By Lucy J. Miller
Reprinted from NASP Communiqué, Vol. 31, #5, February 2003
In October 2002, "A School Psychologist Investigates Sensory Integration Therapies: Promise, Possibility, and the Art of Placebo" was published in Communiqué (Shaw, 2002). The article elicits this question from school psychologists and other readers: "Should I recommend sensory integration therapy for children believed to have sensory processing impairments?"
Children with sensory processing impairments suffer from devastating symptom complexes that significantly affect their self-regulation, self-esteem, social participation (Cohn, Miller, & Tickle-Degnen, 2000) , school performance and other functional abilities (Parham & Mailloux, 2001) . The impact of these impairments on children and their families is substantial. Rigorous research evaluating the effectiveness of sensory integration therapy is crucial and must continue over the upcoming decades. However, these children and families cannot wait for definitive research evidence. They need support, and they need it now.
I have chosen to respond to Shaw's provocative article by focusing on substantive conceptual issues. First, I address the fundamental flaws in Shaw's contention that sensory integration therapy is ineffective. Next, I briefly appraise readers about the current empirical research on sensory processing impairments.
Many studies exist suggesting that the intervention works (e.g., Kinnealey, Koenig, & Huecker, 1999) . Studies also exist suggesting that it does not work (e.g., Polatajko, 1991). An appropriate scientific conclusion is that the effectiveness of Occupational Therapy (OT) using a sensory-based approach is neither proven nor unproven. Therefore, Shaw's deduction is erroneous, and his condemnation of this type of therapy is premature given the lack of rigorous methodology in previous studies.
The knowledge base from research in this field is in its infancy and substantial work is needed before enough rigorous empirical data are available to proffer valid conclusions about the effectiveness of this intervention approach. Empirical evidence characterizing the physiologic and behavioral manifestations of sensory processing impairments is accumulating (Mangeot, Miller, McIntosh et al., 2001; McIntosh, Miller, Shyu et al., 1999b; Miller, McIntosh, McGrath et al., 1999) Furthermore, in a careful pilot intervention, study, we demonstrated significant improvements when OT was administered in a replicable fashion to a homogenous group of children with physiologic indicators of sensory processing impairments (Miller, Brett-Green, Dickinson & James, in preparation). In addition, our current randomized clinical trial is almost complete. Clearly, answering the question, "Is Occupational Therapy effective for children with sensory processing impairments?" requires a long-term program of research addressing efficacy questions as well as questions about the underlying mechanisms of sensory processing impairments and the validity of sensory processing impairments as a separate diagnostic category.
The question, "Is sensory-based OT an effective intervention?" cannot presently be answered by empirical data. Shaw (2002) acknowledges 41 articles on this topic; we found over 80. While opinions differ about the effectiveness of this intervention, no serious scientist would dispute the fact that studies adhering to rigorous criteria for randomized controlled trials have not yet been published. This is not surprising given the inherent difficulty in conducting this type of clinical trial with children. Effectiveness studies administered in natural settings such as a therapy clinic (as contrasted with laboratory settings) are complex and must deal with multiple potential confounds; by contrast, efficacy studies such as those which test the effects of new medications can be much more controlled. Shaw's statement that "There is no evidence that sensory integration therapy is or has ever been an effective treatment for children with ... disability[ies]" (Shaw, 2002, p. 5) is blatantly misleading, given that in this area of social science, only foundations for sophisticated randomized clinical trials have been laid and no rigorous randomized clinical trial that adheres to all required standards has been published yet.
To be considered reliable and valid, randomized clinical trials must adhere to four key standards (Boruch, 1997) : replicable intervention, homogenous sample, sensitive and relevant outcome measures, and rigorous methodology. Of Shaw's 41 related studies and our 80 identified studies, none satisfies all four criteria for a rigorous clinical trial. Each of the existing studies has one or more fatal flaws, and evidence from flawed studies, no matter how many there are, cannot constitute "proof" that intervention is either effective or ineffective. For example, as Shaw notes, random assignment of subjects is crucial. Only three of the 80 studies we reviewed implemented random assignment (see, for example, Wilson, Kaplan, et al., 1992) . Thus, even using his own criterion, Shaw is mistaken when he states that there are "plenty of quality outcome studies (41 as of this writing) ..." (Shaw, 2002, p. 5) . Until a series of rigorous randomized trials are conducted, the only unbiased scientific conclusion is that no definitive, reliable and valid evidence exists either supporting or refuting the effectiveness of this intervention. This does not mean that intervention does not work. A lack of effectiveness data is not equivalent to a negation of effect. The four standards to which future randomized clinical trials must adhere are detailed below.
