Het doel van sociale vaardigheidstraining (oftewel SOVA-training) is om kinderen en jongeren met autisme sociale vaardigheden aan te leren die hen beter in staat stellen contacten aan te gaan met leeftijdgenoten en/of volwassenen. Hiermee wordt bedoeld dat indien kinderen/ouders dat wenselijk achten, een verbreding van het vaardigheidsrepertoire wordt nagestreefd. Nadrukkelijk wordt niet bedoeld om toe te trainen naar een maatschappelijk opgelegde norm.

Autistische kinderen hebben immers net als kinderen en jongeren zonder autisme behoefte aan contact maar zijn vaak minder sociaal vaardig waardoor dit niet lukt. Daarnaast missen leeftijdsgenoten en volwassenen vaak de vaardigheden om goed aan te sluiten bij hen. Ook kan het ontbreken aan randvoorwaarden het contact in de weg staan, zoals de beschikbaarheid van ruimtes die niet te vol en rumoerig zijn.

Gebrek aan contact staat groei en ontwikkeling in de weg. Vriendschappen, bestaand uit het samen dingen doen, gedachten en gevoelens delen en aardige dingen doen voor elkaar, zijn zeker niet vanzelfsprekend bij autisme.  SOVA-training richt zich op het trainen van dergelijke vaardigheden bij het kind met autisme.

SOVA-training wordt met regelmaat aangeboden in het reguliere onderwijs, het speciaal onderwijs, binnen jeugdzorg, jeugdteams, de zorg bij verstandelijke beperking, en binnen ggz-instellingen. Het varieert per school of instelling of SOVA-training gegeven wordt. SOVA-trainingen worden aangeboden voor verschillende leeftijdsgroepen, zowel voor kinderen in de hogere klassen van de basisschool als op de middelbare school en voor (jong)volwassenen. SOVA-trainingen worden veel toegepast om alledaagse sociale vaardigheden te trainen, waaronder oogcontact maken, laten merken hoe je je voelt, complimenten geven en afspraken maken met leeftijdgenoten. SOVA-trainingen kennen een brede verscheidenheid in hun aanpak waaronder groepstrainingen, e-health programma’s, programma’s met een cognitief-gedragstherapeutische aanpak of middels peer support (ESCAP 2020). Tegelijkertijd is er veel overeenkomst tussen verschillende trainingen, zowel qua doel, als qua vaardigheden waarop die zich richten, als qua technieken waarmee ze dat doen (modeling, rollenspellen, huiswerk, gezamenlijke bijeenkomsten, etc).

Voorbeeld van een SOVA-training voor jongeren met autisme is PEERS (Program for the Education and Enrichment of Relational Skills) voor jongeren van 12-18 jaar (Idris, 2021). De PEERS-training is gericht op het maken en behouden van vrienden, hetgeen aansluit op de behoefte van veel jongeren met autisme. Het gaat hierbij om gespreksvaardigheden waaronder het mengen in een gesprek en het afronden van een gesprek, het kiezen van geschikte vrienden en afspreken met vrienden, gepast gebruik van humor en het omgaan met afwijzing, meningsverschillen, geruchten of roddels (Kenniscentrum kinder- en jeugdpsychiatrie, 2021).

Een andere training is de ToM (Theorie of Mind) training, een sociale vaardigheidstraining voor kinderen van 9 t/m 12 jaar met autisme die moeite hebben zich te verplaatsen in de gedachten en gevoelens van anderen, en die moeite hebben met het begrijpen van en omgaan met emoties en/of angstig en agressief zijn in sociale situaties. De ToM-training beoogt de ontwikkeling van het sociaal en emotioneel denken te stimuleren en kent ook een blended versie, de ToM-e (ZonMW 2017).

Hoewel zowel ouders/verzorgers als clinici soms de wens hebben kinderen en jongeren te ondersteunen in hun ontwikkeling op het gebied van sociale en communicatieve vaardigheden, is de vraag in wetenschappelijk onderzoek of het volgen van SOVA-training daaraan bijdraagt in situaties buiten de training. De kwestie die daarbij ook speelt, is in hoeverre SOVA-training aanzet tot het aanpassen van gedrag zonder dat dit uit de interne motivatie van het kind voortkomt. Dat past niet in het huidige tijdsgewricht gericht op inclusie en het verminderen van stigma. En vriendschappen onderhouden `omdat het zo hoort` kan ook veel energie kosten en stress geven.

