Commentary on selected new evidence

With advice from topic experts we selected 4 studies for further commentary.

Autism diagnostic assessment for children and young people

We selected the systematic review and meta-analysis by Kulage et al. (2014) for a full commentary because it adds useful new data to the evidence base reviewed for NICE guideline CG128, which could impact on current recommendations on autism diagnostic assessment.

What the guideline recommends

NICE guideline CG128 recommends that the following should be included in every autism diagnostic assessment:

  • detailed questions about parent's or carer's concerns and, if appropriate, the child's or young person's concerns

  • details of the child's or young person's experiences of home life, education and social care

  • a developmental history, focusing on developmental and behavioural features consistent with the International Classification of Diseases version 10 (ICD-10) or the Diagnostic and Standard Manual version (DSM-IV) criteria (consider using an autism-specific tool to gather this information)

  • assessment (through interaction with and observation of the child or young person) of social and communication skills and behaviours, focusing on features consistent with ICD-10 or DSM-IV criteria (consider using an autism-specific tool to gather this information)

  • a medical history, including prenatal, perinatal and family history, and past and current health conditions

  • a physical examination

  • consideration of the differential diagnosis (see recommendation 1.5.7)

  • systematic assessment for conditions that may coexist with autism (see recommendation 1.5.15)

  • development of a profile of the child's or young person's strengths, skills, impairments and needs that can be used to create a needs-based management plan, taking into account family and educational context

  • communication of assessment findings to the parent or carer and, if appropriate, the child or young person.

NICE guideline CG128 also recommends:

  • That information from all sources, together with clinical judgment, should be used to diagnose autism based on ICD-10 or DSM-IV criteria.

  • That healthcare professionals should be aware that some children and young people will have features of behaviour that are seen in the autism spectrum but do not reach the ICD-10 or DSM-IV diagnostic criteria for definitive diagnosis. Based on their profile, there should be consideration of referral to appropriate services.

Methods

Kulage et al. (2014) reported a systematic review and meta-analysis of 14 prospective and retrospective studies (n=16,548) assessing the impact of changes introduced through the Diagnostic and Standard Manual version 5 (DSM-5) on diagnosis of ASD. The authors conducted 2 meta-analyses:

  • The first meta-analysis included all studies and examined whether the frequency of people diagnosed with ASD differed when using DSM-IV-TR criteria compared with DSM-5. Sensitivity analyses were done by age, country, study design and study quality to address heterogeneity.

  • The second meta-analysis examined differences in autistic spectrum disorder (ASD) subgroup diagnoses (autistic disorder [AD], Asperger's disorder, and pervasive developmental disorder-not otherwise specified [PDD-NOS]) between DSM-IV-TR and DSM-5.

The quality of included studies was evaluated with the Quality Appraisal of Reliability Studies (QAREL), which has 11 items evaluating 7 principles of the reliability of diagnostic tests: 1) appropriateness of subjects, 2) appropriateness of examiners, 3) blinding of examiners, 4) order effects of examination, 5) suitability of the time interval between repeated measures, 6) appropriate test and application, and 7) statistical analysis of inter- or intra-rater agreement.

Results

Fourteen studies were included. Eleven studies used the 2011 DSM-5 draft criteria and 3 studies used the 2010 DSM-5 draft criteria without substantial differences in the study findings. There was a reduction in diagnosis of the following using the full DSM-5 criteria compared to the DSM-IV-TR criteria, specifically:

  • ASD (range 7.3% to 68.4%)

  • AD (range 0% to 40%)

  • Asperger's disorder (range 16.6% to 100%)

  • PDD-NOS (range 50% to 97.5%).

The first meta-analysis included the subgroup of participants diagnosed with ASD using the DSM-IV-TR criteria (n=7517 participants, 14 studies). When DSM-5 criteria were applied, the pooled reduction in ASD diagnosis was 31% (95% confidence interval [CI] 20 to 44, p<0.001). However, heterogeneity between and within studies was high (Q=945, p<0.001, I2=98.6) and sensitivity analyses were done to identify responsible variables:

Age
  • Age ≤3 years (47.8%, 95% CI 44.3 to 51.3; 1 study)

  • Age ≤18 years (25.6%, 95% CI 14.1 to 41.8; 7 studies)

  • Ages 4 to 18 years (53.8%, 95% CI 35.0 to 71.6; 1 study)

  • Age ≥4 years (22.7%, 95% CI 10.5 to 42.4; 3 studies)

  • All ages (48.1%, 95% CI 30.9 to 65.8; 2 studies)

There were significant differences between age subgroups (p<0.001).