Replicable Intervention. School psychologists and others may ask: "What is 'sensory integration therapy'?" Although the originator of the approach used that term decades ago (e.g., Ayres, 1972) , today "best practice" by expert clinicians uses a global intervention -- "occupational therapy" (OT). This difference in terminology is not trivial; and it is not semantic. Shaw's insistence that sensory integration therapy does not work indicates his lack of awareness that current best practice intervention for children with sensory processing impairments is intervention focused on "occupation" as defined below, not treatment using specific sensory techniques in isolation.
OT strives to improve the occupational performance of clients using a transactional approach that considers the person, his/her environment and occupations that are personally meaningful (Baum & Law, 1997) . For example, childhood occupations target daily activities such as eating, dressing, sleeping and playing, as well as school and community activities. Contrary to the uninformed discussion by Shaw, the goal of OT is never to "cure" people. Instead, OT uses purposeful, motivating and meaningful activities in contexts related to life roles to maximize potential and life satisfaction, not to cure individuals. Although OTs may intervene at the impairment level (e.g., to address specific sensory problems in processing tactile, proprioceptive or other sensory stimuli), these interventions are always embedded in occupational functioning (Coster, 1998) . Sensory-based approaches (e.g., OT with a sensory frame of reference or sensory-based OT) are used to establish a foundation for intervention at the level of occupational functioning.
OT is based on dynamic clinical reasoning (Mattingly & Fleming, 1994) . Whenever possible, best practice includes direct treatment to ameliorate sensory processing impairments combined with home, school and community intervention to alter environments/situations ultimately increasing function and life satisfaction. Therapists practicing isolated elements of sensory-based approaches (e.g., strict tactile or auditory protocols alone) are not employing best practice, since they are not intervening within the context of the whole child and his/her environment.
A significant problem in the field is that a published replicable intervention is a sine qua non of a rigorous outcome study. No existing study has published an intervention protocol or a replicable method of establishing fidelity to treatment. A study by Polatajko et al. (1991) highlights the problem created by this lack of replicable intervention protocols. In an otherwise methodologically rigorous study, this team compared children who had received "sensory integration OT" to those who had received "perceptual motor OT," finding that both groups improved. Understanding what differentiates these interventions is impossible without a replicable published fidelity measure for each approach that clearly discriminates the two interventions. Since none of the existing published intervention studies used a replicable intervention protocol, Shaw's conclusion that sensory integration therapy methods are ineffective is unsupportable; his hypothesis must be rejected. Our team has developed a "manualized" (based on a detailed manual) approach to OT for children with sensory processing impairments that has been utilized in our pilot study and our current randomized clinical trial (Miller, Wilbarger, Stackhouse et al., 2001). We have also received a NIH R21 grant to research how well our fidelity measure assesses adherence to the intervention model. Until a randomized trial is conducted with a replicable, manualized protocol, the reliability of previous outcome studies can not be evaluated.
Homogenous Sample. Though Shaw does not detail which of the ~80 studies he selected for his review of 41 studies, few published studies attempted to define a homogeneous sample. The necessity of defining a homogeneous sample is highlighted by a series of studies by Polatajko and Law and colleagues. In one article, they conclude that therapy is ineffective based on the finding that children in both groups made similar changes (Polatajko et al., 1991) . However, in a subsequent article, the authors clarified that one-half of the children made significant changes while the other half did not, concluding that the effectiveness of intervention was obscured by the heterogeneity of the sample (Law, Polatajko, Schaffer et al., 1991) .
