Meta-analysis of parent-mediated interventions for young children with autism spectrum disorder

Abstract

A number of studies of parent-mediated interventions in autism spectrum disorder have been published in the last 15 years. We reviewed 19 randomized clinical trials of parent-mediated interventions for children with autism spectrum disorder between the ages of 1 and 6 years and conducted a meta-analysis on their efficacy. Meta-analysis outcomes were autism spectrum disorder symptom severity, socialization, communication-language, and cognition. Quality of evidence was rated as moderate for autism spectrum disorder symptom severity, communication-language, and cognition, and very low for socialization. Weighted Hedges’ g varied from 0.18 (communication-language) to 0.27 (socialization) and averaged 0.23 across domains. We also examined the relationship between outcome and dose of parent training, type of control group, and type of informant (parent and clinician). Outcomes were not significantly different based on dose of treatment. Comparing parent training to treatment-as-usual did not result in significantly different treatment effects than when parent training was compared to an active comparison group. Based on parent report only, treatment effects were significant for communication-language and non-significant for socialization, yet the opposite was found based on clinician-rated tools. This meta-analysis suggests that while most outcome domains of parent-delivered intervention are associated with small effects, the quality of research is improving.

Keywords

autism spectrum disorder caregivers treatment

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by the presence of pervasive deficits in social-communication skills and restricted and repetitive interests or behaviors. It is currently estimated that 62 in 10,000 children have an ASD (Elsabbagh et al., 2012). Optimal outcomes in development have been associated with intervention targeting the earliest noticeable signs of ASD, specifically deficits in initiation of and responding to joint attention, shared affect, and functional and symbolic aspects of play (Wetherby and Woods, 2006). Targeting these developmental areas has been associated with subsequent improved development in communication-language, cognition, and adaptive behavior (Dawson et al., 2010; Koegel et al., 2014; Smith et al., 2000). Early intervention research has recommended that children receive at least 25 weekly hours of targeted, intensive, early intervention in order to achieve optimal outcomes (National Research Council, 2001; Zwaigenbaum et al., 2009). Unfortunately, this ideal dose is challenging for many families due to limitations in time, financial resources, and insurance coverage. Logistical challenges such as distance required in traveling to treatment services and time required off work to coordinate such care can also limit access to early intervention (Symon, 2001). Increased waitlist times for services due to the growing number of children being diagnosed with ASD at or below 3 years further contributes to difficulties in accessing services.

In considering these barriers, parents present as a promising alternative source of early intervention due to their ability to practice skills with their child throughout the day and across settings. In the United States, Part C of the Individuals with Disabilities Education Act (IDEA, 2004) mandates children with developmental disabilities such as ASD to receive family-focused services to support caregivers’ acquisition of skills that promote child development. Providing parents the skills to effectively cope with their child’s developmental delays can also increase the parents’ sense of competence, reduce stress, and improve family cohesion (Koegel et al., 2002). Therefore, it is of interest to develop models for parent-mediated intervention that are both feasible for parents to implement and result in reduced ASD symptoms along with improved developmental functioning.

The past 15 years have seen increased studies evaluating parent-mediated early intervention. These studies have varied significantly in key methodological considerations such as outcome measures, dose, format of treatment, and sample sizes, which hamper interpretation and generalization of findings. Where conclusions can be drawn, it is unlikely that a single treatment is effective for every child since the ASD population is highly heterogeneous (Baker-Ericzén et al., 2007). For treatments to be effective, they must be flexible enough to apply to individuals with various functioning levels, behavioral profiles, and socioeconomic backgrounds.

A number of reviews on the effectiveness of parent-mediated intervention have been published (Beaudoin et al., 2014; Brookman-Frazee et al., 2006; Lang et al., 2009; Meadan et al., 2009; Oono et al., 2013; Patterson et al., 2012; Schultz et al., 2011; Suppo and Floyd, 2012). Taken together, these studies identified a greater need to report dose of effective treatments (Schultz et al., 2011), social validity outcomes (Meadan et al., 2009), and parent characteristics associated with better outcomes, such as socioeconomic status (SES) or cultural factors (Brookman-Frazee et al., 2006; Lang et al., 2009; Patterson et al., 2012). They called for increased methods to promote generalization of skills (Meadan et al., 2009), reduce parenting stress, and include parents as partners rather than students in the intervention (Beaudoin et al., 2014; Brookman-Frazee et al., 2006; Meadan et al., 2009; Suppo and Floyd, 2012). Reviews have found small positive outcomes for children and greater improvements for parents, such as reduced parenting stress and increased sense of competence (Beaudoin et al., 2014; Kaminski et al., 2008).

Two previous reviews have conducted meta-analyses on data from randomized controlled trials (RCTs; Oono et al., 2013; UK National Institute for Health and Care Excellence (NICE), 2013). Oono et al. (2013) reviewed 17 parent-mediated intervention studies published between 2002 and 2012 for children aged 1 to 6 years 11 months and performed a meta-analysis on 11 of these studies. Small but significant evidence was found for improvements in parent–child interaction only. Results were inconclusive for communication, language, frequency of child initiations, adaptive behavior, and parenting stress. Some improvements were found in parent-reported language comprehension and autism symptom severity, though it was unclear the extent to which these improvements were due to the parent-mediated intervention. None of the included studies used multiple treatment groups. Similar findings were reported in a more recent review that included studies using mixed designs for children aged 0–3 years (Beaudoin et al., 2014).

The majority of the RCTs reviewed in Oono et al. (2013), UK NICE (2013), and Beaudoin et al. (2014) were published between 2010 and 2012. Since 2012, a number of RCTs have been published. It was therefore of interest to conduct an analysis of peer-reviewed studies published between 2000 and 2015 using a structured coding system in light of the high degree of heterogeneity in RCTs. The current meta-analysis was performed on outcomes in the following five primary areas of child impairment: (a) ASD symptom severity, (b) socialization, (c) communication-language, (d) daily living skills, and (e) cognitive functioning. We additionally evaluated the quality of evidence available for each of these outcome areas by applying Grading of Recommendations Assessment, Development, and Evaluation (GRADE) ratings. GRADE is a systematic approach for evaluating the quality of available evidence in reviews of health research, and has become increasingly used in the developmental disability field (e.g. the Cochrane collaboration or NICE guidelines). Decisions regarding quality of evidence are presented as part of a review or meta-analysis to guide future decisions regarding best practice and areas for further research (Guyatt et al., 2011).

Secondary objectives were to determine whether dose of parent-received training in intervention skills (more or less than 20 h of training) was associated with improved child outcomes and whether the effect of the interventions varied according to the type of control group (active comparison vs treatment-as-usual (TAU)). Additionally, the type of informant (parent vs clinician) was investigated due to the potential for response bias across informants. We hypothesized that stronger effects would be observed for studies with more than 20 h of parent training, for comparisons involving a TAU control group, and when parents served as informants on outcome measures.

Method

Search procedure

The primary search engines used were PsycINFO, PubMed, MEDLINE, and Google Scholar in association with WorldCat@OSU. WorldCat@OSU includes the following search engines: Academic OneFile, Academic Search Complete, MasterFILE Complete, WorldCat.org, and Oxford Scholarship Online. Peer reviewed journal articles published in English via journal updates and Online First between 2000 and December 2015 were also searched. The following terms were entered using Boolean operators: parent-mediated, caregiver, caregiver-mediated, parent-based, parent-implemented, parent directed, parent intervention, early intervention, behavioral intervention, and parent training. Each of these terms was entered in combination with “autism” and with and without “randomized controlled trial.” Reference sections of found articles were also reviewed for any additional studies to be included. One additional article was found by reviewing journal update notifications between the period of November 2014 and December 2015.