Country
  • United States (33.4%, 95% CI 23.5 to 45.0; 8 studies)

  • International (28.3%, 95% CI 13.3 to 50.5; 6 studies).

Study design
  • Prospective (33.7%, 95% CI 26.8 to 41.4; 6 studies)

  • Retrospective (28.5%, 95% CI 15.2 to 47.1; 8 studies).

Study quality
  • Met <half quality criteria (28.5%, 95% CI 15.2 to 47.1; 8 studies)

  • Met ≥half quality criteria (34.2%, 95% CI 14.5 to 61.4; 5 studies).

The second meta-analysis included 7 studies with participants meeting the DSM-IV-TR criteria for the ASD subgroups (n=1,227 participants with AD, n=80 with Asperger's disorder, and n=630 with PDD-NOS). When DSM-5 criteria were applied, the pooled reduction was significant for the following:

  • AD diagnosis: 22% (95% CI 16 to 29, p<0.001, heterogeneity: Q=27.7, p<0.001, I2=78.4)

  • PDD-NOS diagnosis: 70% (95% CI 25 to 97, p=0.01, heterogeneity: Q=39.4, p<0.001, I2=87.3).

A non-significant pooled reduction in diagnosis using DSM-5 was observed for the following:

  • Asperger's disorder diagnosis: 70% (95% CI 17 to 96, p=0.38, heterogeneity: Q=18.3, p<0.001, I2=83.6).

Strengths and limitations

Strengths

The main strength was that the authors used the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for their report. At least half QAREL items were met by 5 studies.

Limitations

A limitation was that the population of included studies only partially matched the population looked at in NICE guideline CG128 because 3 studies included children and adults and 2 studies included only adults. This systematic review does not report on how many people who do not have a diagnosis at present would be now included by DSM-5. The authors concluded that this systematic review was underpowered to detect the true impact of DSM-5 for Asperger's disorder (only 4 studies with small samples).

The main weaknesses of the included studies in this systematic review were the lack of reporting about raters' blinding to the results of the DSM-IV-TR and the lack of appropriate statistical measures of agreement such as inter- or intra-rater reliability.

Impact on guideline

This systematic review and meta-analysis provides a comparison between DSM-IV-TR and DSM-5 criteria.

DSM-5 was published after the development of NICE guideline CG128. Therefore, NICE guideline CG128 refers to DSM-IV criteria in recommendation 1.5.5, recommendation 1.5.10 and recommendation 1.5.13. The introduction of the new DSM-5 criteria may have an impact on NICE guideline CG128 because the new criteria seem to be much more strict and therefore fewer people would meet ASD diagnosis.

Specific interventions for the core features of autism – psychosocial interventions

Two studies were selected for this area (Geretsegger 2014; Oono 2013).

What the guideline recommends

NICE guideline CG170 recommends that a specific social-communication intervention should be considered for the core features of autism in children and young people that includes play-based strategies with parents, carers and teachers to increase joint attention, engagement and reciprocal communication in the child or young person. Strategies should:

  • be adjusted to the child or young person's developmental level

  • aim to increase the parents', carers', teachers' or peers' understanding of, and sensitivity and responsiveness to, the child or young person's patterns of communication and interaction

  • include techniques of therapist modelling and video-interaction feedback

  • include techniques to expand the child or young person's communication, interactive play and social routines.

The intervention should be delivered by a trained professional. For preschool children consider parent, carer or teacher mediation. For school-aged children consider peer mediation.

We selected the Cochrane review by Geretsegger et al. (2014) for a full commentary because it provides evidence on music therapy for children with ASD. Music therapy is an emerging intervention that was not considered in NICE guideline CG170 because there was lack of evidence during guideline development.

