Newsletter
Volume 31 | 2008 | Number 3
One Size Does not Fit All: Developing Indivudalized Treatment Protocols for Children with Autism
By Dr. Laura Schreibman, University of California, San Diego
One of the most vexing issues facing those involved in treatment research and treatment implementation for children with autism is the often-reported variability in treatment outcome. Thus while many children improve substantially others show minimal or no improvement. This is true despite the fact that behavioral interventions enjoy substantial empirical support for their effectiveness (e.g., Eikeseth, Smith, Jahr, & Eldevik, 2007; National Research Council, 2001; Sallows & Graupner, 2005). Because of this heterogeneity in outcome, it is the case that the current challenge for behavior analysts is not answering the question of which behavioral intervention is “best” because addressing this question is ultimately likely to do little to solve the problem. Rather, behavior analysts in the area of autism treatment should focus their energies on better understanding the variables affecting treatment outcome and thereby developing strategies allowing us to determine which specific behavioral intervention is best for a given child. In essence, what we need is a strategy, or set of strategies, allowing us to tailor our treatment regimen to the individualized needs of the child and family (Schreibman, 2005).
Developing strategies for individualizing treatments has several important advantages. First, such individualization should allow us to reduce the heterogeneity in treatment response. Second, this should result in an overall increase in the number of children who improve substantially. Third, a technology of individualizing treatment should allow us to be “right” the first time and thus take full advantage of early intervention.
The renowned variability in treatment response even to our most effective treatment strategies suggests the operation of other variables that work to affect outcome. There are several types of such variables that are likely involved. These include child variables (e.g., specific child characteristics and behaviors), parent and family variables (e.g., stress, depression, support), cultural variables (e.g., expectations and treatment acceptability), treatment/target behavior interactions (e.g., some target behaviors may be best approached via a certain type of treatment), and the relatively new area of neurophysiological variables (e.g., certain brain structures or activity patterns may relate to treatment response).
Our laboratory has been focusing on the individualization of treatment for a number of years and what follows is a brief description of four studies in this endeavor. All of these studies have involved child variables as moderators of treatment outcome and serve as examples of the experimental focus on developing a technology for matching treatment to child.
One study (Sherer & Schreibman, 2005) looked at a pattern of child behaviors that predicted outcome with a specific behavioral intervention, Pivotal Response Training (PRT; Koegel, Schreibman, Good, Cerniglia, Murphy, & Koegel, 1989). PRT is a naturalistic behavioral intervention developed in response to identified limitations of more highly-structured, discrete-trial training. (These limitations related mostly to problems with generalization, spontaneity, prompt dependency, and difficulty of implementation.) (e.g., Schreibman, 1988, 2005). PRT is play-based, child directed, conducted within a naturalistic set of interactions, and involves enhancing child motivation and responsivity of multiple cues in the learning situation (see Koegel, et al., 1989 for a complete description).
In this study we looked at the treatment outcome of children with autism who had received PRT during several years of our research. After identifying those children who showed the most positive outcome and those who showed the least positive outcome, we looked at pretreatment videos of the children in a structured laboratory observation to determine any behaviors that discriminated between these groups of participants. In this assessment the child was in a large room with age-appropriate toys and an adult (usually the parent). We scored these observations for various behaviors and identified five specific behaviors that seemed to differentiate the treatment “responders” from the “nonresponders.” We found that compared to treatment nonresponders, treatment responders were those who exhibited more interaction with toys, more social approach behavior, less social avoidance, more verbal stereotypic behavior and less nonverbal stereotypic behavior. We followed this with a prospective study wherein we selected new children with autism aged 4-6 years (matched on IQ, language age, and severity of autism), who either matched the “responder” or the “nonresponder” profile. We then provided them with an intense course (6 months) of PRT. Our dependent measures included measures of language, social behavior, and play. We predicted that those children matching the “responder” profile would show significant positive response to treatment while those matching the “nonresponders” would not. A multiple baseline design across participants confirmed our prediction. Those children matching the “responder” profile made major gains on all dependent measures while those matching the “nonresponder” did not. (In fact, due to the lack of response to treatment, we discontinued the intervention for these children for ethical reasons).