Article titles and abstracts were reviewed for the following inclusion criteria: (a) children were on average between 1 and 6 years old with a confirmed diagnosis of ASD, (b) parents were responsible for implementing the active intervention during the active treatment phase through the assistance of therapist coaching, strategy modeling, feedback, or education, (c) primary outcome variables focused on child functioning, and (d) the study included randomization to a treatment condition (i.e. there was a TAU or active treatment comparison group). Articles appearing to meet these criteria went on for full-text review by the first two authors. Authors initially disagreed on inclusion of 6 out of 43 papers, which were resolved through discussion and additional evaluation by the third author. After review, 24 articles were excluded upon finding evidence that at least one of the above criteria had been violated. Common reasons for exclusion included the sample lacking full randomization, targeting children at risk for, rather than diagnosed with, ASD, and not assessing the impact of treatment on core areas of functioning impacted by ASD. Only studies satisfying each of these four criteria underwent further analysis for inclusion in a qualitative review and meta-analysis. Figure 1 illustrates the search process.

Figure 1.

Search procedure.

Outcome variables

Articles were then included based on their presentation of outcome data for the following five core areas of functioning impacted by ASD: (a) ASD symptom severity, (b) socialization, (c) communication-language, (d) daily living skills, and (e) cognition. The following construct definitions were agreed upon by authors and used to categorize outcome measures in the different studies, independent of how they were originally conceptualized by study authors: (a) ASD severity: a measure reflecting the degree of social communicative impairments and restricted repetitive behaviors that are specific to ASD; (b) Socialization: the ability to engage in social interactions, to demonstrate social awareness of others’ perspectives, and to use social aspects of communication (i.e. pragmatic language) at an age-appropriate level; (c) Communication-language: the ability to produce and comprehend spoken and/or written language at an age-appropriate level, or display play skills that are known precursors of language; (d) Daily Living Skills: the ability to perform self-care activities that are needed to function independently at an age-appropriate level, including feeding, dressing, hygiene, and safety behavior; (e) Cognition: the ability to acquire, retain, attend to, problem-solve, and reason information using concrete objects, abstract ideas, and verbal information at an age-appropriate level.

ASD symptom severity measures overlap to some degree with measures of socialization and communication-language, though we preferred to keep them as separate categories because socialization and communication delays can also be indicative of disorders other than ASD. In addition, these two domains are associated with different patterns of impairment in ASD (e.g. Ray-Subramanian et al., 2011; Yang et al., 2016). A number of measures additionally act as “global” measures of severity (e.g. Autism Diagnostic Observation Schedule (ADOS) Total score, Childhood Autism Rating Scale (CARS) Total score) that we considered inappropriate to categorize differently. Additionally, we opted to include pragmatic language measures (e.g. Early Social Communication Scales) under socialization as they assess for communicative behaviors that are required for social interactions, such as turn-taking and responses to social games.

A total of 54 different instruments were used as outcome measures across the reviewed studies. Of these measures, 14 were excluded because they did not measure an outcome domain of interest (e.g. 4 measured academic achievement, 8 measured emotional and behavior problems, 1 measured diversity of symbolic play, 1 measured repetitive behavior without measuring social-communication). Three additional measures were excluded because they were not considered appropriate as measures of change based on criteria from Anagnostou et al. (2015), which reflect an instrument’s psychometric properties and clinical relevance. Based on these criteria, we excluded the Autism Diagnostic Interview-Revised, CARS, and Social Communication Questionnaire. The remaining 37 measures were independently classified in one of the five categories by the first two authors. Coders agreed on the classification of 32 out of 37 measures and resolved remaining disagreements by discussion. Therefore, data from 37 measures were reviewed. Table 1 shows all outcome measures used and how they were categorized.

Table 1.

Measures used to assess key outcome variables across studies.

Study	ASD severity	Socialization	Communication-language	Daily living skills	Cognition
Aldred et al. (2004)	ADOS-T		MCDIVABS-C
Carter et al. (2011)	PIA-CV-NVC	ESCS; PCFP
Casenhiser et al. (2013)		Modified CBRS	PLS or CASL
Drew et al. (2002)			MCDI-WG		GSID NV IQ
Green et al. (2010)	ADOS-SC	CSBS-SC	PLS-RL, PLS-EL; MCDI RL, EL raw
Hardan (2014)	SRS		MCDI EL longest utterance, W/396; W/680; PLS-EL; VABS-C
Kasari et al. (2010)		Adamson-JE; Child frequency of IJA and RJA
Kasari et al. (2014)		Adamson-JE; ESCS
Kasari et al. (2015)		Adamson-JE; Child frequency of IJA	Play skills; Play level; RDLS
Nefdt et al. (2010)			Number of functional verbal utterances
Pajareya and Nopmaneejumruslers (2011)		FEAS
Poslawsky et al. (2015)		EAS; ESCS
Rickards et al. (2007)		VABS-S			BSID or WPPSI
Roberts et al. (2011)		VABS-S	RDLS-EL, RL; VABS-C; PPEC
Siller et al. (2013)			MSEL-EL
Solomon et al. (2014)			MSEL-RL, MSEL-EL DQs; MCDI-WG; MCDI-WS		MSEL NV IQ
Tonge et al. (2014)	DBC-ASA	VABS-S	VABS-C; RDLS	VABS DLS	PEP-R DQ
Welterlin et al. (2012)		SIB-R SI	MSEL-L DQs; SIB-R RL and EL		MSEL-ELC DQ
Wetherby et al. (2014)	ADOS-SA	CSBS; VABS-S	VABS-C; MSEL-EL, MSEL-RL	VABS DLS	MSEL-ELC

ASD: autism spectrum disorder; Adamson-JE: Parent–child joint engagement coded using Adamson’s (2004) coding scheme; ADOS: Autism Diagnostic Observation Schedule; ASA: Autism Screening Algorithm; BSID: Bayley Scales of Infant Development; C: Communication subscale score; CASL: Childhood Assessment of Spoken Language; CBRS: Child Behavior Rating Scale; CSBS: Child Symbolic Behavior Scales; DBC-SRS: Developmental Behavior Checklist–Social Relating Subscale; DLS: Daily Living Skills subscale score; DQ: Developmental quotient; EAS: Emotional Availability Scales; EL: Expressive Language subscale score; ELC: Early Learning Composite score; ESCS: Early Social Communication Scales; FEAS: Functional Emotional Assessment Scale; GSID: Griffith’s Scale of Infant Development; IJA: Initiation of joint attention; IQ: Intelligence Quotient; MCDI: MacArthur–Bates Communicative Development Inventory; MSEL: Mullen Scales of Early Learning; NV: Non-verbal; NVC: Non-verbal communication dimension; PCFP: Parent–Child Free Play Procedure; PEP-R: Psychoeducational Profile–Revised; PIA-CV: Parent Interview for Autism—Clinical Version; PLS: Preschool Language Scale; PPEC: Pragmatics Profile of Everyday Communication; RDLS: Reynell Developmental Language Scales; RJA: Response to joint attention; RL: Receptive Language subscale score; S: Socialization subscale score; SA: Social Affect subscale; SC: Social communication subscale score; SI: Social Interaction component; SIB-R: Scales of Independent Behavior–Revised; SRS: Social Responsiveness Scale; T: Total score; V: Verbal; VABS: Vineland Adaptive Behavior Scales; W/396: Words produced out of 396; W/680: words produced out of 680; WG: Words and Gestures form; WPPSI: Wechsler Preschool and Primary Scale of Intelligence; WS: Words and Sentences form.

The most frequently reported outcomes were for communication-language (12) and socialization (12), followed by cognition (6), ASD symptom severity (6), and daily living skills (2). A mixture of parent-report measures and clinical assessments were used to measure change in socialization, communication-language, and ASD symptoms. Cognition was assessed using clinical assessments only.

