Archives of Physical Medicine and Rehabilitation journal homepage: www.archives-pmr.org Archives of Physical Medicine and Rehabilitation 2014;95(1 Suppl 1):S33-44
SPECIAL COMMUNICATION
Toward a Theory-Driven Classification of Rehabilitation Treatments Tessa Hart, PhD,a Theodore Tsaousides, PhD,b Jeanne M. Zanca, PhD, MPT,b John Whyte, MD, PhD,a Andrew Packel, MSPT,a Mary Ferraro, PhD, OTR/L,a Marcel P. Dijkers, PhD, FACRMb From the aMoss Rehabilitation Research Institute, Elkins Park, PA; and bDepartment of Rehabilitation Medicine, Icahn School of Medicine at Mount Sinai, New York, NY.
Abstract Rehabilitation is in need of an organized system or taxonomy for classifying treatments to aid in research, practice, training, and interdisciplinary communication. In this article, we describe a work-in-progress effort to create a rehabilitation treatment taxonomy (RTT) for classifying rehabilitation interventions by the underlying treatment theories that explain their effects. In the RTT, treatments are grouped together according to their targets, or measurable aspects of functioning they are intended to change; ingredients, or measurable clinician decisions and behaviors responsible for effecting changes; and the hypothesized mechanisms of action by which ingredients are transformed into changes in the target. Four treatment groupings are proposed: structural tissue properties, organ functions, skilled performances, and cognitive/affective representations, which are similar in the types of targets addressed, ingredients used, and mechanisms of action that account for change. The typical ingredients and examples of clinical treatments associated with each of these groupings are explored, and the challenges of further subdivision are discussed. Although a Linnaean hierarchical tree structure was envisioned at the outset of work on the RTT, further development may necessitate a model with less rigid boundaries between classification groups, and/or a matrix-like structure for organizing active ingredients along selected continua, to allow for both qualitative and quantitative variations of importance to treatment effects. Archives of Physical Medicine and Rehabilitation 2014;95(1 Suppl 1):S33-44 ª 2014 by the American Congress of Rehabilitation Medicine
“Taxonomies are a losing battle, sandcastles shored up against the rising tide of changedbut we fight nevertheless, because they give temporary respite from advancing chaos.”1(pxvi) For some time, rehabilitation researchers and practitioners have called for a classification system by which to organize the controlled chaos of rehabilitation interventions. A sound method for classifying rehabilitation treatments would be useful for consistent identification and labeling of treatments. Consistent labeling of treatments is an essential precursor to quantifying them An audio podcast accompanies this article. Listen at www.archives-pmr.org. Supported by a cooperative agreement from the Department of Education, National Institute on Disability and Rehabilitation Research (grant no. H133A080053). Contents of this article do not necessarily represent the policy of the Department of Education, and you should not assume endorsement by the Federal Government. No commercial party having a direct financial interest in the results of the research supporting this article has conferred or will confer a benefit on the authors or on any organization with which the authors are associated.
(eg, measuring the dose or intensity of particular interventions).2 A classification scheme would also facilitate training and program evaluation, communication and coordination across the disciplines involved in rehabilitation, and both planning and documentation of treatment. Several classification systems have been developed for portions of rehabilitation practice, based on bottom-up or primarily inductive processes.3 That is, treatments have been listed using names or brief descriptions of activities within professional disciplines (eg, occupational therapy, physical therapy).4,5 Although such schemes may be convenient and align well with clinicians’ thinking, they risk separating treatments that are actually similar. For example, treatments aimed at dressing and walking with a cane may share attributes in common to training in sequential activities. Conversely, there is a risk of combining dissimilar activities under the same name, such as the diverse practices that are all called memory (re)training. As well, bottom-up schemes often label treatments by naming the problem that is treated (dressing skills, memory) without specifying how it
0003-9993/14/$36 - see front matter ª 2014 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2013.05.032
S34 is treatedda tautology that helps to keep the mechanisms of effective intervention buried in the black box of rehabilitation.2 In this article, we describe the interim findings of an ongoing effort to develop the structure of a *rehabilitation treatment taxonomy (RTT) based not on names or surface characteristics of interventions but on the theories underlying them. (Words and phrases that are specifically defined in supplemental appendix S1 [available on page A9 of this supplement and online at http:// www.archives-pmr.org/] are marked with an asterisk and italicized when initially used.) In addition to serving the purposes previously described, a classification system based on the theories that explain treatment effects would do much to advance rehabilitation research. Such a system would facilitate the specification, testing, and dissemination of effective treatments; enhance measurement of treatment fidelity; and simplify the process of organizing the knowledge gained (eg, for systematic reviews).6 We have used a number of methods to arrive at this point in the development of the RTT. These include a structured literature search for articles on treatments from all of the main rehabilitation disciplines. We have also sought periodic input from a multidisciplinary advisory committee. However, most of the concepts subsequently described have evolved from discussion among the authors, as we worked through a number of ways to conceptualize treatment theories and possible treatment classification schemes, and tested their resilience to numerous examples of real and hypothetical interventions. Although this work is far from its final form, we present it in the hope that input from the field will provoke more discussion to shape the future of this effort, the challenges of which are subsequently outlined in this article and highlighted in another article in this supplement.7
Theory-driven Taxonomy of Treatments The term *taxonomy can be applied to almost any means of categorizing items in such a way as to show their relations to one another. As an everyday example, a shopping list is a simple taxonomy, classifying items as “things I need,” which could be subdivided by where to find them (grocery store, drug store, etc). A thesaurus is another type of basic taxonomy showing the relations among words and their meanings. A more complex type of taxonomy, a hierarchical tree structure, is illustrated by the familiar Linnaean scheme showing relations among living things, originally based on anatomy and later on genealogy.1 Considering a rehabilitation taxonomy based on theories requires a brief review of what kinds of theories apply to rehabilitation, and how they might be used to sort and interrelate treatments. Rehabilitation is often said to be lacking in theories. Historically, the field has focused on pragmatic rather than theoretical concerns.8 Moreover, the assumption that many rehabilitation *patients would improve regardless of the specifics of the intervention means there has not been pressure to develop, define, and refine treatmentsdactivities that typically force theory developmentdcompared with the situation in fields such as oncology. Nonetheless, rehabilitation does benefit from several types of theory. There is theory focused on explaining how disabling conditions come about, their causes, and risk factors, all of which
T. Hart et al are important for both treatment and prevention efforts. *Enablement/disablement theory9 helps to establish links among impairments and activity/participation restrictions. Understanding these links is important for selecting treatments that will have the maximal impact at the desired level(s) of function, and for refining predictions about the distal or cumulative effects of intervention on a range of outcomes. Enablement theory allows us to hypothesize effects beyond the direct focus of treatment, on *aims that may be down the line from the treatment setting. For example, we may hypothesize that improvement in walking will follow quadriceps strengthening, or improved relationships will follow social skills training. For the RTT, however, we are concerned with *treatment theory, which explains “the actual nature of the process that transforms received therapy into improved health.”10(p34) Treatment theory specifies how and why a treatment works. This type of theory is essential for research that aims to create sound treatments or to make existing treatments stronger, more effective, more efficient, or better suited to the characteristics of specific patients. Treatment theory is essential for isolating and testing the *active ingredients of interventions: these are the attributes of treatment that effect the desired change, and that serve to define and organize treatments and to distinguish them from one another.
Structure of Treatment Theory A brief review of a central concept will place the succeeding material in context. The conceptual framework of the RTT assumes a tripartite structure for every treatment theory.11 Figure 1 illustrates the 3 parts and shows their causal direction (top arrow) in effecting change in the *recipient of treatment, who is commonly a patient but may also be a caregiver or other person. A given treatment may be hypothesized to exert its effects on the *target (a specific aspect of the recipient’s functioning that is selected for change) via some *mechanism of action that is put into play by active ingredients administered by the therapist. The term “ingredients” is borrowed in its medical connotation from pharmacology, although most ingredients in rehabilitation are behavioral. For instance, when a therapist explains a task, praises the patient’s effort, or engages the patient in goal setting, the therapist is administering ingredients just as a physician administers medication or an orthotist provides a brace.
List of abbreviations: ICF International Classification of Functioning, Disability and Health RTT rehabilitation treatment taxonomy
Fig 1 Casual and temporal aspects of the tripartite structure of treatment theory.
