CT FEAT: Families Helping Children Achieve Their Full Potential
  Print This Article    

Autism Treatment based on Applied Behavioral Analysis: 
What Does the Current Research Tell Us?

By Richard Irwin (CTFEAT member)

Research has demonstrated that young children with autism can make dramatic progress in intensive treatment programs based on the principles of Applied Behavioral Analysis (ABA). In fact, some children who begin treatment by age 4 and continue treatment for at least two years progress so much that they become indistinguishable from typical children their age. The most thorough study of the effectiveness of behavioral intervention on children with autism was published in 1987 by Dr. O. Ivar Lovaas of UCLA (Lovaas, O.I. (1987) "Behavioral treatment and normal educational and intellectual functioning in young autistic children," Journal of Consulting and Clinical Psychology, 55, 3-9).

The study compared the progress made by three separate groups of children with autism: one experimental group and two control groups. The experimental group consisted of 19 children who were given an average of 40 hours per week of one-to-one behavioral intervention for a minimum of two years. The first control group consisted of 21 children who were given 10 hours or less per week of behavioral intervention, while the second control group consisted of 21 children not treated by Lovaas and his colleagues. A number of the children in the two control groups received a variety of other interventions as well. All children were diagnosed as autistic by professionals not associated with the study. In addition, the three groups of children were shown by a number of standard test measures to be virtually identical prior to treatment. The behavioral treatment addressed all the deficits normally associated with autism spectrum disorders: cognitive, social, behavioral and communication.

In the end, substantial and measurable differences were seen between the experimental group and the two control groups. The experimental group children, as a whole, showed an average gain of 20 IQ points while the two control groups showed no gain at all. Nine children in the experimental group (47%) successfully completed regular first grade without any supports and obtained IQ scores in the average to above average range. These nine children had an average gain of over 30 IQ points and by all measures were normal functioning.

Eight of the remaining ten children in the experimental group demonstrated substantial gains in all areas of development, but were unable to attend school without any support. They completed first grade in special education or language-delayed classes. The remaining two children were placed in classes for autistic or mentally retarded children..
In contrast, only one child in the two control groups completed regular first grade and had an IQ score in the average range. Of the children in the control groups, 53% were placed in classes for autistic or mentally retarded children, with the rest completing first grade in special education or language-delayed classes.

Lovaas and his colleagues published a follow-up study (McEachin, J.J, Smith, T., & Lovaas, O.I. (1993). "Long-term outcome for children with autism who received early intensive behavioral treatment," American Journal on Mental Retardation, 4, 359-372) in which they reevaluated the 9 best-outcome children from the original study when they were about 13 years old (These "children" are today, in 1998, an average of 26 years old since the data for this study was gathered in 1984-1985). In addition to measuring the IQ of these children at follow-up, two other tests, the Vineland Adaptive Behavior Scales and the Personality Inventory for Children, were used to evaluate this group. These tests are designed to detect any psychological disturbances and to determine if a child has the behaviors needed to cope with everyday life. All tests were administered by professionals who did not know the children’s personal or treatment histories.

The results of the follow-up study demonstrated that the gains made by these children persisted. Eight of the 9 children continued to succeed in normal education classes. One child had been placed in special education classes subsequent to the original study, but one child originally placed in special education classes had later been moved to regular education classes. The IQ scores of the 9 best-outcome children were the same as at the end of the original study. Independent examiners were given a mixed group of these best outcome children and typical children to test, using all the measures discussed above. The examiners were given no information whatsoever on the children being tested. These "blind" examiners could not distinguish the best-outcome children from their typical peers on measures of cognitive, academic, social or adaptive skills.

Dr. Jay Birnbrauer and Dr. David Leach of Murdoch University published the best effort thus far at replication of the original study by Lovaas (Birnbrauer, J.S. & Leach, D.J. (1993) "The Murdoch early intervention program after 2 years," Behaviour Change, 10, 63-74). They found that 4 of 9 children receiving behavioral treatment made significant progress and were approaching normal levels of functioning, while only 1 of 5 children receiving no behavioral treatment made significant progress. This study was limited in three ways. First, there was a limited number of children available for the study. Second, the children receiving behavioral intervention were only supplied 20 hours of treatment per week on average as opposed to the 40 hours per week supplied in the Lovaas study. Last, due to a lack of funding, the study was only able to continue for two years. Despite all this, the data is consistent with the results reported in the original Lovaas study.