Data analysis

Findings from the included studies were combined using meta-analytic methods to determine the average effect of parent-mediated interventions as compared to control groups. For studies that included both an active comparison and TAU group, results from the parent training and TAU groups only were included to increase comparability across studies, since the majority of reviewed studies used a TAU control group. Standardized mean differences for post-treatment measures were calculated using formulas for Hedges’ g. For the majority of the included effect sizes, Hedges’ g was calculated from reported means and standard deviations for the treatment and control group. In a few circumstances, this information was not reported, and Hedges’ g was calculated using other methods. Specifically, the socialization and communication-language effect sizes for Pajareya and Nopmaneejumruslers (2011) and Rickards et al. (2007) were calculated utilizing reported p-values. The formulas recommended by Rosenthal (1991) and Card (2012) were used for all effect size computations. Cohen’s (1988) criteria for trivial (<0.2), small (0.2–0.49), medium (0.5–0.79), and large (⩾0.8) effect sizes were then applied to determine magnitude of effect.

Effect sizes across outcome areas were treated independently. However, while some studies reported only one outcome measure within a category, other studies reported multiple outcome measures within a category. Including multiple effect sizes in the same category from the same study would violate the independence assumption, which in turn risked inflating the sample size of the statistical tests and effect sizes beyond what is actually included in the meta-analysis (Wolf, 1986). When studies reported multiple outcome measures within one category, the effect sizes were averaged to allow each study to contribute one effect size to the overall average effect sizes. The calculated effect sizes across studies were then averaged within each outcome domain and weighted to control for sample size. Hedges’ test for homogeneity (Hedges’ Q-test) was used to determine the degree of heterogeneity of effect sizes (Hedges and Olkin, 1985); statistical significance indicates that there is significant heterogeneity among the effect sizes contributing to the mean weighted effect size. The I² index was used to determine the magnitude of heterogeneity among studies when significant heterogeneity was found (Higgins and Thompson, 2002; Huedo-Medina et al., 2006). This statistic uses the following cutoffs: 25% = low heterogeneity, 50% = moderate heterogeneity, and 75% = large heterogeneity.

Next, the first two authors reviewed studies for quality of design, risks of bias, and overall evidence to come to agreement on GRADE-quality ratings through discussion (Guyatt et al., 2011). This method involves rating quality of evidence of the intervention for specific outcomes of interest within and across studies on five areas: (a) Indirectness: did the study look at the variables and population of interest based on the research question? (b) Risk of bias: did the study’s methodology account for biases, such as failure to blind or selective reporting? (c) Imprecision: did the study account for random error in measurement of the outcome variable? (d) Inconsistency: was there high heterogeneity in findings within a specific outcome across studies? and (e) Publication bias: was there evidence of publication bias in the selection of reviewed studies? Evidence for each outcome is rated as very low, low, moderate, or high. Once a rating has been made for each outcome of interest within each study, summary ratings are made across studies to determine the overall quality of evidence available for each outcome. The methodology for applying these ratings can be found through a series of publications published by the GRADE Working Group (e.g. Guyatt et al., 2011).

To address secondary goals, articles were reviewed for the number of hours the parents were trained in intervention delivery. The dose of parent training was coded based on whether parent received training for 20 h or more, or less than 20 h. This cut point was chosen as it allowed for the division of studies into relatively even groups coded as either “high” or “low” dose. We did not differentiate between individual and group time in instruction. For instance, if a study provided two 1-h 1:1 coaching sessions and one 2-h group education session, the parent training dose was considered 4 h. Separate analyses were performed to examine the effect of dose on outcome variables. Table 2 shows the categorization of studies according to dose of parent training received. Separate analyses were also conducted based on whether effect sizes were derived from active comparison versus TAU control groups. We then explored differences in outcomes based on whether parent versus clinician-rated assessments were used. Analyses were only conducted when there were three or more studies on a given outcome variable domain.

Table 2.

Design characteristics of reviewed studies.

Study author	TX program	N (TX, CT)	TX group mean age (months) (targeted range)	Dose and content of CT condition	Dose of parent training	Dose coding	Delivery format (group, 1:1, modeling)
Aldred et al. (2004)	Child’s Talk Project	14, 14	44.5^a (29–60)	TAU	12 h^a: 9 sessions (monthly therapist sessions for 6 months followed by three bimonthly consolidation sessions)	<20	Group-based PE, 1:1 parent–child center-based sessions, video feedback, personalized written programs provided to parents and school.
Carter et al. (2011)	HMTW	32, 30	21.1 (15–25)	TAU	23 h^a: 8 2.5-h weekly group sessions and three 1:1 sessions	⩾20	Group-based PE, 1:1 parent–child sessions with therapist coaching.
Casenhiser et al. (2013)	DIR/Floortime	25, 26	42.5 (24–59)	TAU	104 h: 2 h a week for 1 year	⩾20	Child therapy and parent coaching, video feedback.
Drew et al. (2002)	Parent Focus Training	12, 12	21.4 (⩾20)	TAU	24 h: 1 3-h session a week every 6 weeks for 12 months	⩾20	1:1 in-home therapist consultation, modeling, PE.
Green et al. (2010)	PACT	77, 75	45.0 (24–59)	TAU	96 h: 6 months of biweekly 2-h therapy sessions and 6 months of booster sessions	⩾20	1:1 clinic-based coaching using video feedback.
Hardan (2014)	PRT	27, 26	49.2 (24–72)	17 h: 10 PE groups on ASD and two parent–child clinician consults	16 h over 12 once-a-week sessions	<20	Group sessions and 1:1 parent–child sessions.
Kasari et al. (2010)	JASPER	19, 19	30.4 (21–36)	TAU	12 h: 3 sessions a week for 8 weeks	<20	In home 1:1 direct instruction, modeling, guided practice, feedback, summary handouts.
Kasari et al. (2014)	JASPER	60, 52	41.9 (24–60)	24 h; 12 weekly 2-h group meeting to discuss JASPER modules	24 h: 2 1-h sessions a week for 12 weeks	⩾20	In home 1:1 direct instruction, modeling, guided practice, summary handouts.
Kasari et al. (2015)	JASPER	43, 43	30.7 (22–36)	10 h: 1:1 parent informational meetings	10 h: 2 30-min sessions per week for 10 weeks	<20	1:1 active coaching and feedback
Nefdt et al. (2010)	PRT	13, 14	38.9 (<60)	WL	2.25 h: 66-min DVD viewed an average of 2.04 times	<20	Self-directed learning module via DVD instruction and manual, chapter tests, video models, interactive learning tasks of videos to grade for treatment fidelity.
Pajareya and Nopmaneejumruslers (2011)	DIR/Floortime	16, 16	56.6 (24–72)	TAU	13.5 h^a: one education workshop, 1 training session, and 1 follow-up session	<20	Group-based PE workshop, 3-h DVD workshop recording, manual, 1:1 home-based training, modeling, guided practice, feedback.
Poslawsky et al. (2015)	VIPP-AUTI	40, 38	42.2 (16–61)	9.5 h^a: 7.5 h of home visits and two PE groups on ASD	7–9.5 h^a: 5 1- to 1.5-h semi-monthly sessions over 3 months, 2 group PE sessions on ASD sessions	<20	1:1 home-based training, video feedback, clinic-based PE group sessions.
Rickards et al. (2007)	Home-based program	18, 21	70.7 (36–60)	5 h center-based individualized intervention delivery by teacher	45–65 h: control condition followed by 1–1.5 h per week for 40 weeks of intervention implementation	⩾20	Center-based team consultation, 1:1 in-home PE, written summaries, modeling, environmental supports when appropriate (e.g. communication aids).
Roberts et al. (2011)	Building Blocks	27, 28	41.5 (24–60)	1. 80 h center-based PE and support group and concurrent child playgroups with therapists to target pre-academic skills2. TAU	40 h: 2-h semi-monthly sessions over 40 weeks	⩾20	1:1 home-based therapist training, modeling, feedback, discussion of issues, preschool observation and consultation.
Siller et al. (2013)	Focused Playtime Intervention	36, 34	58.3 (<60)	6 h of Parent Advocacy Coaching over 4 monthly sessions on treatment and education planning	18 h: 1 90-min weekly session for 12 weeks and 3 90-min Parent Advocacy Coaching sessions	<20	1:1 home-based modeling, feedback, and guided practice, PE using a workbook, video feedback, instruction, and homework.
Solomon et al. (2014)	PLAY Project	64, 64	49.9 (32–60)	TAU	36 h: 3-h monthly sessions for 1 year (est. mean = 31.56 h; SD = 9.03)	⩾20	Self-directed 2-h DVD training, manual, 1:1 home-based, modeling, coaching, feedback, written video feedback.
Tonge et al. (2014)	Preschoolers with Autism	33, 35	46.0 (30–60)	1. 15 h: 10 PE and counseling sessions2. TAU	25 h: 10 1-h behavior management sessions alternated with 10 1.5-h PE and counseling sessions over 20 weeks	⩾20	1:1 center-based instruction, practice, modeling, video examples, homework, and feedback.PE: small group and center-based PE and discussion.
Welterlin et al. (2012)	TEACCH	10, 10	30.5 (24–36)	TAU	18 h: 1.5 h a week for 12 weeks	<20	1:1 home-based clinician-implemented parent training, modeling, PE, homework, observation, coaching, and feedback.
Wetherby et al. (2014)	SCERTS	42, 40	19.6 (16–20)	28.8 h (mean): small group treatment once a week for 9 months	88.56 h (mean): active training of 3 sessions a week for 6 months. Maintenance: 2 sessions a week for 3 months	⩾20	Active training: 2:1 ratio of home and clinic-based using direct teaching, modeling, guided practice, feedback, and independent practice with faded supports. Generalization sessions occurred in the home and community.

