Disability and Rehabilitation, October 2006; 28(19): 1169 – 1176
REVIEW
Advancing rehabilitation research: An interactionist perspective to guide question and design
DOREEN J. BARTLETT1, JENNIFER MACNAB2, COLIN MACARTHUR3, ANGIE MANDICH4, JOYCE MAGILL-EVANS5, NANCY L. YOUNG6, DERYK BEAL7, ANGELA CONTI-BECKER8 & HELENE J. POLATAJKO9 1
School of Physical Therapy, Faculty of Health Sciences, The University of Western Ontario, 2Department of Epidemiology and Biostatistics, Faculty of Medicine, The University of Western Ontario, 3Bloorview Research Institute, and Department of Paediatrics, Faculty of Medicine, University of Toronto, 4School of Occupational Therapy, Faculty of Health Sciences, The University of Western Ontario, 5Department of Occupational Therapy, Faculty of Rehabilitation Medicine, University of Alberta, 6Northern Ontario School of Medicine, Laurentian University, 7The Hospital of Sick Children’s Research Institute and Department of Communication Disorders, Faculty of Medicine, University of Toronto, 8The Faculty of Health Sciences, The University of Western Ontario, and 9Department of Occupational Therapy, Faculty of Medicine, University of Toronto, Toronto, Canada Accepted December 2005
Abstract The purpose of this position statement is to propose an interactionist framework to bring together the existing literature and provide a unifying direction for rehabilitation research. The framework comprises three components: the conceptual model, the research question, and the research design. The interactionist conceptual model has been adapted from the World Health Organization International Classification of Functioning, Disability, and Health. The model forms the starting point that guides the specification of the research question, which, in turn, guides the selection of research design. This approach demands that the question takes precedence and that there be an extensive repertoire of research designs, each of which is valued for its ‘goodness-of-fit’ with the question, rather than an a priori, single hierarchical ordering of designs. Research designs must be appropriate for questions that examine the disability experience, development over the lifespan, multifaceted interventions, low incidence conditions, and development of new interventions. Analytical challenges include dealing with confounding, mediating, and moderating variables. Rehabilitation researchers – and those who fund their work – should consider and value the use of diverse research methods to best answer the questions posed from the interactionist perspective.
Keywords: Rehabilitation research, interactionist perspective, International classification of functioning, Disability and Health
Introduction Rehabilitation researchers share a common aspiration: to optimize the lives of individuals living with disabilities. Rehabilitation research, however, has been driven by many perspectives, resulting in a diverse and often disconnected body of knowledge. In 2001, the Ontario Rehabilitation Research Network (ORRN), a multidisciplinary group of rehabilitation scientists, clients, health care administrators, decision makers, and policy analysts, was established. Key objectives of ORRN were to initiate a network for the advancement
of rehabilitation research, set research priorities, advocate for funding of rehabilitation research, and increase the transfer of knowledge into practice and policy. In February 2004, ORRN provided a forum in which to refine the planning of rehabilitation research to better meet the needs of clients receiving rehabilitation services. A subgroup of childhood rehabilitation researchers with backgrounds in biostatistics, epidemiology, kinesiology, occupational therapy, physical therapy, medicine, psychology, rehabilitation science, and speech-language pathology worked together to create a position paper on rehabilitation research.
Correspondence: Doreen Bartlett, PhD, PT, School of Physical Therapy, 1588 Elborn College, The University of Western Ontario, London, Ontario, Canada N6G 1H1. Tel: 1 519 661 2111, ext. 88953. Fax: 1 519 661 3866. E-mail:
[email protected] ISSN 0963-8288 print/ISSN 1464-5165 online ª 2006 Informa UK Ltd. DOI: 10.1080/09638280600551567
