Introduction
Persistent challenges in social communication and social interaction across multiple contexts are main characteristics of autism spectrum disorder (ASD) [
1]. Effective interventions that address these difficulties are desirable, and should aim to provide autistic individuals with skills that enable them to gain improved self-confidence and control over their social lives. This could, in turn, prevent negative outcomes associated with ASD such as comorbid mental disorders [
2], bullying [
3], loneliness [
4], and school absenteeism [
5]. Despite widespread demand, however, access to evidence-based interventions for individuals with ASD remains limited.
Social skills group training (SSGT) is an umbrella term for interventions applying socially instructive techniques and behavioral modification principles in group settings to improve social skills, typically used in the clinical management of children and adolescents with ASD in the normative intellectual range. The interventions vary considerably in their content and structure. A recent systematic review, including 18 trials comprising a total of 745 participants, indicated small to moderate effect sizes for parent- and observer-report (
g = 0.47 and 0.40, respectively) and non-significant effects for teacher-report [
6]. Two subsequently published large-scale randomized controlled trials (RCTs) also suggested that the gains in terms of social skills are limited [
7,
8]. The first of these trials was a multi-center trial performed at six German universities with specialized ASD outpatient clinics, in which 228 children and adolescents were randomized to 12 sessions of a group-based cognitive behavioral program or treatment as usual. This study reported a small effect size (
d = 0.33) three months after the intervention ended [
8]. The second trial, which is the largest to date, was a multi-center trial conducted at 13 child and adolescent psychiatry outpatient units in Sweden and coordinated by our research center. A total of 296 children and adolescents were randomized to a 12-week version of the SSGT ‘KONTAKT’ or standard care only. The effect, as measured by the parent-rated Social Responsiveness Scale–Second Edition (SRS-2) [
9] three months after completed training, was small (
d = 0.16) and only statistically significant for the adolescent subsample (
d = 0.33) and for girls (
d = 0.40) [
7].
The scientific evidence base for SSGT in ASD has evolved from initial small pilot trials conducted in university settings to more recent pragmatic multi-center trials performed in real-world clinical settings. Despite this progress, there are still major gaps in the literature. For instance, the SSGT interventions have rarely been tested against active comparators, outcome measures are often unblinded, and the key components and mechanisms of change are poorly understood [
6,
10]. Specially, one important and largely unanswered issue is to what extent more training provides additional benefits [
6]. By understanding the potential benefits of longer periods of training, service providers would be able to optimize the training in terms of both costs and effects. The duration of most SSGT programs investigated in previous studies was typically no more than 3–4 months. A few lasted only 4–5 weeks, while two programs lasted as long as 20 weeks [
10]. Even though the intensity varied across different programs, their relatively short duration is in stark contrast to many other common interventions for individuals with ASD, such as early intensive behavioral intervention for young children with ASD which typically involves many hours of training per week over years. Added benefits from longer periods of training would certainly make sense from a theoretical perspective. Prolonged periods of focused practice are generally necessary for humans to acquire and maintain complex skills [
11]. It is unlikely that social skills are exceptions. While such skills are acquired implicitly in typical development, individuals with ASD may need to learn them explicitly. Still, with enduring practice a transition from explicit to implicit processing might occur [
12], making the performance of the new skills more natural and effective over time. In particular, individuals with learning challenges in certain areas are likely to benefit from programs that allow for overlearning so that a skill can slowly become an integrated part of their repertoire [
13]. On the other hand, more intense interventions might increase the risk of fatigue, refusal, and drop-out. It should also be noted that long-term training does not necessarily mean more of the same. In particular, longer period of training allows for a gradual shift in content from the acquisition of new skills towards the application of these skills in situations of relevance for the participants’ everyday lives. Additionally, an incremental increase in the tailoring of the intervention for each unique participant is possible.
This study aimed to estimate the effects of a longer version of SSGT KONTAKT in children and adolescents with ASD. The program consisted of 24 weekly sessions, with a gradual increase in tailoring and focus on complex skills in real-world situations. We hypothesized that children and adolescents with ASD who received the program would show increased social communication skills and improved daily adaptation compared to a control group receiving standard care only. We also hypothesized that the intervention would reduce perceived stress, general symptom severity, and increase the global level of functioning.