Published studies in this field have included diverse participants, such as persons with cognitive disabilities (Huff & Harris, 1987) ; motor disabilities (Jenkins, Fewell, & Harris, 1983) ; aphasia (Jenkins et al., 1983) ; schizophrenia (Bailey, 1978) ; and learning disabilities (Humphries, Snider, & McDougall, 1993) . Without rigorously defined sample inclusion and exclusion criteria, a determination that variation in outcomes is due to either sample or intervention variables cannot be made. In our recent research, we found that even among the relatively homogeneous group identified with idiopathic sensory processing impairments, several distinct phenotypes emerge (Miller, Lane, & James, 2002) . Future research must include specific markers discriminating particular phenotypic features of samples before randomized clinical trials will be reliable (i.e., replicable).
The question "Does sensory integration therapy work?" is na•ve. A more learned question is, "What effects are evident for a specific group of individuals receiving a specifically defined intervention compared to another intervention?" Making a global statement based upon studies with widely divergent samples is unsophisticated. Only a statement related to the effectiveness of the approach utilized in a study related to a specified population would be informative. A careful reviewer would observe that the inclusion criteria in most studies are insufficiently clear to inform conclusions. Thus, the knowledge base defining the phenotypic characteristics of sensory processing impairments must be refined before statements related to the effectiveness of intervention for specific subgroups can adhere to scientific standards.
Sensitive and appropriate outcome measures. Recent qualitative research has shown three child-centered priorities that should be included in outcome studies on this population: self-regulation, self-esteem and social participation (Cohn et al., 2000) . Existing studies focus on: motor (e.g., Huff et al., 1987 ) , academic (e.g., Humphries et al., 1993) , language (e.g., Schroeder, 1982) and sensory-perceptual skills (e.g., Werry, Scaletti, & Mills, 1990) . These studies lack a sound theoretical basis for using these specific performance markers as outcomes tied to the specified intervention.
The 41 studies reviewed by Shaw mainly utilize performance markers as dependent measures, not outcomes that are functional and of importance to children, their parents and their teachers (i.e., they lack social validity). Even though a child's motor or language skills as measured by a standardized test may not show significant improvement, the intervention may have had a positive impact on other important qualities. Since previous studies evaluate neither impairment (e.g., underlying physiologic processing) nor functional changes, Shaw's global statement negating the effectiveness of intervention is precipitate.
Rigorous Methodology. The lack of rigor that characterizes previous studies reflects both the difficulty that all social sciences have in conducting randomized trials in natural (e.g., non-laboratory) settings and the relatively recent shift toward research in the field of OT. Previous studies have contributed to the knowledge base; however, future research must correct existing methodologic flaws to provide results that can be interpreted with greater confidence. Although Shaw correctly identified design impediments, he implies that flaws exist only in studies showing that interventions were effective. His opinion is both biased and incorrect. All existing studies, including the studies that demonstrate that this intervention approach is effective, display one or more of the following three methodologic flaws:
1) Studies of OT effectiveness generally lack randomized assignment to more than one treatment group. For example, of the ~ 80 previous studies, most had only one treatment group; only 13 compared the targeted intervention to no-treatment, and only 11 compared the intervention to an alternative treatment. Only three studies used randomization.
2) The studies generally lack blind assignment to treatment group and blind assessments before and after intervention. For example, only four studies used blind assignment to treatment groups or specified that evaluators were blind to group inclusion.
3) The studies generally lack adequate statistical power to detect significant effects. Most existing studies have such small sample sizes that statistical significance is unlikely to be found even if the intervention has a large effect (Ottenbacher & Maas, 1999) . Shaw seems unaware of the basic introductory principle that much more power is needed to show that an intervention is not effective than to demonstrate it is effective. In other words, the study must have a narrow confidence interval that is close to zero to demonstrate that an intervention does not work, otherwise Type II errors cannot be prevented. Since the sample sizes in these studies were typically small, in each study there would be a high probability that the null hypothesis would not be rejected, even if the population effect sizes were appreciable. Shaw's conclusion that evidence exists demonstrating that intervention does not work is flawed because the power needed to demonstrate this premise is not approached by any existing study. In future studies, power could be improved with extremely large samples, or homogenous samples with exceptionally sensitive outcome measures.
An additional caution to those trying to interpret existing results is that statistical significance must not be confused with clinical significance. Obtaining a statistically significant result requires a large enough sample and sensitive enough measures to obtain (for example) a p value of .05 or less; information of clinical significance is often obtained by examining the trends in data and asking the question, "how do existing studies inform our knowledge base?" Specific questions might include: what outcome measures are the most sensitive indicators of change, which specific populations are the most effected by intervention, which specific intervention approaches are most efficacious and over what period of time. A constructive addition to the existing literature would be a detailed examination of the current effectiveness studies to explicate what has been learned from each study, rather than trying to answer a naive binary query, "Does this intervention work?"