1170
D. J. Bartlett et al.
The position paper was endorsed by ORRN in February 2005; a summary of the paper, a power point presentation, and a ‘tool kit’ for rehabilitation research trainees are available at: http://icarus.med. utoronto.ca/crrn/initiatives.asp. Although the focus of the group’s work was on children and youth, we believe that the issues are generalizable to rehabilitation researchers working with people of all ages. As a subgroup, we began by endorsing the interactionist perspective proposed by the childhood rehabilitation theme leaders, Polatajko, Cameron and Rigby [1] as a key orientation for childhood rehabilitation research. We believe that client outcomes are generally influenced by the interdependence (i.e., interaction) of factors, rather than by a single factor [2]. The interactionist perspective focuses on the individual and family in the context of the broader physical, social, and attitudinal environments [3]. This interactionist perspective is reflected in a range of theories or approaches intended to explain status at one point, or changes in status over time of any physical or biological phenomenon. These approaches include general systems theory as it relates to biology or physical systems such as the weather [4], the theory of human ecology as an explanation of how children develop in the context of their families and communities [5], dynamic systems theory as it has been applied to the development of cognition and functional movement in young children [6], and theories of occupational performance as explanations of the multiple factors that contribute to what people do in their lives [7 – 9]. For instance, Bronfenbrenner described the relationship between a child and the environment as a system with properties and momentum of its own resulting from the interaction between the two. In a second example, performance is believed to arise from the interaction among the person, the environment, and the occupation (or task in which the person is engaged). In brief, the idea behind the interactionist perspective is that the whole (i.e., the outcome of interest) is much more than a sum of the influential parts; the interaction among multiple contributing factors is critical in shaping the outcome. At the outset, our group also identified key values related to rehabilitation research (see Table I) to which we are committed, although we do not elaborate on all of these values in the manuscript. In particular, we believe that it is critical to incorporate the perspectives of clients and families in the planning, interpretation, and dissemination phases of rehabilitation research. Most methodological resources describing quantitative research designs do not explicitly incorporate this essential value. Qualitative research designs, in general, and participation action research, in particular, are most comptabile with this value.
Table I. ORRN childhood rehabilitation group values relating to childhood rehabilitation research. We value . the importance of theory [10] . the inclusion of clients and families in the planning of research [11] . the incorporation of the cultural context of children and families in planning research [12,13] . the perspectives of all stakeholders (practitioners, payers, policy makers, and researchers, in addition to service recipients and their families) in planning research [14] . the investigation of all components of the ICF in research designs to better understand complex and chronic childhood conditions [15] . the use of a broad range of research designs and analytical methods [16,17]
The objective of this position paper is to present a framework for rehabilitation research with special emphasis on children and youth. Specifically, we define key terms, describe an interactionist framework of functioning using the International Classification of Functioning, Disability, and Health (ICF) [18], and discuss how this framework may be applied to rehabilitation research to guide the selection of research questions and designs. We discuss an approach to selection of research designs emphasizing a fit with the research question, rather than selection of the topranked design option from a level of evidence table. Finally, we identify key methodological and analytical challenges and solutions. Definitions The ICF model defines function (the ‘positive’ outcome) and disability (the ‘negative’ outcome) from an interactionist perspective. Examples to clarify how these terms are interactionist phenomenon follow in the next paragraph. As an all-encompassing model, neither function nor disability depends on the identification of a specific disease or disorder. Function is the outcome that arises out of the interaction among body structures (anatomical integrity), body functions (both physiological and psychological), activity (the execution of a task or action by an individual), and participation (involvement in a life situation). In contrast, disability is the outcome that arises out of the interaction among impairments, activity limitations, and participation restrictions. Function and disability are modified by context, which is not only internal (personal) and external (environmental) factors, but also the interaction among these components. Personal factors comprise characteristics of the person independent of the health condition. Environmental factors include the physical, social, and attitudinal environments in which people live. Thus, both function and disability are interactionist