www.archives-pmr.org
Toward theory-driven classification As shown in the bottom arrow of figure 1, the process of defining (planning) a treatment flows in reverse temporal order to the causal chain. That is, the therapist starts by considering the target (What aspect of this person’s functioning am I trying to change?), then considers the mechanism (How might changes be effected in this target?), and finally proceeds to assemble and implement the ingredients (What will I do to put the mechanism of action in motion?). In this framework, both the target and the ingredients are measurable, but the mechanism of action is generally unobservable and must be inferred. This is especially the case for interventions that rely on learning. Learning is assumed ultimately to entail some alteration in synaptic transmission in the brain as a result of experience. Because these changes are invisible at the current state of the science, it may be more fruitful to examine certain ingredients, or clusters of ingredients, for which there is empirical evidence for their causing behavior change. For example, feedback is an active ingredient in learning for which the evidence is incontrovertible. We cannot see the effects of feedback on the brain in real time, but we can explore the mechanisms by which it works by examining the effects of qualitative and quantitative variations in feedback on behavior. We can approach a mechanistic understanding of feedback by discovering, for example, that knowledge of performance and knowledge of results are differentially effective on different kinds of tasks,12,13 or that external feedback enhances some kinds of learning and inhibits others.14,15 Therefore, in the subsequent discussion, we focus more on ingredients than on mechanisms; however, it is clearly beyond our scope to review all literature relevant to either of these. What does it mean to say that ingredients, like targets, are measurable? Ingredients may be measured as the result of selections and decisions on the part of the clinician. Ingredients are delivered through drugs and devices, training in tasks and strategies, and other, specific clinician behaviors that are meant to effect changes in patient functioning. As shown in figure 1, therapists work from selecting a measurable target, to hypothesizing the mechanisms that will change the target, to selecting clusters of ingredients (ways of manipulating tissues, stimulating organ systems, transmitting information, or delivering structured experiences) that engage the hypothesized mechanism of action. The efficacy of treatment can only be determined if the target is measurable, meaning that the intended change, or lack thereof, can be detected. Ingredients must also be measurable if they are to be systematically varied to bring about the optimal change in patient function. Variations in ingredients may be qualitative (eg, when a therapist administers a certain type of feedback) or quantitative (eg, when a therapist changes a *dosing parameter, eg, the frequency or schedule of feedback). Ingredients may be visible and measurable in a single *session of treatment, or they may be reflected in systematic variations over a *course of treatment involving multiple sessions. It will be obvious to clinicians that the linear relations conveyed in figure 1 represent a vast simplification of the actual process that unfolds during a treatment episode. During that process, the interpersonal interaction between the treater and the recipient of treatment adds many ingredients or modifies the delivery and the effect of ingredients already in the equation. In real time, clinicians modify their decisions continually according to the reactions and characteristics of the person treated. By presenting the elements of treatment theory as unidirectional in figure 1, we do not intend to imply that patients are passive recipients of treatment. Indeed, the fact that this is seldom the case www.archives-pmr.org
S35 adds to the richness and complexity of rehabilitation, and contributes numerous challenges to any attempt to classify its treatments.16 Similarly, the use of the term “target” to represent the specific aspect of functioning intended for change is not meant to imply that people treated in rehabilitation should be objectified or that clinicians can or should disregard the whole person. In introducing this scheme, which will inevitably strike some readers as overly reductionistic, the intent is not to explain everything that happens in rehabilitation, but to foster discussion of the utility of the tripartite structure to organize one’s thinking about treatments and the way they act to produce positive changes in human functioning. Some of the issues surrounding the tension between mechanistic models and clinical reasoning are addressed in another article in this supplement as well.7
Target Domains and Treatment Groupings of Rehabilitation In developing the framework of the RTT, we have identified 4 superordinate *treatment groupings that differ from one another in all aspects of treatment theory. That is, they are addressed to different target domains, work via different mechanisms of action, and are characterized by different *essential ingredients. These and other attributes of the 4 groupings are subsequently discussed and are summarized in table 1 with respect to their typical targets, mechanisms, ingredients, and dosing parameters. Table 1 also includes examples of familiar clinical interventions that rely heavily on each treatment grouping. We have labeled these 4 treatment groupings by the types or domains of targets that they include, in recognition of the target-first thinking that characterizes treatment planning (see fig 1). At first glance, this may seem to resemble classifying treatments by goals or deficit areas (mobility, memory, dressing, etc), a practice we previously identified as insufficient. However, our definition of the target of treatment refers to a more specific aspect of functioning than is captured by terms such as these. In addition, these 4 groupings represent domains of targets, which we expect to respond to coherent classes of treatment ingredients, guided by coherent classes of treatment theories. That is, we conceptualize these groupings as containing many different treatments, which are related by virtue of similarity in their tripartite structure (targets, mechanisms, and ingredients).
Structural tissue properties As the name implies, treatments in this grouping are intended to change the structural qualities of tissue (eg, the size or elasticity of muscle or soft tissue) by applying various forms of energy according to a prescribed schedule. As shown in table 1, many of these treatments serve to elongate or enlarge tissues. Examples include the various methods of stretching tendons (eg, serial casting and passive range of motion) and wound healing treatments. Another example in the rehabilitation literature is a bladder training intervention in which increased fluid intake stretches the bladder muscle and helps to restore muscle tone. This, coupled with a voiding schedule, has been used to restore continence.17 Massage to reduce tissue adhesions also fits in this grouping. Although a structural change is the target of this treatment, range of motion and/or pain may be expected to be improved indirectly, via enablement theory.
S36 Table 1
T. Hart et al Treatment groupings by target domains in a preliminary taxonomy of rehabilitation treatments
Attributes of Groupings
Structural Tissue Properties
Typical targets
Organ Functions
Skilled Performances
Size (eg, length) Shape Flexibility
Output Efficiency Capacity (eg, percent of normative function) Response dynamics of system
Mechanisms of action
Tissue remodeling processes (macroor microscopic)
Speed/efficiency of performance Quality (compared with standard of completeness, independence, appropriateness to context, etc) Automaticity Learning: mix of implicit and explicit mechanisms
Essential ingredient
Application of energy to tissues
Up- or down-regulation Nonvolitional learning mechanisms (habituation, classical conditioning) Substitution of function Change in functional Facilitation of performance output of organ (system) on part of recipient
Active ingredients (examples)
Type(s) of energy Methods to enhance applied (thermal, effort on part of mechanical, recipient (eg, electromagnetic, etc) motivational interventions)
Typical dosing parameters
Amount, intensity, schedule of energy applied Progression in physical demands on tissue to maximize change Clinical examples Tendon lengthening (ranging, casting) Wound healing
Progression in demand on system to maintain optimal challenge level
Muscle strengthening Cardiovascular endurance exercise Deep brain stimulation Tilt table (up-regulate baroreceptor system) Transcranial direct current stimulation Prosthetic limbs
Within this treatment grouping, decisions about ingredients that may be empirically pitted against one another in the search for the best treatments for a given type of problem and patient involve consideration of several factors. These include the type(s) of energy that may be used, other attributes of devices that may supply the energy, dosing parameters (eg, the amount and schedule of application), and the type and speed of treatment *progression. Progression refers to algorithms by which demands on tissue properties are gradually increased. For example, accelerating the tissue stretching demands of serial casting by using 1 to 4 days versus 5 to 7 days between cast placements results in faster improvements in range of motion and less pain and fewer pressure ulcers.18 It will be seen that progression is a critical parameter for other treatment groupings as
Cognitive/Affective Representations Amount, completeness, accuracy of knowledge Changes in emotional reactions, emotional nuances of cognition (attitudes, beliefs, etc)
Semantic/affective information processing Semantic memory
Facilitation of acquisition of information on part of the recipient Set manipulations (instructions, Attributes of information rationales, motivational aids) presented (modality, Coaching guidance/cues organization, complexity, etc) during performance Methods of facilitating Feedback, reinforcement, acquisition (didactic, response to error socratic, prompting, Provision of strategies modeling, persuasion, etc) Methods to promote Methods to enhance generalization comprehension/retention/use Progression in demands Amount of information on performance to maintain per unit time optimal challenge level Amount/spacing of Schedules of practice repetition and rehearsal (intensity, number/ duration, patterning) Training in gait, activities of daily living, use of memory orthotics, alternative responses to anger triggers, etc Dexterity/motor control exercises Mental rehearsal of physical movements
Patient/caregiver education Adjustment counseling/ psychotherapy Referrals to community services How to (without practice)
well, and underlies the important clinical principle that a systematic, but not overwhelming, increase in demand is a powerful agent of functional change. Further subdivision in this category will most likely be according to the types of energy delivered to tissues and, hence, the types of tissue changes that are likely to result. For example, mechanical energy may result in elongation or enlargement of tissues, whereas heat energy may affect tissue flexibility or elasticity. In some real-world situations, multiple forms of energy are combined or used sequentially to produce the maximum clinical effect. For example, heat could be used to make tissue more pliable, followed by mechanical energy to stretch it out. In our scheme these would count as separate treatments, each with its own target, essential ingredient, and mechanism of action. www.archives-pmr.org
Toward theory-driven classification