Methods

Geretsegger et al. (2014) conducted a Cochrane review of 9 randomised controlled trials and 1 'counterbalanced' trial (n=165 children, 10 studies) assessing the effects of music therapy compared to placebo therapy or standard care for people with autism spectrum disorder. Music therapy interventions were delivered by professional music therapists through regular sessions including free and structured improvisation, playing music, singing songs, and listening (one-to-one and family-based settings). The duration of music therapy interventions ranged from 1 or 2 weeks (daily basis) to 7months (weekly basis). Participants had a diagnosis of pervasive developmental disorder as defined by ICD-10 or DSM-IV or DSM-IV-TR. The primary outcomes were social interaction, non-verbal and verbal communicative skills, initiating behaviour, social-emotional reciprocity and adverse effects. Secondary outcomes included social adaptation skills.

Results

Participants were children between 2 and 9 years old. Children received a diagnosis of ASD with a standardised tool including the Childhood Autism Rating Scale (CARS), the Autism Diagnostic Interview Revised (ADI-R), or the Social Responsiveness Scale (SRS). Outcome measures included non-generalised outcomes (changes in child's non-generalised behaviour in the same setting of the intervention) and generalised outcomes (changes observed in other settings).

  • There was a greater effect on non-generalised social interaction skills following music therapy (standardised mean difference [SMD] 1.06, 95% CI 0.02 to 2.10, p=0.046; 1 study, n=10).

  • There was a greater effect on generalised social interaction skills following music therapy (SMD 0.71, 95% CI 0.18 to 1.25, p=0.0092; 3 studies, n=57).

  • There was a greater effect on non-generalised communicative skills (non-verbal) following music therapy (SMD 0.57, 95% CI 0.29 to 0.85, p=0.000068; 3 studies, n=30).

  • Generalised communicative skills (non-verbal) were not significantly higher after music therapy compared to control therapy (SMD 0.48, 95% CI −0.02 to 0.98, p=0.060; 3 studies, n=57).

  • There was a greater effect on non-generalised communicative skills (verbal) following music therapy (SMD 0.33, 95% CI 0.16 to 0.50, p=0.00015; 4 studies, n=92).

  • Generalised communicative skills (verbal) were not significantly higher after music therapy compared to control therapy (SMD 0.30, 95% CI -0.28 to 0.89, p=0.31; 2 studies, n=47).

  • There was a greater effect on non-generalised initiating behaviour following music therapy (SMD 0.73, 95% CI 0.36 to 1.11, p=0.00011; 3 studies, n=22).

  • There was a greater effect on non-generalised social-emotion reciprocity following music therapy (SMD 2.28, 95% CI 0.73 to 3.83, p=0.0039; 1 study, n=10).

  • There was a greater effect on non-generalised social adaptation following music therapy (SMD 1.15, 95% CI 0.69 to 1.61, p<0.00001; 3 studies, n=22).

  • There was a greater effect on generalised social adaptation following music therapy (SMD 0.24, 95% CI 0.02 to 0.46, p=0.029; 1 study, n=4).

  • None of the studies reported adverse effects.

Strengths and limitations

Strengths

The main strength was that the study used the Cochrane methodology and had low risk of bias. The authors judged that more than 75% of studies had low risk of attrition bias, reporting bias and other bias. If heterogeneity was present, it was not significant apart from a combined meta-analysis of non-generalised and generalised outcomes.

Limitations

A limitation of this systematic review was the small sample size of the included studies (6 studies with 10 or fewer participants and 4 studies with 50 or fewer participants). Sample size limitation was partially compensated by most of the studies using crossover designs. The authors judged that half or more of the studies had unclear risk of selection bias, detection bias and performance bias.

Impact on guideline

This systematic review and meta-analysis provides evidence that music therapy may have positive effects on social interaction and communication skills in children with ASD. Two of the 10 included studies in this systematic review were also considered during guideline development under art-based interventions. However, there are no specific recommendations in NICE guideline CG170 for the use of art-based interventions for the treatment of autism. Guideline committee members commented during this surveillance review that there was not a clear opinion about music therapy in the current guideline. It was concluded that the evidence from this Cochrane review was not enough to update NICE guideline CG170 in this area because the evidence was from studies with small sample sizes and unclear risk of relevant bias.

We selected the Cochrane review by Oono et al. (2013) for a full commentary because it provides evidence that children with ASD may make gains in language skills following parent-mediated interventions and topic experts felt that this evidence is very applicable to recommendation 1.3.1.