Advancing rehabilitation research phenomena because they arise out of an interaction among multiple components, the product of which is more than the sum of the individual contributions. The interactionist perspective captured by the ICF contrasts with a more traditional model that construes disability as arising predominantly from the individual. The traditional model holds that disability arises from a deviation from norms of body structure or function (impairments) subsequent to trauma or a health condition [19]. Rehabilitation, based on the traditional model, used a ‘fix the child’ approach [20]. Research was primarily directed to investigations of impairment, such as understanding the mechanisms of acquisition, the nature of the phenomenon, and strategies to approximate normality [19]. The ICF takes a much broader perspective of disability. For example, a child might have an impairment of motor coordination, an activity limitation in throwing a ball at a target, and a participation restriction by being excluded from playing basketball with classmates at recess. Contextual factors associated with this child’s function might include personal factors (age) and environmental factors (peer attitudes). Disability is the broad concept resulting from the interaction of all these factors. The ICF model also articulates multidirectional influences and consequences. For example, a child’s inability to play with classmates at recess might contribute to secondary low levels of fitness and self-esteem. Application of the ICF and the interactionist model to rehabilitation research and practice Several authors have described the applicability of the ICF, a framework developed through universal consensus, for rehabilitation research and practice. Stucki and colleagues [21] suggest although the ICF has potential to advance rehabilitation and improve communication among practitioners, clients, and researchers, success will depend on its compatibility with measures of the constructs and its ‘practicability’ [21]. The ICF is comprehensive and the time required to complete the checklist makes it impractical for application in either research or practice in its current form. Rentsch and colleagues [22] simplified the framework and developed checklists for use by a multidisciplinary team to successfully organize work in a neurorehabilitation setting. The quality of interdisciplinary work processes improved with a more systematic approach to rehabilitation tasks by team members following the introduction of the ICF. Rehabilitation based on the interactionist perspective encourages the investigation of multiple interacting factors (body structure and function, activity, and participation) in the context of personal and environmental factors. For example, the ICF takes into account the individual’s physical, social, and
1171
attitudinal environments and considers ‘functioning’ as participation on a continuum and not in terms of normality [19]. Using this model, the broad range of factors that may influence function can be identified and therefore research may be directed towards body structure and function (impairments), activity (activity limitations), participation (participation restrictions) or the associated personal and environmental contextual factors, alone or in combination. It is our belief that an interactionist perspective, as captured by the ICF, describes the complexity of disability. However, if the ICF is to provide a framework for researchers that will ultimately provide useful information for practicing clinicians, it must differentiate between fixed and modifiable contextual factors. Furthermore, in its current form, its applicability to analysis is not readily apparent. We propose an interactionist framework that includes a modification of the ICF model specific to rehabilitation research. The primary modification of the ICF is differentiation between fixed and modifiable personal and environmental factors. Modifiable factors are the targets for interventions in rehabilitation and knowledge of the influence of fixed factors can provide prognostic information to assist in realistic goal setting. Figure 1 (adapted from the World Health Organization ICF (Ref. [18] p. 18) with permission) depicts the modified ICF, referred to as the ICF-r (i.e., the ICF for rehabilitation). The ICF-r includes the constructs of primary concern to rehabilitation researchers. Both personal and environmental factors are identified as having fixed and modifiable components. The arrows from the environmental and personal factors have been repositioned to indicate an interaction between body function and structure and activity, and between activity and participation. Although we believe that direct effects are possible, we wanted to highlight the interaction effects. Both an association and interaction (indicated by the ‘X’) have been added between personal and environmental constructs. The shaded portions have been added to the ICF’s original version to highlight the concept that both function and disability are interactionist phenomena among all aspects of the ICF-r. As applied to research, the interactionist framework comprises three parts: the model (described above), the question, and the design. The model is the starting point that brings together the existing knowledge base and guides the specification of new research questions. As always, the research question, in the context of the model, determines the design and analysis. Regardless of the topic, researchers should clearly position their research question in the context of the ICF-r, in addition to framing it within a theoretical perspective [10]. As indicated earlier, many theoretical perspectives are compatible with the
1172
D. J. Bartlett et al.
Figure 1. The ICF-r: an interactionist model of the relationships among all components of Function, and its negative counterpart, Disability (adapted from [18] p. 18, with permission from the World Health Organization).
