Randomized, controlled trial of an intervention for toddlers with autism: The Early Start Denver Model.
Dawson G, Rogers S, Munson J, Smith M, Winter J, Greenson J, Donaldson A, Varley J. Pediatrics. 2010 Jan;125(1):e17-23.

The Early Start Denver Model (ESDM) is the first comprehensive behavioral intervention for ASD shown to be effective for toddlers. This model uses a relationship-based approach that draws on elements of Applied Behavior Analysis (ABA) – a style of therapy that emphasizes breaking activities into small, measurable units that are then reinforced with reward. In this study, researchers performed a randomized, controlled trial of the intervention over two years and showed that it led to improvements in IQ, language abilities, and adaptive functioning (i.e., the skills necessary for everyday activities). It also reduced the overall severity of ASD, with a substantial proportion of the children changing from a diagnosis of autism to a milder classification of Pervasive Developmental Disability, Not Otherwise Specified (PDD-NOS). During the study, 48 toddlers diagnosed with ASD, 1^½ to 2^½ years of age, were randomly assigned to receive the Early Start Denver Model or an intervention commonly available in the community. Children in the Early Start Denver Model group received an average of 15 hours per week working one-on-one with a therapist trained in the technique. Parents were taught to continue the therapy at home for at least 5 hours a week. The other 24 toddlers worked with therapists in the area, receiving a range of conventional therapies, for an average of 9 hours a week and received an additional 9 hours of therapy in a specialized preschool or other group program. Overall, the average number of intervention hours was similar between the two groups. After two years, the children receiving the Early Start Denver Model intervention showed a 17.6 point increase in IQ compared to a 7 point increase in the comparison group working with community therapists. The children in the Early Start Denver Model group also showed continued growth in adaptive behaviors (e.g., brushing teeth, getting dressed, eating with utensils, etc.), while the other children fell further behind their typically developing peers. Notably, 7 of the 24 children in the Early Start Denver Model group had significant enough improvements to change from a diagnosis of ASD to PDD-NOS. Only one of the children in the comparison group had a change in diagnosis. This study is the first randomized, controlled trial of an intervention for toddlers with ASD and adds to the body of evidence supporting the importance of early diagnosis and treatment.

Randomized controlled caregiver mediated joint engagement intervention for toddlers with autism.
Kasari C, Gulsrud AC, Wong C, Kwon S, Locke J. J Autism Dev Disord. 2010 Sep;40(9):1045-56.

A study found that a parent-led intervention for toddlers with autism improved the children's responsiveness to joint engagement and increased the variety of their play. Joint attention skills involve the ability to engage others using eye contact or gestures to communicate nonverbally (e.g., pointing to a toy, looking between an event and a parent to share interest). Research has shown that children with better joint attention skills go on to have better language skills, suggesting that teaching joint engagement may be an important focus of early ASD interventions. During the study, parents were trained to lead 24 therapy sessions over 8 weeks. The interventions incorporated aspects of Applied Behavior Analysis (ABA) and focused on following the child's lead in activities and maintaining his interest. Parents were taught to engage his or her child by imitating his actions, talking about what the child was doing, repeating back and expanding on things the child said, and making eye contact throughout the session. Researchers assessed how strictly parents adhered to the treatment principles and how comfortably they led the session. After completing the parent-led intervention, the 19 toddlers showed that they had successfully moved from play that was primarily object-focused to play with increased levels of joint engagement. However, the intervention only improved the children's responsiveness to joint engagement initiated by the parent and did not improve the children's own ability to initiate joint engagement. While the study showed that most of the parents strictly adhered to the treatment protocol, the level of fidelity did not predict how well his or her child responded to the intervention. The quality of the parent's interaction during the session also did not affect outcome, nor did the type or amount of additional treatment the child was receiving. Notably, the study is one of the first to show that a parent-mediated intervention can improve joint attention skills and diversity of play. The authors note that it will be important to gauge how the effectiveness of parent-mediated interventions compares to those led by therapists.

Evaluation of comprehensive treatment models for individuals with autism spectrum disorders.
Odom SL, Boyd BA, Hall LJ, Hume K. J Autism Dev Disord. 2010 Apr;40(4):425-36.

A study evaluated 30 treatment models for ASD, assessing whether the procedures were well-documented, how many times the model had been replicated, and how much evidence existed to support the effectiveness of the model's interventions. Overall, the study found that the procedures for using the models, most of which were based on aspects of Applied Behavior Analysis, were well-documented in manuals or guides. However, few of the models had developed measures to assess whether the model was well-implemented. Most of the models had been replicated by others – almost half had been repeated two or more times at independent sites. The weakest aspect was proof of effectiveness – 16 of the 30 models had never published research on their effectiveness in a peer-reviewed journal. The authors evaluated the models using journal articles, book chapters, web pages, procedural manuals, and phone interviews with the models' developers. After assessing the evidence, the researchers found that five models scored highly on at least four of the six dimensions that were evaluated: the Denver Model, LEAP (Learning Experiences and Alternative Programs for Preschoolers and Their Parents), Lovaas Institute, May Institute, and PCDI (Princeton Child Development Institute). Models with low evaluation scores included the following: Hanen, Higashi, Eden, Summit, Lancaster, and Son Rise. These rankings can be used as a guide for researchers, service providers, and families faced with treatment decisions. In the future, model developers will need to create measures to verify that the model has been effectively implemented. Additional research will also be necessary to document that the interventions are effective.

Evidence-based practices in interventions for children and youth with autism spectrum disorders.
Odom S, Collet-Klingenberg L, Rogers SJ, Hatton DD. Preventing School Failure. 2010;54(4):275-82.

Investigators reviewed the research literature to identify educational techniques for students with ASD that have sufficient evidence to support their effectiveness. While education programs are expected to be built upon evidence-based practices, currently there are no guidelines to identify whether an intervention is adequately supported by research. After developing a set of criteria, the researchers identified 24 intervention practices that qualify as "evidence-based." The study evaluated individual instructional practices or strategies (e.g., social skills training groups, positive behavioral support strategies, etc.), rather than comprehensive treatment models like the Denver Model or Lovaas Model. For studies to be considered as evidence of the practice's effectiveness, they had to include children with ASD younger than 22 years of age, measure outcome, show that gains in a targeted area clearly followed the use of the practice, and have sufficient controls for comparison. The practice needed support from at least two experimental studies (subjects are randomly assigned to groups) or observational studies (subjects are not randomly assigned but researchers can draw inferences about the effect of the treatment from the data) carried out by independent researchers. Practices that could only be supported by single-subject design – where subjects function as their own control group – would be deemed evidence-based only if there were at least five studies from three independent research groups. (A combination of several single-subject design and experimental/observational studies was also allowed.) After identifying the list of evidence-based practices, the researchers developed a chart matching the practice with the improved skill areas (academic, behavior, communication, play, social, and transitions). Practitioners and families can use the chart to identify which evidence-based practice has been shown effective in the skill area related to the student's learning objectives. The team also constructed web-based modules that include implementation instructions and procedural guidelines for each of the practices. The authors note that the number of evidence-based practices is expected to grow as more literature on ASD interventions is published.