3.1. Participant Characteristics
Participants included 27 males and 14 females 1 to 46 years of age (M = 14.4 ± 10.2 years (1.6–46.5)). One male and one female were excluded from the analyses due to a lack of usable scores on the ASIEP (see Feasibility below), leaving 39 participants in total. The sample was primarily Caucasian and non-Hispanic. The group had a range of cognitive abilities (SB-V ABIQ: M(36) = 58.9 ± 17.3, Mdn = 50, range (47–109); SB-V FSIQz score: M(30) = 56.98 ± 21.74, Mdn = 56.04, range (10.48–98.01)), behaviorial symptoms (ABC-C Total M(32) = 37.0 ± 27.3, range (0–98)), adaptive skills (Vineland-3 ABC M(38) = 55.4 ± 22.1, range (20–98)), autism symptoms (SCQ M(32) = 13.6 ± 6.9, range (3–26)); SRS-2 Total M(24) = 67.7 ± 12.1, range (47–88)), and clinical autism diagnoses (ASD 56% (22/39; 20 males, 2 females) (see Table 1
). Females had significantly higher IQ scores (p
< 0.05 for ABIQ and p
< 0.01 for FSIQ and all deviation IQ scores) and a lower rate of ASD (p
< 0.001). There were no significant differences between males and females in age, adaptive skills, behavioral problems, or total T-score on the SRS-2, although there were trends toward improved functioning across all measures in females.
3.2.1. ASIEP-3 Feasibility and Reliability
Feasibility. The ASIEP-3 was successfully administered and completed by all but two of the forty-one total participants enrolled in the study (with one non-compliance due to naptime and one unstandardized administration with the child requiring the parent in the room for completion), resulting in 39 ASIEP-3 Interaction assessments for analysis. Thus, the feasibility for scorable administration among FXS participants in a clinical research setting is estimated to be 95% or greater.
Administration Reliability. The senior psychologist (MB) provided feedback to the evaluators, and reliable, accurate administration was achieved after the three practice and two additional ASIEP-3 administrations. The most common areas of feedback were (1) adjusting the examiner’s language to the developmental age of the child, (2) providing directives during the Direct Cues phase at the appropriate intervals and frequency, and (3) ensuring that the participant was exposed to a minimum of three play activities before the switch to the passive phase.
Scoring Inter-rater Reliability. Using the senior psychologist (MB) as the standard, across different domains of the ASIEP-3, the ICCs with the bachelor’s-level RAs ranged from 0.24 to 0.85, with the widest 95% confidence interval at −0.35 to 0.64 and the highest and narrowest 95% confidence interval at 0.61–0.94. ICCs with the master’s level psychometrist (LC) and doctoral-level clinical psychologist (EC) ranged from 0.79 to 0.93, with the widest confidence interval at 0.43 to 0.94 for one subdomain, although all other CIs ranged from 0.73 (or above) to 0.98. All four of the ASIEP-3 score codes (Interaction, CIP, No Response, and Aggressive) were highly correlated among the three raters (MB, LC, and EC) (all r values > 0.70, all p values < 0.001), indicating robust agreement among raters.
3.2.2. ASIEP-3 Score Profiles
A total of 39 ASIEP-3 administrations were analyzed (26 males, 13 females). Overall, there was a wide range of performance in three of the four ASIEP-3 score categories (Interaction, CIP, and No Response), as well as the overall Autistic Interaction Score (AIS) (see Table 2
). There was a low incidence of Aggression scores throughout the dataset, resulting in a minimal distribution for the statistical tests. The highest average percentage of ASIEP-3 scores was in the Interaction category for both males (M = 36.5% ± 14.5%) and females (M = 48.7% ± 14.3%), followed by the Constructive Independent Play category (males M = 33.8% + 21.3 and females M = 42.8% ± 16.2).
Gender. When evaluating the differences between males and females in the ASIEP-3 categories and for the overall AIS score, males showed a significantly higher average percentage of No Response scores (p < 0.001) and overall AIS scores (p < 0.01) compared to females, and there was also a trend toward a significantly higher percentage in the Aggressive score domain among males (p = 0.084). Females showed a higher mean ASIEP-3 Interaction score (p < 0.05) and a lower ASIEP-3 No Response score (p < 0.001) compared to males. The distribution of AIS between males and females differed by such a magnitude that none of the correlations between AIS and the other variables remained significant when examined by gender.
Age. For the entire group, the Interaction score was positively correlated with age (rho(39) = 0.409, p < 0.01). While there was a negative relationship between age and AIS for the entire sample (rho(39) = −0.340, p < 0.05), this was no longer significant when examined by gender.
As the ASIEP-3 was originally developed for use up to age thirteen, we examined the ASIEP-3 scores by age group to determine if the participants outside of the original standardization range had reasonable score profiles. The group was divided into participants thirteen years old and younger (N = 22) and those aged fourteen years and older (N = 17). There were no significant differences in the distribution of any ASIEP-3 scores between the younger and older participants. When examining potential relationships between age and the ASIEP-3 scores for the younger and older age groups separately, there were no significant relationships. Finally, there were no significant differences in the distribution of gender or autism diagnosis between the younger and older age groups.