Serious scientists use previous studies to improve future studies, moving the research forward. Careful examination of the ~80 previous studies reveals that important contributions to the study of sensory processing impairments have been made. Shaw appears eager to "prove" a particular approach doesn't work rather than to determine what has been learned from previous research. He is blatantly incorrect in stating, "This is not one of those common cases where there is not enough information upon which to effectively evaluate the treatment." (Shaw, 2002, p. 5) . In fact, none of his references "supporting" that statement refers to a randomized clinical trial. Given the current level of research, scientists should not be surprised with diverse findings. This inconsistency is predictable, given the variation in sample characteristics, intervention methods and duration, and outcomes measured. For these methodologic reasons, we again conclude that no current study "proves" that this therapy approach does not work.
The goal of my discussion here is not to discredit previous studies. While all existing studies have flaws, these limitations inform future studies that address the effectiveness of OT for children with sensory processing impairments. No reliable and valid evidence supports Shaw's contention that sensory integration intervention is "unethical" or based on a "pseudo-science." Shaw's insistence that if "it works," success must be due to a placebo effect, highlights an interesting contradiction. Despite his undeviating contentions that OT is not effective, this odd juxtaposition of a long discussion of placebo effect suggests that he has evidence of positive outcomes from OT.
Before high quality effectiveness studies can be conducted, more objective and direct methods are required to characterize the population with sensory processing impairments. Thus, our research team has developed a laboratory paradigm to measure physiologic responses to sensory stimulation using electrodermal reactivity (EDR) during a sensory challenge protocol (Miller, 1999; 2001). EDR assesses responses to sensory stimuli by measuring electrical changes in the skin. Skin conducts electricity when eccrine sweat glands are activated by cholinergic fibers of the sympathetic nervous system (Dawson, Chodirker, & Chudley, 1995) . Thus, EDR provides an index or marker of sympathetic activity. We have found that the EDR of children with severe sensory processing dysfunction (i.e., diagnosed with fragile X syndrome) differs significantly (p. <.01) from the EDR of typically developing children after sensory stimulation (Miller et al., 1999) . Also, children with idiopathic sensory processing impairments and no other diagnosis demonstrated significantly abnormal EDR after sensory stimulation (p <.01) (McIntosh et al., 1999b) .
In addition, our team uses vagal tone (VT) to measure parasympathetic nervous system functioning (Schaaf, Miller, Sewell et al., in press) . In these studies, we have found that children with sensory processing impairments displayed significantly lower vagal tone than typically developing participants, consistent with other studies that found decreased parasympathetic functioning associated with stress vulnerability, developmental and cognitive delays, and emotional and behavioral over-reactivity. Finding this combination of atypical sympathetic and parasympathetic markers of central nervous system functioning has implications for underlying mechanisms that may be disordered in children with sensory processing impairments.
Furthermore, our research team has developed a short version of The Sensory Profile (Dunn, 1999) that has a coherent factor structure and high internal reliability (McIntosh, Miller, Shyu, et al., 1999a) , discriminating children with sensory processing impairments from typically developing children (p < .01). Parent ratings of sensory processing impairments on the Short Sensory Profile were related to physiologic measures (EDR) of sensory reactivity (p < .01).
Empirical evidence from the psychophysiologic and parent-report tools is critical to future outcome studies. These methods alleviate limitations of previous studies by providing quantitative methods to select a homogenous sample for future outcome studies.
Significant limitations exist in knowledge related to sensory processing impairments and the effectiveness of interventions that ameliorate the underlying deficits and overt manifestations of this condition. Hence, we have a clear and exciting challenge to action. The field offers tremendous promise. We and others are implementing a series of studies to elucidate the underlying mechanisms of sensory processing impairments, to help define the phenotypic characteristics of sensory processing impairments, and to evaluate the effectiveness of OT in remediating the dysfunction. New research with stronger empirical standards is forthcoming. We are on the cusp of an explosion of knowledge in this area. Rigorous scientific study is urgently needed to move the field forward.