ASD: autism spectrum disorder; CT: Control; DIR: Developmental, Individual-differences, and Relationship–based model; HMTW: Hanen’s More Than Words; JASPER: Joint Attention Symbolic Play Engagement and Regulation; PACT: Parent-mediated Communication-focused Treatment; PE: parent education; PECS: Picture Exchange Communication System; PRT: Pivotal Response Training; SCERTS: Social Communication, Emotional Regulation, and Transactional Supports; TAU: treatment-as-usual; PLAY: Play and Language for Autistic Youngsters; TEACCH: Treatment and Education of Autistic and Related Communication Handicapped Children; TX: Treatment; VIPP-AUTI: Video Intervention to promote Positive Parenting for children with Autism; SD: standard deviation; WL: waitlist.

Data unavailable, value estimated.

Results

Search results

A total of 19 RCTs were located. Across these studies, 608 children and their parents received active treatment and 597 received the treatment in the control condition. Within studies, sample sizes ranged from 10 to 77 participants in active treatment and from 10 to 75 in the control condition. The average age of child participants was 42 months, ranging from 15 to 72 months. Half of the studies were conducted in the United States (Carter et al., 2011; Hardan et al., 2015; Kasari et al., 2010, 2014, 2015; Nefdt et al., 2010; Siller et al., 2013; Solomon et al., 2014; Welterlin et al., 2012; Wetherby et al., 2014). The remaining studies were conducted in Australia (Rickards et al., 2007; Roberts et al., 2011; Tonge et al., 2014), the United Kingdom (Aldred et al., 2004; Drew et al., 2002; Green et al., 2010), Canada (Casenhiser et al., 2013), Asia (Pajareya and Nopmaneejumruslers, 2011), and the Netherlands (Poslawsky et al., 2015).

Data on parent socioeconomic factors were missing for one study (Drew et al., 2002). Children were largely from middle-income, two-parent families. The participating parent was primarily the mother and unemployed. Parents were primarily in their 30s. For studies that reported parent education, the majority had at least some college education, with the exception of Kasari et al. (2014), which targeted low-resourced families, and Rickards et al. (2007) that had almost 50% of parents with less than 12 years of education. Samples were mostly Caucasian, except for Pajareya and Nopmaneejumruslers (2011) which was based in Thailand, Siller et al. (2013) which was composed of 41% to 47% Hispanic and Latino, and Kasari et al. (2014) which contained 66% ethnic minorities.

Intervention model

Studies primarily categorized their intervention models as social-communication based (Aldred et al., 2004; Carter et al., 2011; Casenhiser et al., 2013; Green et al., 2010; Pajareya and Nopmaneejumruslers, 2011; Poslawsky et al., 2015; Siller et al., 2013) or naturalistic developmental behavioral (Drew et al., 2002; Hardan et al., 2015; Kasari et al., 2010, 2014, 2015; Nefdt et al., 2010; Rickards et al., 2007; Roberts et al., 2011; Solomon et al., 2014; Tonge et al., 2014; Wetherby et al., 2014). Categorizations of naturalistic developmental behavioral were based on the definition from Schreibman et al. (2015). One study used principles primarily from structured teaching (Welterlin et al., 2012).

Control condition

The control conditions were TAU (k = 9), waitlist (k = 1), and active comparison (k = 9). Two studies using an active comparison also included a second TAU control group. Active comparison conditions consisted of parent education or instruction (Hardan et al., 2015; Kasari et al., 2014, 2015; Poslawsky et al., 2015; Rickards et al., 2007; Siller et al., 2013; Tonge et al., 2014) or concurrent clinician-implemented intervention in addition to parent education (Roberts et al., 2011). Wetherby et al. (2014) provided clinician-implemented intervention only in a group format. Based on dose of control condition, three studies provided a high (over 20 h) comparison treatment (Kasari et al., 2014; Roberts et al., 2011; Wetherby et al., 2014).

Dose of parent-delivered intervention

Only 7 of 19 studies presented information on how much intervention parents were requested to deliver to their child outside of parent training sessions (Aldred et al., 2004; Casenhiser et al., 2013; Drew et al., 2002; Green et al., 2010; Pajareya and Nopmaneejumruslers, 2011; Solomon et al., 2014; Wetherby et al., 2014). Three studies requested that parents deliver 30 min or less of intervention per day (Aldred et al., 2004; Casenhiser et al., 2013; Green et al., 2010). Drew et al. (2002) requested 1 h daily and Solomon et al. (2014) requested 2 h per day. Pajareya and Nopmaneejumruslers (2011) and Wetherby et al. (2014) requested higher doses, at 20 and 25 h per week, respectively. None of the studies presented data on the average dose of intervention that parents actually delivered; therefore, the relationship between dose of parent-delivered intervention and child outcomes could not be analyzed in the current meta-analysis. Additional information on key characteristics of the included studies is presented in Table 2.

Parent training format

The majority of parent training in intervention delivery occurred in a 1:1 format within the family’s home to promote generalization and maintenance. Groups were typically small (four to five parents per group) when trainings were group-based. Meetings were often short (30–60 min) and focused on teaching from an intervention manual. A number of studies requested that families take videos of parent–child interaction outside of training sessions or took videos during a session for the therapist to provide feedback. A number of studies also emphasized the use of parent coaching, which involved facilitating the parent in identifying goals, child cues, and appropriate responses, rather than directly instructing these skills.

Study quality

GRADE-quality ratings for each outcome area can be found in Table 3. Quality of evidence was rated as moderate for ASD symptom severity, communication-language, and cognition, and very low for socialization. Regarding risk of bias, all of the reviewed studies used randomization. The majority of studies were single blinded and achieved allocation concealment. Parents were not blind to their treatment condition, therefore parent-rated measures risked potential for bias. Due to strict inclusion criteria, indirectness was minimal across studies. Imprecision of studies was related to confidence intervals (CIs) almost always including zero and most studies being underpowered. Publication bias was likely among included studies since the search procedure was focused on peer-reviewed studies published in English. Table 3 contains additional detail on study quality. Complete ratings on all dimensions for every study are available by contacting the authors.

Table 3.

Summary of findings table.