Organ functions This grouping includes treatments in which the functions of organs or organ systems are modified, often by systematic stimulation in order to increase or decrease system output, or to bring function to a desired level through natural adaptive processes. The latter may include up- or down-regulation, and habituation to repetitive stimuli. For example, a tilt table may be used to upregulate the baroreceptor system; systematic exposure to sudden noises may reduce startle responses through habituation. Certain forms of massage may also fall into this grouping because massage may alter the sensitivity of pain receptors. This example is notable because massage was also described in the structural tissue properties grouping. The inclusion of similar treatments in 2 groupings illustrates that classification is indeed based on the specific treatment theory hypothesized to account for treatment effects. That is, massage might be hypothesized to reduce pain directly by altering the output of pain receptors, a mechanism belonging to the organ functions group, or the hypothesized mechanism may exert indirect effects by altering the structural properties of soft tissues that cause pain (structural tissue properties group). This is but one example of a case in which empirical research would be needed to clarify the mechanism by which pain is relieved. It may be that massage-related ingredients invoke 2 different mechanisms simultaneously in every case, or it may be that different ingredients or dosing parameters in massage (eg, amount or location of pressure) can shift the balance toward one or the other mechanism that may prove more effective for a given clinical situation. The changes in organ function produced by treatments in this grouping may be permanent or short-lived. For example, both psychostimulant medication and electrical stimulation of the brain from outside the skull (eg, transcranial direct current stimulation) invoke temporary changes in brain function. A different type of temporary modification in brain function is exemplified by immersion in a virtual reality snow world for reduction of pain during wound treatments (eg, in patients with severe burns).19 Here, the presumed mechanism is to capture the attention system by stimulation strong enough to divert it, at least partially, from the pain signal. Another example of a treatment capitalizing on a similar mechanism is playing soothing music through headphones to patients with dyspnea while they are walking.20 The music is hypothesized to divert attention from unpleasant and anxiety-provoking bodily stimuli (eg, shortness of breath), allowing patients to walk with less discomfort. Other treatments in this grouping include those whose mechanism involves substituting for the normative function of an organ (system) by artificial means, presumably for longer-lasting change. Cochlear implants, prosthetic limbs, and deep brain stimulation for treatment of Parkinson’s disease, which provides an artificial stimulus for neurotransmitter release, are some examples of this (see table 1). The musculoskeletal and cardiovascular systems are each the focus of change in many common rehabilitation treatments, such as progressive resistance exercise for muscle strengthening and aerobic exercise for cardiovascular endurance. In aerobic exercise, the aim is to improve the efficiency of oxygen use by gradually pushing the system’s capacity via a systematic increase in demand. This serves to maintain a steady degree of cardiovascular challenge as the function of the system improves. The same principle is in play for progressive resistance exercises for strengthening specific muscle groups. As the
www.archives-pmr.org
S37 muscles adapt to the demand and their force production capacity increases, the weight and/or number of repetitions must also be increased to induce further improvements in strength. Thus, as for the structural tissue properties grouping, dosing parameters that characterize the nature of treatment progression are important for many treatments in the organ functions category (see table 1). This grouping includes both *volitional treatments and *nonvolitional treatments, that is, some of these treatments do not require any active participation (beyond a minimal level of cooperation) to exert their effects on the recipient. Medications and brain stimulation fall into this category. However, others (eg, exercise-based treatments) do require effort on the part of the recipient. Their effects are, therefore, likely to be influenced by active ingredients that are directed toward motivation, as shown in table 1. Such ingredients are not essential for this group because it is the organ stimulation at a given level of challenge that induces the desired change, and this can be achieved by both active and passive means. For example, muscles can be strengthened through volitional exercise or by electrical stimulation, although the effects on the muscle are not identical with these 2 methods. Although learning is not required for all treatments in this grouping, it is common to pair them with learning-based treatments in clinical practice. For example, a therapist may provide a home program by which a patient can carry out muscle strengthening exercises. Learning the routine is not essential for the mechanism of action to work on the target of muscle strengthening. As long as progressive resistance is applied, the muscle will become stronger whether the patient does the program at home or under the continual instruction of the therapist. To help ensure that the patient can continue the program outside of therapy, the therapist might administer other, volitional learning treatments (which we situate in different treatment groupings, which will be subsequently described) to affect other targets beyond muscle strengthening. These might include patients’ understanding of the importance of the muscle-related treatment, their ability to remember to perform it independently, and their ability to self-monitor progress. Although volitional learning is not involved in this treatment grouping, some passive types of learning are inherent in certain treatments in this grouping where the brain is the organ expected to change. For example, rehabilitation treatments may include systematically pairing (or uncoupling) stimuli to alter the circumstances leading to an automatic (unconditioned) response. A familiar example is the case in which a patient develops fear responses to clinicians and/or treatment settings that are associated with pain. In addition to trying to reduce the pain itself, either by addressing the cause or by diminishing the sensation, the rehabilitation team can also work to attenuate the fear response by systematically changing the clinicians or settings that are linked to that experience (counterconditioning). If learning is defined as relatively long-lasting behavior change gained through experience,21 we might also consider sensory habituation, which can be used for vestibular treatments in addition to noise accommodation, as previously described, as a primitive form of learning. All of the learning processes in this treatment grouping are nonvolitional. They are considered primitive in that they are shared by many nonhuman species and require no verbal mediation. Although wakefulness is required for nonvolitional learning, it is thought to occur without the need for conscious awareness of the learning episodes or the stimulus-response pairings involved, although this remains controversial.22
S38 At first glance, these examples of organ function treatments may appear too disparate to belong in 1 grouping. The commonality of these treatments, reflected in their essential ingredient (see table 1), is that they alter the functional output of the organ in question. This alteration may occur either by changing the dynamic relation between inputs and outputs of an organ or system, or by substituting for its functional output. In the example of aerobic exercise, input-output dynamics are changed in that the same level of physical demand (input) results in less heart rate increase (output) after exercise training than before. After habituation or classical conditioning, the same stimulus (input) may produce less autonomic stimulation or an attenuated response (output). As previously noted, substitution (by artificial means) for the normative output of a damaged system occurs with treatments such as prosthetic limbs and deep brain stimulation. Further subdivision within the organ functions grouping may involve first specifying the bodily system selected for change, and then the methods for enacting the desired functions. However, principles that may apply to general mechanisms for enhancing change (eg, the progression of functional challenges or patterning of stimuli to effect optimal adaptation) may cut across organ systems.
Border Zone Between Structural Tissue Properties and Organ Functions At this point, many readers will recognize a quandary that we have faced repeatedly over several years of categorizing and recategorizing treatments: no matter what treatment groupings we compose, the boundaries between them are permeable. Determining whether the target involves changing tissue properties, versus changing organ functions, is difficult because organs are made of tissues, and because the structure of tissue can be intimately related to its function. For example, a leg amputation removes a body structure and also impairs a body function (weight bearing). Providing a prosthetic limb substitutes for the structure but also enhances the function. Similarly, one may seek to stretch the detrusor muscle of the bladder, which would represent a structural change in the bladder’s architecture. However, the same change also increases the bladder’s storage capacity, that is, changes its function. When organ function is changed without a macroscopic change in structure, it is easier to make the distinction. However, when both structure and function are changed, then we need a clearer method of selecting the most relevant treatment target. This and similar challenges to the future and nature of the RTT are subsequently discussed in this article and in another article in this supplement.7
Treatments Based on Volitional Learning We now consider 2 treatment groupings in which all mechanisms of action involve learning and all treatments are volitional. That is, they require active engagement and effort on the part of the recipient, who might be a patient, caregiver, or employer. The interventions in these 2 categories make up what has been termed experience-based treatment.16,23 As clinicians, we expect patients and caregivers to emerge from the experience of rehabilitation having learned new facts, concepts, skills, attitudes, and habits, and to have adapted their prior learning to the new context of disability. However, few rehabilitation practitioners receive a
T. Hart et al strong foundation in learning theories during their training,24,25 and little rehabilitation research is based on explicit models of learning. In developing the theory-driven RTT, we have considered the implications of theoretical frameworks developed over the last century to explain human learning, such as behavioral learning theory, social learning theory, cognitivism, and theories of motor learning. These macrolevel theories differ in emphasis according to their historical contexts, the phenomena of particular interest to their developers, and other factors, such as whether the theories were also meant to explain learning in nonhuman species. There is no accepted scheme by which the relations of these theories to one another may be readily shown, and no straightforward way to overlay them onto the learning-based treatments used in rehabilitation.26 In keeping with our intention to form initial treatment groupings by the types of targets addressed, we have attempted to categorize the types of targets that are likely to be promoted by volitional learning in rehabilitation, regardless of the learning theories that may have been historically invoked to explain various kinds of changes in functioning or behavior. As subsequently discussed, the attempt to classify learningbased treatments into mutually exclusive groups has been extremely challenging. The 2 very broad groupings subsequently described include components of many complex interventions that will require further organization to sort their active ingredients and dosing parameters into combinations useful for both research and practice.