Discussion
The present study examined the effects of 24 weekly sessions of SSGT KONTAKT compared with standard care for children and adolescents in the normative IQ range with ASD. The study was conducted in real-world clinical settings with a sample of children and adolescents typical of clients with ASD referred to child and adolescent psychiatry. The positive effect on parent-rated social communication and other autistic trait related social skills was large, both posttreatment and 3 months after completion. The point estimate suggested a substantially larger effect than what has been reported for shorter SSGT programs. Thus, this study endorses the potential benefit of prolonged training previously voiced by young individuals undertaking shorter social skills training programs [
31].
Parallel to the present study, a large-scale RCT evaluating the 12-week version of the KONTAKT program was conducted by the same research group [
7]. The procedures of the two studies were largely identical: they were conducted during the same time period within the same clinical settings, used staff with comparable training, recruited samples in the same manner, and included samples with highly similar characteristics. While the evaluation of the shorter version of KONTAKT led us to conclude that the intervention is feasible and safe in routine care, the estimated effects were modest and inconsistent. Three months after completion the participants undertaking the 12-week training on average had improved 11 points from baseline on the parent-rated SRS. As a comparison, those participating in the 24-week training had on average improved 23 points 3 months after completion. Thus, the average change score was more than twice the magnitude for the extended version. A few additional RCTs evaluating SSGT programs have included parent-rated SRS as an outcome measure, allowing for a rough comparison of the change scores in the treatment groups. A German multi-center study evaluating a 14-week SSGT-program derived from KONTAKT, reported a mean reduction of 15 points on the parent-rated SRS in the treatment group 3 months after the intervention [
8]. Two studies of an intensive 5-week program in 7–12-year-olds yielded an average improvement of less than 10 points [
32,
33]. The same was true for a small study of a 15-week social skills program based on cognitive behavioral principles [
34], and a pilot study comparing two different 4-week social skills interventions in a limited sample [
35]. A somewhat larger change (an average improvement of 14 points) was observed in a trial evaluating a multimodal anxiety and social skills intervention for adolescents [
36]. Finally, two separate RCTs have evaluated the 14-week Program for the Education and Enrichment of Relational Skills (PEERS) in children and adolescents aged 11–18 years [
37,
38]. The treatment group yielded an average improvement of 10 points in one of these studies [
38] and 22 points in the other [
37]. The latter was the only previous study we could find with results that were on a par with ours.
Group differences in teacher-rated SRS and secondary outcome measures did not reach statistical significance in the present study, with the exception of trainer-rated DD-CGAS posttreatment. However, the estimated improvement on these measures was comparable to those reported for the shorter version of KONTAKT. The previous evaluation of the shorter version indicated statistically significant effects in favor of the intervention on several of the secondary measures [
7], which highlights the limited precision of the present study and stresses the need for larger samples to better estimate the true effect of long-term training. The results on adverse effects in the present trial were comparable to those previously reported for the shorter version [
7] and underscore the importance of identifying subgroups that are unlikely to benefit from this particular intervention and monitor closely the group dynamics. Some adverse reactions (e.g., related to intragroup dynamics and gained insights) might be avoided by minor adjustments to the intervention. Overall, our data on adverse events serve to illustrate the value of such information, which has not been routinely monitored and reported in psychosocial intervention research [
40].