The results of this study led to additional questions. First, was the Sherer and Schreibman (2005) profile predictive of response to any treatment or was it specific to PRT? Second, how important were each of the individual profile behaviors in the prediction of effectiveness? Third, what can we do about the majority of children who fall between “responders” and “nonresponders?”
To address these questions, we conducted a preliminary study (Schreibman, Stahmer, Cestone, & Dufek, submitted) wherein we identified young (age 24-47 mo.) children with autism who matched the Sherer and Schreibman (2005) nonresponders profile except for one of the individual profile behaviors. Three of the children matched the nonresponder profile except they engaged in higher toy contact than those children in the original study and three matched the nonresponder profile except they were less avoidant. These children were then provided with three weeks of intensive PRT treatment within a multiple baseline across participants design. Dependent measures included cued and spontaneous communicative vocalizations. In general, we found that these children responded at a level above that of the nonresponders in the original study but below that of the responders in that study. It was apparent that those children with higher toy contact did respond at a higher level to PRT than those who were less avoidant suggesting that avoidance was not as important as an individual predictor.
To address the issue of specificity of the profile to PRT, the children were also provided with a three-week course of discrete trial training (DTT), a more highly structured behavioral intervention (Maurice, Green, & Luce, 1996). Our preliminary results suggest that the variability in response to DTT supports the specificity of the original profile to PRT. Further research is needed to replicate these findings and to assess the potential role of other individual profile behaviors.
We conducted another investigation looking at the role of peer social avoidance in predicting the effectiveness of an inclusive classroom program on the acquisition of language skills in toddlers with autism (Ingersoll, Schreibman, & Stahmer, 2001). Six children with autism and three typically developing children (all aged 26-41 mo.) participated. In a pretreatment assessment three of the children with autism were found to be highly avoidant of peers and three were minimally or nonavoidant. The children then participated for six months in an integrated toddler classroom. Dependent measures included subsequent levels of peer avoidance and language use. Results indicated that the children who were initially assessed as high peer avoiders a) remained highly peer avoidant after six months and made minimal language gains during this period. In contrast, the children who were initially assessed as low peer avoidant remained so and exhibited substantial gains in language use. These results suggest that peer social avoidance may be a predictor of improvement (at least in language) in an integrated educational setting for young children. Such information may be useful in that it suggests that either highly avoidant children may not be good candidates for such inclusive environments or such children may require intervention to reduce peer avoidance before being placed in such settings.
As noted above, one of the major advantages of individualizing treatments for children with autism is to take best advantage of early intervention. Thus if we can determine which form of intervention is best for a specific very young child, we can be “right” the first time and make the most of the early intervention window. In a large project now in its final stages we are looking at the relative effectiveness of two forms of communication training (PRT and the Picture Exchange Communication System, PECS) for very young (24-48 mo.) children with autism. Both of these communication systems are naturalistic interventions based on principles derived from applied behavior analysis. PRT (Koegel, et al., 1989) is vocally based whereas PECS (Bondy & Frost, 2001) is visually based. With PECS the child uses picture icons to communicate with a communicative partner. Discrimination and use of single icons is followed by “sentence strips” wherein the child expresses desires, uses adjectives, comments on the environment, etc. Both PECS and PRT are commonly used with this population but which intervention is used with any individual child is typically based upon other factors besides specific matching of child with intervention. We are interested in two main questions: Is one of these treatment protocols better in terms of teaching communication skills to young, minimally verbal or nonverbal children with autism? What are the pretreatment behavioral characteristics associated with responsivity to each of the two strategies?
