Researching the Effectiveness of Sensory Integration

Lucy Jane Miller, Ph.D., OTR, and Moya Kinnealey, Ph.D., OTR/L

Sensory integration has provided the profession of occupational therapy with more research studies than any other theory or treatment approach. Although there are some methodological concerns which need to be addressed over the next decades, it is remarkable that in the short time since the development of sensory integration theory and treatment, so many research projects have been implemented and reported. This article provides a brief overview of issues germane to evaluating sensory integration efficacy research.

Characteristics Of SI Procedures

Sensory integration treatment is a complex treatment modality. Its complexity is due to the fact that there are innumerable variations in sensory integration treatment, based upon the needs of the client, the background of the therapist and availability of treatment services. It is critical to understand what sensory integration treatment is before undertaking an analysis of the efficacy of the treatment as reported in the literature.

A summary of characteristics of sensory integration treatment developed by Kimball (1988) includes the following: active participation by the individual being treated; client-directed activity; treatment which is individualized; activities which are purposeful and require an adaptive response; activities which emphasize sensory stimulation; treatment based on improving underlying neurological processing and organization; and treatment which is provided by a therapist with advanced training in specific sensory integration treatment techniques.

Discriminating Sensory Stimulation From Sensory Integration In Therapy

Distinguishing intervention which includes the characteristics of SI treatment from sensory stimulation, is critical in conducting or reviewing research in sensory integration. Kimball (1988) most specifically addresses this issue. If sensory integration is based on the individual needs and responses of the client, how can treatment be standardized so that it can be studied?

A quantitative researcher strives to have independent variables which are observable, measurable and replicable. This approach may result in what Ottenbacher (1991) refers to as a linear or reductionistic approach, that is, an approach which tries to simplify the treatment to a small number of strictly defined inputs.

An example of a study which specifically defines the treatment input is Ottenbacher, Short & Watson (1981) in which a group of children was administered rotary vestibular stimulation, with head position, speed of rotation and duration of treatment specifically controlled. Is this sensory integration treatment or merely sensory stimulation?

If the tenets described by Ayres (1972), Kimball (1988), Koomar & Bundy (1991) and others are adhered to, it can be seen that research which focuses on one narrow type of sensory stimulation is too limited to be called "sensory integration treatment." It is not client-directed, individualized to the needs and responses of the client, or purposeful, (i.e., requiring an adaptive response by the client).

Although this reductionistic or "narrow" approach has certain methodological advantages from the point of view of conducting a research study, it has severe limitation from the point of view of supplying knowledge about sensory integration treatment. Thus the reader may be able to arrive at conclusions regarding the effect of specific sensory stimulation on a certain patient population, but the findings are not generalizable to sensory integration treatment as a whole.

Ottenbacher (1991) provides a thought-provoking discussion of this issue. "Sensory integration is a multifaceted intervention approach that is difficult to reduce to component parts. To ensure the integrity of sensory integration as an independent variable requires special planning. The description of the treatment, the training of individuals responsible for implementing the intervention, and the development of procedures to monitor delivery of treatment are vital to any successful investigation" (p. 390).

Of particular import to Ottenbacher’s discussion is the need to both operationally define the treatment being provided, and to ascertain that the treatment actually represents the construct which the investigator claims to be measuring. Is a study which provided one type of controlled tactile stimulation to clients for a specified duration (e.g., Wells and Smith, 1983), actually a SI study?

Another important distinction highlighted by Cermak and Henderson (1989, 1990) is the differentiation between perceptual-motor and sensory integration treatment effect studies. "Perceptual-motor programs…tend to be preplanned, therapist directed, structured programs." However, distinctions between these programs are not always provided by the authors in descriptions of their studies, making it difficult to determine if the treatment provided was actually sensory integration. Some basic guidelines to use are to determine if the therapy session was structured, therapist directed, or provided in a group (thus more likely to be perceptual-motor treatment) and if the treatment providers were specifically trained in sensory integration treatment techniques.