While our preliminary results suggest that there are benefits of extended training, there are also costs such as personnel and other resources as well as time and effort on the part of the client. Notably, a majority of clinics involved in the evaluation of KONTAKT were not prepared to implement the longer version for the present trial. To use available resources effectively, service providers must have access to information that enables them to make informed decisions about the optimal length of the intervention in terms of both costs and benefits. This would require not only estimates of higher precision than the ones we provide here, but also a more detailed “dose–response” characterization. This is a key component of the development and evaluation of pharmaceutical products, where any given dose provides a mixture of desirable and undesirable effects [
39]. Dose–response curves for both effectiveness and unwanted effects can help to clarify if a dose is beyond further beneficial effects, or if the risk of undesirable effects increase or decrease with the length or intensity of the intervention. So far, this has not been an integral part of the development of psychosocial interventions. While we believe that the “dose–response” analogue could be useful also in this research field, inherent differences between social skills training and pharmacological treatment must also be considered. A “higher dose” of social skills training, for instance, is not necessarily just more of the same. There is a natural progress in the acquisition of a new skill, from practicing the skill in a controlled and safe environment to applying the skill in everyday life. Longer training will inevitably allow for more applied and individualized training, once the basic skills are acquired. The more favorable outcome of the extended version of KONTAKT suggested by the present study might, therefore, partly be explained by qualitative rather than quantitative differences between the two versions of the intervention.
Future research should attempt to clarify further the mechanisms involved in SSGT and its effects. While longer duration allows for overlearning and consolidation of acquired skills, other aspects of the intervention might be just as important. It is, for instance, not clear if more intense treatments over shorter periods of time would have the same benefits [
32,
33]. Further, the role of parent involvement in maintaining the training and the acquired skills is an important focus of future investigation. It is, for instance, possible that the quality of parental involvement improves over time, which would favor longer interventions. To move beyond the current standard of social skills training, including the KONTAKT program, future research also needs to identify characteristics of the training where there still is potential for improvement. Modern approaches to skills acquisition in general (e.g., deliberate practice) would point to aspects such as high-quality feedback, methods to maintain focused practice and motivation, and strategies to move past plateaus [
11]. Another possibility would be to find ways to effectively shape the individual’s mental representation of social situations held in the long-term memory, making it possible to respond quickly and adequately in such situation despite the inherent limitations of the short-term memory [
41]. These are not explicit components of the KONTAKT program in its current form.
Limitations
The present study has some limitations that deserve to be addressed explicitly. First, the blinded teacher ratings were deemed unreliable due to a large amount of missing data. Thus, our conclusions are mainly based on unblinded parent ratings, and we cannot rule out that their assessments were biased. Teachers, on the other hand, reported that they sometimes did not have sufficient insight to be qualified to assess the abilities of individual students. This is unfortunate since gains observed by teachers would suggest that the intervention effects have generalized outside the home environment. Many studies in this field rely solely on unblinded measures which underscores the need for blinded assessment tools that are both valid and obtainable [
42,
43]. Second, no head-to-head comparison with a shorter SSGT program was included in the trial. We are fairly confident that the indirect comparison with the short version of KONTAKT presented in the discussion is valid, given the almost identical procedures used in the two trials. A major difference between the trials was that the present trial was conducted at only two of the 13 clinics involved in the evaluation of the short version, but we find it unlikely that this discrepancy had a major impact on the outcome. Third, as mentioned above, the relatively small sample size resulted in estimates of limited precision. Thus, the point estimates reported here must be interpreted with caution. In addition, the limited sample did not permit analyses of moderators and mediators of the training effect. The previous evaluation of KONTAKT suggested that the effect partly was moderated by age and sex [
7]. It would be of paramount interest to clarify if these differences also remain when the training period is prolonged, or whether different moderators apply. Finally, some of the interventions received as part of standard care (e.g., CBT and counseling) might have content that partly overlaps with KONTAKT. This might have led to an underestimation of the true effect of KONTAKT, although the low number of individuals receiving such interventions would suggest that any such effect was minimal. Similarly, four participants in the KONTAKT group (and none in the control group) received habilitation services such as heavy weighted blankets and cognitive aids for structuring one’s daily living. We have no reason to believe that this had a noteworthy impact on the results and found no such indications.
Acknowledgements
We thank the children, adolescents, parents and teachers who participated in the study. We are also thankful to the leads of child and adolescent psychiatry in Stockholm (Peter Ericson, Paula Liljeberg, Charlotta Wiberg Spangenberg, Karin Forler, Alkisti Nikolayidis Linderholm). Finally, we direct sincere thanks to colleagues, Åsa Garetzos, Jacob Björkvist, BUP Brommaplan and the administrative personnel at the child and adolescent psychiatry Stockholm units (BUP-KIND).
Compliance with ethical standards