“Less stress”: a pilot study on a cognitive behavioral treatment program for anxiety in children with autism spectrum disorders

Feasibility was measured by collecting feedback from therapists and users after the treatment was completed. The therapists completed questionnaires adapted for “Less stress” after every session. They evaluated to what extent the child had carried out the assigned home assignment (rated from 0 = not at all to 6 = to a very large extent), and a self-rated adherence (3 items) / competence (1 item) evaluation (rated from 0 = not at all to 4 = to a very large extent, see Table 2 for the questions/ratings).

The participating children and parents also completed questionnaires adapted for “Less Stress” after the end of treatment, including both ratings of user satisfaction and open-ended questions like “What I liked the most/the least with “Less Stress”, (see Table 3 for the questionnaire ratings).

As previously noted, the children were diagnosed with ASD at their local CAMHS.

The Social Responsiveness Scale; SRS (34) was used as a measure of ASD severity. SRS is a validated questionnaire that assesses behavioral and social-communicative traits (35) with a total score that has been justified in factor analyses (36). The SRS scores are reported to be continuously distributed in the general population, but it is advised to consider a diagnostic interview for ASD with a t-score > 70 (34).

To assess comorbidity and school refusal, mothers were interviewed with the revised version of the semi-structured Schedule for Affective Disorders and Schizophrenia for School-Aged Children: Present and Lifetime Version (K-SADS-PL DSM-5) (37) at baseline and after treatment. One K-SADS item measures school refusal and is scored as present if the child: “Protests intensely about going to school, or sent home or refusal to go at least one day per week and/or persistent reluctance or refusal to go to school”.

Satisfactory convergent and discriminative validity and acceptable values for inter-rater reliability have been reported for the former US and Nordic versions of K-SADS-PL (38-40). For the DSM-5 version of K-SADS-PL, a recent Icelandic study found inter-rater reliability estimates ranging from fair to excellent (kappa = 0.57-0.90), with most diagnoses in the excellent (> 0.75) range (41).

The second author, an experienced child and adolescent psychiatrist, conducted the parental interviews at baseline. She knew that the children had been diagnosed with ASD but was blind to questionnaire ratings of anxiety symptoms. The interviews were completed by telephone due to large geographical distances, an administration method previously found to be valid in differentiating children with and without anxiety disorders (42). The same child and adolescent psychiatrist conducted interviews after treatment. As such, the diagnostic procedure was not blind, but baseline interviews were not available when the after-treatment interviews were conducted.

Children’s Global Assessment Scale; CGAS, is a clinician-rated tool which was used to indicate the lowest overall level of the child’s psychosocial functioning (at home, at school, with peers) (43). Scores range from 0 to 100 with high scores indicating better functioning. CGAS is divided into 10-point intervals with a description of the child’s level of functioning for each interval. Moderate to excellent inter-rater reliability has been found when used with untrained- [.73] and expert [.92] raters, respectively (44). The child and adolescent psychiatrist rated CGAS after conducting the K-SADS interviews.

Child anxiety symptoms were measured with the Revised Child Anxiety and Depression Scale (RCADS) (33), using the approved Norwegian translation (available at https://www.childfirst.ucla.edu/resources/). Symptoms were rated at baseline and after treatment by both parents and children. The RCADS has been developed to correspond more closely to selected DSM-IV anxiety disorders and includes a scale for major depression. Good test-retest reliability has been established (33) and convergent and discriminant validity as well as good internal consistency has been demonstrated for the RCADS in a clinical sample (45). The developer established ‘clinical thresholds’ for the overall RCADS-score using the anxiety disorders interview schedule for DSM-IV, child and parent versions (46) as a comparison, and normative data for the subscale thresholds. A t-score of 65 and 70 means that the score is roughly in the top 7% and 2%, respectively, of scores of un-referred young people of the same age (described as borderline clinical- and clinical threshold by the developer) (45). A review found the RCADS to show robust internal consistency reliability in different assessment settings, countries, and languages (47).

The Child Anxiety Life Interference Scale (CALIS) (48) was rated by both parents. CALIS is a measure of life interference due to anxiety experienced by the child (9 items) and parents (16 items, including items on the interference experienced by the parents in their own life). In children with ASD, anxiety has been found to impact upon the child’s engagement in activities both in and outside of home and upon parent life, as rated by CALIS (10). CALIS is scored from 0 (not at all) to 4 (a great deal). CALIS was found to have good internal consistency, moderate-to-high test re-test reliability, significant inter-rater reliability, good convergent and divergent validity and was sensitive to treatment change in the initial psychometric study (48). Satisfactory psychometric findings have been reported from Nordic studies (49, 50).