Understanding and Describing the Sample

Many research studies are difficult to interpret or to replicate due to the lack of specificity of sample description. In an attempt to synthesize the efficacy of research in pediatric occupational therapy, Clark and Pierce (1988) divide research studies into categories based upon type of procedures employed, diagnosis and type of design. Analysis of the groupings demonstrates that in the cerebral palsy studies, subjects were diagnosed as having spastic hemiplegia, spastic quadriplegia, neuromotor disorder, mental retardation, bilateral spasticity of upper extremities and mixed types.

The second grouping, early intervention studies, includes diagnoses of amputation, Down syndrome, "disadvantaged," "normal," thumb-sucking, cancer and cerebral palsy.

The third group, sensory integration studies, includes attention deficit disorders, learning disabilities, aphasia, autism, mental retardation, "normal," minimal brain dysfunction, prematurity, behavior disorders, developmental apraxia, profound retardation, "at-risk," and cerebral palsy. This overview of sensory integration efficacy studies highlights the diversity of diagnostic categories to which sensory integration treatment is applied. A careful review of primary sources for these studies reveals that few of the studies provide operational definitions for the diagnoses of subjects included.

Learning disabilities, for example, is a broad group. Are the authors studying children who have the classic definition of learning disabilities, i.e., a difference between intelligence and achievement of one standard deviation or more? In many cases, it is impossible to determine from the description noted by the authors. In fact, all of the populations above, including cerebral palsy, "normal," etc., are heterogeneous groupings. There is no reason to assume that the children within any of these diagnostic groupings have sensory integrative disorders. Specific assessment of sensory integrative function is needed in order to make this determination. Therefore, treatment programs designed to alleviate sensory integrative disorders may not affect those children who have a specific diagnosis but do not have sensory integrative dysfunction. "It should come as no surprise that studies using varying definitions and conflicting criteria to identify children as dysfunctional yield different or conflicting results. This occurs despite the application of similar measurement procedures and treatment techniques" (Ottenbacher, 1991, p. 395).

Defining and Controlling the Approach

Although it is not impossible to conduct methodologically sound research on the efficacy of sensory integration treatment, there are some special challenges due to the nature of the groups being studied and the complexity of the treatment approach. What then is recommended for the researcher who wishes to conduct a study which will expand the knowledge base in sensory integration?

Researchers must begin to define and control aspects of their treatment approach, so that the research is replicable and reliable. Kimball’s (1988) criteria are useful in detailing aspects of the treatment approach which should be delineated in research on sensory integration.

1. How much active participation is encouraged? How is it documented and qualified?

2. How much of the therapy session is child-directed? How is that controlled for and documented?

3. Is the treatment approach individualized? Is there a separate treatment plan for each individual included in the study? Are the results of treatment recorded separately or aggregated by the group?

4. Is purposeful activity incorporated into the treatment approach, and is an adaptive response encouraged? How is the effort to achieve these goals documented?

5. Is the sensory integrative input varied based upon the child’s response? To what extent is the treatment plan varied? Is this determined during therapy or after? How is it documented?

6. Are activities designed to provide sensory input with goals of affecting underlying organizational or processing abilities? How is the provision of sensory stimulation documented and how is achievement of the underlying goals measured?

7. How is the training of the therapists quantified? Are they experienced in sensory integration treatment or are they actually administering perceptual-motor techniques?

Choosing an Appropriate Design

There are numerous texts and articles on research design that will be of interest to the investigator planning a research project. Of particular relevance to sensory integration is the applicability of case studies, single subject research designs, meta-analysis of existing studies, and careful (and probably small) studies of equivalent groups.

Three additional techniques hold promise for producing information which will help establish the efficacy of sensory integration: the cross-lag research design, as described by Ottenbacher (1991, p. 393-394); the quasi-experimental research designs (Cook & Campbell, 1979) which have the potential to illuminate relative merits of treatment approaches when compared to one another; and qualitative research approaches (Goetz and LeCompte, 1984).

Space limitations preclude a comprehensive discussion of research design considerations. However, the investigator is urged to carefully think through the research question and the hypotheses, prior to deciding on a design. Many sensory integration studies are hindered by authors who "think it works," but have not clarified the hypotheses. Thus, the study becomes an exploratory investigation of numerous treatment procedures and even more numerous outcome measures, and the likelihood of finding significant results is decreased. Collecting additional data does not necessarily improve the research design!