interactionist orientation of the ICF-r. Although some people have expressed frustration that the ICF, and by association, the ICF-r, does not clearly identify a point of entry (e.g., body structure or function) or desired outcome (e.g., full societal participation), the corollary is that researchers have the freedom to select the focus and direction of investigation. Bartlett and Lucy [15] describe the flexibility offered by the ICF in planning comprehensive outcomes research in rehabilitation. If all rehabilitation researchers were to describe how their proposed research fits into the interactionist model, such as Cieza and Stucki [23] do, rehabilitation research would be integrated in a way that has not yet been accomplished. As a result, existing knowledge would be easier to summarize and gaps easier to identify; thereby advancing rehabilitation research systematically. Methodological challenges and solutions in rehabilitation research The diversity of research questions derived from the ICF-r demands an extensive repertoire of research designs, each of which is valued for its ‘goodness-offit’ with the question rather than an a priori, single hierarchical ordering of designs. The work of clinical epidemiologists in specifying a hierarchy of research designs for a variety of clinical problems [24] is of value to rehabilitation researchers. Nevertheless, this hierarchy has limited application given the complexity and breadth of questions in rehabilitation sciences that are generated by the interactionist phenomena of function and disability. As one example, the Levels of Evidence Table produced by the Oxford Centre for Evidence-based Medicine [24] places ‘outcomes research’ at level 2c evidence (i.e., in the lower half of the table in terms of strength of evidence), when it
might be precisely the design that best matches a key question of interest to rehabilitation researchers [15]. We support others [25,26] in advocating for an approach to selecting the optimal design based on its ‘goodness of fit’ to the research question, rather than its placement in the hierarchy of levels of evidence. Importantly, the use of both major scientific traditions of inquiry, qualitative and quantitative, should be considered in the endeavour to optimize the lives of individuals living with disabilities. Design challenges The website described in the introduction contains a Research toolkit of design options for rehabilitation researchers and researchers in training. The site contains specific examples of research questions relating to understanding the lived experience, theory building, measurement development, describing the pattern of development, investigating risk factors for a certain condition, determining resources needed to provide assessment and ongoing rehabilitation, understanding predictors of outcomes, and determining the effectiveness of rehabilitation interventions. Design options are provided for each research question, complemented with a glossary of terms used. Methodological and statistical resources are listed along with ‘quick tips’ about planning and conducting rehabilitation research, with an emphasis on issues for those who study low incidence populations or, for other reasons, have access only to small samples of participants. Although many questions of interest to rehabilitation researchers are appropriately investigated with research designs used by our colleagues in other health disciplines (as elaborated upon in the Toolkit), there are areas that present unique design challenges. They include understanding the disability
Advancing rehabilitation research experience, development over the lifespan, multifaceted interventions, low incidence conditions, and development of new interventions. Health policy and program development needs to be based on solid research through a variety of both qualitative and quantitative methods. Understanding the experience of disability Families, service providers, and policy makers have many questions relating to a better understanding of clients’ experiences and perspectives at all levels of the ICF-r. Questions such as ‘What is the culture of children with congenital hearing impairments?’ or ‘What is it like to acquire a permanent disability in adolescence?’ are answered best using qualitative research designs such as ethnography, phenomenology, life history, or grounded theory. Questions about the extent and breadth of the experience of disability such as ‘What are the most common barriers to full participation in the school setting?’ may be best answered using a combination of qualitative in-depth interviews and cross-sectional surveys. Understanding development Families, service providers, and policy makers need detailed descriptive information about the functioning over time of children with disabilities in order to understand the multiple influences on functioning as these children grow and develop. Questions such as ‘What is the life course of a child and his or her family after an acquired brain injury?’ can be answered using ethnography. Detailed case studies can offer rich information about unique conditions and novel interventions. Longitudinal quantitative observational studies can provide the best designs if the phenomenon of interest involves interaction with personal or environmental factors or other extraneous influences that either cannot be controlled experimentally or are themselves of interest to study. Many research questions involve the influences of naturally occurring phenomena, such as the nature of family support to improve outcomes, which cannot be randomly allocated to groups. We need studies that are designed to understand the ‘web of causation’ in explaining the complex inter-relationships among the individual, family, and society involved in functioning. A focus on understanding both fixed and modifiable personal and environmental factors associated with outcomes important to clients will better guide rehabilitation decision-making and intervention. Studying complex, multifaceted interventions Designs such as the randomized controlled trial (RCT) are most feasible if the question involves
1173
studying the effectiveness of an intervention that is unidimensional, discrete, non-individualized, and controllable through a highly refined protocol; if the interest is in the group response; and if individual differences can either be controlled or eliminated through randomization [27,28]. These studies are more easily conducted on phenomena at the body structure and function level than the participation level. Frequently, however, the RCT may not be the best fit, given the heterogeneity of the populations and the breadth of interventions required by individuals with complex conditions and their families. If the research question is determining the most effective intervention, and the nature of the intervention is such that it is multidimensional and individualized (rather than unidimensional and standardized), quasi-experimental and analytical observational designs can provide the best options, if a sufficiently large sample is available [29]. Investigators might also consider qualitative designs to ascertain the nature of individualized components of intervention, client and family preferences with respect to interventions, and explorations of the reasons that interventions do not work for some clients [30]. In contrast to the highly controlled intervention protocols required by RCTs, rehabilitation services are typically focused on client’s individual needs, and involve multiple disciplines. In other words, rehabilitation interventions are individualized, complex, and multidimensional; the outcome is not a result of a single intervention, but is a product of multiple interventions [26,31,32]. This is less appropriate for study with an RCT and is better suited to an approach incorporating both qualitative and quantitative methods and a program evaluation design strategy, as suggested by McWilliam and colleagues [33]. In addition to determining the optimal intervention for groups of people, rehabilitation practitioners are interested in knowing about multiple factors that modify intervention effects for individuals [10]. Although the influence of extraneous variables can be built into the design of an RCT, understanding the influences of multiple interacting factors requires a design that makes use of, rather than eliminates, the natural variability that is commonly associated with the clients we see in practice (see, for example, work by Dishman and colleagues [34,35]). Furthermore, the level of client participation in intervention [10], as well as the interaction between the client and clinician [31], differs by ICF level, being relatively greater for interventions targeting participation than impairments. The strategy of double-blinding, in which both the client and care-provider are unaware of treatment allocation, is difficult – if not impossible – to implement in rehabilitation research because clients are active participants in intervention [10].