Methods

Oono et al. (2013) conducted a Cochrane review of 17 randomised controlled trials (n=919) assessing the effectiveness of parent-mediated early interventions in terms of the benefits for both children and their parents. Control groups included no treatment, treatment as usual, waiting list, alternative child-centred intervention not mediated by parents or an alternative parent-mediated intervention different to the intervention under study. Parent-mediated early interventions were delivered by professionals (including group or individual training) to improve the management of their children's ASD-related difficulties in areas such as communication, social development, learning and behaviours. The duration of the interventions ranged from 1 week to 2 years. Participants were children with ASD (aged between 17 months to 6 years with varied levels of functioning). The primary outcomes were child communication and social development (including language development [comprehension and expression], social communication skills and skills in interaction with parent) and parents' level of stress. Sensitivity analyses were done including studies with low and unclear risk of bias in 4 domains: sequence generation, allocation concealment, blinding of outcome, and attrition.

Results

Meta-analyses included 10 studies that evaluated interventions focusing on parent interaction style in facilitating children's communication compared to 'treatment as usual'. The rest of the studies could not be compared directly because they were different in their theoretical basis, interventions, and outcome measures.

Significant improvements were observed in the following parent-mediated intervention groups:

  • Language development (comprehension [parent report]) (mean difference [MD] 36.26, 95% CI 1.31 to 71.20, p=0.042; 3 studies, n=204).

  • Autism severity (SMD -0.30, 95% CI -0.52 to -0.08, p=0.0081; 6 studies, n=316).

  • Shared or joint interaction (coding of parent-child interactions) (SMD 0.41, 95% CI 0.14 to 0.68, p=0.0032; 3 studies, n=215).

  • Parent synchrony (coding of parent-child interactions) (SMD 0.90, 95% CI 0.56 to 1.23, p<0.00001; 3 studies, n=244).

For the following outcomes, no significant differences were observed between parent-mediated interventions and control groups:

  • Language development (comprehension [direct or independent assessment]) (SMD 0.29, 95% CI −0.20 to 0.78, p=0.25; 2 studies, n=200).

  • Language expression (direct or independent assessment) (SMD 0.14, 95% CI −0.16 to 0.45, p=0.36; 3 studies, n=264).

  • Language expression (parent report) (MD 29.44, 95% CI −14.99 to 73.86, p=0.19; 3 studies, n=204).

  • Joint language (direct or independent assessment) (SMD 0.45, 95% CI −0.05 to 0.95, p=0.077; 2 studies, n=64).

  • Child communication (parent or teacher report) (MD 5.31, 95% CI −6.77 to 17.39, p=0.39; 3 studies, n=228).

  • Child initiations (coding of parent-child interactions) (SMD 0.38, 95% CI −0.07 to 0.82, p=0.095; 4 studies, n=268).

  • Parent stress (SMD −0.17, 95% CI −0.70 to 0.36, p=0.52; 2 studies, n=55).

Regarding social communication skills, meta-analysis was not performed for this outcome and studies reported mixed results (no differences using a directly observed assessment measure, improvements with more intensive treatment, and significant improvements on teacher-reported social and language skills).

Strengths and limitations

Strengths

The main strength was that the study used the Cochrane methodology and had low risk of bias. The authors judged that between 50% and 75% of the included studies had low risk of selection bias (randomisation), detection bias, attrition bias and reporting bias. Eleven meta-analyses were performed including the primary outcomes without heterogeneity in 5 studies and non-significant heterogeneity in 4 studies.

Limitations

A limitation was the variation in outcomes measures which limited the number of studies included in the meta-analyses. The authors judged that there was high risk of allocation concealment and performance bias in most of the included studies.

Impact on guideline

This systematic review and meta-analysis provides evidence that children with ASD may make gains in language skills following parent-mediated interventions. Four of the 17 included studies in this systematic review were also considered during guideline development under behavioural interventions. They also commented about the limitations and low quality of evidence from this systematic review. It was concluded that the evidence from this Cochrane review was not enough to update NICE guideline CG170 in this area because the evidence had high risk of allocation concealment and performance bias.

Research recommendation 2.2 Managing behaviour that challenges in children and young people with autism

We selected the randomised controlled trial by Bearss et al. (2015) for a full commentary because it partially addresses research recommendation 2.2 in the guideline.