De werkgroep wil in deze module dan ook meer zicht krijgen op de meerwaarde van SOVA-training in vergelijking met het niet aanbieden hiervan.

1. Social performance (critical)

2. Interaction (critical)

3. Quality of friendships (critical)

4. Number of friends (important)

5. Social skills (important)

6. Quality of life (important), 7. Patient satisfaction (important)

Description of studies

Gilmore (2022) performed a systematic review and meta-analysis to evaluate the effectiveness of group social skills interventions at improving social functioning in adolescents with congenital, acquired or developmental disabilities. The electronic databases PubMed, CINAHL, Embase, Cochrane Central Register of Controlled Trials, PsycINFO, and Web of Science were searched with relevant search terms from inception to March 2020. Inclusion criteria were 1) RCTs evaluating group social skills interventions in adolescents (mean age 11-17 years) with acquired, congenital or developmental disabilities, 2) reporting of outcome measures addressing social knowledge, social competence or social participation, 3) availability of treatment manual for the intervention, and 4) comparison was no intervention or another intervention. Studies were excluded if 1) the group social skills training was delivered in schools only, 2) no English publication was available, and 3) the focus was on adolescents with a primary mental health disorder. A total of sixteen studies were included in the systematic review, of which twelve adhered to our selection criteria and PICO, and were therefore described and analyzed in the current literature analysis (Choque-Olsson, 2017; Jonsson, 2019; Corbett, 2016; Schohl, 2014; Laugeson, 2009; Matthews, 2018; Rabin, 2021; Shum, 2019; Yoo, 2014; Freitag, 2016; White, 2013; Vernon, 2018).

Ashori (2019) conducted an experimental study with pre-test post-test design and control group to examine the effect of video modeling on the social skills of children with ASD. A total of 24 male children aged 6-8 years with a diagnosis of ASD with high performance (IQ between 70-85) in the first or second grade of school were included. Children were excluded if 1) there was evidence of a neuro-developmental disorder except ASD, and/or 2) they currently followed a similar intervention program. Participants were randomly assigned to either the video modeling experimental group or the waitlist control group. This study was conducted at two elementary schools in Tehran, Iran, and did not receive funding from any public, commercial, or non-profit agency.

Corbett (2019) performed an RCT to examine the impact of SENSE Theater intervention on behavior and social cognition in youth with ASD. In total, 87 children aged 8-16 years with a diagnosis of ASD according to the DSM-5 and an IQ ≥70, participated. Children were excluded if, over the last six months, they displayed aggression defined as verbal or physical threats to harm other children or adults. Participants were randomly divided into the SENSE Theater intervention group or the waitlist control group. This study was conducted at the Vanderbilt University Medical Center, Nashville, Tennessee, USA. Funding was received from NIMH R34 MH097793 (Corbett), NICHD Grant to Vanderbilt Kennedy Center, a VKC Hobbs Discovery Award, and donations to SENSE Theater to assist some families with travel costs.

Dekker (2019) conducted an RCT to investigate the effectiveness of a manualized group social skills training for high-functioning children with ASD. A total of 122 high-functioning (IQ ≥80) children in the last two and half years of primary education with a diagnosis of ASD according to the DSM-IV-TR were included. Exclusion criteria were 1) physical condition affecting participation, and 2) child could not travel to the child mental health center for training. Participants were randomly assigned to the social skills training group, the social skills training with additional parent and teacher involvement group, or the care as usual group. The study was conducted at a child mental health center in Groningen, the Netherlands, and was funded by the Netherlands Organization for Health Research and Development (ZonMw). 

Ko (2019) conducted a study to expand the evidence of the START socialization program for adolescents with ASD. This study was performed in the same sample of participants as the study conducted by Vernon (2018), which previously examined results from the parent and self-report survey measures. A total of 35 participants were randomized to either the START intervention program or the waitlist control group. The study was conducted in Santa Barbara, USA and was funded by the Organization for Autism Research.

