differences in baseline scores were not significantly different between the groups, on some outcomes differences within groups at baseline would be considered clinically meaningful. To account for variability in baseline scores, we used change scores in the analyses. Given the range of pre-treatment scores, within-person change was viewed as the most clinically relevant post-treatment score. An alpha of 0.05 was used for all comparisons. Data reported represents numbers of subjects with complete data sets (see reasons for attrition in Fig. 1).

The primary outcome for this study was the score obtained on GAS and we found that the children receiving the treatment scored significantly higher (p = 0.003) than the controls on goal attainment with an effect size of 1.2. Our finding is consistent with Pfeiffer et al. (2011) who found that children with ASD who participated in 6-week program of occupational therapy using sensory integration made significantly greater gains in their individualized goal attainment scale scores in comparison to those who received a fine motor intervention. Goal attainment scaling is a method to individualize and quantify goals for clinical populations. It has been used extensively in the clinical literature, and is recommended as an outcome measure in randomized control trials of psychosocial interventions in ASD (Ruble et al. 2012) such as the current study. Of note, the procedures we utilized in constructing scaled goals were consistent with the recently published recommendations of for its use (Ruble et al. 2012) including that: benchmarks were carefully constructed, goals were scaled at equal intervals, and rating of goals post intervention was based on parent interview by an independent evaluator blind to study condition. Although adherence to these conditions increases the reliability and validity of GAS (Ruble, et al. 2012) our findings must be interpreted with caution given that the parents were not blind to the intervention.

Two valuable aspects of GAS are that it provided a means to individualize goals based on each child’s individual needs and to identify areas that are important to the parents. Individualization is an important aspect of treatment given the heterogeneity and developmental nature of ASD as it is likely that each child has a unique set of pre-treatment characteristics that impact the choice of goals and outcomes (Stahmer et al. 2011). Further, utilization of goals that are important and meaningful to the parents assures that the primary stakeholders (families of children with ASD) needs are being addressed. This is an important aspect of any intervention and is in keeping with contemporary trends in intervention research (PCORI 2013; Melnyk and Morrison-Beedy 2012). For the current study, individual goals were based on parent-identified areas of need and assessment data that were established prior to treatment allocation. Many of the goals for the treatment and UC controls were similar in type as shown in Table 2. The most frequent goals were related to self-care including goals such as greater independence in dressing, feeding, toileting or grooming activities (treatment = 27, UC = 25). The second and third most frequent goals were about play (treatment = 16, UC = 15); and sitting for participation in activities such as synagogue or dinner (OT/SI = 12, UC = 9). There were some minor differences in goal type between the groups; the treatment group had more goals related to fine motor skills (OT/SI = 5, UC = 1) and participation in daily routines (treatment = 7, UC = 1); whereas the usual care group had more goals related to emotional regulation (treatment = 3, UC = 6), sleep (treatment = 1, UC = 3) and inappropriate touching (treatment = 0, UC = 4).

A second aspect of individualization that is important for ASD research and practice is that intervention strategies were tailored to each child’s assessed areas of need. In this study, the Data Driven Decision Making Process (Schaaf, in press; Schaaf et al. 2012; Schaaf and Benevides 2011; Schaaf and Blanche 2012) was used to individually tailor treatment activities to address the specific sensory-motor factors that were hypothesized to be affecting each participant’s goal attainment and functional skills. The treatment utilizes individually tailored sensory motor activities at the just right challenge with a playful approach to facilitate the child’s adaptation to promote function. Thus, the focus of treatment is on each individual’s sensory motor factors hypothesized to be impacting function, but importantly, the expected outcomes are functional behaviors. It is likely that this individualization was an important aspect of the positive findings of this study, and should be modeled in future studies.

In terms of functional behaviors, the children in the treatment group significantly decreased their need for caregiver assistance on self-care and social activities in comparison to the UC controls on the PEDI. In addition, they also showed a trend toward improvement in self-care and social skills. Thus, not only did the caregivers rate the children in the treatment group as needing less assistance from them in these activities, they also rated their skill level higher. These secondary outcome data should be interpreted with caution given that we completed multiple comparisons, however, these findings are consistent with the philosophy of the treatment approach—that adequate processing and integration of sensory information provides an important foundation for participation in functional, meaningful activities (Ayres 2005). Of note, the PEDI has been shown to have good reliability and validity as an outcome measure of functional behaviors (Nichols and Case-Smith 1996).

In terms of the sensory-motor factors that may underlie these findings, in this cohort many participants in both groups showed deficits in sensory modulation and praxis (measured via the Sensory Profile and the SIPT), and thus, the intervention was tailored to address these areas. Improvements in sensory modulation and praxis skills therefore, may underlie the gains seen in self-care and social skills. In regard to sensory modulation (over or under-reactivity to typical levels of sensation), the individually-tailored treatment for these children included a focus on activities that facilitated sensory modulation and regulation of behavioral responses to these sensory experiences. As the child’s ability to modulate sensation improved, it is likely that their behavioral regulation also improved and subsequently they were better able to participate in self-care and social activities. Interestingly, the subjects in the treatment group did show a decreasing trend of negative sensory behaviors on the Sensory Perceptual Behaviors Subscale of the PDDBI and this approached significance (p = 0.064), supporting this interpretation.