What the guideline recommends

The research recommendation suggests an evaluation of a group-based parent training intervention for parents or carers of children and young people with autism in reducing early and emerging behaviour that challenges in the short- and medium-term compared with treatment as usual. The guideline committee considered that a randomised controlled trial design should be used assessing short- and medium-term reduction in behaviour that challenges, parental and sibling stress, quality of life and the child or young person's adaptive function, medium-term use of medication, and cost effectiveness of a wide range of services, such as additional educational support and social services, and health service use by families.

Methods

Bearss et al. (2015) conducted a 24-week randomised controlled trial (n=180) assessing whether parent training was superior to parent education for reducing behavioural problems in children with ASD confirmed by DSM-IV-TR (aged 3 years to 6 years 11 months). Exclusion criteria were children in whom there would be treatment changes, children with receptive language <18 months, not enrolled in a school programme, living in a household without an English-speaking caregiver, with a diagnosis of Rett disorder or childhood disintegrative disorder, presence of a known serious medical condition, a current psychiatric disorder requiring alternative treatment, or children whose parents participated in a structured parent training programme in the past 2 years previous to this randomised controlled trial (RCT). The study was conducted in 6 sites in the United States. Parent training was delivered individually in 11 core sessions over 16 weeks covering the identification of children's behaviours, strategies to manage behaviours, and maintenance of improvements. Parent education was delivered in 12 sessions over 24 weeks covering information on ASD without any instruction on behaviour management. The primary outcomes were the parent-rated Aberrant Behaviour Checklist-Irritability subscale (ABC-I) and the parent-rated Home Situations Questionnaire – Autism Spectrum Disorder (HSQ-ASD). It was pre-specified that 25% reduction of both ABC-I and HSQ-ASD indicated clinically meaningful improvement. The secondary outcomes were the Improvement item of the Clinical Global Impression scale (CGI-I), adaptive functioning and parent-child interactions (measured using the Standardised Observational Analogue Procedure [SOAP]). Adverse events were assessed by an independent evaluator. Outcomes were measured at baseline, week 12 and week 24.

Results

Effect sizes were calculated by taking the difference in the least squares means at week 24 and diving by the pooled standard deviation at baseline.

The 24‑week parent training programme led to a greater reduction in disruptive behaviour on parent-reported outcomes compared with parent education but this reduction was not clinically meaningful:

  • The ABC-I decreased 47.7% (from 23.7 to 12.4) in the parent training group and 31.8% (from 23.9 to 16.3) in the parent education group (least squares mean difference −3.9, 95% CI −6.2 to −1.7, p<0.001, effect size=0.62).

  • The HSQ-ASD decreased 55.0% (from 4.0 at baseline to 1.8 by week 24) in the parent training group and 34.2% (from 3.8 to 2.5) in the parent education group (least squares mean difference −0.7, 95% CI −1.1 to −0.3, p<0.001, effect size=0.45).

The 24‑week parent training programme led to a greater overall improvement compared with parent education rated by an independent clinician blinded to treatment assignment:

  • The CGI-I was rated much improved or very much improved by 68.5% of participants in the parent training group compared with 39.6% of participants allocated to the parent education group (p<0.001). The number needed to treat was 4.

  • The most frequent adverse events were cough and rhinitis (around 50% in each group) and diarrhoea (around 30% in each group). No significant differences were found in adverse events between parent training and parent education groups.

Strengths and limitations

Strengths

The main strengths of this study were the low risk of selection bias and reporting bias as well as that this study was focused on children which make the results applicable to NICE guideline CG170.

Limitations

A limitation of this study was the high risk of performance bias and detection bias. Although some outcome assessors were blinded, this blinding was restricted to the secondary outcomes only as the parents could not be blinded for the parent-rated outcomes. These limitations have an impact on the applicability of the results to NICE guideline CG170.

Impact on guideline

This RCT partially addresses NICE guideline CG170 research recommendation 2.2 because it did not measure parental and sibling stress and quality of life and it was individual not group based. Although adaptive skills were measured, this publication only shows baseline data but it is mentioned that these results will be presented in a separate report. There was not an assessment of medium-term use of medication or a cost-effectiveness analysis which was an additional criterion of the research recommendation.


This page was last updated: 22 September 2016