Startling is Shaw's use of terms such as "verdict," "unethical behavior," "Guinea pigs," "indifference to facts," "emotionally appealing hypothesis," and "deliberately creating mysteries" (Shaw, 2002, p. 5-6) . Serious scientists engage in conceptual debates, while remaining objective. In a scholarly review, an objective position should be maintained. Unequivocal and emotional statements do not have a role in appraising extant data in social science research.
While controversial articles such as Shaw's promote discussion, changes in practice should occur only as a result of new science and not as a result of emotion-laden appeals on either "side" of a discussion. Unfortunately, insurance companies and school districts may deny services to children in response to unsubstantiated published opinions. By focusing on current empirical research, I hope a more balanced perspective has been presented. Children and families deserve and desperately need help. This help is most effective when it comes from collaborative teams of service providers including both school psychologists and OTs.
The scientific study of sensory processing impairments is extremely challenging. Much remains to be accomplished. Further systematic inquiry requires eliminating emotion and belief from professional dialogues, instead relying on the scientific method to increase our knowledge base. Scientists and practitioners alike must promote research that leads to better diagnoses and effective interventions, improving the lives of children and their families.
Ayres, A. J. (1972). Sensory integration and learning disorders. Los Angeles: Western Psychological Services.
Bailey, D. M. (1978). The effects of vestibular stimulation on verbalization in chronic schizophrenics. American Journal of Occupational Therapy, 32, 445-450.
Baum, C. M., & Law, M. (1997). Occupational therapy practice: Focusing on occupational performance. American Journal of Occupational Therapy, 51, 277-288.
Boruch, R. F. (1997). Randomized experiments for planning and evaluation: A practical guide. Thousand Oaks, CA: SAGE Publications.
Cohn, E., Miller, L. J., & Tickle-Degnen, L. (2000). Parental hopes for therapy outcomes: Children with sensory modulation disorders. American Journal of Occupational Therapy, 54(1), 36-43.
Coster, W. (1998). Occupation-centered assessment of children. American Journal of Occupational Therapy, 52(5), 337-344.
Dawson, A. J., Chodirker, B. N., & Chudley, A. E. (1995). Frequency of FMR1 premutations in a consecutive newborn population by PCR screening of Guthrie blood spots. Biochem. Mol. Med., 56, 63-69.
Dunn, W. (1999). The Sensory Profile: Examiner's Manual. San Antonio, TX: The Psychological Corporation.
Huff, D. M., & Harris, S. C. (1987). Using sensorimotor integrative treatment with mentally retarded adults. American Journal of Occupational Therapy, 41, 227-231.
Humphries, T. W., Snider, L., & McDougall, B. (1993). Clinical evaluation of the effectiveness of sensory integrative and perceptual motor therapy in improving sensory integrative function in children with learning disabilities. Occupational Therapy Journal of Research, 13, 163-182.
Jenkins, J. R., Fewell, R., & Harris, S. R. (1983). Comparison of sensory integrative therapy and motor programming. American Journal of Mental Deficiency, 88, 221-224.
Kinnealey, M., Koenig, K. P., & Huecker, G. E. (1999). Changes in special needs children following intensive short-term intervention. The Journal of Developmental and Learning Disorders, 3, 85-103.
Law, M., Polatajko, H. J., Schaffer, R., Miller, J., & Macnab, J. (1991). The impact of heterogeneity in a clinical trial: Motor outcomes after sensory integration therapy. Occupational Therapy Journal of Research, 11, 177-189.
Mangeot, S. D., Miller, L. J., McIntosh, D. N., McGrath-Clarke, J., Simon, J., Hagerman, R. J., & Goldson, E. (2001). Sensory modulation dysfunction in children with attention deficit hyperactivity disorder. Developmental Medicine and Child Neurology, 43, 399-406.
Mattingly, C., & Fleming, M. H. (1994). Clinical reasoning: Forms of inquiry in a therapeutic practice. Philadelphia: S.A. Davis Company.
McIntosh, D. N., Miller, L. J., Shyu, V., & Dunn, W. (1999a). Overview of the Short Sensory Profile (SSP). In W. Dunn (Ed.), The Sensory Profile: Examiner's manual (pp. 59-73). San Antonio, TX: The Psychological Corporation.
