Of the 82 participants who completed the study, 75 completed ADOS assessments before and after 6 months of treatment. There was a significant improvement in the ADOS total calibrated severity scores (CSS) of these participants (M=-.56, SD = .17, Sum(x) = 42, p = 0.003, Fig. 1). Separating the ADOS CSS into social affect (SA) and restricted and repetitive behavior (RRB) components revealed that changes were driven by large improvements in ADOS SA CSS (M=-.49, SD = .18, Sum(x) = 37, p = 0.001) and weak improvements in ADOS RRB CSS that were not significant (M=-.47, SD = .26, Sum(x) = 42, p = 0.08).
Next, we performed a multiple regression analysis to determine whether pre/post treatment changes in ADOS CSS were associated with the age, final dosage, and initial ADOS CSS of the participants. The results revealed that only initial ADOS CSS were significantly associated with change in the total ADOS CSS (β = − .37, p = .002; Fig. 2.) such that participants with higher initial ADOS CSS (i.e., more severe initial symptoms) exhibited larger improvements. Age (β = .03, p = .78) and final dosage (β = .03, p = .8) were not significantly associated with changes in the ADOS CSS. Equivalent findings were also apparent for the ADOS SA CSS and RRB CSS (see supplementary material for detailed regression analysis of all measures).
Of the 82 subjects who completed the study, parents of 61 completed the SRS-2 questionnaire before and after treatment. Parents reported a significant improvement in core ASD symptoms of children who completed treatment (M=-3.29, SD = 1.13, Sum(x) = 201, p = 0.043, Fig. 3). This was true for both the social sub-scale scores (M=-2.51, SD = 1.19, Sum(x) = 153, p = 0.038, Fig. 3) and the RRB sub-scale scores (M=-2.88, SD = 1.14, Sum(x) = 176, p = 0.014). A multiple regression analysis revealed that pre/post changes in SRS scores were significantly associated with the initial social (β = − .49, p < .001) and RRB (β = − .39, p = .003) scores both not with age (β = .04, p = .72) or final dosage (β = .08, p = .53). Hence, higher initial SRS scores predicted larger improvements following treatment.
Of the 82 subjects who completed the study, 76 completed the Vineland questionnaire before and after 6 months of treatment. There was a significant improvement in total Vineland scores in children who completed treatment (M = 4.37, SD = 1.18, Sum(x) = 332, p < 0.001, Fig. 4). Improvements were apparent in the communication (M = 4.37, SD = 1.61, Sum(x) = 332, p = 0.008), daily living (M = 4, SD = 1.47, Sum(x) = 305, p = 0.007), and socialization (M = 5.66, SD = 1.5, Sum(x) = 430, p < 0.001) sub-scales. A multiple regression analysis revealed that pre/post treatment changes in the socialization sub-scale were significantly associated with initial socialization sub-scale scores (β = − .41, p < .001), but not with age (β = .06, p = .26) or dosage level (β = − .01, p = .65). Participants with lower initial socialization scores exhibited larger improvements. Changes in other sub-scales of the Vineland questionnaire were not associated with any of the covariates.
Of the 82 participants who completed the study, 76 completed cognitive assessments before and after 6 months of treatment. Equivalent analyses did not reveal any significant impact of treatment on any of the cognitive subtests: Block design perceptual (M = .13, SD = .25, Sum(x) = 10, p = 0.65, Fig. 5), Matrix perceptual (M=-.14, SD = .19, Sum(x) = 11, p = 0.5), Vocabulary verbal (M = .29, SD = .28, Sum(x) = 22, p = 0.37), Similarities & Information verbal (M = .04, SD = .22, Sum(x) = 3, p = 0.91), or Coding processing speed (M = .22, SD = .18, Sum(x) = 11, p = 0.24).
Next, we examined the relationships between the SRS, Vineland, and ADOS, which measure similar ASD symptom domains in different manners (Fig. 6). There were significant correlations between scores from the socialization sub-scale of the Vineland and the SRS social scale both before (r(55)=-0.42, p = 0.001; Fig. 6) and after treatment (r(55)=-0.44, p < 0.001). However, there were no significant correlations in pre-post treatment changes across the two scales (r(55)=-0.015, p = 0.27) indicating low reliability in parental reported changes across the two measures.
Similarly, there were significant correlations between the ADOS-2 SA CSS and the Vineland socialization sub-scale scores before (r(67)=-0.28, p = 0.02) and after (r(67)=-0.3, p = 0.01) treatment, but there was no correlation in pre-post treatment changes across the two scales (r(67)=-0.07, p = 0.56). There were no significant correlations between the ADOS-2 SA CSS and SRS social scores before (r(54)=-0.14, p = 0.3) or after (r(54)=-0.15, p = 0.28) treatment and no correlation in pre-post treatment changes across the two measures (r(54) = 0.04, p = 0.77). This demonstrates low reliability between parental and clinical reports of changes in the severity of social symptoms following treatment.