Skilled performances This grouping includes treatments that have in common, in a nutshell, learning by doing. Throughout a lifetime, humans (and other animals) improve the speed, accuracy, and quality of performances through repetition. This improvement with practice may occur in the course of everyday life, as we become better at daily tasks (eg, reading, parenting, getting along with others, and performing various kinds of work), or we may set out to achieve improvement in music, sports, or other skill sets through conscientious practice or drills, and/or by getting coaching from an expert. Rehabilitation clinicians provide the expert coaching and the concentrated practice that patients need to learn new skills (eg, wheelchair management) or to perform previously learned skills (eg, walking and dressing) with a new body. A list of all the skills trained in rehabilitation would be impossible to compile; most of the therapy activities performed in physical, occupational, and speech therapy sessions would fall into this category. Familiar examples include training in gait and other aspects of mobility, and activities and routines of daily living. Also in this grouping would be social skills training, coaching job interviewing skills, prompting mental rehearsal to improve motor recovery, training the use of compensatory devices to keep appointments, and prompting the role playing of anger management strategies. As shown in table 1, the essential ingredient of this category is facilitation of performance on the part of the recipient. This can refer to any type of performance and any type of facilitation on the part of the therapist, from saying “go ahead and start” to prompt a walk down the hall, to providing elaborate environmental set up, cues, and feedback throughout a multistep task. Although repeated training or practice in a particular skill is often necessary to achieve the improvements desired in rehabilitation, it is not an essential ingredient because we assume that even one performance of a skill will result in some change, albeit a small one. However, www.archives-pmr.org
Toward theory-driven classification schedules of practice (massed vs distributed, etc) represent an important set of dosing parameters for this group of treatments.27 Another important dosing parameter is discernible across treatment episodes, and is also found in interventions targeting structural tissue properties and organ functions. This is the progression of treatment demands to maintain an optimal level of challenge. In the skilled performance domain, optimal improvement may occur when the recipient is systematically pushed beyond, but not too far beyond, his or her capacity. Writing mostly of cognitive development, Vygotsky28 termed this critical range of performance the zone of proximal development, in which learning is stimulated but not overwhelmed by challenge. A trainer or therapist can use a variety of means to keep performance in this zone, emphasizing the dimension along which progress is wanted (speed, quality judged against some standard of accuracy or independence, etc). These progression methods may be explicit, as when the therapist engages the patient in setting goals that represent advances in performance, urges the patient to go faster or to pay attention to additional factors bearing on quality, or becomes more selective about the aspects of performance that are cued or reinforced (shaping). The progression may also be more implicit, such as when the therapist gradually decreases the amount of guidance and support while expecting the patient to do more on his or her own, or sets up the task environment to increase difficulty by, for example, introducing background noise. It is important to note that progression means an increase in the level of challenge that is systematic. By implication, this means that progression is typically planned in advance and is not strictly reactive to patient performance. If a therapist decreases the level of cueing or guidance because the patient has improved and needs less help, this is not progression because the demands on performance are unchanged. However, even if progression is not emphasized or if an asymptote is reached, repeated performance can still lead to improved automaticity. In this case, performance may remain at a steady state of quality or speed but will become less demanding of conscious resources, permitting more attention to be directed to concurrent tasks or, if desired, to further improvement in qualitative aspects of performance. The skilled performances category includes many different types of human activities. The performances that are trained may be relatively smooth and continuous (eg, wheelchair propulsion) or accomplished in discrete steps (eg, wheelchair-to-bed transfer). They may involve new or old learning, as previously noted, and they may involve both physical and mental routines. They may encompass treatment approaches that are considered restorative (eg, training hemispatial attention by forcing a response to stimuli further and further away from midline) or compensatory (eg, teaching the use of a cane in a gait sequence). The materials used in treatment sessions may be meaningful (ie, real-world tasks or portions thereof practiced in context) or meaningless (eg, repetitive motor tasks intended to promote dexterity). We have explored some of these dimensions as possible ways to further subdivide this large category, but thus far none have produced results. Some of the distinctions do not hold up well enough for precise definition. Restorative and compensatory approaches can readily shade into one another depending on the precision of measurement of the desired behavior,29 and old versus new learning may be highly dependent on the personal experiences of each patient. For example, the use of a planner or cell phone schedule to aid memory may be second nature to some patients but entirely unfamiliar to others. Even wheelchair locomotion, which is a new set of skills for most learners, taps into previously acquired navigation www.archives-pmr.org
S39 skills and perhaps knowledge of general principles about how wheels move to propel a vehicle forward. The treatment targets in the skilled performances grouping may also involve either functions or activities, as classified by the International Classification of Functioning, Disability and Health (ICF).30 We made a thorough attempt to subdivide this grouping on the basis of ICF levels, but it ultimately broke down for the following reasons. Functions, as conceptualized in the ICF, cannot be practiced in isolation because performance always requires some integrated combination of perceptual, cognitive, and motor functions. Thus, even performance-based treatments aimed at improving functions must use activities as the delivery vehicle of practice opportunities.11 As we learned in an empirical study of therapy coding, included in this supplement,31 it is difficult, if not impossible, to make a nonarbitrary distinction between these levels in a treatment context. The ICF itself does not always clearly differentiate the function and activity levels for a given performance domain. For example, gait patterns are classified as a body function, whereas walking, which is similarly defined, is an activity. Moreover, except for a few specific techniques for teaching discrete sequences (eg, chaining and certain kinds of mnemonics), diverse skilled performances tend to respond to the same active ingredients. Some of these are presented in table 1 and subsequently discussed. The active ingredients that have a bearing on skilled performances are almost as varied as the performances themselves. Although some ingredients are measurable in a single session, others might require a course of treatment to play out, as in the example of progression of demands along a particular dimension to improve speed, accuracy, or quality of performance. Another example of ingredients visible during a course of treatment would be systematic variations in the environments or stimulus conditions in which a skill is practiced, with the aim of enhancing generalization. Within a session, therapists use a number of active ingredients that are partly orthogonal to one another, that is, that can be varied more or less independently. As shown in table 1, some ingredients are generally administered prior to a therapeutic activity or task to help the recipient get into set. These include various kinds of explanations about the task, verbal instructions or modeling to help ensure accurate performance, pointers about which aspects to attend to in performance, and rationales as to why the task is important. The latter may help to raise motivation and the level of effort when the task is performed. Other ingredients that may be used to increase effort and engagement include various goalsetting techniques, a set of potentially powerful manipulations which may be underused in rehabilitation.32 Among the variations in goals that may be manipulated is their specificity, the level of challenge imposed, and the degree to which goals are set by the patient, therapist, or negotiated between the 2 people. Many therapists also try to influence motivation by selecting materials and tasks of personal relevance to the recipient, with the implicit hypothesis that these ingredients will enhance effort. Feedback is a nearly ubiquitous ingredient in skilled performance treatments. In fact, it could be argued that even in the absence of external feedback from the clinician, the intrinsic feedback from the performance is a key component of learningby-doing, possibly even part of the mechanism of action. For external feedback, therapists select the type, amount, timing, and nature and quantity of any reward or reinforcement. External feedback varies along a number of dimensions, such as modality (eg, verbal information from therapist vs visual feedback from a
S40 mirror or video), valence (eg, positive feedback vs error correction feedback), and timing relative to the task (concurrent or terminal; knowledge of performance vs knowledge of results). Therapists’ decisions about feedback also include the degree to which they will prompt self-evaluation from the patient, and how errors will be handled. Errors may be prevented to the extent possible and ignored when they occur, as in errorless learning33; or processed with the patient in an effort to enhance self-evaluation and prevent similar errors in the future. Quantitative variations, that is, schedules of feedback and reinforcement, are important dosing parameters. In addition to feedback, therapists use a host of ingredients during a session to manipulate performance, such as verbal or physical guidance (cues), deliberate arrangements of materials to change the challenge level of various task components, and provision of internal and external strategies. The mechanisms of action involved in improving skilled performances are assumed to involve a mix of implicit and explicit learning, which consists of partially dissociable learning systems specialized to process and store different kinds of information. Implicit learning appears to be phylogenetically older, more diffusely represented in the brain, and may be thought of as preverbal. This system stores information slowly and gradually over multiple episodes or trials; it is considered primarily responsible for habit learning, which is extremely durable but is marked by inflexibility (ie, task specificity). In contrast, the explicit learning system represents learning in verbal/symbolic codes, producing fast, even 1-trial, learning.22 This system is localized in hippocampal/medial temporal lobe structures,34,35 and processes memory for events, words, places, and factsdarbitrary associations that we are generally aware of acquiring in specific learning episodes, that we can verbalize when necessary, and that we can readily generalize across situations. The distinctions between implicit and explicit learning also seemed to open up possibilities for subdividing the treatments within the skilled performances grouping. It was tempting to think that therapeutic activities could be sorted into one or the other category on the basis of their emphasis on procedural skills versus factual instruction, or repetitive practice versus consciously processed strategies. However, the implicit-explicit distinction proved to be less of a dichotomy and more of a continuum. Although it is possible to tease these systems apart in laboratory settings, and they are affected differentially by certain kinds of brain damage, in everyday functioning the systems are always operating together.36 It even appears that in the absence of brain injury, different individuals depend on one or the other system to learn a given task, for reasons unknown.37 Moreover, we realized that different mixes of implicit and explicit learning occur depending not only on the task, but also on the stage of learning. This opened up the possibility that a single treatment could cross the line between implicit and explicit at a point that was very difficult to define. For a good everyday example of this dilemma, consider learning a new sport, such as tennis. At first, explicit learning predominates as the learner consciously absorbs verbalizable rules for how to act (eg, how to hold the racquet, how to place the feet) and begins to put them into practice. At this stage, performance may be slow and awkward, and marked by conscious processing of every move as the performance gradually acquires the desired shape. Later, implicit learning predominates as the learner repeatedly practices the (ideally) correct actions to stamp them in as habits. With continued practice, the performance becomes smooth and automatic, allowing more cognitive space for fine
T. Hart et al tuning at a more explicit level. Experienced musicians and athletes, for example, talk of their performances seeming to slow down as they acquire expertise, affording them the opportunity to make on-the-spot performance decisions that previously would have been impossible. In rehabilitation, repeated practice of activities, such as walking, may result in automatization of the skill at a certain level; however, explicit learning may be reinvoked at any point that the focus of training is shifted by the therapist to a less familiar target, such as qualitative improvement in the gait pattern or use of a device that provides less support for weight bearing or balance. The continual interplay of the implicit and explicit systems in this and other examples led us to conclude that there was not a meaningful distinction to be made at the level of treatment groupings. Although we do not believe that a distinction between implicit and explicit learning will support a taxonomic subdivision into mutually exclusive categories, the implicit-explicit continuum might still be of value in further efforts to organize ingredients in clusters or dimensions that will help to foster systematic treatment planning. As in the example of learning tennis, clinicians should consider the stage of skill acquisition in deciding whether to try to promote or emphasize learning by one system or the other. The continuum is also helpful in clinical decisions about how to evaluate progress and how to structure cues. Task and learner characteristics may also play a role. For example, implicit learning may be relatively preserved in patients with severe explicit memory impairments, including frank disorientation.38-40 For these learners, strategies that promote implicit learning (eg, minimizing the probability of error during skill acquisition) often result in superior performance.33,41 For all learners, implicit learning is best tested by evaluating performance rather than the verbal accounts that are typically used to assess explicit learning. Implicit learning may be best tapped for tasks that can be performed under invariant stimulus conditions (eg, fixed sequences), and for repeated practice after a skill has been nearly mastered. When a task is seen to involve primarily implicit learning, performance may be aided, but is sometimes hindered, by verbalizable strategies or instructions.42 Interference may occur if the explicit instructions contradict what the performer is learning implicitly, and there may not be a benefit to urging the learner to verbalize what has been learned. In contrast, explicit knowledge that is discovered by, rather than presented to, the learner does not appear to disrupt implicit learning.36