1174
D. J. Bartlett et al.
The fact that rehabilitation is a dynamic process, in which the client plays a participatory role, underscores the importance of understanding the nature of individual differences among clients in moderating outcomes. Studying intervention in low incidence or highly heterogeneous conditions In rare, low incidence or highly heterogeneous populations, investigators can consider single subject experimental designs. These designs offer a practical methodology for exploring cause-and-effect relationships and sources of variation in treatment response. A variety of designs exist to minimize threats to internal validity. External validity can be dealt with by replicating results across participants (i.e., direct replication) and with varying interventions for individuals with different characteristics (i.e., clinical replication). Developing new interventions In childhood rehabilitation, as for rehabilitation researchers across the lifespan, we are still at the stage of developing theories to support new intervention programs. Questions such as ‘How can we construct an advocacy program to encourage full social participation of children with special needs?’ can be addressed using a grounded theory approach. Single subject experimental designs are recommended as the first step in the development and testing of new interventions. Analytical challenges The ICF-r demands careful consideration of the role of the factors under investigation in each studyspecific conceptual model. In research guided by the interactionist perspective, it is important not only to control for confounding variables, but also to make the distinction among confounding, mediating, and moderating variables. Conceptually, a confounder can be thought of as an alternative cause of the outcome of interest that is not intermediate on the pathway between the factor under study and the outcome [36]. A mediator behaves exactly the same way as a confounder but is distinct in that the mediator falls intermediate on the causal pathway. The mediator represents a process or mechanism that either fully or partially explains the observed effect [37,38]. For example, activity limitations may mediate the association between impairment and participation. A key feature of the interactionist perspective is its acknowledgement of the role of both the environmental and personal context in modifying performance. A moderator specifies the conditions under which a
given effect will occur and the conditions under which the direction or strength of an association may vary. A mediator explains how an effect occurs whereas a moderator explains when an effect will occur. In quasi-experimental designs and observational studies, confounding is generally controlled by multivariable analysis that incorporates possible confounding variables into the model. The effect of any single variable can then be estimated after adjusting or controlling for the contribution of other factors. A common mistake is the inclusion of mediating variables in multivariable models intended to control for confounding. This frequently results in a ‘masking’ of the effect under investigation. When variables are measured at different points in time, establishing the temporal sequence is relatively straightforward. However, when variables are measured simultaneously, the problem is to derive the logical precedence of the variables [39]. The role of the conceptual model guided by the ICF-r is to specify the sequence of variables measured at the same time so as to serve as a guide in model building. For example, in a crosssectional study examining the relationship between motor impairment and participation in basketball, visual acuity may be a confounding variable; however, ball skills would be a mediating variable. Personal factors, such as age and gender, or environmental factors, such as school policies, may moderate the association between motor impairment and participation in basketball. Moderation can be assessed by testing for statistical interaction. Modifiable factors in the environment and person that moderate performance should be considered as key targets for intervention. In contrast, moderators that are fixed personal and environmental factors can be used to assist with prognostic decision-making and realistic goal setting. Sjogren-Ronka and colleagues [40] provide an example of model testing; however, investigations fully exploring mediating and moderating constructs and differentiating fixed and modifiable contextual factors have not yet been published. The final complexity in attempting to apply the ICF framework to our understanding of rehabilitation relates to the frequently hypothesized bidirectional relationships among the many factors. This two-way relationship is not easily modeled using traditional regression techniques and, as such, there is increased use of Structural Equation Modeling to model these more complex relationships. According to Peek [41], Structural Equation Modeling is ideally suited to rehabilitation research designs to test complex conceptual models such as those suggested by the framework we present. The caveat is that as the relationships we wish to model become more complex, so does the analysis. Research teams interested
Advancing rehabilitation research in more comprehensive models of functioning and participation need to include a biostatistician skilled in the development and interpretation of these models and have access to a large enough sample. Summary We encourage all researchers working in the field of rehabilitation to use the interactionist framework presented here. The holistic and comprehensive interactionist perspective captured by modification of the ICF, the ICF-r, provides a conceptual model from which to understand the broad scope of function and disability and rehabilitation research. Legitimate areas of inquiry in rehabilitation research include all components of the ICF-r. Researchers in rehabilitation should continue to ensure that the research question is established first, and then the optimal design is selected from a wide repertoire of designs that provide the best evidence for the various purposes. Researchers should not feel pressured to select the ‘gold standard’ of research evidence (i.e., the RCT) without ensuring that it is the optimal design to answer the question posed in the context of the ICF-r. Using an approach that capitalizes, rather than eliminates, natural variation in our populations of interest and using observational analytical techniques fits well with the multifactorial nature of disability and functioning and the interactionist perspective described by the ICF-r. We encourage funding agencies and grant and manuscript reviewers to consider diverse research methods, both qualitative and quantitative, provided that investigative teams provide strong and clear rationales that the selected method is the best way to address an important research question in the context of expressed values and current knowledge. Acknowledgements We acknowledge the Ontario Rehabilitation Research Network (Childhood Rehabilitation Theme) for their financial support. References 1. Polatajko HJ, Cameron D, Rigby P. The working vision: The interactionist perspective. Presentation at the Ontario Rehabilitation Research Network Workshop: Rehab Research . . . It’s our move, 20 February 2004. 2. Last JM. A Dictionary of Epidemiology. 4th ed. Oxford: Oxford University Press; 2001. 3. Simmeonsson RJ, Leonardi M, Lollar D, Bjorck-Akesson E, Hollenweger J, Martinuzzi A. Applying the International Classification of Functioning, Disability and Health (ICF) to measure childhood disability. Disability and Rehabilitation 2003;25:602 – 610. 4. Von Bertalanffy L. General systems theory. New York: Brazillier; 1968.