McIntosh, D. N., Miller, L. J., Shyu, V., & Hagerman, R. (1999b). Sensory-modulation disruption, electrodermal responses, and functional behaviors. Developmental Medicine and Child Neurology, 41, 608-615.
Miller, L. J., Brett-Green, B., Dickinson, M., & James, K. (in process). Effectiveness of occupational therapy for children with sensory processing impairments: A pilot study.
Miller, L. J., Lane, A. E., & James, K. (2002). Defining the behavioral phenotype of sensory processing dysfunction. Paper presented at the University of Colorado Health Sciences Center Developmental Psychobiology Research Group 12th Biennial Retreat, "Behavioral phenotypes in developmental disabilities", Estes Park, CO.
Miller, L. J., McIntosh, D. N., McGrath, J., Shyu, V., Lampe, M., Taylor, A. K., Tassone, F., Neitzel, K., Stackhouse, T., & Hagerman, R. (1999). Electrodermal responses to sensory stimuli in individuals with fragile X syndrome: A preliminary report. American Journal of Medical Genetics, 83, 268-279.
Miller, L. J., Reisman, J. E., McIntosh, D. N., & Simon, J. (2001). An ecological model of sensory modulation: Performance of children with Fragile X Syndrome, Autism, Attention-Deficit/Hyperactivity Disorder, and Sensory Modulation Dysfunction. In S. S. Roley, E. I. Blanche & R. C. Schaaf (Eds.), Understanding the nature of sensory integration with diverse populations (pp. 57-88). San Antonio, TX: Therapy Skill Builders.
Miller, L. J., Wilbarger, J. L., Stackhouse, T. M., & Trunnell, S. L. (2002). Use of clinical reasoning in occupational therapy: The STEP-SI Model of Treatment of Sensory Modulation Dysfunction. In A. C. Bundy, S. J. Lane & E. A. Murray (Eds.), Sensory integration: Theory and practice (2nd ed., pp. 435-451). Philadelphia, PA: F.A. Davis Company.
Ottenbacher, K. J., & Maas, F. (1999). How to detect effects: Statistical power and evidence-based practice in occupational therapy research. American Journal of Occupational Therapy, 53, 181-188.
Parham, L. D., & Mailloux, Z. (2001). Sensory integration. In J. Case-Smith (Ed.), Occupational therapy for children (4th ed., pp. 329-381). St. Louis, MO: Mosby, Inc.
Polatajko, H. J., Law, M., Miller, J., Schaffer, R., & Macnab, J. (1991). The effect of a sensory integration program on academic achievement, motor performance, and self-esteem in children identified as learning disabled: Results of a clinical trial. Occupational Therapy Journal of Research, 11, 155-176.
Schaaf, R. C., Miller, L. J., Sewell, D., & O'Keefe, S. (in press). Children with disturbances in sensory processing: A pilot study examining the role of the parasympathetic nervous system. American Journal of Occupational Therapy.
Schroeder, E. (1982). Improvement in academic achievement through enhancement of perceptual and sensory integrative functioning. School Psychology International, 3, 97-104.
Shaw, S. R. (2002, October). A school psychologist investigates sensory integration therapies: Promise, possibility, and the art of placebo. National Association of School Psychologists Communiqué, 31 (2), 5-6.
Werry, J. S., Scaletti, R., & Mills, F. (1990). Sensory integration and teacher-judged learning problems: A controlled intervention trial. Journal of Pediatrics & Child Health, 26(1), 31-35.
Wilson, B. N., Kaplan, B. J., Fellowes, S., Gruchy, C., & Faris, P. (1992). The efficacy of sensory integration treatment compared to tutoring. Physical & Occupational Therapy in Pediatrics, 12(1), 1-36.
Dr. Lucy Jane Miller, Ph.D., OTR, is an Associate Professor of Rehabilitation Medicine and Pediatrics at the University of Colorado Health Sciences Center in Denver, CO. She directs the Sensory Integration Dysfunction Treatment and Research (STAR) Center at The Children's Hospital in Denver, CO and is the Executive Director of The Foundation for Knowledge In Development (KID Foundation) in Littleton, CO. The author and Editor are deeply indebted to Steve Landau (Research Editor) for his editorial suggestions.