Cermak and Henderson (1989, 1990) provide a compelling argument for research on the short-term effects of sensory integration treatment, which for the most part has been unreported in the literature. They note that most of the sensory integration research has been based on a pre-test/post-test model where change over time is measured. However, they point out that frequently teachers and parents comment on the immediate improvements in the behavior or learning of the child which indicate that "there is an immediate, or perhaps a very short-term effect of sensory integration procedures" (p. 7). They suggest that immediate changes in behavior, e.g., attention and organization, could be investigated through continuous performance tasks, or paired associated learning before and after sensory integration treatment sessions.

There is no "right" way to study the effectiveness of sensory integration treatment. The correct research design depends on the research question, the sophistication of the investigator and numerous factors intrinsic to the study. Exploration of numerous related questions utilizing a wide variety of designs should be encouraged. The use of consultants with expertise in research design and statistics and knowledge of the constructs of sensory integration is highly recommended.

"A diversity in research approaches ensures that important scientific questions will be addressed from multiple vantage points. The consensus derived from different empirical angles will assist in converting research findings into professional agreement and public confidence" (Ottenbacher, 1991).

Interpreting The Results

There are two critical aspects of interpretation which must be addressed: the limitations of available instruments in occupational therapy; and the importance and variety of outcome measures that are used.

The importance of understanding the psychometric properties and intended purpose of scales which are utilized in a study cannot be overemphasized. Before undertaking research, and even prior to finalizing the research questions, it is useful to explore the options available for measuring the constructs to be studied. Existing research is hampered by a serious lack of well-standardized, reliable and valid measurement tools.

The choice of outcome measures is based on two factors: first, the research question, and second, the validity of the measure. If the question relates to the effectiveness of sensory integration treatment in producing changes in school performance, then academic measures would be appropriate. A review of the literature highlights the fact that in many instances the instrumentation is described, but the initial research hypothesis is not elaborated. It is not of interest that children who receive SI treatment demonstrate changes in test scores, unless documentation can be supplied that indicates that the test scores are indicative of changes in performance or the behavioral constructs which are under study.

Does a change in a test score on an intelligence scale mean that a child has become more intelligent? Probably not. Does the change mean that the child will be performing better in school? Not necessarily, although changes in the child which caused the child’s test scores to improve may be related to changes which also affect school performance. Thus it can be seen that researchers need to clarify the choice of outcome measures and relate that choice back to the hypothesis of the study.

The psychometric properties and purpose of the outcome measures also need to be considered. Is a criterion referenced scale appropriate for measuring change over time in a group of children? It may be valuable to provide descriptive documentation, especially if the scores obtained represent small, discriminating categories; however, if the purpose of the study is to document quantitative changes over time or compare the effectiveness of two treatment approaches, then a scale with a reliable and valid standard score must be used.

Unfortunately, there are few well-standardized scales in occupational therapy available for this purpose. Thus, researchers are cautioned to limit the research questions to those which can be answered using existing measures as well as scales developed outside the profession which are designed to measure related concepts. This does not mean that only standardized tests can be utilized in sensory integrative research, but rather that the choice of instrumentation will in part determine the generalizability of results, and the format of the research question. The obvious implications for occupational therapy with regard to the importance of allocating resources to the development of standardized tests is emphasized once more by this discussion.

Cermak and Henderson (1989, 1990) include the following list of domains in which change in response to sensory integration treatment may be demonstrated: "organization, learning rate, attention, affect, exploratory behavior, biologic rhythm (sleep-wake cycle), sensory responsivity, play skills, self-esteem, peer interaction and family adjustment" (p. 7).

Although intriguing, these domains represent a challenge to the profession in terms of existing measurement technologies. With proper planning, implementation and analysis, it can be expected that progress will be made over the next decade in measuring these important areas.

Building an Empirical Consensus Based on Collective Research

Ottenbacher (1991) raises a variety of complex issues, including the importance of developing a consensual science which supports and documents the effective elements of sensory integration.