Luckhardt (2018) performed a study on a subsample of the multicenter RCT conducted by Freitag (2016). Children and adolescents aged 8-20 years with a diagnosis of ASD participated in the SOSTA-net trial (Freitag, 2016) and were randomized to receive either add-on group-based social skills training SOSTA-FRA or treatment as usual. Participants of the study were included in the Frankfurt subsample of the full SOSTA-net trial. A total of 33 participants were included. Exclusion criteria were 1) IQ <70, 2) schizophrenia, bipolar disorder, social phobia, obsessive compulsive disorder, major depressive disorder with suicidal ideation or any personality disorder as well as aggressive behavior 3) any severe neurological or medical condition interfering with group therapy, 4) group-based social skills training during the last 6 months prior to inclusion, and 5) in-patient treatment. The study was conducted at the Department of Child and Adolescent Psychiatry, University Hospital Frankfurt, Germany. Funding was received from the German Research Foundation.

Shih (2019) performed a multisite RCT to examine the effectiveness of a social skills intervention called Remaking Recess (RR) in elementary school-age children with ASD. Children were included if 1) their parents and primary classroom teacher provided informed consent to participate, 2) they met the ADOS research criteria for ASD, 3) were aged 5-11 years, and 4) if they were included in a general education classroom for at least 51% of the school day. No exclusion criteria were reported. A total of eighty children met the inclusion criteria and were randomly assigned to either the RR intervention group or the waitlist control group. The study was conducted at three large urban school districts: 1) Los Angeles Unified School District, 2) School District of Philadelphia, and 3) Rochester City School District. This trial was supported by a grant from the Health Resources and Services Administration of the US Department of Health and Human Services and NIMH K01MH1001099 (Locke).

Study characteristics of the included studies are shown in Table 1

Results

1. Social performance

Nine studies included by Gilmore (2022) reported on social performance (by using questionnaires), assessed by using the Social Responsiveness Scale (SRS) (range: 0-195) (Choque-Olsson, 2017; Corbett, 2016; Schohl, 2014; Freitag, 2016; White, 2013) or SRS-2 (range: 65-260) (Jonsson, 2019; Rabin, 2021; Shum, 2019; Vernon, 2018). In addition, Luckhardt (2018) also reported on social performance assessed by using the SRS. For all scales, higher scores indicate greater severity of social impairment. Choque-Olsson (2017), Jonsson (2019), Rabin (2021), Shum (2019) and Freitag (2016) provided change data from baseline to postintervention.

Data of two studies could not be pooled because the authors provided estimated marginal means (Corbett, 2016) or because data was provided based on if participants received the facial emotion recognition task or the biological motion task (Luckhardt, 2018).

• Corbett (2016) reported an estimated marginal mean (SD) of 73.53 (7.83) for the intervention group (n=17) and 80.23 (7.81) for the control group (n=16).
• For the facial emotion recognition task, Luckhardt (2018) reported a postintervention mean (SD) SRS Parent score of 92.6 (27.3) for the intervention group (n=17) and 94.5 (27.7) for the control group (n=14). Mean difference was -1.90 (95%CI -21.37 to 17.57) in favor of the intervention group, which was considered not clinically relevant. The authors reported a postintervention mean (SD) SRS Parent score of 88.0 (24.4) for the intervention group (n=17) and 96.6 (27.6) for the control group (n=14) for the sample that received the biological motion discrimination task. Mean difference was -8.60 (95%CI -27.14 to 9.94) in favor of the intervention group, which was considered not clinically relevant.

Figure 1 shows the pooled data. Both standardized mean differences (SMDs) are in favor of the social skills intervention. The SMD of the SRS Parent of -0.55 (95%CI -0.77 to -0.32) was considered clinically relevant.

Figure 1. The effect of social skills training on social performance in children and adolescents with ASD (postintervention)

_{Z: p-value of the pooled effect; df: degrees of freedom; I2: statistic heterogeneity; CI: confidence interval}

2. Interaction

Three studies reported on the outcome measure interaction (by observation), assessed by using the Peer Interaction Paradigm (PIP) (Corbett, 2019), the Playground Observation of Peer Engagement (POPE) time-interval coding system (Shih, 2018) and by performing video-recorded conversations (Ko, 2019).