	Nr of participants (studies)	Effect size (g) (95% CI)	Quality of evidence (GRADE)	Explanation
ASD symptom severity	420 (6)	0.22 (0.03–0.41)	Moderate due to imprecision	All studies are underpowered for given effect size; CI included 0 for all but one study.
Socialization	846 (13)	0.22 (0.09–0.36)	Very low due to bias, inconsistency, and imprecision	Majority and highest quality studies were limited in design due to loss to follow-up, ambiguity regarding blinding, or lack of allocation concealment.Variance in point estimates, several studies did not show an overlap in CI, the statistical test for heterogeneity was significant, and the I² value was large.Majority of studies were underpowered, CIs were wide and contained 0.
Communication-language	785 (13)	0.16 (0.02–0.31)	Moderate due to imprecision	Majority of studies were underpowered, CIs were wide and contained 0.
Cognition	334 (6)	0.24 (0.03–0.46)	Moderate due to imprecision	Majority of studies were underpowered, CIs were wide and contained 0.

CI: confidence interval; GRADE: Grading of Recommendations Assessment, Development, and Evaluation; ASD: autism spectrum disorder.

Meta-analysis

Overall effect of parent-mediated intervention

Tables 4 to 7 show summary data for individual studies and Forest plots for different outcome areas. Since only two studies reported outcome data for adaptive behavior (Tonge et al., 2014; Wetherby et al., 2014), meta-analyses could not be performed on this outcome. Effect sizes for these two studies were small (unweighted g = 0.47, 95% CI = −0.01–0.94; Tonge et al., 2014) and medium (unweighted g = 0.58, 95% CI = 0.14–1.02; Wetherby et al., 2014). The tables show effect sizes of the interventions as well as unweighted average effect sizes and weighted mean overall effect size (Hedges’ g). They additionally show sample sizes (n) from each study contributing to overall effect, each study’s weight within the overall effect size, and 95% CI. The significance of the effect is determined by the significance of its corresponding z-score (p < 0.05). Heterogeneity test values are also shown (Hedges’ Q-test and I² index). Given the variability in sample sizes contributing to each outcome, conclusions are based on weighted average effect sizes.

Table 4.

Summary data and Forest plot for ASD symptom severity outcomes across studies.

Study	INT	CT	Weight	g	95% CI
	n	n			Lower	Upper
Aldred et al. (2004)	14	14	6.3%	0.78	0.01	1.55
Carter et al. (2011)	27	20	11.1%	0.001	−0.58	0.58
Green et al. (2010)	74	72	35.1%	0.14	−0.19	0.46
Hardan (2015)	25	22	11.0%	0.49	−0.09	1.07
Tonge et al. (2014)	35	35	16.8%	0.22	−0.25	0.69
Wetherby et al. (2014)	42	40	19.7%	0.15	−0.28	0.58
Total (95% CI)	217	203	100.0%	0.22	0.03	0.41

ASD: autism spectrum disorder; INT: intervention; CT: control; CI: confidence interval.

Heterogeneity: Hedges’ Q-test = 3.79, df = 5, p = 0.58, I² = 0%.

Test for overall effect: Z = 2.21, p = 0.01.

Table 5.

Summary data and Forest plot for socialization outcomes across studies.

Study	INT	CT	Weight	g	95% CI
	n	n			Lower	Upper
Carter et al. (2011)	26	21	6%	−0.05	−0.61	0.52
Casenheiser et al. (2013)	25	26	6%	0.30	−0.26	0.85
Green et al. (2010)	74	72	17%	0.39	0.06	0.72
Kasari et al. (2010)	19	19	4%	0.70	0.05	1.36
Kasari et al. (2014)	59	48	13%	0.20	−0.18	0.59
Kasari et al. (2015)	43	43	10%	0.79	0.35	1.23
Pajareya and Nopmaneejumruslers (2011)	15	16	4%	0.12	−0.58	0.83
Poslawsky et al. (2015)	39	35	9%	−0.16	−0.62	0.30
Rickards et al. (2007)	18	21	5%	0.12	−0.51	0.75
Roberts et al. (2011)	27	28	6%	−0.71	−1.26	−0.17
Tonge et al. (2014)	35	35	8%	0.36	−0.11	0.83
Welterlin et al. (2012)	10	10	2%	−1.21	−2.17	−0.26
Wetherby et al. (2014)	42	40	10%	0.57	0.13	1.01
Total (95% CI)	432	414	100%	0.22	0.09	0.36

CI: confidence interval; INT: intervention; CT: control.

Heterogeneity: Hedges’ Q-test = 35.90, df = 12, p < 0.001, I² = 66.57%.

Test for overall effect: Z = 3.25, p < 0.01.

Table 6.

Summary data and Forest plot of communication-language outcomes across studies.

Study	INT	CT	Weight	g	95% CI
	n	n			Lower	Upper
Aldred et al. (2004)	14	14	4%	0.15	−0.59	0.89
Casenheiser et al. (2013)	25	26	7%	0.22	−0.33	0.78
Drew et al. (2002)	12	12	3%	0.64	−0.18	1.46
Green et al. (2010)	74	72	19%	0.11	−0.21	0.44
Hardan et al. (2015)	25	22	6%	0.17	−0.41	0.74
Kasari et al. (2015)	43	43	11%	0.26	−0.16	0.69
Nefdt et al. (2010)	13	14	3%	0.92	0.13	1.72
Roberts et al. (2010)	26	26	7%	−0.34	−0.88	0.20
Siller et al. (2013)	34	30	8%	0.13	−0.36	0.62
Solomon et al. (2014)	44	44	11%	0.12	−0.30	0.54
Tonge et al. (2014)	35	35	9%	−0.07	−0.54	0.40
Welterlin et al. (2012)	10	10	3%	0.11	−0.76	0.99
Wetherby et al. (2014)	42	40	10%	0.48	0.04	0.92
Total (95% CI)	397	388	100%	0.16	0.02	0.31

CI: confidence interval; INT: intervention; CT: control; df: degrees of freedom.

Heterogeneity: Hedges’ Q-test = 11.50, df = 12, p = 0.49, I² = 0%.

Test for overall effect: Z = 2.44, p < 0.01.

Table 7.

Summary data and Forest plot of cognition outcomes across studies.

Study	INT	CT	Weight	g	95% CI
	n	n			Lower	Upper
Drew et al. (2002)	12	12	7%	0.74	−0.09	1.56
Rickards et al. (2007)	18	21	12%	0.36	−0.28	0.99
Solomon et al. (2014)	52	47	30%	0.23	−0.17	0.63
Tonge et al. (2014)	35	35	21%	0.17	−0.30	0.64
Welterlin et al. (2012)	10	10	6%	0.26	−0.62	1.14
Wetherby et al. (2014)	42	40	25%	0.13	−0.30	0.56
Total (95% CI)	169	165	100%	0.24	0.03	0.46

CI: confidence interval; INT: intervention; CT: control; df: degrees of freedom.

Heterogeneity: Hedges’ Q-test = 1.86, df = 5, p = 0.76, I² = 0%.

Test for overall effect: Z = 2.21, p = 0.01.

Intervention effects on all outcome variables were significantly different from zero. Small treatment effects were found for ASD symptom severity (weighted g = 0.22, 95% CI: 0.03–0.41, p < 0.05), cognition (weighted g = 0.24, 95% CI: 0.03–0.46, p < 0.05), and socialization (weighted g = 0.23, 95% CI: 0.09–0.36, p < 0.05). A trivial effect was found for the communication-language domain (weighted g = 0.18, 95% CI: 0.03–0.32, p < 0.05). The communication-language outcome, as well as the socialization, domain had the most contributing studies and most participants to contribute to the meta-analysis. The socialization domain was the only domain that demonstrated significant heterogeneity in contributing effect sizes (as seen in the significant Hedges’ Q-test), with a moderate level of heterogeneity between effect sizes, suggesting the possibility of moderator effects.

