Interventions aimed at restricted and repetitive behaviours (RRBs) in Autism Spectrum Disorder: systematic review and meta-analysis

Background: Despite being considered a core feature of ASD, restricted and repetitive behaviours (RRBs) have received less attention, if compared to the domain of social interaction and communication, and less frequently targeted by interventions. This is surprising, given their role as major management challenges, obstacles to adaptive functioning and cause of distress for subjects and their families. We conducted a systematic, exhaustive, and up-to-date systematic review and meta-analysis of randomized controlled interventions (RCTs) aimed specifically at RRBs. To avoid methodological limitations found in other reviews, no limitations to the age of the sample, timespan of search and type of intervention were set. Methods: Web of Knowledge database (including Web of Science, MEDLINE®, KCI – Korean Journal Database, Russian Science Citation Index and SciELO Citation Index) was searched from inception up to January 1st, 2020. Randomized controlled trials in ASD individuals, specifically aimed at RRBs or both core domains were included, following PRISMA guidelines. In a systematic review we analysed the main characteristics of included studies, such as mean age, sample sizes, mean follow up duration, diagnosis of ASD, assessments of IQ and psychiatric comorbidities. Primary outcome of the meta-analysis was the mean reduction of RRBs; effect sizes reported as Hedges’ g and 95% CIs, calculated as differences, from baseline to endpoint, between two compared interventions. Assessments of biases, comprising publication bias, and cumulative analyses were also performed. Results: Overall, 80 studies (3114 subjects) were included in the systematic review and 46 studies (1339 subjects) in the meta-analytic phase. Included studies were published between 1992 and 2019, mean sample size was 40 patients (in intervention arm), with mean age of 10.5 years and average 19% female participants. Mean follow up was 4 months. IQ assessment was unclear in half of the studies, other psychiatric comorbidities were not disclosed in 61% of the studies. Risk of bias was low in 14 studies (17.5%). Overall effect size for interventions aimed at RRBs was small, but significantly beneficial ( g = -0.37, Cis -0.26 to -0.47), heterogeneity was moderate (I2 = 43.92%, p < 0.01). Subgroup analyses revealed similar results in the three subtypes of interventions analysed: pharmacological (g = -0.45, CI -0.26 to -0.64), psychotherapy and education (g = -0.42, CI -0.19 to -0.65) and complementary interventions (g = -0.25, CI -0.12 to -0.38). Differences were not significant among intervention types (p = 0.16). Inspection of forest and funnel plots revealed the presence of five outliers, which exclusion reduced heterogeneity significantly, but did not affected substantially the magnitude of overall and subgroup effect sizes. Results were also not affected by small-study effects, but publication bias was probably present since grey/unpublished literature was not searched. Conclusions: on the basis of current literature aimed specifically at RRBs, there is no robust evidence to favour any specific intervention for improving RRBs in subjects with ASD, even if small effects were detected for any intervention type analysed