Cognitive/affective representations This last grouping includes treatments aimed at changing internal (ie, central nervous system) representations related to cognitions and affects. Cognitions are referred to here as thoughts and ideas, whereas affect is used as a shorthand term to encompass both automatic and more reflective aspects of emotional experience and response. As shown in table 1, typical targets in this treatment grouping include the amount, completeness, and accuracy of the recipient’s fund of knowledge. Didactic-style patient/ family education interventions are familiar examples. However, the targets in this group are not confined to dry facts, but include the personal relevance or emotional valence of the information. Such factors are at play when the therapist works to change the recipient’s attitudes (propensities to act), beliefs,43 or emotional responses to particular thoughts or events. Thus, this treatment grouping includes many elements of therapies aimed at www.archives-pmr.org
Toward theory-driven classification adjustment to disability. Also included are counseling or psychotherapy interventions aimed at changes in mood states, improvement in insight and self-awareness, and altered interpretations of perceptions and events. For the moment, both cognitive and affective representations are included without attempting to separate them in different subgroups because of the way they are intertwined in normative function. We cannot say, for example, that a particular treatment aims to change cognition directly, with subsequent effects on affect hypothesized to occur via enablement theory. At the state of the science, we cannot pinpoint the causal or temporal relations among internal events that might be termed emotion, cognition, belief, and attribution; in many cases, we cannot specify whether these events precede or follow measurable actions.44 The essential ingredient in this treatment grouping is the facilitation of acquisition of information on the part of the recipient. This information could be novel to the recipient, or it could mean a reinterpretation of information that is already known, as a result of reevaluation of oneself or one’s environment.45 Thus, the common factor is that the recipient gains salient information in cognitive and/or affective form, through some means of facilitation that is set up by the therapist. In some treatments, most of the information is selected and provided by the therapist with the recipient listening and/or viewing materials and asking questions. As previously noted, familiar clinical examples include education sessions in which the therapist informs the recipient about the disability and/or its treatment. Other treatments in this grouping, however, are facilitated by the clinician, and the relevant information is self-discovered by the recipient. For example, the therapist may ask the patient to try several different strategies for energy conservation or fatigue management in the natural environment, with the results to be reviewed in subsequent sessions. As in cognitive behavior therapy, the therapist may also assign the patient to do experiments outside of treatment sessions to test the validity of beliefs about oneself (cognitive reappraisal); or to improve the level of self-awareness about some aspect of the disability. The mechanisms of action for treatments in this category are difficult to specify beyond the rather tautological statement that changes take place via semantic and emotional learning and information processing. Further work to impose theoretical order in this category may be assisted by efforts in related fields concerned with changes in motivation, attitude, and behavior. Psychotherapy outcome research would seem an ideal source of inspiration for conceptualizing the mechanisms that drive changes in this grouping. Recently, however, noting Mark Twain’s observation that everyone talks about the weather but nobody does anything about it, former American Psychological Association President Alan Kazdin remarked that “mechanism is the weather of psychotherapy research.”46(p3) For the same reason that cognitive, affective, and behavioral processes cannot be neatly subdivided in this treatment grouping, the causes and effects within complex treatment programs (eg, psychotherapy) are difficult to tease apart. Nonetheless, there has been progress in identifying specific ingredients associated with specific changes in recipient functioning. For example, in motivational interviewing, a nonconfrontational method used primarily in addictions counseling to explore ambivalence and promote change, measurable recipient behaviors (instances of change talk) have been linked to later changes in addiction behavior. Thus, therapist-supplied ingredients that lead to increased change talk are considered key in the effectiveness of this method.47 www.archives-pmr.org
S41 In another related field, health psychology, Michie,6,48 Abraham,49 Schultz,50 and colleagues have developed taxonomies of therapist behaviors that are used to promote change, such as providing information on the links between health and behavior, prompting identification of barriers, and using various kinds of persuasion. (Both taxonomies also include interventions for facilitating practice of recipient skills, which would fall into the skilled performances grouping of the current scheme.) Michie’s and Abraham’s work, in particular, reflects an attempt to link operationally defined therapist behaviors to empirically supported theories explaining how health-related behaviors may be changed and maintained over time. Self-management interventions for chronic health conditions, such as the programs developed by Lorig and Holman,51 also include a mix of information, selfregulation training, and other skill building. These programs have been studied with respect to their possible mechanisms of action in bringing about a variety of health outcomes. For example, changes in self-efficacy, that is, one’s confidence in the ability to solve health-related problems, may explain positive changes in symptom report apart from actual changes in behavior. Related work from within the field of rehabilitation is exemplified by Siemonsma et al,52,53 who examined cognitive restructuring and both mental and physical experimentation as key components of a treatment theory underlying an intervention for low back pain. The active ingredients in the cognitive/affective representations grouping are those that foster changes in cognition, emotion, and/or propensity for behavior via acquisition of information. As shown in table 1, clinicians presenting information to recipients should consider the attributes of the information itself (length, complexity, etc) and its modality (written, oral, visual, etc). In modern rehabilitation, modality could additionally refer to presentation over the Internet or telephone versus face-to-face. The therapist has, at his or her disposal, many other active ingredients that would be hypothesized to increase the likelihood that the recipient will understand, remember, and use information. The therapist may organize the material in various ways, such as chunking it into manageable packets, or providing a scaffold of basic knowledge on which new information may be overlaid. The therapist may also prompt the recipient to engage in active processing of the material, hypothesizing that this will result in better retention. Active processing may be accomplished through asking the recipient to rehearse or paraphrase the material, to recount a related personal experience, or to ask questions. Other active ingredients include prompting the use of memory aids, such as taking notes, to enhance later access to information. For treatments that are based in self-discovery of knowledge, the therapist may select ingredients from a number of modes of prompting such discovery. These could include asking the recipient to think about and perhaps reinterpret past experiences or concepts of oneself, or providing behavioral homework assignments that involve seeking particular knowledge, to be reviewed in a later session. For treatments that aim to change attitudes, beliefs, or feelings, various methods of exploration and persuasion may be counted as active ingredients, along with provision of normative information and appeals to fear. The therapist providing treatments in this grouping may also select among ingredients sometimes associated with therapist style. The therapist can be more or less directive or persuasive depending on the nature of the treatment target and other factors (eg, recipient characteristics). Depending on the material to be presented or facilitated, the therapist may deliver a preset
S42 curriculum, or flex as they discover the boundaries of the recipient’s knowledge, feelings, and attitudes and resistance to or readiness for change. This is analogous to a therapist who coaches a skilled performance by providing guidance according to the results of a moment-by-moment assessment of the performer’s strengths and weaknesses relative to the task at hand. The dosing parameters in this category are somewhat different from those that affect skilled performances. Changing cognitive and affective representations are certainly enhanced by repetition, not in the sense of practice but in terms of repeated exposure to, and rehearsal of, information. Dosing in this category may also be varied by the density of the material (amount per unit time of exposure). Although the therapist may progress this type of learning by presenting easy before complex material, approaching neutral before emotionally laden material, or assigning material that builds on information received earlier, there is no real analogy to the concept of progression in which the therapist’s systematic increase in demands actually accounts for some of the change in structural tissue properties, organ functions, and skilled performances.
Border Zone Between Skilled Performances and Cognitive/Affective Representations Just as it is difficult to sort treatments cleanly into those aimed at structural changes to tissue versus functional changes to organs, the volitional learning categories present some boundary difficulties. Early on in considering these 2 groupings, we recognized that a very large number of clinical interventions consist of admixtures of skill training and practice, along with the provision of information of various kinds. For example, clinicians constantly provide explanations, rationales, and feedback before, during, and after skilled performances. How extensive would these need to be to qualify as separate treatments in the cognitive/affective representations category? How would we recognize them as such? It was previously mentioned that skilled performances may shift from a more automatic to a more conscious mode at any time that explicit information necessitates a reconsideration of strategy or approach. Does this mean that when a therapist talks to a patient about the patient’s gait pattern while walking on a treadmill that the therapist is using a cognitive/affective representation treatment, and the moment the therapist walks away the patient is engaged in a skilled performances treatment? For the sake of simplicity, we would prefer to say no, that the verbal guidance is an ingredient added to the skilled performances treatment, but a moment’s reflection will reveal many examples where the boundaries are muddy. If the patient suddenly refuses to walk and the therapist needs to provide a rationale for walking, persuasion to walk, or a goal-setting manipulation to engage motivation, is this a different treatment with a different target (motivation) or a motivational ingredient overlaid on a treatment whose target is smoothness of gait? In this instance we lean toward calling it a different treatment, but clearly there needs to be some rationaledpreferably not an arbitrary onedfor justifying this and many other similar decisions. Another example of an unclear boundary between these 2 groupings is reflected in the homework dilemma. Many times, as previously noted, therapists ask recipients to practice skills learned in therapy or to complete various assignments between sessions. The dilemma in classifying homework is that the
T. Hart et al change in the target selected by the therapist in assigning the homework, and its associated mechanism of action, are out of the therapist’s control. The recipient is in charge of whether and how the homework gets done. The therapist may provide ingredients (eg, instructions, log sheets, promise of rewards) to try to influence when, where, and how often the homework is accomplished. As long as the therapist is not supervising the performance, however, all of these ingredients are aimed to targets within the cognitive/affective treatment grouping (knowledge, motivation, etc). However, if the homework itself addresses targets in the skilled performances grouping, is it important to capture that fact to adequately characterize the treatment administered by this therapist? We think yes, but the precise method for doing so is not clear cut. This is but one example of issues that would need to be resolved if the general ideas for treatment classification expressed here are to be translated into concrete schemes for coding and measuring the components of actual treatments.11
Shape of Things to Come Clearly, with these 4 treatment groupings, we now have only the rudiments of a taxonomy. We have alluded to some of the challenges that arise at the boundaries of the current groups. We have also noted the difficulties encountered in finding logical subdivisions with which to move forward in creating the remainder of the RTT, especially in the categories related to learning. What is the best format for this taxonomy, and how should we proceed to achieve it? Before beginning this work, we envisioned that the RTT would be a hierarchical tree structure, of the type Linnaeus used to classify living things. This type of taxonomy shows relations from general categories, of which there are no exemplars because they are abstractions (eg, there are no generic mammals, only individuals belonging to species of mammals), to the most specific categories which contain the actual items. The relations between superordinate and subordinate classes are consistent throughout the tree and may be expressed by concepts such as contained with, caused by, or shared history with.1 We anticipated that the RTT would also meet the criteria for a scientific hierarchy, like that of Linnaeus, which stipulates that the top categories contain all of the subordinates, that the relations between each 2 levels of the tree are the same, and that each specific item in the taxonomy belongs to 1, and only 1, class. The original proposal for the RTT was that it be developed in part by the creation of rules, a blueprint, for defining treatments and recipients along the lines described elsewhere.11 These rules would be tested by the creation of 2 minor branches of the RTT (ambulation interventions and executive function interventions), which we termed local taxonomies. We expected these branches to rely on substantially different treatment theories. Development of the local taxonomies was expected to highlight remaining conceptual issues needing to be addressed by the blueprint, and to result in the beginnings of a framework that we could replicate for other types of treatments, to fill out the remaining branches of the tree. The expected form of this work is depicted in figure 2. What actually happened is shown in figure 3. In trying to classify treatments according to the tripartite structure of their underlying theories, we recognized that treatments focused on ambulation or on executive function would not all belong together, as they might in a more traditional scheme focused on www.archives-pmr.org
Toward theory-driven classification
Fig 2
Proposed development of an RTT based on local taxonomies.