Criticisms of the UCLA Studies
It is unlikely that any two studies in the field of autism have generated as much excitement and undergone as much scrutiny as the original behavioral intervention study published in 1987 and its follow-up published in 1993. A number of criticisms have been leveled at these two studies, some of which identify legitimate areas for improvement in the studies and some of which are no more that distortions of the truth. The following is a list of commonly seen criticisms of the Lovaas studies. This list is by no means comprehensive. It should be pointed out that five experts in the field of autism wrote commentaries which were published alongside the 1993 article. Each agreed that the 9 best-outcome children appeared to have made significant gains and that these gains could be attributed to the behavioral treatment.

Children were not randomly assigned to the experimental and control groups. This is absolutely true. The suggestion made by critics is that the experimental group was intentionally loaded with higher functioning children and all the lower functioning children were placed in the control group. In fact, distribution between the experimental and control groups was made solely on the basis of resource availability. If there were adequate resources to give a child 40 hours per week of one-to-one behavioral treatment, then the child was assigned to the experimental group. If these resources were not available at the time of referral, then the child was assigned to the control group receiving 10 hours or less of behavioral treatment. The children in the second control group were not assigned to this group by Lovaas and his colleagues but were part of a group being treated by other professionals. Lovaas and his colleagues could have assigned these children on a completely random basis; however, most reputable institutions consider it unethical to assign to groups in this manner. Most ethics review boards prefer children to be assigned based on the availability of treatment. More importantly, the three groups of children were shown by all test measures to be essentially identical prior to treatment, eliminating the argument that the study was rigged for success.

The children in the study were not representative of autistic children as a whole. Some critics have suggested that the children in the experimental group had abnormally high IQ scores at intake and were, therefore, high functioning children who would have done well regardless of treatment. This is just not true for several reasons. First, a recent publication comparing treatment programs ("The Effectiveness of Early Intervention," edited by M. J. Guralnick, (Paul H. Brookes Publishing Co., 1997)) listed the average IQ scores at intake for eight different programs, including the UCLA program. The average intake IQ scores reported for these programs were essentially identical. Second, the experimental group was shown to be identical to the two control groups prior to treatment and yet the two control groups did not make any significant progress. Third, there is no evidence in the literature that higher functioning children with autism make gains regardless of treatment.
Lovaas and his colleagues claimed they cured children of autism. Lovaas and his colleagues demonstrated that it was possible for children with autism to achieve "normal functioning" through intensive behavioral treatment. Nowhere in any of the literature published by the Lovaas group is the claim made that behavioral intervention is a cure for autism.

The best-outcome group did not achieve "normal functioning." Some critics have questioned whether the best-outcome group achieved normal functioning or just attained high functioning status, retaining some residual features of autism. In the 1993 follow-up study, a battery of test measures were used to determine cognitive, social, communicative and behavioral functioning. These test measures were applied to the best-outcome group as well as to typical children by professionals blind to the identity or background of the children. These professionals were unable to detect any evidence of autism.

The 1987 study relied heavily on the use of aversives. Aversives - a sharp "no" or a light slap on the thigh - were used as part of the treatment procedure in a few cases where children exhibited high rates of aggression and self-stimulatory behaviors. New methods, devised by Lovaas and other researchers, have replaced the use of aversives. Currently, no reputable program uses aversives.
Other treatment programs are just as effective. A chapter in a recently published book ("Early Intervention in Autism" by G. Dawson and J. Osterling, in The Effectiveness of Early Intervention," edited by M. J. Guralnick, (Paul H. Brookes Publishing Co., 1997)) reviewed eight early intervention programs for children with autism. Dawson and Osterling state that "there exists little evidence that the philosophy of the program is critical for ensuring a positive outcome as long as certain fundamental program features are present." However, the review is both misleading and factually incorrect in some instances. Of the eight programs reviewed, four (including the UCLA program) utilize the same intensive one-to-one behavioral (discrete trials) intervention used by Lovaas in his 1987 study, while a majority of the rest use behavioral intervention in one form or another. These facts were not made clear in the review. The four programs not using one-to-one behavioral teaching (LEAP, TEACCH, Colorado Health Sciences and Walden Preschool), have not published any data comparing outcomes of children in the program to those in control groups. Comments on the outcomes for these programs, as reported by Dawson and Osterling, are as follows:

TEACCH. It was reported that 4 year old autistic children gained 15-19 IQ points by 9 years of age. In fact, this gain was seen only by the most severely retarded autistic children. In the end, their IQ scores were still in the mentally retarded range. When all of the children examined were included in the comparison (severely retarded, mildly retarded and nonretarded), it was found that no gain in IQ was made from age 4 to age 9. Despite being in existence since 1972, and having published numerous articles in the field of autism, the faculty at TEACCH have yet to publish any peer-reviewed research supporting the effectiveness of their treatment approach. The only evidence offered to demonstrate the effectiveness of the TEACCH program are the results from some parent satisfaction surveys.
Colorado Health Science. According to Dawson and Osterling, this program resulted in the "doubling of developmental rate in several areas." This increase in developmental rate was determined by using a method known as prediction analysis. Prediction analysis attempts to determine what progress is due to treatment and not just the result of normal maturation. Prior to treatment, a child is tested to determine their skill levels in areas such as language or cognition. These measured skill levels are expressed, in months, as the developmental level of the child. For example, a 40 month old autistic child may be determined to have language skills equivalent to a 20 month old typical child prior to treatment.

The developmental level is divided by the child’s actual age to determine a baseline development rate. In the case given here, the child’s developmental rate for language is 0.5, the result of dividing the 20 month developmental level by the child’s actual age of 40 months. Prediction analysis says that this is the rate of development expected of an autistic child in the absence of any treatment. At an age of 50 months the child is expected to have language skills equivalent to a 25 month old typical child. This prediction is made by multiplying the child’s actual age by the development rate (0.5 x 50 months = 25 months).

After a period of treatment, the predicted developmental level is compared to the measured developmental level. If the measured developmental level is higher than the predicted developmental level, then the treatment is considered effective by prediction analysis. If at an actual age of 50 months the child was determined to have language skills equivalent to a 30 month old typical child, then the treatment would be considered effective because the skill level was greater than that predicted (25 months).

Prediction analysis is not always reliable and is not a rigorous evaluation of a treatment. In addition, though the children in the Colorado Health Science program made some progress in the areas of cognition, language, fine motor skills and social skills, these skills never came close to reaching age appropriate levels.

LEAP. Children in the LEAP program were described as making "significant" gains in language, cognitive and motor skills. Like the Colorado Health Science outcomes, these gains were based on a form of prediction analysis. It was also claimed that 50% of the LEAP children go on to placement in "regular education classrooms." However, it is not clear if these children required any supports in these classes. Placement in regular education classes can be a misleading statistic as the children could require extensive supports or none at all (as in the case of the 9 best-outcome children in the Lovaas study). Some school districts mandate full inclusion.

Walden Preschool. The Walden program also used school placement (86%) as fundamental proof of the treatment effectiveness. Walden also observed that language use tripled after treatment. However, if a child was using only a few words prior to treatment, this may not be a significant gain. No measures were taken to show if improvements were made in the areas of cognitive function, social skills and behavior.

In conclusion, none of these four programs have published treatment results that come close to matching those achieved by Lovaas and his colleagues. Most treatments today claim to be effective and to help autistic children "progress." However, since the basis for these claims can vary greatly, these treatments must be held to a higher standard. Parents should ask: What kind of outcomes, supported by peer-reviewed research, does this treatment produce? The goal should not be "any progress" but the maximum possible measurable progress that children with autism can make.

Back to Top


CT FEAT, P.O. Box 7063 Prospect, CT 06712-0063
Copyright 2010 CT FEAT CTContact Us