To account for heterogeneity across studies in treatment approaches, analyses were repeated excluding the study that was not identified as social-communication or developmental behaviorally oriented (Welterlin et al., 2012). This resulted in a small treatment effect for socialization (weighted g = 0.26, p < 0.05, 95% CI = 0.12–0.40) though this effect was not significantly different from what was found in the original analysis. All other outcomes remained the same; therefore, this study was retained in the study selection for further analyses. Because of the inherent subjectivity associated with the classification of measures, two additional analyses with revised groupings were conducted: the first combined pragmatic measures of language with the socialization group, maintained the structural language measures in the communication-language group, and maintained all ASD symptom severity measures as originally categorized. This resulted in a small effect for socialization (weighted g = 0.24, 95% CI: 0.11–0.36) and a trivial effect for communication-language (weighted g = 0.11, 95% CI: −0.03 to 0.25). The second analysis extended upon the first by also adding ASD symptom severity measures into the socialization group. This did not change the magnitude of the treatment effect for socialization (weighted g = 0.22, 95% CI: 0.09–0.35).

Effect of intervention by dose of parent training

Dose of active intervention ranged from 2.3 to 104 h. Studies were coded as providing less than 20 h (k = 9) or 20 h or more (k = 10) of parent training while in the active treatment group. Results of subgroup meta-analyses based on dose are shown in Table 8. For studies with less than 20 h of parent training, socialization and communication-language was associated with small treatment effects. Analyses were not performed for cognition or ASD symptom severity because there was only one study assessing change in cognition and two studies assessing change in ASD symptom severity. Across studies with doses at or above 20 h, small effects were observed for socialization and cognition, and trivial non-significant effects were observed for ASD symptom severity and communication-language. Outcomes were not significantly different based on dose of treatment. Hedges’-Q homogeneity tests were non-significant across outcomes.

Table 8.

Effects of treatment based on dose of parent training.

Outcome variable	Dose (hours)	Contributing studies	N	Unweighted g	Weighted g	Z	95% CI	Hedges’ Q-test
ASD severity	Less than 20	2	39	–	–	–	–	–
	20 or higher	4	178	0.13	0.14	1.28	−0.07 to 0.35
Socialization	Less than 20	5	126	0.05	0.25	1.93	−0.004 to 0.51	0.95
	20 or higher	8	306	0.15	0.22	2.63**	0.06 to 0.38
Communication-language	Less than 20	6	139	0.29	0.25	2.05*	0.01 to 0.49	0.59
	20 or higher	7	258	0.17	0.14	1.53	−0.04 to 0.31
Cognition	Less than 20	1	10	–	–	–	–	–
	20 or higher	5	159	0.32	0.24	2.12*	0.02 to 0.46

ASD: autism spectrum disorder; CI: confidence interval.

N represents the number of participants in the treatment group.

Significant at *p < 0.05; **p < 0.01; ***p < 0.001; right-tailed.

Effect of intervention by type of control group

Results of subgroup meta-analyses based on type of control group are presented in Table 9. There were no significant differences in treatment effects comparing parent-mediated intervention to TAU versus parent-mediated intervention to active comparison across different outcome areas. Comparing parent-mediated intervention to TAU, effects were small for socialization and cognition. Effects were significant yet trivial for communication-language and ASD severity. When the control condition was an active comparison, effect sizes were small and significant for socialization, and trivial and significant for communication-language. Hedges’ homogeneity tests were non-significant for all outcomes.

Table 9.

Effects of treatment in comparison to TAU versus active control.

Outcome variable	Control condition	Contributing studies	N	Unweighted g	Weighted g	Z	95% CI	Hedges’ Q-test
ASD severity	TAU	4	150	0.29	0.19	1.64**	−0.04 to 0.42	–
	AC	–	–	–	–	–	–
Socialization	TAU	7	204	0.09	0.26	2.55**	0.06 to 0.46	0.17
	AC	6	228	0.14	0.20	2.06**	0.01 to 0.39
Communication-language	TAU	8	227	0.28	0.17	1.82**	−0.01 to 0.36	0.004
	AC	5	170	0.14	0.18	1.63**	−0.04 to 0.40
Cognition	TAU	4	109	0.35	0.27	1.92*	−0.01 to 0.54	–
	AC	2	60	–	–	–	–

CI: confidence interval; ASD: autism spectrum disorder; TAU: treatment-as-usual; AC: active comparison.

N represents the number of participants in the treatment group tail.

Significant at the p < 0.05 level: *p < 0.05; **p < 0.01; ***p < 0.001; right-tailed.

Effect of intervention by informant type

Given the wide range of socialization and communication-language assessment tools found across studies, it was of interest to look at treatment effects based on the type of informant providing the information. Effect sizes were calculated for the subgroup of studies reporting results using parent-rated measures, and then again for the subgroup reporting results from clinician-administered assessments. This was not done for other domains because they were largely consistent in the use of clinician- or parent-rated measures. In the socialization domain, 8 studies used at least one parent-rated measure, whereas 6 studies used clinician-rated instruments. In the communication-language domain, 9 studies used at least one parent-rated measure and 11 studies used at least one clinician-rated instrument. Based on parent report only, treatment effects were significant for communication-language (weighted mean g = 0.18, 95% CI: 0.01–0.34, z = 2.08, p < 0.05) and non-significant for socialization (weighted mean g = 0.12, 95% CI: −0.05–0.29, z = 1.35, p = 0.18). In contrast, a significant treatment effect was found for socialization (weighted mean g = 0.27, 95% CI: 0.06–0.47, z = 2.57, p < 0.05) but not for communication-language (weighted mean g = 0.14, 95% CI: −0.01–0.29, z = 1.90, p = 0.06) using clinician-rated tools. Tests for heterogeneity were significant for socialization domain based on parent report (Hedges’ Q = 28.38, p < 0.001) and clinician-rated tools (Hedges’ Q = 11.91, p < 0.05), suggesting the possibility of moderator effects.

Discussion

The current report reviews RCTs of parent-mediated interventions for young children with ASD and presents results of a meta-analysis on outcomes in four primary areas of child impairment. Secondary goals were to explore whether dose of parent training in intervention delivery was associated with child outcomes, whether type of control group used impacted treatment effects, and whether type of informant was associated with the magnitude of treatment effects. We extended upon previous reviews of parent-mediated interventions by focusing on different child outcomes as well as other key methodological variables. Of the studies analyzed, only 3 overlapped with Beaudoin et al. (2014), which included mixed method designs (Carter et al., 2011; Drew et al., 2002; Kasari et al., 2010), and 11 were included in a literature review completed by Oono and colleagues (Aldred et al., 2004; Carter et al., 2011; Casenhiser et al., 2013; Drew et al., 2002; Green et al., 2010; Kasari et al., 2010; Nefdt et al., 2010; Pajareya and Nopmaneejumruslers, 2011; Rickards et al., 2007; Roberts et al., 2011; Siller et al., 2013). Eight new RCTs have been included in the current meta-analysis.

Methodologies varied widely across studies. Combined samples contributing to different outcome areas were large, with samples ranging from 334 to 846. A number of studies presented with risk of low power, having included less than 20 participants receiving active intervention (Aldred et al., 2004; Drew et al., 2002; Kasari et al., 2010; Nefdt et al., 2010; Pajareya and Nopmaneejumruslers, 2011; Rickards et al., 2007; Welterlin et al., 2012). Over half of RCTs did not have an active comparison group. Additionally, there was wide variability in dose of parent training in intervention delivery, ranging from 2 to 104 h. Finally, there was a wide range of outcome measures used, which inspired the comparison of treatment effects based on type of informant. For example, 4 out of the 13 studies that assessed socialization used the Vineland Adaptive Behavior Scales (VABS)—Socialization domain (Sparrow et al., 2005) while the other 9 focused on skills such as conventional gesture use, play, and reciprocal interactions using structured observations or parent report. Of note is that we found the quality of evidence of reviewed studies to be improved compared to previous GRADE-quality ratings (e.g. Oono et al., 2013).