• Corbett (2019) reported on interaction, defined as cooperative play and verbal interaction. The authors performed a 20-minute semi-structured interaction with the participant with ASD. Continuous timed-event coding of cooperative play (= percentage of time the participant was engaged in a reciprocal activity for enjoyment that involved the participation of ≥2 children) and verbal interaction (= engagement between ≥2 children including the participant with ASD that begins with a verbal overture and continues in a reciprocal sequence of to-and-from communication) was conducted.
• Shih (2018) reported on interaction, defined as joint engagement. The authors used the POPE time-interval coding system that identifies durations of joint engagement (e.g., playing a game or having a conversation) with peers.
• Ko (2019) reported on interaction, defined as questions asked, positive facial expressions and mutual engagement. Video-recorded conversations were systematically coded for each of these variables.

Table 2. Results on interaction

3. Quality of friendships

Two studies included in Gilmore (2022) reported on quality of friendships, assessed by using the Friendship Quality Scale (FQS) (Schohl, 2014; Laugeson, 2009). The FQS is a self-report measure to assess the quality of children’s and early adolescents’ friendships. Both studies used different versions of the FQS. Schohl (2014) used the FQS (range: 0-23) that consists of 23 yes/no questions from five different subscales. Laugeson (2009) used the FQS (range: 23-115) that consists of 23 items each answered on a 5-point Likert scale ranging from 1 ‘not true’ to 5 ‘really true’. For both versions, higher scores reflect better quality friendships.

• Schohl (2014) reported a mean (SD) FQS score of 82.45 (15.41) for participants in the social skills intervention group (n=29) and 82.65 (19.42) for participants in the control group (n=29). Mean difference was -0.2 (95%CI -9.22 to 8.82) in favor of the control group. This difference was considered not clinically relevant. 
• Laugeson (2009) reported a mean (SD) FQS score of 17.2 (4.0) for participants in the social skills intervention group (n=17) and 16.6 (4.6) for participants in the control group (n=16). Mean difference was 0.60 (95%CI -2.35 to 3.55) in favor of the social skills intervention group. This difference was considered not clinically relevant.

4. Number of friends

Five studies included in Gilmore (2022) reported on number of friends, assessed using the Quality of Socialization Questionnaire (QSQ) or the previous version, namely the Quality of Play Scale (QPQ) (Schohl, 2014; Laugeson, 2009; Matthews, 2018; Shum, 2019; Yoo, 2014). The QSQ is a 12-item self- and caregiver-report measure adapted from the QPQ, of which two items assess the frequency of hosted and invited get-togethers over the previous month. Higher scores indicate participation in more hosted and invited get-togethers with friends. Results are presented in Table 3.

Figure 2 shows the pooled data per assessment measure and subscale. For the means of this guideline, raw scores as presented in the figure were transformed (+/-). All mean differences are in favor of the intervention, except one regarding QPQ Adolescent total get-togethers which is in favor of the control group.

Table 3. Results on the frequency of hosted and invited get-togethers

Figure 2. The effect of social skills training on the frequency of hosted and invited get-togethers in children and adolescents with ASD (postintervention)

_{Z: p-value of the pooled effect; df: degrees of freedom; I2: statistic heterogeneity; CI: confidence interval}

5. Social skills

Nine studies included by Gilmore (2022) reported on social skills, assessed by using either the social parent and social teacher subscales of the Adaptive Behavior Assessment System to measure everyday skills related to social functioning (ABAS-II) (Choque-Olsson, 2017; Jonsson, 2019; Corbett, 2016; Shum, 2019) or the social parent, teacher and adolescent subscales of the Social Skills Rating Scales (SSRS)/the Social Skills Improvement System Rating Scales (SSIS-RS) (Schohl, 2014; Laugeson, 2009; Matthews, 2018; Rabin, 2021; Vernon, 2018). In addition, two other studies reported on social skills, assessed by the Autism Social Skills Profile (ASSP) (Ashori, 2019) and the SSRS (Dekker, 2019). For all scales, higher scores indicate better social skills. Choque-Olsson (2017), Jonsson (2019) and Shum (2019) provided change data from baseline to postintervention.

