Evaluation of the Effect of a Social Skills Program for Children and Adolescents With ASD

Participants were recruited by psychologists and psychiatrists from a Multidisciplinary Unit for Autism Spectrum Disorder, in accordance the 2000 Helsinki declaration.

A total sample of 94 participants was divided in two age-groups: children (8 to 12 years old) and adolescents (13-17 years old). These groups participated in the study over a three-year period (2017, 2018, and 2019). Each year the investigators recruited approximately 30 participants (children and adolescents). Subjects were paired according to age, gender and intelligence quotient; then each pair was randomized between the two groups (experimental group and waiting list control group) using R software with a fixed random seed.

The experimental group of each year received treatment from January to April. The waiting list control group received the treatment from April to July, after participating in the trial. For both groups, pre-treatment evaluation was performed in January and post-treatment evaluation was done in April. The treatment groups were conducted by one psychologist, two co-therapists (psychiatrists or mental health nurses), and with the assistance of two graduate students in psychology. Fifteen participants were excluded due to missing data.

Materials and design Diagnostic instruments The Autism Diagnostic Observation Schedule-2 was administered to all participants. To be included in the study, participants had to score above the cut-off point. Intellectual abilities and verbal comprehension were measured with the Wechsler Intelligence Scale for Children and Adolescents.

Outcome measurement instruments

Pre- and post-questionnaires for parents were administered to assess comorbid symptomatology:

Child Behavior Checklist (6-18): This instrument is a questionnaire filled out by parents to assess behaviors and emotional problems in subjects aged 1 year 6 months to 5-years old, and 6 to 18 years old. Both internalizing behaviors (e.g., anxiety, depression, etc.) and externalizing behaviors (e.g., aggression, hyperactivity, etc.) are evaluated. The present study focused specifically on: Anxious/Depressed, Social Problems and Thought Problems.

The Spence Children's Anxiety Scale: This instrument is a self-report that measures anxiety disorders in childhood and adolescence. It consists of 44 items that assess separation anxiety, obsessive-compulsive disorder, panic, social phobia, generalized anxiety, etc. It is evaluated from a Likert scale, with a maximum score of 114, average 57, and minimum 0.

Experimental Coding of Social Behavior: To assess social behaviors, the investigators coded 14 minutes of free playtime from session #2 and session #10 (last session). Free playtime consisted of participants playing with each other in a non-structured format, using different games that were offered or remaining alone if they wanted to. Therapists and co-therapists did not interact during free playtime. Coded videos were recorded with two different cameras located at two different fixed angles. Observational data corresponded to participants that attended the specific sessions (i.e., sessions 2 and 10) needed for these analyses and they allowed coders (based on audiovisual characteristics) sufficient coding time and angles to code the target behaviors. In order to obtain baseline observational data from the control group, this group was scheduled for two free playtime sessions (coinciding with sessions 1 and 2 from the experimental group) the second session was videocoded. To obtain follow-up data from the control group, another free playtime session was scheduled three months later.

Participants were informed about the recording, the research purpose of the videos, and the location of the cameras.

In order to quantitatively code social behaviors, an observational instrument was developed and administered. This instrument was inspired by a previous observational instrument and by the approach of social difficulties of the authors of the Autism Diagnostic Observation-2. This instrument allows the observer to quantify the number of occurrences of target behaviors according to pre-established categories. The scale includes six defined dimensions and category systems that meet the requirements of exhaustively and mutual exclusivity.

A systematic observation was performed to obtain a code matrix. Two coders analyzed and coded the entire set of videos. In order to assure reliability between coders, 20% of videos were coded for these propose and coders were blind to the randomized session number (sessions #2 and #10) that they were assigned to code. Two blinded coders performed a training phase until reaching a moderate-to-high (76-89%) Cohen's Kappa agreement rate, prior to coding the videos that were then analyzed for this study.

The observational design was nomothetic (several participants were observed), included follow-up (an initial session and the program's last session), and multidimensional (several dimensions of the observation instrument were considered suitable) (Nomothetic/follow-up/Multidimensional). As the therapeutic process extended to several sessions, the groups were considered as a plurality of units. We worked with two levels of response: verbal and non-verbal.

Procedures Parents were invited to attend an informative session in which the study was presented. Afterwards, signed consent and informed assent was obtained from each child. Sessions took place in a dedicated group-intervention room and had a length of 90 minutes per session.

The social skills intervention program was based on a consented adaptation of the Social Adjustment Enhancement Intervention of the University of California. The original design contains more than 20 sessions, but in order to be feasible in a Hospital-based service, we adapted the program to 10 sessions and eliminated the parent component. According to previous community research, evidenced-based intensive intervention programs are challenging to implement and they have long waiting lists. Therefore, there is a tendency to limit the length of group intervention programs to a fewer number of sessions in order to provide higher access to services. For this study, the number of sessions was decreased and the program was simplified in order to make it feasible to our hospital setting. Session structure was consistent and included initial greeting, free play time, didactic, joke telling, and closing activities. Topics of didactic sessions were related to social competence, such as empathy, talking about feelings, and solving social problems. Each session had the purpose of practicing different social competence skills that are typically affected in a young population with Autism Spectrum Disorder.

Data analysis The Statistical Package for the Social Sciences (SPSS v23.0) was used for statistical analyses. Descriptive analyses were conducted to characterize the sample. To test hypotheses, 2x2 ANOVA analyses were conducted, comparing differences between groups (experimental vs control) and time (baseline vs follow up). The amount of social behaviors and psychological symptoms were compared.

All statistical tests were performed with a bilateral contrast and the level of significance was set at 0.05, while marginally significant was understood to be below 0.10 and above 0.05, so as to better understand our results and potential discussion. Based on the previous literature, the analyses described above were replicated stratifying by verbal IQ and age.