Cognitive Skills. There were weak to moderate, significant relationships between IQ and the ASIEP Interaction score (ABIQ rho(36) = 0.394, p < 0.05), the ASIEP No Response score (FSIQ rho(31) = −0.405, p < 0.05), and the ASIEP AIS (ABIQ rho(36) = −0.401, p < 0.05). This indicates that those with a higher IQ tend to have higher Interaction scores and lower No Response score and AIS. All deviation IQ scores (NVIQz, VIQz, and FSIQz) were negatively correlated with the ASIEP-3 No Response score (rho(30) = −0.376 to −0.415, p < 0.05) such that those with lower IQ scores tended to have more No Response codes during the ASIEP-3.
3.2.3. ASIEP Validity with ASD Diagnosis and other Measures
ASIEP scores were examined based on a clinical autism diagnosis (see Figure 3
). Among those with an autism diagnosis (FXS+ASD N
= 22), the average percentages of Interaction scores were significantly lower (M(22) = 34.8% versus M(15) = 49.9%, p
= 0.002), and the No Response scores were significantly higher (M(22) = 29.5% versus M(15) = 7.6%, p
< 0.001), compared to those with FXS−ASD. Therefore, while the FXS-ASD group earned half of their total score in the Interaction category, those with FXS+ASD earned just over one-third of their scores in the Interaction category. Furthermore, the FXS−ASD group had very low No Response scores (M(15) = 7.6%) compared to the FXS+ASD group who earned one-third of their codes in the No Response category (M(22) = 29.6%). Constructive Independent Play and Aggressive scores were not significantly different based on autism diagnosis (all p
ABC-C. There were significant positive relationships between ABC-C Social Avoidance and the two ASIEP-3 scores. Both the ASIEP-3 No Response score (rho(32) = 0.384, p < 0.05) and ASIEP AIS (rho(32) = 0.368, p < 0.05) were correlated with ABC-C Social Avoidance, indicating relationships between the parent-reports of Social Avoidance and the observed behaviors of social avoidance during the ASIEP-3 session.
SRS-2. While there were no significant correlations between the SRS-2 Total T-score and any of the ASIEP-3 scores, the SRS-2 Social Motivation T-score was negatively correlated with the ASIEP-3 Interaction score (rho(24) = −0.434, p < 0.05). As higher T-scores on the SRS-2 indicate greater autism symptomatology, an increase in Social Motivation deficits was expected to relate to a lower Interaction score.
SCQ. The SCQ total score was moderately correlated with ASIEP-3 AIS (rho(28) = 0.435, p < 0.05) and approached a significant negative correlation with the ASIEP-3 Interaction score (rho(28) = −0.356, p = 0.063).
Vineland-3. There were no significant correlations between any of the Vineland-3 scores and any of the ASIEP-3 scores when considering the entire sample, despite the Vineland-3 Socialization standard score being significantly higher among those without autism (FXS−ASD = M(13) = 70.62 + 26.41 versus FXS + ASD = M(24) = 53.58 + 17.10); t(35) = 2.382, p < 0.05). Among those with an autism diagnosis, the ASIEP-3 Aggression score was negatively correlated with the Vineland-3 Communication standard score (rho(24) = −0.415, p < 0.05) and approached a significant negative relationship with the Vineland-3 Socialization standard score (rho(24) = −0.377, p = 0.069). The Vineland-3 Socialization standard score was negatively correlated with the ABC-C Social Avoidance score (rho(45) = −0.364, p < 0.01) and all of the SRS-2 subscale T-scores (rho(43)= −0.441 to −0.616, all p < 0.01).
ADOS-2. Contemporaneous ADOS-2 scores were available for a subset (N = 10) of participants. All but one of the ADOS-2 administrations resulted in a score in the autism range. In this small homogenous sample, the ADOS-2 Social Affect score did not correlate with the ASIEP-3 scores but did correlate with the parent-report measure of social interaction (SCQ: rho(8) = 0.709, p = 0.049). ADOS-2 RRB total was strongly correlated with ASIEP-3 No Response score (rho(10) = 0.642, p = 0.045) and ASIEP-3 AIS (rho(10) = 0.635, p = 0.049). The negative relationship between the ADOS-2 Total score and the ASIEP-3 CIP score approached significance (rho(10) = −0.584, p = 0.073). Most likely due to the small, homogeneous sub-sample and limited score range, the ADOS-2 Comparison score was not significantly correlated with any ASIEP-3 score or other parent-report measures of social interaction.
3.2.4. Exploratory and Ad-hoc Analyses
Passive Phase Switch. When considering the ASIEP-3 Administration Checklist results by autism diagnosis (FXS−ASD or FXS+ASD), 87% of those without ASD noticed the switch compared to 58% of those with ASD. Among those that did notice the switch, 90% of those without ASD noticed it immediately, compared to 50% of those with ASD. Those who noticed the switch to the passive phase were significantly older (t(37)= −2.530, p = 0.016), and their average ASIEP-3 Interaction scores were significantly higher (t(37)= −2.506, p = 0.017).
Aggression. ABC-C (N = 32) item 4 (M=.75 ± 0.88; range (0–3)), item 10 (M= 0.94 ± 0.88; range (0–3), item 47 (M = 0.75 ± 0.88; range (0–3)), item 57 (M=1.03 ± 1.03; range (0–3)), and the sum of these four items (M = 3.47 ± 3.09; range (0–12)) showed the range of the FXS phenotype, despite very low ASIEP-3 Aggression scores. These scores were not significantly different based on autism diagnosis (all p > 0.05). Further, 15–30% of the participants were rated as moderate or severe in these aggressive behaviors (item 4 = 15.7%, item 10 = 28.1%, item 47 = 21.9%, and item 57 = 34.4%).