Data of Corbett (2016) was not pooled because they provided estimated marginal means. The authors reported an estimated marginal mean (SD) of 4.61 (2.22) for the intervention group (n=17) and 2.88 (2.27) for the control group (n=16).

Figure 3 shows the pooled data per assessment measure and subscale. For the means of this guideline, raw scores as presented in the figure were transformed (+/-). All standardized mean differences (SMD) are in favor of the social skills intervention. Two SMD of -0.53 (95%CI -1.47 to 0.42) for the ABAS-II Social Teacher subscale and -2.71 (95%CI -3.87 to -1.55) for the ASSP were considered clinically relevant. The others were considered not clinically relevant.

Figure 3. The effect of social skills training on social skills in children and adolescents with ASD (postintervention)

_{Z: p-value of the pooled effect; df: degrees of freedom; I2: statistic heterogeneity; CI: confidence interval}

6. Quality of life, 7. Patient satisfaction

None of the studies reported on the outcome measures quality of life and patient satisfaction.

Level of evidence of the literature

1. Social performance

The level of evidence regarding social performance was downgraded by three levels to very low because the unblinded nature of all studies (risk of bias: -1), conflicting results (inconsistency: -1), and the confidence interval is crossing the border of clinical relevance (imprecision: -1).

2. Interaction

The level of evidence regarding interaction was downgraded by three levels to very low because the unblinded nature of all studies (risk of bias: -1), the low number of included patients and the wide confidence intervals of individual studies (imprecision: -2).

3. Quality of friendships

The level of evidence regarding quality of friendships was downgraded by three levels to very low because the unblinded nature of all studies (risk of bias: -1), the confidence intervals are crossing the borders of clinical relevance and the low number of included patients (imprecision: -2).

4. Number of friends

The level of evidence regarding number of friends was downgraded by two levels to low because the unblinded nature of all studies (risk of bias: -1) and the low number of included patients (imprecision: -1).

5. Social skills

The level of evidence regarding social skills was downgraded by two levels to low because the unblinded nature of all studies (risk of bias: -1), and the confidence intervals are crossing the border of clinical relevance (imprecision: -1).

6. Quality of life, 7. Patient satisfaction

None of the studies reported on the outcome measures quality of life and patient satisfaction and could therefore not be graded.

A systematic review of the literature was performed to answer the following question:

What is the effectiveness of (add-on) social skills training compared with waitlist control, placebo attention or no add-on social skills training for children and adolescents with autism spectrum disorder (ASD)?

Relevant outcome measures

The guideline development group considered social performance, interaction, and quality of friendships as critical outcome measures for decision making and social skills, number of friends, quality of life, and patient satisfaction as important outcome measures for decision making.

A priori, the working group did not define the outcome measures listed above but used the definitions used in the studies.

The working group defined a 25% difference for dichotomous outcomes (0.8 ≥ RR ≥ 1.25), and 0.5 SD (baseline) or -0.5 > SMD > 0.5 for continuous outcomes as a minimal clinically (patient) important difference.

Search and select (Methods)

The databases Medline (via OVID), Embase (via Embase.com), and PsycINFO were searched with relevant search terms from 2017 until 11 July 2023. The detailed search strategy is depicted under the tab Methods. The systematic literature search resulted in 216 hits. Studies were selected based on the following criteria:

• Systematic reviews (searched in at least two databases, detailed search strategy, risk of bias assessment and results of individual studies available), or randomized controlled trials (RCTs);
• Studies including twenty or more (ten per arm) participants;
• Studies with a maximum follow-up of six months;
• Studies according to the PICO; and
• Studies with English language full-text publication.

58 studies were initially selected based on title and abstract screening. After reading the full text, 51 studies were excluded (see the table with reasons for exclusion under the tab Methods), and seven studies were included.

Results

Seven studies (one systematic review and six RCTs) were included in the analysis of the literature. The assessment of the risk of bias is summarized in the risk of bias tables. Blinding of participants was not included in the risk of bias assessment performed by Gilmore (2022), but for the development of this guideline blinding was taken into consideration.