Because sensory integration is a complex issue, Ottenbacher notes that there is an "absence of a unifying research paradigm," and cautions that "research is only one component of science, and in fact, can produce little of lasting value unless it grows out of consensus supported by theory" (p. 398).

Clark (1991) notes that Ayres, in her more than thirty year career, "discovered a new paradigm for explaining a variety of neurological disorders in children." Dr. Ayres was a proponent of change; and it is a challenge for all students of sensory integration to follow her lead in asking questions and seeking answers. Our current challenge is to grow beyond the existing knowledge which she dedicated her life to developing, and continue to provide new knowledge which will help all the children and clients whom we hope to service in order to validate the new "paradigm" which she constructed.

Sensory integration is a complex topic. Research which adds to the body of knowledge must be conducted over time, with a large variety of well-defined samples, by numerous researchers, representing large numbers of programs. Questions about the effectiveness of sensory integration treatment may not be answered in an isolated study.

Researchers may be tempted to conduct studies with large numbers of subjects, to answer "important and meaningful" questions. However, information can be obtained from a series of smaller research projects, all of which have a common goal of adding to the body of research in an area.


American Occupational Therapy Association (1988). Efficacy data brief: Research supports efficacy of sensory integration procedures. 3, 5.

Ayres, A.J. (1972). Sensory integration and learning disorders. Los Angeles: Western Psychological Services.

Cermak, S.A. & Henderson, A. (1989). The efficacy of sensory integration procedures. Sensory Integration Quarterly Newsletter, XVII, (4).

Cermak, S.A. & Henderson, A. (1990). The efficacy of sensory integration procedures. Sensory Integration Quarterly Newsletter, XVIII, (1).

Clark, F. & Pierce, D. (1988). Synopsis of pediatric occupational therapy effectiveness. Sensory Integration News, 16(2).

Cook, T.D. & Campbell, D.T. (1979). Quasi-experimentation: Design and analysis issues for field settings. Boston: Houghton Mifflin.

Fisher, A.G. & Bundy, A.C. (1991). The interpretation process. In A.G. Fisher, E.A. Murray & A.C. Bundy (Eds.), Sensory integration: Theory and practice. Philadelphia: F.A. Davis Co.

Goetz, J. & LeCompte, M. (1984). Ethnography and qualitative design in educational research. New York: Academic Press.

Kimball, J. (1988). The emphasis is on integration, not sensory. American Journal on Mental Retardation. 92(5), 423-424.

Koomar, J.A. & Bundy, A.C. (1991). The art and science of creating direct intervention from theory. In A.G. Fisher, E.A. Murray & A.C. Bundy (Eds.), Sensory integration: Theory and practice. Philadelphia: F.A. Davis Co.

Madsen, P.S. & Conte, J.R. (1980). Single subject research in occupational therapy: A case illustration. American Journal of Occupational Therapy. 34, 263-267.

Ottenbacher, K. (1982). Sensory integration therapy: Affect or effect? American Journal of Occupational Therapy. 36, 571-578.

Ottenbacher, K. (1991). Research in sensory integration: Empirical perceptions and progress. In A.G. Fisher, E.A. Murray & A.C. Bundy (Eds.), Sensory integration: Theory and practice. Philadelphia: F.A. Davis Co.

Ottenbacher, K., Short, M.A., Watson, P.J. (1981). The effects of a clinically applied program of vestibular stimulation on the neuromotor performance of children with severe developmental disability. Physical And Occupational Therapy In Pediatrics 1, 1-11.

Wells, M.E., & Smith, D.W. (1983). Reduction of self-injurious behavior in mentally retarded persons using sensory integrative techniques. American Journal of Mental Deficiency. 87, 664-666.

From Miller, LJ & Kinnealey, M (1993), Sensory Integration International, 21(2), 1,3,5-7. Originally published in 1993 by Sensory Integration International, Volume XXI, Number 2. Reprinted with permission.

From Miller, LJ & Kinnealey, M (1993), Sensory Integration International, 21(2), 1,3,5-7. Originally published in 1993 by Sensory Integration International, Volume XXI, Number 2. Reprinted with permission.

Back to Library