Participants include 39 children with autism (mean age: 29.21 months) who came into treatment with fewer than 10 functional words. The children were matched on early word use (no words at intake versus some words), developmental level, and chronological age. They were then randomly assigned to either PRT or PECS training. The children received 258 hours of treatment distributed over 23 weeks and follow-up assessments were conducted after three months. While our dependent measures include a number of standardized and behavioral assessments, covering language, social, cognitive, play, joint attention, and imitation behaviors, many of these have not been analyzed to date. For presentation here we can discuss the results from the MacArthur Communicative Developmental Inventory (CDI; Fenson, et al., 1993) as these data have been analyzed. The CDI is a standardized parent report measure of early language competence that has been widely used in research with typical and atypical children up to 30 months. We used the CDI Words and Gestures protocol which contains a vocabulary checklist of 346 words frequently found in children’s early vocabularies. The parent indicates which words they believe the child can comprehends and which words they believe the child comprehends and says. For our study we have analyzed raw scores for words produced on this vocabulary checklist as our measure of spoken words. PECS participants were assessed in terms of which phase of the PECS protocol they had mastered at the end of treatment and follow-up.
Our preliminary results suggest that there was no significant overall difference between the conditions on children’s acquisition of spoken words. However we have found that with those children who entered treatment with some words (i.e., 1 to 10 functional words) both PRT and PECS were very effective in teaching spoken words. However, those children who entered treatment with no functional words were much less likely to gain speech. We also have found that for those children who entered treatment with no words, PECS was effective in teaching them many more words they could use functionally (with the icons) than with PRT suggesting that for young children entering treatment with no functional words, PECS might be a promising place to begin. Thus, it looks as if one child characteristic, word use, may be helpful in making a decision as to which communication strategy to employ. However a caveat must be introduced here in that analyses of other dependent measures must be completed before major conclusions can be drawn in this area.
These four studies are offered as beginning stages in a directed line of research aimed at understanding moderating variables that affect treatment outcome for behavioral interventions for children with autism. Our hope is that identifying strategies allowing us to tailor treatment to individual children with autism will produce an overall higher level of significant positive treatment outcome for this population.
References
- Bondy, A. & Frost, L. (2001). The Picture Exchange Communication System. Behavior Modification, 25, 725-744.
- Eikeseth, S., Smith, T., Jahr, E., Eldevik, S. (2007). Outcome for children with autism who began intensive behavioral treatment between ages 4 and 7: A comparison controlled study. Behavior Modification, 31, 264-278.
- Fenson, L., Dale, P., Reznick, S., Thal, D., Bates, E., Hartung, J., et al. (1993). MacArthur Communicative Developmental Inventory. San Diego, CA: Singular Publishing.
- Ingersoll, B., Schreibman, L., & Stahmer, A. C. (2001). Brief report: Differential treatment outcomes for children with autistic spectrum disorder based on level of peer social avoidance. Journal of Autism and Developmental Disorders, 31, 343-349.
- Koegel, R. L., Schreibman, L., Good, A. B., Cerniglia, L., Murpjy, C., & Koegel, L. K. (1989). How to teach pivotal behaviors to autistic children. University of California, Santa Barbara.
- Maurice, C., Green, G., & Luce, S. C. (Eds.). Behavioral intervention for young children with autism: A manual for parents and professionals. Austin, TX: PRO-ED, Inc.
- National Research Council (2001). Educating children with autism. C. Lord & J. McGee (Eds.). Washington, D.C.: National Academy Press.
- Sallows, G. O., & Graupner, T . D. (2005). Intensive behavioral treatment for children with autism: Four-year outcome and predictors. American Journal on Mental Retardation, 110, 417-438.
- Schreibman, L. (1988). Autism. Newbury Park, CA: Sage Publications.
- Schreibman, L. (2005). The science and fiction of autism. Cambridge, MA: Harvard University Press.
- Schreibman, L., Stahmer, A. C., Barlett, & Dufek, S. (submitted). Brief Report: Toward refinement of a predictive behavioral profile for treatment outcome in children with autism.
- Sherer, M. R. & Schreibman, L. (2005). Individual behavioral profiles and predictors of treatment effectiveness for children with autism. Journal of Consulting and Clinical Psychology, 75, 525-538.