deficits. For example, some of the treatments used to improve both ambulation and executive function entail skilled performances, which all respond to many of the same active ingredients and would, therefore, reside together in the taxonomy. Because the treatments involved in ambulation and those involved in executive function would each end up scattered on different branches of the RTT, the local taxonomies would not allow for construction of discrete parts of the tree. We made other attempts at creating a tree structure that bypassed the idea of local taxonomies, but these ran aground because it was difficult to apply consistent criteria for subdividing the groups. Some of our early subdivisions were based on presumed mechanisms of action, as when we identified groupings based on functional regulation and stimulus-response links (all of which are now together in the organ functions grouping). However, others were primarily based on types of targets, as when we ended up with a grouping called extrinsic changes. This included environmental modifications, devices, prostheses, and memory aids in an unwieldy grouping of dubious utility. Still others were based on no identifiable portion of the tripartite
Fig 3 What thwarted the attempt to develop the RTT using local taxonomies.
www.archives-pmr.org
S43 structure, as when we created an initial split between treatments that require volition on the part of the recipient and those that do not. This had a strong intuitive appeal but could not be sustained on any grounds that would also support the remainder of the taxonomy. It may well be that living things are among the few entities that fit reasonably well into a Linnaean hierarchical tree structure, which a taxonomy expert has termed a pipe dream for the goal of organizing knowledge or information.1 For the RTT, we may need to consider, instead of trees, multidimensional matrices for organizing treatments along clusters of ingredients in addition to target classes. Perhaps concept maps could represent key relations in a less linear, more intuitive way, or we may need to create multiple groupings (eg, clades), which show interrelations among items with common features without encompassing an entire universe of items in one view.54 Our challenge is to incorporate input from the field, one of the primary motivations for this series of articles, to develop an RTT that is simple and useful, consistent yet expandable, to allow for classification of future treatments. This may be a pipe dream, but it is one worthy of pursuit.
Keywords Classification; Rehabilitation; Therapeutics
Corresponding author Tessa Hart, PhD, Moss Rehabilitation Research Institute, 50 Township Line Rd, Elkins Park, PA 19027. E-mail address: [email protected].
References 1. Lambe P. Organising knowledge: taxonomies, knowledge and organisational effectiveness. Oxford: Chandos Publishing; 2007. 2. Dijkers MP, Hart T, Tsaouside T, Whyte J, Zanca JM. Treatment taxonomy for rehabilitation: past, present and prospects. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S6-16. 3. DeJong G, Horn S, Gassaway J, Slavin M, Dijkers M. Toward a taxonomy of rehabilitation interventions: using an inductive approach to examine the “black box” of rehabilitation. Arch Phys Med Rehabil 2004;85:678-86. 4. Bode RK, Heinemann AW, Semik P, Mallinson T. Patterns of therapy activities across length of stay and impairment levels: peering inside the black box of inpatient stroke rehabilitation. Arch Phys Med Rehabil 2004;85:1901-8. 5. Horn SD, DeJong G, Ryser DK, Veazie PJ, Teraoka J. Another look at observational studies in rehabilitation research: going beyond the holy grail of the randomized controlled trial. Arch Phys Med Rehabil 2005;86(12 Suppl 2):S8-15. 6. Michie S, Prestwich A. Are interventions theory-based? Development of a theory coding scheme. Health Psychol 2010;29:1-8. 7. Dijkers MP, Hart T, Whyte J, Zanca JM, Packel A, Tsaousides T. Rehabilitation treatment taxonomy: implications and continuations. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S45-54. 8. Keith RA. The role of treatment theory. In: Fuhrer MJ, editor. Assessing medical rehabilitation practices: the promise of outcomes research. Baltimore: Paul H. Brookes Publishing; 1997. p 257-74. 9. Whyte J. Contributions of treatment theory and enablement theory to rehabilitation research theory and practice. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S17-23.
S44 10. Keith RA, Lipsey M. The role of theory in rehabilitation assessment, treatment, and outcomes. In: Glueckauf R, Sechrest L, Bond G, McDonel E, editors. Improving assessment in rehabilitation and health. Newbury Park: Sage; 1993. p 33-58. 11. Whyte J, Dijkers MP, Hart T, et al. Development of a theory-driven rehabilitation treatment taxonomy: conceptual issues. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S24-32. 12. Cirstea MC, Levin MF. Improvement of arm movement patterns and endpoint control depends on type of feedback during practice in stroke survivors. Neurorehabil Neural Repair 2007;21:398-411. 13. Peled A, Karniel A. Knowledge of performance is insufficient for implicit visuomotor rotation adaptation. J Mot Behav 2012;44: 185-94. 14. Boyd LA, Winstein CJ. Impact of explicit information on implicit motor-sequence learning following middle cerebral artery stroke. Phys Ther 2003;83:976-89. 15. Wulf G, Chiviacowsky S, Schiller E, Avila LT. Frequent externalfocus feedback enhances motor learning. Front Psychol 2010;1:190. 16. Hart T. Treatment definition in complex rehabilitation interventions. Neuropsychol Rehabil 2009;19:824-40. 17. Glenn J. Restorative nursing bladder training program: recommending a strategy. Rehabil Nurs 2003;28:15-22. 18. Pohl M, Ruckriem S, Mehrholz J, Ritschel C, Strik H, Pause MR. Effectiveness of serial casting in patients with severe cerebral spasticity: a comparison study. Arch Phys Med Rehabil 2002;83:784-90. 19. Patterson DR, Jensen MP, Wiechman SA, Sharar SR. Virtual reality hypnosis for pain associated with recovery from physical trauma. Int J Clin Exp Hypn 2010;58:288-300. 20. Brooks D, Sidani S, Graydon J, McBride S, Hall L, Weinacht K. Evaluating the effects of music on dyspnea during exercise in individuals with chronic obstructive pulmonary disease: a pilot study. Rehabil Nurs 2003;28:192-6. 21. Ormrod JE. Human learning. 3rd ed. Upper Saddle River: PrenticeHall; 1999. 22. Shanks DR. Learning: from association to cognition. Annu Rev Psychol 2010;61:273-301. 23. Hart T, Fann J, Novack T. The dilemma of the control condition in experience-based cognitive and behavioral treatment research. Neuropsychol Rehabil 2008;18:1-21. 24. Fuhrer MJ, Keith RA. Facilitating patient learning during medical rehabilitation. Am J Phys Med Rehabil 1998;77:557-61. 25. Ehlhardt LA, Sohlberg MM, Kennedy M, et al. Evidence-based practice guidelines for instructing individuals with neurogenic memory impairments: what have we learned in the past 20 years? Neuropsychol Rehabil 2008;18:300-42. 26. Hart T, Powell JM. Principles of learning in TBI rehabilitation. J Head Trauma Rehabil 2011;26:179-81. 27. Schweighofer N, Lee JY, Goh HT, et al. Mechanisms of the contextual interference effect in individuals poststroke. J Neurophysiol 2011;106:2632-41. 28. Vygotsky L. Mind and society: the development of higher psychological processes. Cambridge: Harvard Univ Pr; 1978. 29. Massie C, Malcolm MP, Greene D, Thaut M. The effects of constraint-induced therapy on kinematic outcomes and compensatory movement patterns: an exploratory study. Arch Phys Med Rehabil 2009;90:571-9. 30. World Health Organization. International Classification of Functioning, Disability and Health (ICF). Geneva: World Health Organization; 2001. 31. Hart T, Ferraro M, Myers R, Ellis CA. Opening the black box: lessons learned from an interdisciplinary inquiry into the learning-based contents of brain injury rehabilitation. Arch Phys Med Rehabil 2014; 95(1 Suppl 1):S66-73. 32. Hart T, Evans J. Self-regulation and goal theories in brain injury rehabilitation. J Head Trauma Rehabil 2006;21:142-55.