1175
5. Bronfenbrenner U. The ecology of human development: Experiments by nature and design. Cambridge, MA: Harvard University Press; 1979. 6. Thelen E, Smith LB. A dynamic systems approach to the development of cognition and action. Cambridge, MA: MIT Press; 1994. 7. Canadian Association of Occupational Therapists. Enabling occupation: An occupational therapy perspective. Ottawa, ON: CAOT Publications ACE; 1997. 8. Kielhofner G. Dynamics of human occupation. In: Kielhofner G, editor. Model of human occupation. 3rd ed. Baltimore, MD: Lippincott, Williams, & Wilkins; 2002. pp 28 – 43. 9. Polatajko HJ. Muriel Driver Lecture 1992: Naming and framing occupational therapy: A lecture dedicated to the life of Nancy B. Canadian Journal of Occupational Therapy 1992; 59:189 – 200. 10. Miller LT, Lee CJ. Gathering and evaluating evidence in clinical decision making. Journal of Speech-Language Pathology and Audiology 2004;28:96 – 99. 11. Meyers AR, Andresen EM. Enabling our instruments: Accommodation, universal design, and access to participation in research. Archives of Physical Medicine and Rehabilitation 2000;81:(Suppl 2):S5 – 9. 12. Campbell SK, Wilhelm IJ. (Eds). Meaning of culture in pediatric rehabilitation and health care. Physical and Occupational Therapy in Pediatrics 1991;11(4):1 – 42. 13. Zawaiza T, Walker S, Ball S. Diversity matters: Infusing issues of people with disabilities from underserved communities into a trans-disciplinary research agenda in the behavioural and social sciences. Bridging Gaps: Refining the Disability Research Agenda fro Rehabilitation and Social Sciences. Washington, DC: Howard University Research and Training Center; 2002. 14. Furher M. Conference report: An agenda for medical rehabilitation outcomes research. American Journal of Physical Medicine and Rehabilitation 1995;74:243 – 248. 15. Bartlett DJ, Lucy SD. A comprehensive approach to outcomes research in rehabilitation. Physiotherapy Canada 2004;56: 237 – 247. 16. Lee CJ, Miller LT. The process of evidence-based clinical decision making in occupational therapy. American Journal of Occupational Therapy 2003;57:473 – 477. 17. Polatajko HJ. The study of occupation. In: Christiansen CH, Townsend EA, editors. Introduction to occupation. The art and science of living. Upper Saddle River, NJ: Prentice Hall; 2004. pp 29 – 45. 18. World Health Organization. International Classification of Functioning, Disability, and Health. Geneva: WHO; 2001. 19. Bickenbach JE, Chatterji S, Badley EM, Ustun TB. Models of disablement, universalism and the international classification of impairments, disabilities, and handicaps. Social Science and Medicine 1999;48:1173 – 1187. 20. King G, McDougall J, Kertoy M, Currie M. What is childhood disability? Focus on . . . distributed by the Research Alliance for Children with Special Needs. 2001;Vol 1, Issue 1. 21. Stucki G, Ewert T, Cieza A. Value and application of the ICF in rehabilitation. Disability and Rehabilitation. 2002;24: 932 – 938. 22. Rentsch HP, Bucher P, Dommen Nyffeler I, Wolf C, Hefti H, Fluri E, Wenger U, Walti C, Boyer I. The implementation of the ‘International Classification of Functioning, Disability and Helath’ (ICF) in daily practice of neurorehabilitation: An interdisciplinary project at the Kantonsspital of Lucerne, Switzerland. Disability and Rehabilitation 2003;25: 411 – 421. 23. Cieza A, Stucki G. Understanding functioning, disability, and health in rheumatoid arthritis: The basis for rehabilitation care. Current Opinion in Rheumatology 2005;17:183 – 189.
