TABLE 2

Summary of Evidence

FindingsStrength of Evidence
Comprehensive programs targeting multiple core deficits
Behavioral programs based on Lovaas/ applied behavioral analysis (ABA)
  We found 16 previous systematic reviews and meta-analyses14–17,19,20,42–51; most reported significant effects in the areas of language, adaptive skills, and IQ. These reviews included only 2 randomized controlled trials; the other studies were either nonrandomized trials or observational studies. There was not enough evidence to suggest the superiority of 1 behavioral curriculum over others.Moderate
  A recent meta-analysis19 found a dose-response relationship for intensive behavioral interventions on both language and adaptive skills outcomes. Higher intensity (hours per week) and higher duration (in months or years) led to better outcomes. Twenty hours per week was the minimum intensity of most comprehensive programs.Moderate
Developmental
  We found one 10-month nonrandomized trial of the Scottish Autism Center comprehensive program.52 Improvements in socialization, daily living skills, and motor and adaptive behavior were reported.Low
Integrative: Behavioral + Developmental
  In a high-quality randomized controlled trial53 and several observational studies, the Early Start Denver Model has reported significant gains in cognitive ability and other core deficits in preschool-aged children.Moderate
  Other programs such as STAR,54 the Walden Toddler Program,55 and ABA combined with TEACCH56 showed improvements in core deficits in 1 uncontrolled observational study each.Low
Environmental support
  There were 2 poor-quality nonrandomized controlled trials of TEACCH to nonspecific educational programs.57,58 Improvements in cognitive function, social skills, and adaptive behavior were reported. Both small studies were conducted in Italy by the same researchers. In both studies, TEACCH was conducted in a residential setting; 1 study also had a TEACCH arm in a “natural setting.”Low
Social skills programs for higher-functioning children and adolescents
  Many controlled trials21,23,24,29,59–65 and observational studies25,26,30,31,66–70 of social skills programs have been conducted. We conducted several meta-analyses on social skills studies that used similar outcome measures. There was moderate, consistent evidence that social skills programs as a whole are effective for both children and adolescents. Effect sizes tend to be significant and fairly large. However, our analyses could not determine which approaches are best for which children. Effective interventions took place in both individual and group settings.Moderate
Interventions for children with no or limited language
Picture Exchange Communication System (PECS)
  One previous systematic review32 reported on 2 randomized controlled trials, 1 nonrandomized controlled trial, and 3 uncontrolled observational studies. Results in communication/social skills were consistently positive in the short term but inconsistent in the long term. The outcome effect sizes varied across studies.Moderate
 Augmentative and alternative communication devices
  There are no controlled trials or observational studies on the efficacy or effectiveness of Augmentative and Alternative Communication (AAC) interventions8; only single-subject studies have been reported.Insufficient
Auditory integration training
  A previous systematic review71 reported no significant improvements in sound sensitivity in 3 controlled trials. None of these trials reported significant improvement in core deficits. There were 2 trials published after the review. One found no improvements in core deficits35; the other found no improvement in sound sensitivity, but did find improvements in language, intelligence, and social skills.72Moderate: for ineffectiveness