Quality of life was rated by the children using the Inventory of Life Quality in Children and Adolescents (ILC) (51, 52). A review of Norwegian ILC studies found satisfactory norms and measures of validity and reliability (53). The ILC Life Quality total score (LQ_0–28) yield scores ranging from 0-28, where higher values indicate better life quality. The ILC problem score (PR) is computed by dichotomizing each of seven subscales to indicate whether a problem is present (1) or not (0) on the subscale. A mean ILC problem score can then be calculated, with a 0-7 range. A mean LQ_0–28 score of 22.6 (SD 3.9) and a mean PR score of 1.3 (SD = 1.6) has been reported by typically developing Norwegian youth aged 8–16 years (n = 1987) (51).

At baseline, the assessment with K-SADS revealed the number of comorbid disorders to be mean 3.2 per patient (range 2-6): ADHD (7 of 10 children, Enuresis nocturna in one child, and moderate (n = 2) and severe (n = 1) Depressive episodes in three children).

There were a mean number of 2.2 anxiety disorders diagnosed per child (range 1 - 5): Specific Phobia (n = 7), Separation Anxiety Disorder (n = 6), Social Phobia (n = 2), Generalized Anxiety Disorder (n = 2), Agoraphobia (n = 2), Panic Disorder (n = 1), Selective Mutism (n = 1) and Unspecified Anxiety Disorder (n = 1). (The specific phobias were: Darkness, heights, tunnels, flying, toilets, skin wounds, animals, crowds, car queues and closed rooms. Two children had one phobia, three children had two, and two children more than five phobias). Furthermore, school refusal was present in four of the ten children.

After treatment, seven of the eight completers benefited from the program. Five of the seven children did not meet criteria for any anxiety disorders and two had fewer anxiety disorders (both had one remaining Specific Phobia and one had only a partly remission of the Selective Mutism). One child met criteria for the same anxiety disorders found at baseline. Furthermore, the three diagnosed Depressive episodes and school refusal were no longer present, with no change in the diagnoses of ADHD and Enuresis nocturna. When comparing questionnaire ratings on baseline and after treatment, the mean scores from all informants had improved, and most differences reached statistical significance (see Table 5).

The therapists found the “Less stress” program feasible in a sample of children with ASD and comorbid anxiety. However, the therapists generally perceived that administering the program for those with comorbid ADHD could pose an extra challenge. Adaptations were needed to cope with the short attention span and hyperactive-impulsivity, which affected their ability to sit quietly and be attentive over time. It is plausible that the visually oriented “Less stress” was particularly important for those with ADHD, along with repetition of concepts, and frequent practice of strategies across contexts, as these adaptations have been found to aid children with this condition (17). Feedback from therapists suggested that the children generally worked well with their home assignments, with an increase over time, indicating that these were perceived as meaningful/useful by the participants. Non-completers did not carry out their home assignments (exposures). In addition, one child did not do his home assignment exposures but enjoyed coming to the sessions. All participants had a high user satisfaction, and the parents generally expressed confidence in their ability to apply the “Less stress” toolbox after the program.

Two children did not complete the program. Our baseline measures show that they had few anxiety disorders relative to the other participants (one each) and were better functioning on other measures than the completers were. This could suggest that the “Less stress“ program is well-suited for those severely affected by anxiety. The non-completers might have had a lack of motivation due to a lower level of anxiety, but as child motivation was not measured separately, we cannot say whether that was the reason. However, it is worth noticing that the therapists informed that all but one of the participating children came somewhat reluctantly to the sessions and only because their parents wanted them to come. Thus, motivation as such could be generally low when judged by child behavior alone.

Although a review did not report unambiguous effect of parental involvement in CBT for anxiety (54), the therapists experienced that parental involvement was crucial in our study. They considered it highly unlikely that progress had been made without the continuous parental prompting of both attendance to sessions and the completion of the required home assignments (exposures).

According to the diagnostic interview with the mothers, seven of eight completers improved after treatment with substantially fewer anxiety disorders than before treatment, suggesting that the “Less stress” program gave improvement as intended.