The meta-analysis indicated that parent-mediated intervention resulted in only small improvements in ASD symptom severity, socialization, and cognition. Only trivial improvements were observed in communication-language skills. The effect sizes found in this review were of similar magnitude to those reported by Oono et al. (2013), other than language comprehension. We categorized both receptive and expressive language measures into one outcome area for the current analyses. The treatment effect for communication-language as a whole using only data included in the Oono et al. (2013) review was trivial, and was comparable to the size of the effect found in the current review. Dose of parent training in intervention delivery was marginally associated with larger effects. Contrary to our predictions, studies in which parents received less than 20 h of training were associated with small treatment effects for communication-language. Receiving 20 or more hours of parent training was associated with small treatment effects for socialization skills and cognition and only trivial, non-significant improvements in language and ASD severity. Participating in low dose interventions may allow children to enroll in more hours of other, community-based treatments such as targeted speech therapy to result in larger improvements over time that are unrelated to the active treatment. Comparing parent-mediated intervention to TAU, small improvements were found for socialization and cognition. Larger treatment effects were observed when the intervention was compared to TAU rather than active comparison for socialization, though this difference did not reach statistical significance. Approximately equal treatment effects were found for communication-language across comparison groups. Despite our expectation that parents in the intervention group would indicate greater change than clinicians due to their awareness of treatment condition, improvements in socialization were trivial based on parent report, while improvements in communication-language were small. Clinician-rated assessments indicated small improvements in socialization and trivial improvements in communication-language. This suggests that the inability to blind parents may have had minimal effects on outcomes, a conclusion consistent with Oono et al.’s (2013) review.

While the field has seen a large increase in RCTs evaluating the effects of parent-mediated ASD intervention in the last 3 years, current results suggested that overall effects of parent-mediated interventions on child outcomes are small. Turning to the practical significance of these findings, an effect size of d = 0.2 corresponds to such a high degree of overlap between groups that differences are likely trivial clinically (Magnusson, 2014). It could be that parent-implemented interventions may be better suited for reducing challenging behavior than addressing core ASD symptoms (Bearss et al., 2015).

The current review had limitations. Since meta-analyses are often prone to publication bias, this review may be over-reporting true effect sizes due to the inclusion of only published studies (Bakker et al., 2012; Brand et al., 2008). One result (the ASD symptom severity measure from Solomon et al., 2014) could not be converted into a statistic compatible for inclusion in the meta-analysis. Additionally, data from the active comparison group were not included in the meta-analyses when a second TAU control group was also used in the same study, in order to increase comparability across all studies (Roberts et al., 2011; Tonge et al., 2014). Multiple studies used both parent report and clinician-rated assessments of socialization and communication-language; the overlap in variability across studies prevented us from directly comparing outcomes based on informant type. Also, since studies did not consistently report amount of intervention hours delivered to the child, we were unable to quantify this effect on child outcomes. Finally, the categorization of outcome measures presented challenges due to the conceptual overlap between socialization, communication, and language domains. We thought it would be more precise to distinguish between “general” measures of ASD like the Social Responsiveness Scale (SRS) from more “narrow” measures like the VABS Communication. We were able to account for this limitation by having high agreement on independent categorizations of measures by two of the authors and finding similar effect sizes for outcomes using two separate categorization systems.

The significant variety in key methodological considerations and interventions complicates our ability to interpret and generalize findings on the effectiveness of parent-mediated interventions. Although the calculation of standardized effect sizes is intended to account for variability across measures, a number of tools have limited evidence as valid outcome measures, which makes results more difficult to interpret. Having a standardized battery of measures and a method for accurately measuring the dose of intervention the child actually receives would improve the confidence in our conclusions. Only once we have developed methods to measure dose will we be able to determine the minimum amount of intervention needed to maximize child outcomes while also considering caregiver burden associated with intervention delivery. Consistent use of an active comparison group may also reduce risk of responder bias on parent-rated outcome measures. When TAU is used, comparisons are further limited by the great variability in the quality of community-based services received within and across settings.

Conclusion

Experts have agreed on the benefit of parents acting as interventionists for their young children with ASD (National Autism Center, 2015; National Research Council, 2001). The current review found small improvements in ASD symptom severity, socialization, and cognition, and trivial improvements in communication-language. These sobering results are likely impacted by inconsistencies across studies in the quality and quantity of parent training, as well as the methodology across trials. Improvements in the methodology of parent training studies will help produce clearer answers regarding their impact on the core functioning areas of ASD. On a more positive note, our review of quality of evidence confirms that the field has made great strides in the last 15 years, both in the quantity and quality of published intervention studies evaluating parent-mediated interventions for young children with ASD.

Footnotes

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Adamson

Bakeman

Deckner

(2004) The development of symbol-infused joint engagement. Child Development 75(4): 1171–1187.

*Aldred

Green

Adams

(2004) A new social communication intervention for children with autism: pilot randomised controlled treatment study suggesting effectiveness. Journal of Child Psychology and Psychiatry 45(8): 1420–1430.

Anagnostou

Jones

Huerta

. (2015) Measuring social communication behaviors as a treatment endpoint in individuals with autism spectrum disorder. Autism 19(5): 622–636.

Baker-Ericzén

Stahmer

Burns

(2007) Child demographics associated with outcomes in a community-based pivotal response training program. Journal of Positive Behavior Interventions 9(1): 52–60.

Bakker

van Dijk

Wicherts

(2012) The rules of the game called psychological science. Perspectives on Psychological Science: A Journal of the Association for Psychological Science 7(6): 543–554.

Bearss

Johnson

Smith

. (2015) Effect of parent training vs parent education on behavioral problems in children with autism spectrum disorder: a randomized clinical trial. Journal of American Medical Association 313(15): 1524–1533.

Beaudoin

Sebire

Couture

(2014) Parent training interventions for toddlers with autism spectrum disorder. Autism Research and Treatment 2014: e839890.

Brand

Bradley

Best

. (2008) Accuracy of effect size estimates from published psychological research. Perceptual and Motor Skills 106(2): 645–649.

Brookman-Frazee

Stahmer

Baker-Ericzén

. (2006) Parenting interventions for children with autism spectrum and disruptive behavior disorders: opportunities for cross-fertilization. Clinical Child and Family Psychology Review 9(3–4): 181–200.

10.

Card

(2012) Applied Meta-Analysis for Social Science Research. New York: Guilford Press.

11.

*Carter

Messinger

Stone

. (2011) A randomized controlled trial of Hanen’s “more than words” in toddlers with early autism symptoms. Journal of Child Psychology and Psychiatry and Allied Disciplines 52(7): 741–752.

12.

*Casenhiser

Shanker

Stieben

(2013) Learning through interaction in children with autism: preliminary data from a social-communication-based intervention. Autism 17(2): 220–241.

13.

Cohen

(1988) Statistical Power Analysis for the Behavioral Sciences. 2nd ed. Hillsdale, NJ: Lawrence Erlbaum.

14.

Dawson

Rogers

Munson

. (2010) Randomized, controlled trial of an intervention for toddlers with autism: the early start Denver model. Pediatrics 125(1): e17–e23.

15.

*Drew

Baird

Baron-Cohen

. (2002) A pilot randomised control trial of a parent training intervention for pre-school children with autism. Preliminary findings and methodological challenges. European Child & Adolescent Psychiatry 11(6): 266–272.

16.

Elsabbagh

Divan

Koh

Y-J

. (2012) Global prevalence of autism and other pervasive developmental disorders. Autism Research 5(3): 160–179.

17.

Gehanno

Rollin

Darmoni

(2013) Is the coverage of Google Scholar enough to be used alone for systematic reviews. BMC Medical Informatics and Decision Making 13: 7.

18.

*Green

Charman

McConachie

.; the PACT Consortium (2010) Parent-mediated communication-focused treatment in children with autism (PACT): a randomised controlled trial. The Lancet 375(9732): 2152–2160.

19.

Guyatt

Oxman

Akl

. (2011) GRADE guidelines: 1. Introduction—GRADE evidence profiles and summary of findings tables. Journal of Clinical Epidemiology 64(4): 383–394.

20.

*Hardan

Gengoux

Berquist

. (2015) A randomized controlled trial of Pivotal Response Treatment Group for parents of children with autism. Journal of Child Psychology and Psychiatry 56(8): 884–892.

