Examining Factors Related to Response to Treatment in Autism Spectrum Disorders

Background:  Research consistently demonstrates that up to 50% of children with autism spectrum disorders (ASD) enrolled in various forms of early intensive behavioural intervention (EIBI) demonstrate significant improvements (Howlin et al., 2009; Smith et al., 2010). The remaining children respond less optimally; the reasons for this variability in response to treatment remain unclear. Research examining the specific factors responsible is needed (Lord et al., 2005; Rogers & Vismara, 2008). One approach is to examine predictors of treatment response. Single-subject research has suggested that children who display higher levels of toy contact (TC), approach and stereotyped and repetitive verbalizations (SRV), along with lower levels of avoidance and stereotyped and repetitive non-verbal behaviours (SRNVB), respond better to an empirically supported intervention, Pivotal Response Treatment (PRT; Schreibman et al., 2009). Other PRT research also suggests that affect may play a role in predicting treatment response (Koegel et al., 1988).

Objectives:  To examine, using group data, whether previously established treatment response variables and affect predicted communication gains over 12 months of treatment in the PRT-based Nova Scotia EIBI program (Bryson et al., 2007).

Methods: Participating families were enrolled in a larger EIBI effectiveness study (Smith et al., 2010). Eligibility for the clinical program was based only on age under 6 years and a clinical diagnosis of ASD. Data were collected at baseline, and after 6 and 12 months of intervention, including assessments of cognitive ability (M-P-R), both receptive and expressive communication (PLS-IV, VABS) and autism symptoms (SRS). Age Equivalents (AE) were used as the unit of measurement for outcome variables. Behavioural predictors were coded from video-recorded interactions between the child and a therapist collected as probe data in the context of the intervention program. Using Landis and Koch’s (1977) criteria, only those video-coded predictor variables that achieved “substantial” reliability were included in the main analyses. Only approach (ICC = .51) and SRNVB (ICC = .31) were excluded using these criteria. Reliability of the remaining variables ranged from .63 to .75. Data from all three time points were available for N = 27 (M age = 51.26 mos, SD = 9.63; M cog AE = 26.89, SD = 9.66), and multi-level modeling was used to examine the hypotheses.

Results: Baseline cognitive AE, F (2, 22.77) = 4.00, p = .03, chronological age, F (2, 23.16) = 7.38, p = .00, and affect, F (2, 22.73) = 9.09, p = .00, were all associated with changes in expressive communication over time. None of the previously established treatment response predictors (i.e., TC, approach, avoidance, SRV and NVSRB) were associated with changes in either receptive or expressive communication.

Conclusions: Surprisingly, group data collected for children enrolled in a PRT-based intervention did not find that previously established treatment response variables predicted changes in communication scores over time. As found in other EIBI studies, both younger age and higher cognitive abilities at baseline did predict change in communication outcomes. Moreover, baseline levels of positive affect predicted increases in expressive communication over 12 months of intervention, a novel finding.