The diagnostic findings were corroborated on the anxiety questionnaires, as a significant mean reduction of anxiety symptoms was found after treatment from all three informants (about two SD’s for the parents and one SD for the children on the RCADS total t-score). The mean ratings of anxiety and impairment from both parents are in line with the diagnostic interview and this is worth noticing, as the scores from the fathers can be considered more independent of the interview. All informants have scores in the same direction, although not all reach statistical significance, and in line with previous literature (55), the children report lower levels of anxiety compared to the parents. These patterns of parent and child ratings could result from children with ASD and comorbid anxiety having difficulty in reporting their anxiety while the parents are more aware of them. Such an interpretation is consistent with previous research suggesting that children with ASD underreport their anxiety symptoms (56, 57) also when anxiety is present as determined by clinician consensus (58, 59). Perhaps also the questions did not convey meaning to them, in line with difficulties in identifying and communicating ones’ own emotions, known to affect children with ASD to a relative high extent (60-62).

Beyond improvement in diagnostic status/anxiety symptoms, we found a positive mean change in two broader measures: the clinician-rated CGAS and the child-rated life quality. However, the children were still impaired by ASD and other comorbid disorders after treatment, as the CGAS score (mean 48) still indicated a moderately affected functioning (43), and the scores on life quality were lower than those reported for the general child population (51).

Our sample included one child with mild mental retardation and some with somewhat reduced IQ and/or severe language problems, suggesting that the program may be suitable also in this subgroup, given relevant adaptations. Not withholding our small sample size, this is encouraging as there is little research investigating the effects of CBT for this population (11).

Most children in our sample had more than one anxiety disorder (mean 2.2, range 1-5), in line with previous findings on children with ASD (2, 63). The "Less stress" program was equally effective for the different anxiety disorders in the present study, with the exception of selective mutism, where only partial remission was found. Partial remission is a tentative definition of the increased occasional use of language sometimes found in children treated for selective mutism also when using defocused communication and graded exposure in an especially designed home-and school based intervention for this condition (64). Furthermore, a majority had comorbid ADHD, and three of the ten children had a depressive episode, in line with a study showing that children with ASD and comorbid anxiety have more ADHD and other comorbid disorders compared to children with ASD and no anxiety disorder (55). The comorbidity did not seem to affect outcome in our study, as also shown in a recent study reporting no clear differential impact of co-occurring psychiatric disorders on response to anxiety treatment (65). However, as previously noted, implementing the program for children with comorbid ADHD were perceived as posing some extra challenges for the therapists.

Our study with “Less stress” did not include specific work on depression or school refusal. Still the children with depressive episodes did not meet diagnostic criteria for depression after treatment, and school refusal was also no longer present. Anxiety and depression often coexist and is then reported to represent greater symptom severity (66), and school performance has been found to be the life domain most impacted by anxiety in children with ASD (9, 10). It is thus encouraging that working with their anxiety problems and mastering difficult exposure tasks in “Less stress” might have counteracted depression and school refusal among the participating children.

Contrary to a study, reporting that a notable portion of youth with externalizing disorders, such as ADHD comorbid to ASD and anxiety, did not meet criteria for these externalizing disorders after treatment for anxiety (65), we found no change in the presence of ADHD after “Less stress”.

We have no definite explanation for this difference, but while ADHD was not described as an area of concern in that study, this was in contrast to the children participating in “Less stress” where the participating children had a high rate of severe ADHD. This might suggest that this outcome difference is related to difference in baseline severity of ADHD symptoms.

A strength of the study was that it was carried out in an ordinary clinical setting including local therapists with some (n = 3) and extensive (n = 2) CBT experience.

Limitations include that it was a pilot study without a control group and with a small sample size. In addition, although we have no information about other ongoing treatment during “Less stress”, we cannot rule out that improvement took place due to other factors in the everyday life of the children. Furthermore, there were not blinded diagnostic evaluation after treatment and no measures of objective treatment adherence by the therapists. As all therapists had clinical experience with children with ASD, we are unable to infer how feasible the intervention would be with clinicians who have no experience working with children with ASD.

Although measures of therapist-, child- and parent-reported feasibility yielded positive results (i.e., high ratings of therapist rated adherence and competence and high user satisfaction from parents and children), such measures could be subject to bias.