1176
D. J. Bartlett et al.
24. Oxford Centre for Evidence-based Medicine Levels of Evidence [Internet]; May 2001 – [cited 2004 August 3]. Available from http://www.cebm.net 25. Tickle-Degnen L, Bedell G. Heterarchy and hierarchy: A critical appraisal of the ‘levels of evidence’ as tool for decision making. American Journal of Occupational Therapy 2003;57: 234 – 237. 26. Hyde P. Fool’s gold: Examining the use of gold standards in the production of research evidence. British Journal of Occupational Therapy 2004;67:89 – 94. 27. Fuhrer MJ. Overview of clinical trials in medical rehabilitation: Impetuses, challenges and needed future directions. American Journal of Physical Medicine and Rehabilitation 2003;82:(Suppl 10):S8 – 15. 28. Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinical epidemiology. A basic science for clinical medicine. Boston, MA: Little, Brown and Company; 1991. 29. Horn SD. Clinical practice improvement methodology: Implementation and evaluation. New York: Faulkner and Grey; 1997. 30. Schumacher KL, Koresawa S, West C, Dodd M, Paul SM, Tripathy D, Koo P, Miaskowski C. Qualitative research contribution to a randomized clinical trial. Research in Nursing and Health 2005;28:268 – 280. 31. Johnston MV. Desiderata for clinical trials in medical rehabilitation. American Journal of Physical Medicine and Rehabilitation 2003;82(Suppl 10):S3 – 7. 32. Whyte J. Clinical trials in rehabilitation. What are the obstacles? American Journal of Physical Medicine and Rehabilitation 2003;82(Suppl 10):S16 – 21. 33. McWilliam CL, Stewart M, Vingilis E, Hoch J, Ward-Griffin C, Donner A, Browne G, Coyte P, Anderson K. Flexible client-driven in-home case management: An option to consider. Care Management Journals 2004;5:73 – 86.
34. Dishman RK, Motl RW, Saunders R, Felton G, Ward DS, Dowda M, Pate RR. Self-efficacy partially mediates the effect of a school-based physical-activity intervention among adolescent girls. Preventative Medicine 2004;38:628 – 636. 35. Dishman RK, Motl RW, Saunders R, Felton G, Ward DS, Dowda M, Pate RR. Enjoyment mediates effects of a schoolbased physical-activity intervention. Medicine and Science in Sports and Science 2005;37:478 – 487. 36. Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia, PA: Lippincott-Raven; 1998. 37. Baron RM, Kenny DA. The moderator-mediator variable distinction in psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology 1986;51:1173 – 1182. 38. MacKinnon DP. Technical assistance report: mediation analysis. [Internet; cited 2005 March 22]. Available from: [http://www.public.asu.edu/*davidpm/rip1/Mediation_Analysis. PDF]. 39. Susser M, Stein Z. Third variable analysis: Application to causal sequences among nutrient intake, maternal weight, birthweight, placental weight, and gestation. Statistics in Medicine 1982;1:105 – 120. 40. Sjogren-Ronka T, Ojanen MT, Leskinen EK, Tmustalampi S, Malkia EA. Physical and psychosocial prerequisites of functioning in relation to work ability and general subjective well-being among office workers. Scandinavian Journal of Work, Environment & Health 2002;28:184 – 190. 41. Peek MK. Structural equation modeling and rehabilitation research. American Journal of Physical Medicine and Rehabilitation 2000;79:301 – 309.