Sensory integration therapy for children with autism and sensory processing difficulties: the SenITA RCT
Elizabeth Randell,Melissa Wright,Sarah Milosevic,David Gillespie,Lucy Brookes-Howell,Monica Busse-Morris,Richard Hastings,Wakunyambo Maboshe,Rhys Williams-Thomas,Laura Mills,Renee Romeo,Nahel Yaziji,Anne Marie McKigney,Alka Ahuja,Gemma Warren,Eleni Glarou,Sue Delport,Rachel McNamara
DOI: https://doi.org/10.3310/TQGE0020
Abstract:Background: Carers report unmet need for occupational therapy services addressing sensory difficulties in autism, yet insufficient evidence exists to recommend a therapeutic approach. Objectives: Our aim was to determine the clinical effectiveness and cost-effectiveness of sensory integration therapy for children with autism and sensory difficulties across behavioural, functional and quality-of-life outcomes. Design: We carried out a parallel-group randomised controlled trial, incorporating an internal pilot and a process evaluation. Randomisation utilised random permuted blocks. Setting and participants: Children were recruited via services and self-referral in Wales and England. Inclusion criteria were having an autism diagnosis, being in mainstream primary education and having definite/probable sensory processing difficulties. Exclusion criteria were having current/previous sensory integration therapy and current applied behaviour analysis therapy. Intervention: The intervention was manualised sensory integration therapy delivered over 26 weeks and the comparator was usual care. Outcomes: The primary outcome was problem behaviours (determined using the Aberrant Behavior Checklist), including irritability/agitation, at 6 months. Secondary outcomes were adaptive behaviour, functioning and socialisation (using the Vineland Adaptive Behavior Scales); carer stress (measured using the Autism Parenting Stress Index); quality of life (measured using the EuroQol-5 Dimensions and Carer Quality of Life); functional change (according to the Canadian Occupational Performance Measure); sensory processing (determined using the Sensory Processing Measure™ at screening and at 6 months to examine mediation effects); and cost-effectiveness (assessed using the Client Service Receipt Inventory). Every effort was made to ensure that outcome assessors were blind to allocation. Results: A total of 138 participants were randomised (n = 69 per group). Usual care was significantly different from the intervention, which was delivered with good fidelity and adherence and minimal contamination, and was associated with no adverse effects. Trial procedures and outcome measures were acceptable. Carers and therapists reported improvement in daily functioning. The primary analysis included 106 participants. There were no significant main effects of the intervention at 6 or 12 months. The adjusted mean difference between groups on the Aberrant Behavior Checklist - irritability at 6 months post randomisation was 0.40 (95% confidence interval -2.33 to 3.14; p = 0.77). Subgroup differences in irritability/agitation at 6 months were observed for sex of child (intervention × female = 6.42, 95% confidence interval 0.00 to 12.85; p = 0.050) and attention deficit hyperactivity disorder (intervention × attention deficit hyperactivity disorder = -6.77, 95% confidence interval -13.55 to -0.01; p = 0.050). There was an effect on carer stress at 6 months by region (intervention × South England = 7.01, 95% confidence interval 0.45 to 13.56; p = 0.04) and other neurodevelopmental/genetic conditions (intervention × neurodevelopmental/genetic condition = -9.53, 95% confidence interval -18.08 to -0.98; p = 0.030). Carer-rated goal performance and satisfaction increased across sessions (p < 0.001), with a mean change of 2.75 (95% confidence interval 2.14 to 3.37) for performance and a mean change of 3.34 (95% confidence interval 2.63 to 4.40) for satisfaction. Health economic evaluation suggests that sensory integration therapy is not cost-effective compared with usual care alone. Limitations: Limitations included variability of the intervention setting (i.e. NHS vs. private), delay for some receiving therapy, an error in administration of Vineland Adaptive Behavior Scales and no measurement of comparator arm goal performance. Conclusions: The intervention did not demonstrate clinical benefit above standard care. Subgroup effects are hypothesis-generating only. The intervention is likely to be effective for individualised performance goals, although it is unclear whether effects were in addition to standard care or were maintained. Future work: Further investigation of subgroup effects is needed. Trial registration: This trial is registered as ISRCTN14716440. Funding: This project was funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 26, No. 29. See the NIHR Journals Library website for further project information.