T. Hart et al 33. Clare L, Jones RS. Errorless learning in the rehabilitation of memory impairment: a critical review. Neuropsychol Rev 2008;18:1-23. 34. Poldrack RA, Packard MG. Competition among multiple memory systems: converging evidence from animal and human brain studies. Neuropsychologia 2003;41:245-51. 35. Murre JM. Implicit and explicit memory in amnesia: some explanations and predictions by the tracelink model. Memory 1997;5:21332. 36. Sun R, Slusarz P, Terry C. The interaction of the explicit and the implicit in skill learning: a dual-process approach. Psychol Rev 2005; 112:159-92. 37. Willingham DB, Nissen MJ, Bullemer P. On the development of procedural knowledge. J Exp Psychol Learn Mem Cogn 1989;15: 1047-60. 38. Ewert J, Levin HS, Watson MG, Kalisky Z. Procedural memory during posttraumatic amnesia in survivors of severe closed head injury: implications for rehabilitation. Arch Neurol 1989;46: 911-6. 39. Ptak R, Gurbrod K, Schnider A. Association learning in the acute confusional state. J Neurol Neurosurg Psychiatry 1998;65:390-2. 40. Cavaco S, Anderson SW, Allen JS, Castro-Caldas A, Damasio H. The scope of preserved procedural memory in amnesia. Brain 2004;127: 1853-67. 41. Ptak R, der Linden MV, Schnider A. Cognitive rehabilitation of episodic memory disorders: from theory to practice. Front Hum Neurosci 2010;4. 42. Vidoni ED, Boyd LA. Achieving enlightenment: what do we know about the implicit learning system and its interaction with explicit knowledge? J Neurol Phys Ther 2007;31:145-54. 43. Meeus M, Nijs J, Van Oosterwijck J, Van Alsenoy V, Truijen S. Pain physiology education improves pain beliefs in patients with chronic fatigue syndrome compared with pacing and self-management education: a double-blind randomized controlled trial. Arch Phys Med Rehabil 2010;91:1153-9. 44. Lowe R, Ziemke T. The feeling of action tendencies: on the emotional regulation of goal-directed behavior. Front Psychol 2011; 2:346. 45. Prochaska JO, DiClemente CC. The transtheoretical approach. In: Norcross JC, Goldfried MR, editors. Handbook of psychotherapy integration. New York: Oxford Univ Pr; 2005. p 147-71. 46. Kazdin AE. Mediators and mechanisms of change in psychotherapy research. Annu Rev Clin Psychol 2007;3:1-27. 47. Miller WR, Rose GS. Toward a theory of motivational interviewing. Am Psychol 2009;64:527-37. 48. Abraham C, Michie S. A taxonomy of behavior change techniques used in interventions. Health Psychol 2008;27:379-87. 49. Michie S, van Stralen MM, West R. The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci 2011;6:42. 50. Schultz R, Czaja SJ, McKay JR, Ory MG, Belle SH. Intervention taxonomy (ITAX): describing essential features of interventions. Am J Health Behav 2010;34:811-21. 51. Lorig K, Holman H. Self-management education: history, definition, outcomes, and mechanisms. Ann Behav Med 2003;26:1-7. 52. Siemonsma PC, Schroder CD, Dekker JH, Lettinga AT. The benefits of theory for clinical practice: cognitive treatment for chronic low back pain patients as an illustrative example. Disabil Rehabil 2008;30:1309-17. 53. Siemonsma PC, Schroder CD, Roorda LD, Lettinga AT. Benefits of treatment theory in the design of explanatory trials: cognitive treatment of illness perception in chronic low back pain rehabilitation as an illustrative example. J Rehabil Med 2010;42:111-6. 54. Blanchard DL. The ABC’s of animal taxonomy. The Cold Blooded News (Newsletter of the Colorado Herpetological Society). Available at: http:// www.dlblanc.com/coloherp/cb-news/archive/taxonomy/Taxonomy.php. Accessed November 19, 2013.
www.archives-pmr.org
Toward theory-driven classification
Supplemental Appendix S1 Glossary of Terms Active ingredients e Attributes of a treatment, selected or delivered by the clinician, that play a role in the treatment’s effects on the target of treatment. These include the essential ingredients associated with the treatment’s known or hypothesized mechanism of action, and any other ingredients that moderate the treatment’s effect(s) but may be common to multiple treatments. Aim(s) (of treatment) e Aspect(s) of the patient’s or other recipient’s functioning or personal factors that is predicted to change indirectly (via mechanisms specified in enablement/disablement theory) as a result of the treatment-induced change in the treatment target. A single treatment may have multiple aims, and there may be a chain of treatment aims (eg, exercises leading to increased strength leading to improved ambulation leading to greater community participation) with at least 1 (in the case of treatments delivered to other recipients) involving the patient’s functioning. Although highly relevant to the ultimate clinical value of a treatment, these distal treatment aims are not relevant to the definition or classification of the treatment. (As previously noted, when the target of treatment is clinically and functionally significant in its own right, we avoid calling it a treatment aim to avoid confusion about direct vs indirect effects of treatment.) Course of treatment e Series of treatment sessions or therapistrecipient contacts that are pursuing change in a specific treatment target, sometimes with some form of treatment progression within or between contacts. Dosing parameters e Quantitative variations in the strength, intensity, frequency, and/or quantity of treatment ingredients; these are often expressed as an amount of time during which the recipient is continuously exposed to the ingredient, the number of times a discrete ingredient is administered, or the magnitude of the ingredient on a quantifiable scale. Enablement/disablement theory e Conceptual system that specifies how change in one aspect of a patient’s functioning (eg, at the level of an International Classification of Functioning, Disability and Health component: body structure, body functioning, activity/activity limitation, participation/participation restriction, personal factor, or environment) will translate into changes in another aspect, specifically a characteristic classified elsewhere in the framework being used. Essential ingredients (of a treatment) e Active ingredients, selected or delivered by the clinician, that define a particular treatment and distinguish it from other treatments. The essential ingredients are those that are specified by the corresponding treatment theory, and are hypothesized or known to be necessary for the treatment’s effects on the treatment target. Inactive ingredients (of a treatment) e Attributes of a treatment that do not define or moderate the impact of the treatment on the target. Ingredients may be presumed to be inactive because they are not addressed by a treatment theory (eg, the building in which the treatment is conducted) or have been empirically determined to be inactive. Ingredients
e
See Treatment ingredients
Mechanism of action e Process by which the treatment’s essential ingredients induce change in the target of treatment. A treatment theory should specify how the essential ingredients engage mechanisms of action to bring about desired treatment effects, that is,
www.archives-pmr.org
S44.e1 specification of the mechanism of action explains how the essential ingredients alter the treatment target within the framework of the treatment theory. Similarly, additional mechanisms of action specify how other active ingredients moderate the effects of the treatment. Nonvolitional treatments e Treatments whose hypothesized mechanisms of action require no effort on the part of the recipient (other than cooperation/nonresistance). Unlike volitional treatments (subsequently defined), the recipient of nonvolitional treatment is always the patient/client undergoing rehabilitation, not a third party (eg, caregiver). Patient/client e Person with a disability or at risk of a disability who is the intended beneficiary of treatment. Progression e Clinician’s deliberate, systematic alteration of the treatment to maintain, over time, the degree of challenge to the body system/behavior(s) selected for change. Progression is often triggered by improvements in the target of treatment; therefore, the pace of progression (within a single treatment contact or a course of treatment) typically depends on the pace of change in the treatment target. The form that treatment progression takes (and hence the nature of the challenge that is being maintained) is often specified by the treatment theory; therefore, multiple treatment sessions may need to be observed to distinguish between one treatment and another. Recipient (of treatment) e Individual whose function/behavior is intended to be changed directly as a result of treatment. In most cases this is the person with a disability (patient/client recipient), but in some instances a caregiver or employer may be the other recipient who is changed by the intervention (eg, to provide care or intervention to the patient/client or to create a more supportive environment for the patient/client). Rehabilitation treatment taxonomy e System of classifying rehabilitation treatments based on a principle or set of principles that allows for distinctions between treatments, which have practical and/or theoretical utility. Session (of treatment) e Individual episode of treatment (typically minutes to a few hours), which may be repeated during a course of treatment. Target of treatment/treatment target e Aspect of the recipient’s functioning, or personal factor, that is predicted to be directly changed by the treatment’s mechanism of action. Specification of the target of treatment in a theory in terms of an International Classification of Functioning, Disability and Health variable(s) helps to define the scope of the treatment/treatment theory. (The aim of treatment refers to changes in functioning obtained in indirect waysdsee Aim(s) (of treatment), above. Although there are instances where the target of treatment is functionally important in its own right, without reference to distal enablement effects, we nevertheless reserve the term treatment aim only for the functionally relevant clinical effects that are distal to the treatment target.) Taxonomy e System of classification or categorization based on characteristics that have important pragmatic or theoretical implications. Treatment e Action taken by a health professional, in the context of contact with a treatment recipient, to alter the functioning of an individual with a disability or at risk of a disability. Treatment is defined broadly to include provision of information, devices, and referrals, specific active experiences, and passive interventions.