Similarly, it is possible that the intervention also had an impact on praxis. Praxis involves the ability to conceive of, plan, and organize goal-directed motor actions (Ayres 1989; Dziuk et al. 2007) and is related to adequate processing and integration of body sensory information (tactile, vestibular and proprioception). The intervention aimed to facilitate body awareness and praxis through individually-tailored, active, sensory-motor activities rich in tactile, proprioceptive and vestibular sensations. Many self-care activities such as dressing require adequate body awareness and thoughtful planning and execution of motor skills. Thus, it is possible that improved body awareness and praxis had a positive impact on ability to carry out these self-care tasks. Similarly, social interactions require constant processing of varied, often unpredictable sensations and the need for spontaneous responses (i.e.: praxis) and are likely affected by difficulty processing and integrating sensory information related to the body (Hilton et al. 2007, 2010; Baker et al. 2008; Ashburner et al. 2008; Reynolds et al. 2011; Watson et al. 2011; Hochhauser and Engel-Yeger 2010). Thus, as the children’s praxis improved, it is plausible that their ability to adaptively plan and carry out social interaction activities also improved and they became more independent. Further testing of these potential relationship is needed and in order to validate the idea that improvements in sensory modulation and praxis were related to improvements in functional skills, it will be important in future studies to specifically measure any changes in in these factors and their relationship to changes in functional skills such as self-care and socialization. In this study we were limited by the lack of instruments validated to measure change in these factors for this population within our 10-week intervention period. The SIPT is not recommended as a pre-post-test measure for periods shorter than 8–12 months (Ayres 1989) and its utility for shorter intervention periods has not been tested. Similarly, the Sensory Profile has not been validated for use as a pre-post assessment (Dunn 1999) although there is emerging data that test–retest reliability of certain sub-scores may be utilized in this way. Until these measures are validated for use as outcome measures in shorter intervention periods, or outcome measure to evaluate change in sensory functions are validated, future studies should consider a longer intervention period so that these assessments can be used to measure change in sensory-motor skills and determine their relationship to any changes in functional skills and adaptive behavior.

In terms of the proposed mechanism underlying the positive findings in this study, one explanation is that the intervention impacted neuroplasticity—the ability of the nervous system to be shaped and influenced by experience. It is well regarded in the neuro-developmental literature that early sensory motor experiences promote neuroplasticity and enhance the capacity of the brain to adapt to environmental challenges (Shonkoff and Phillips 2000; Ayres 1972; Dawson et al. 2012). Thus, it is possible that through the process of neuroplasticity that the children became more independent in their functional skills as their ability to process and integrate sensory information improved. Further testing of this assumption is needed using methods that evaluate nervous system activity pre and post intervention. There is some preliminary evidence that change in neural activities results from enriched environments. For example, Dawson et al. (2012) showed that more organized EEG activity occurred in children with ASD who also made gains in the Denver Early Start Program; and Miller et al. (2007) showed that electrodermal activity, a measure of sympathetic nervous system activity, showed a trend to decrease (expected direction) following a sensory-enriched intervention in subjects who were previously sensory hyper-reactive. An important next step in this research will be to measure changes in brain activity that may be concurrent with improvements in adaptive behaviors and individual goals as suggested by Schaaf et al. (2013).

In summary, our data provide preliminary support for the efficacy of a manualized intervention designed to address difficulties processing and integrating sensory information for children with ASD. We show improvements in our primary outcome—Goal Attainment as well as our secondary outcome measures showing improvements in self-care and social activities reflected by decreased caregiver assistance. These findings should be interpreted cautiously until they are replicated in a larger sample size. In addition, in future studies it would be useful to include additional outcome measures that rely on direct observation of goal attainment and sensory behaviors to provide further validation of GAS findings. It will be important to supplement parent reported data with direct observational measures. It will also be important to include a longer intervention period in future studies and follow-up testing to determine if the observed changes are maintained. Finally, although we randomized subjects based on autism severity and cognition, we were not able to include these strata in our analysis due to our sample size. Future studies would be strengthened by the inclusion of a larger sample so that impact of potentially confounding variables on treatment outcomes can be evaluated. Of note, almost all of our participants (30 of 32 children, or 94 %) demonstrated high severity of autism, and 22 or 68.75 % also had high IQ. It would be useful if future studies utilizing this intervention include children with low severity and/or low cognition to determine if the findings from this study are replicated with this sample. Similarly, our sample of convenience resulted in a sample with little ethnic diversity and future studies should make an effort to include participants from more diverse backgrounds. Despite these limitations, this study provides evidence that this intervention may be a useful adjunct to a comprehensive intervention program for individuals with ASD who have functional and behavioral challenges related to difficulty processing and integrating of sensory information.