Early Intensive Behavioural Intervention (UCLA YAP Model) and Autism Ranking:

Status Research

There are a number of limitations to all of the research studies published to date. For example

• Some of the studies (such as Perry, Cohen and DeCarlo, 1995; Smith et al, 2000; Butter Mulick and Metz, 2006) had a small number of participants (less than 10).
• Some of the studies (such as Granpeesheh et al, 2009; and Perry et al, 2008) had large numbers of participants (several hundred in each case) but were retrospective case studies.
• Some of the studies (such as MacDonald et al, 2014; and Smith et al. 2000) used single case design methodologies (such as multiple baseline or pre-post test).
• Some of the studies (such as Farrell, Trigonaki and Webster, 2005) used comparative designs (comparing one intervention against another) but did not ensure that the participants in each intervention group were evenly matched.
• Most of the controlled studies (such as Hayward et al, 2009; Magiati, Charman and Howlin, 2007; Remington et al, 2007) where participants were evenly matched were non-randomised.
• The two randomised controlled studies (Sallows et al, 2005; Smith et al, 2000 b) were non–blinded, although the study by Smith did use blinded assessors who were independent of the research team.
• Some of the studies (such as Eikeseth et al, 2007; and  McEachin, Smith, & Lovaas, 1993 were follow up studies of previous studies and reported on the same participants from those other studies.
• Some of the studies (such as Mudford et al, 2001) did not report any useful outcome data.
• Some of the studies (such as Birnbrauer and Leach, 1993) examined EIBI programmes that are similar to but not identical to the UCLA YAP model (such as the Murdoch Early Intervention Program). In other studies (such as Bibby et al, 2001) which looked at programmes which approximated to the UCLA YAP model, it was not clear how closely the programmes actually matched that model.
• Some of the studies (Reed, Osborne and Corness, 2007) compared a number of ABA-based interventions, (such as the UCLA YAP model and the Complete Application of Behavior Analysis to Schools approach) with a range of other programmes but without differentiating the results between the different ABA-based interventions. This makes it difficult to evaluate the efficacy of the UCLA-YAP results.
• Some of the studies (such as Eldevik et al, 2006) looked at less intensive forms of the UCLA YAP model (12 hours per week) making it difficult to know if the  usual, intensive version of the model is effective.
• Some of the identified trials (Eikeseth et al, 2002; Lovaas, 1987; McEachin et al, 1993; Smith et al, 2000) were carried out by staff at the Lovaas Institute or by former colleagues of Lovaas. Those researchers may therefore have been biased towards the intervention, however unconsciously.

A number of systematic reviews have reported the following limitations to some of the studies:

Eldevik S. et al (2009) “First, although not accurately reported in some of the studies, they estimated that the EIBI groups on average received significantly more hours of intervention than did control groups. Second, a variety of assessment instruments were used across children and studies that made it difficult to compare results across studies and may have led to results being spuriously positive. Third, in some studies it was unclear at what points in time the assessments were conducted, particularly at posttreatment when in some cases assessments were undertaken years after treatment had ended. Fourth, the studies reported test scores in different ways that included standard scores, age equivalents, and raw scores.”

Reichow and Wolery (2009) “In the comparative studies reviewed, little is known about the comparison conditions, and little uniformity appears to exist across studies. These groups often lacked standardization within the group, were poorly defined, had no measures of procedural fidelity, and had no data on whether participants received supplemental treatments. Some studies of this synthesis described the comparison group as eclectic (e.g., Cohen et al. 2006; Eikeseth et al. 2007 ; Eldevik et al. 2006 ). By definition, eclectic treatments can vary across participants within a group, thus creating variability within the comparison group. This variability creates a situation in which the treatment effect might be over- or underestimated, and does not create a situation where treatment components can be compared.”

For a comprehensive list of potential flaws in research studies, please see "Why some autism research studies are flawed."

Updated

16 Jun 2022

Last Review

01 Sep 2016

Next Review

01 Jan 2023