21.

Hedges

Olkin

(1985) Statistical Methods for Metaanalysis. Orlando, FL: Academic Press.

22.

Higgins

JPT

Thompson

(2002) Quantifying heterogeneity in a meta-analysis. Statistics in Medicine 21(11): 1539–1558.

23.

Huedo-Medina

Sánchez-Meca

Marín-Martínez

. (2006) Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychological Methods 11(2): 193–206.

24.

Individuals with Disabilities Education Act (IDEA) (2004) 20 U.S.C. § 1400.

25.

Kaminski

Valle

Filene

. (2008) A meta-analytic review of components associated with parent training program effectiveness. Journal of Abnormal Child Psychology 36(4): 567–589.

26.

*Kasari

Gulsrud

Paparella

. (2015) Randomized comparative efficacy study of parent-mediated interventions for toddlers with autism. Journal of Consulting and Clinical Psychology 83(3): 554–563.

27.

*Kasari

Gulsrud

Wong

. (2010) Randomized controlled caregiver mediated joint engagement intervention for toddlers with autism. Journal of Autism and Developmental Disorders 40(9): 1045–1056.

28.

*Kasari

Lawton

Shih

. (2014) Caregiver-mediated intervention for low-resourced preschoolers with autism: an RCT. Pediatrics 134(1): e72–e79.

29.

Khabsa

Giles

(2014) The number of scholarly documents on the public web. PLoS ONE 9(5): e93949. Available at: http://doi.org/10.1371/journal.pone.0093949

30.

Koegel

Ashbaugh

. (2014) The importance of early identification and intervention for children with or at risk for autism spectrum disorders. International Journal of Speech-Language Pathology 16(1): 50–56.

31.

Koegel

Symon

Koegel

(2002) Parent education for families of children with autism living in geographically distant areas. Journal of Positive Behavior Interventions 4(2): 88–103.

32.

Lang

Machalicek

Rispoli

. (2009) Training parents to implement communication interventions for children with autism spectrum disorders (ASD): a systematic review. Evidence-Based Communication Assessment and Intervention 3(3): 174–190.

33.

Magnusson

(2014) Interpreting Cohen’s d effect size. Available at: http://rpsychologist.com/d3/cohend/ (accessed 15 January 2016).

34.

Meadan

Ostrosky

Zaghlawan

. (2009) Promoting the social and communicative behavior of young children with autism spectrum disorders: a review of parent-implemented intervention studies. Topics in Early Childhood Special Education 29(2): 90–104.

35.

National Autism Center (2015) National Standards Report. Randolph, MA: National Autism Center. Available at: http://www.nationalautismcenter.org/national-standards-project/phase-2/

36.

National Research Council (2001) Education Children with Autism. Washington, DC: National Academy Press.

37.

*Nefdt

Koegel

Singer

. (2010) The use of a self-directed learning program to provide introductory training in pivotal response treatment to parents of children with autism. Journal of Positive Behavior Interventions 12(1): 23–32.

38.

Oono

Honey

McConachie

(2013) Parent-mediated early intervention for young children with autism spectrum disorders (ASD). Evidence-Based Child Health: A Cochrane Review Journal 8(6): 2380–2479.

39.

*Pajareya

Nopmaneejumruslers

(2011) A pilot randomized controlled trial of DIR/Floortime™ parent training intervention for pre-school children with autistic spectrum disorders. Autism 15(5): 563–577.

40.

Patterson

Smith

Mirenda

(2012) A systematic review of training programs for parents of children with autism spectrum disorders: single subject contributions. Autism 16(5): 498–522.

41.

*Poslawsky

Naber

FBA

Bakermans-Kranenburg

. (2015) Video-feedback Intervention to promote Positive Parenting adapted to Autism (VIPP-AUTI): a randomized controlled trial. Autism 19(5): 588–603.

42.

Ray-Subramanian

Huai

Weismer

(2011) Brief report: adaptive behavior and cognitive skills for toddlers on the autism spectrum. Journal of Autism and Developmental Disorders 41(5): 679–684.

43.

*Rickards

Walstab

Wright-Rossi

. (2007) A randomized, controlled trial of a home-based intervention program for children with autism and developmental delay. Journal of Developmental and Behavioral Pediatrics 28(4): 308–316.

44.

*Roberts

Williams

Carter

. (2011) A randomised controlled trial of two early intervention programs for young children with autism: centre-based with parent program and home-based. Research in autism spectrum disorders. 5(4): 1553–1566.

45.

Rosenthal

(1991) Meta-Analytic Procedures for Social Research. Newbury Park, CA: SAGE.

46.

Schreibman

Dawson

Stahmer

. (2015) Naturalistic developmental behavioral interventions: empirically validated treatments for autism spectrum disorder. Journal of Autism and Developmental Disorders 45(8): 2411–2428.

47.

Schultz

Schmidt

Stichter

(2011) A review of parent education programs for parents of children with autism spectrum disorders. Focus on Autism and Other Developmental Disabilities 26(2): 96–104.

48.

*Siller

Hutman

Sigman

(2013) A parent-mediated intervention to increase responsive parental behaviors and child communication in children with ASD: a randomized clinical trial. Journal of Autism and Developmental Disorders 43(3): 540–555.

49.

Smith

Buch

Gamby

(2000) Parent-directed, intensive early intervention for children with pervasive developmental disorder. Research in Developmental Disabilities 21(4): 297–309.

50.

*Solomon

Van Egeren

Mahoney

. (2014) PLAY project home consultation intervention program for young children with autism spectrum disorders: a randomized controlled trial. Journal of Developmental and Behavioral Pediatrics 35(8): 475–485.

51.

Sparrow

Cicchetti

Balla

(2005) The Vineland Adaptive Behavior Scales. 2nd ed. Los Angeles, CA: Pearson.

52.

Suppo

Floyd

(2012) Parent training for families who have children with autism: a review of the literature. Rural Special Education Quarterly 31(2): 12–26.

53.

Symon

(2001) Parent education for autism: issues in providing services at a distance. Journal of Positive Behavior Interventions 3(3): 160–174.

54.

Thompson

McFerran

Gold

(2014) Family-centred music therapy to promote social engagement in young children with severe autism spectrum disorder: a randomized controlled study. Child: Care, Health and Development 40(6): 840–852.

55.

*Tonge

Brereton

Kiomall

. (2014) A randomised group comparison controlled trial of “preschoolers with autism”: a parent education and skills training intervention for young children with autistic disorder. Autism 18(2): 166–177.

56.

UK National Institute for Health and Care Excellence (NICE) (2013) Autism in under 19s: support and management. Available at: https://www.nice.org.uk/guidance/cg170

57.

Warren

McPheeters

Sathe

. (2011) A systematic review of early intensive intervention for autism spectrum disorders. Pediatrics 127(5): e1303–e1311.

58.

*Welterlin

Turner-Brown

Harris

. (2012) The home TEACCHing program for toddlers with autism. Journal of Autism and Developmental Disorders 42(9): 1827–1835.

59.

Wetherby

Woods

(2006) Early social interaction project for children with autism spectrum disorders beginning in the second year of life: a preliminary study. Topics in Early Childhood Special Education 26(2): 67–82.

60.

*Wetherby

Guthrie

Woods

. (2014) Parent-implemented social intervention for toddlers with autism: an RCT. Pediatrics 134(6): 1084–1093.

61.

Wolf

(1986) Meta-Analysis: Quantitative Methods for Research Synthesis. Newbury Park, CA: SAGE.

62.

Yang

Paynter

Gilmore

(2016) Vineland adaptive behavior scales: II profile of young children with autism spectrum disorder. Journal of Autism and Developmental Disorders 46(1): 64–73.

63.

Zwaigenbaum

Bryson

Lord

. (2009) Clinical assessment and management of toddlers with suspected autism spectrum disorder: insights from studies of high-risk infants. Pediatrics 123(5): 1383–1391.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.10 MB