S44.e2 Treatment grouping e Broad class of treatments that is similar in essential ingredients (eg, forms of energy) and is able to act on a class of similar treatment targets (eg, tissue properties). Treatment ingredients e Observable (and, therefore, in principle, measurable) actions, chemicals, devices, or forms of energy that are selected or delivered by the clinician. See also Active ingredients, Dosing parameters, Essential ingredients, and Inactive ingredients.
T. Hart et al Treatment theory e Conceptual system that predicts the effects of specific forms of treatment on their targets, specifying the law(s) that expresses the relations between essential ingredients and treatment target changes. (This is similar to Mechanism of action but may be broader and more inclusive.) Volitional treatments e Treatments where a hypothesized mechanism(s) of action requires some effort on the part of the treatment recipient.
www.archives-pmr.org
SPECIAL COMMUNICATION
Toward a Theory-Driven Classification of Rehabilitation Treatments Tessa Hart, PhD,a Theodore Tsaousides, PhD,b Jeanne M. Zanca, PhD, MPT,b John Whyte, MD, PhD,a Andrew Packel, MSPT,a Mary Ferraro, PhD, OTR/L,a Marcel P. Dijkers, PhD, FACRMb From the aMoss Rehabilitation Research Institute, Elkins Park, PA; and bDepartment of Rehabilitation Medicine, Icahn School of Medicine at Mount Sinai, New York, NY.
Abstract Rehabilitation is in need of an organized system or taxonomy for classifying treatments to aid in research, practice, training, and interdisciplinary communication. In this article, we describe a work-in-progress effort to create a rehabilitation treatment taxonomy (RTT) for classifying rehabilitation interventions by the underlying treatment theories that explain their effects. In the RTT, treatments are grouped together according to their targets, or measurable aspects of functioning they are intended to change; ingredients, or measurable clinician decisions and behaviors responsible for effecting changes; and the hypothesized mechanisms of action by which ingredients are transformed into changes in the target. Four treatment groupings are proposed: structural tissue properties, organ functions, skilled performances, and cognitive/affective representations, which are similar in the types of targets addressed, ingredients used, and mechanisms of action that account for change. The typical ingredients and examples of clinical treatments associated with each of these groupings are explored, and the challenges of further subdivision are discussed. Although a Linnaean hierarchical tree structure was envisioned at the outset of work on the RTT, further development may necessitate a model with less rigid boundaries between classification groups, and/or a matrix-like structure for organizing active ingredients along selected continua, to allow for both qualitative and quantitative variations of importance to treatment effects. Archives of Physical Medicine and Rehabilitation 2014;95(1 Suppl 1):S33-44 ª 2014 by the American Congress of Rehabilitation Medicine
“Taxonomies are a losing battle, sandcastles shored up against the rising tide of changedbut we fight nevertheless, because they give temporary respite from advancing chaos.”1(pxvi) For some time, rehabilitation researchers and practitioners have called for a classification system by which to organize the controlled chaos of rehabilitation interventions. A sound method for classifying rehabilitation treatments would be useful for consistent identification and labeling of treatments. Consistent labeling of treatments is an essential precursor to quantifying them An audio podcast accompanies this article. Listen at www.archives-pmr.org. Supported by a cooperative agreement from the Department of Education, National Institute on Disability and Rehabilitation Research (grant no. H133A080053). Contents of this article do not necessarily represent the policy of the Department of Education, and you should not assume endorsement by the Federal Government. No commercial party having a direct financial interest in the results of the research supporting this article has conferred or will confer a benefit on the authors or on any organization with which the authors are associated.
(eg, measuring the dose or intensity of particular interventions).2 A classification scheme would also facilitate training and program evaluation, communication and coordination across the disciplines involved in rehabilitation, and both planning and documentation of treatment. Several classification systems have been developed for portions of rehabilitation practice, based on bottom-up or primarily inductive processes.3 That is, treatments have been listed using names or brief descriptions of activities within professional disciplines (eg, occupational therapy, physical therapy).4,5 Although such schemes may be convenient and align well with clinicians’ thinking, they risk separating treatments that are actually similar. For example, treatments aimed at dressing and walking with a cane may share attributes in common to training in sequential activities. Conversely, there is a risk of combining dissimilar activities under the same name, such as the diverse practices that are all called memory (re)training. As well, bottom-up schemes often label treatments by naming the problem that is treated (dressing skills, memory) without specifying how it
0003-9993/14/$36 - see front matter ª 2014 by the American Congress of Rehabilitation Medicine http://dx.doi.org/10.1016/j.apmr.2013.05.032
S34 is treatedda tautology that helps to keep the mechanisms of effective intervention buried in the black box of rehabilitation.2 In this article, we describe the interim findings of an ongoing effort to develop the structure of a *rehabilitation treatment taxonomy (RTT) based not on names or surface characteristics of interventions but on the theories underlying them. (Words and phrases that are specifically defined in supplemental appendix S1 [available on page A9 of this supplement and online at http:// www.archives-pmr.org/] are marked with an asterisk and italicized when initially used.) In addition to serving the purposes previously described, a classification system based on the theories that explain treatment effects would do much to advance rehabilitation research. Such a system would facilitate the specification, testing, and dissemination of effective treatments; enhance measurement of treatment fidelity; and simplify the process of organizing the knowledge gained (eg, for systematic reviews).6 We have used a number of methods to arrive at this point in the development of the RTT. These include a structured literature search for articles on treatments from all of the main rehabilitation disciplines. We have also sought periodic input from a multidisciplinary advisory committee. However, most of the concepts subsequently described have evolved from discussion among the authors, as we worked through a number of ways to conceptualize treatment theories and possible treatment classification schemes, and tested their resilience to numerous examples of real and hypothetical interventions. Although this work is far from its final form, we present it in the hope that input from the field will provoke more discussion to shape the future of this effort, the challenges of which are subsequently outlined in this article and highlighted in another article in this supplement.7
Theory-driven Taxonomy of Treatments The term *taxonomy can be applied to almost any means of categorizing items in such a way as to show their relations to one another. As an everyday example, a shopping list is a simple taxonomy, classifying items as “things I need,” which could be subdivided by where to find them (grocery store, drug store, etc). A thesaurus is another type of basic taxonomy showing the relations among words and their meanings. A more complex type of taxonomy, a hierarchical tree structure, is illustrated by the familiar Linnaean scheme showing relations among living things, originally based on anatomy and later on genealogy.1 Considering a rehabilitation taxonomy based on theories requires a brief review of what kinds of theories apply to rehabilitation, and how they might be used to sort and interrelate treatments. Rehabilitation is often said to be lacking in theories. Historically, the field has focused on pragmatic rather than theoretical concerns.8 Moreover, the assumption that many rehabilitation *patients would improve regardless of the specifics of the intervention means there has not been pressure to develop, define, and refine treatmentsdactivities that typically force theory developmentdcompared with the situation in fields such as oncology. Nonetheless, rehabilitation does benefit from several types of theory. There is theory focused on explaining how disabling conditions come about, their causes, and risk factors, all of which
T. Hart et al are important for both treatment and prevention efforts. *Enablement/disablement theory9 helps to establish links among impairments and activity/participation restrictions. Understanding these links is important for selecting treatments that will have the maximal impact at the desired level(s) of function, and for refining predictions about the distal or cumulative effects of intervention on a range of outcomes. Enablement theory allows us to hypothesize effects beyond the direct focus of treatment, on *aims that may be down the line from the treatment setting. For example, we may hypothesize that improvement in walking will follow quadriceps strengthening, or improved relationships will follow social skills training. For the RTT, however, we are concerned with *treatment theory, which explains “the actual nature of the process that transforms received therapy into improved health.”10(p34) Treatment theory specifies how and why a treatment works. This type of theory is essential for research that aims to create sound treatments or to make existing treatments stronger, more effective, more efficient, or better suited to the characteristics of specific patients. Treatment theory is essential for isolating and testing the *active ingredients of interventions: these are the attributes of treatment that effect the desired change, and that serve to define and organize treatments and to distinguish them from one another.
Structure of Treatment Theory A brief review of a central concept will place the succeeding material in context. The conceptual framework of the RTT assumes a tripartite structure for every treatment theory.11 Figure 1 illustrates the 3 parts and shows their causal direction (top arrow) in effecting change in the *recipient of treatment, who is commonly a patient but may also be a caregiver or other person. A given treatment may be hypothesized to exert its effects on the *target (a specific aspect of the recipient’s functioning that is selected for change) via some *mechanism of action that is put into play by active ingredients administered by the therapist. The term “ingredients” is borrowed in its medical connotation from pharmacology, although most ingredients in rehabilitation are behavioral. For instance, when a therapist explains a task, praises the patient’s effort, or engages the patient in goal setting, the therapist is administering ingredients just as a physician administers medication or an orthotist provides a brace.
List of abbreviations: ICF International Classification of Functioning, Disability and Health RTT rehabilitation treatment taxonomy
Fig 1 Casual and temporal aspects of the tripartite structure of treatment theory.
www.archives-pmr.org
Toward theory-driven classification As shown in the bottom arrow of figure 1, the process of defining (planning) a treatment flows in reverse temporal order to the causal chain. That is, the therapist starts by considering the target (What aspect of this person’s functioning am I trying to change?), then considers the mechanism (How might changes be effected in this target?), and finally proceeds to assemble and implement the ingredients (What will I do to put the mechanism of action in motion?). In this framework, both the target and the ingredients are measurable, but the mechanism of action is generally unobservable and must be inferred. This is especially the case for interventions that rely on learning. Learning is assumed ultimately to entail some alteration in synaptic transmission in the brain as a result of experience. Because these changes are invisible at the current state of the science, it may be more fruitful to examine certain ingredients, or clusters of ingredients, for which there is empirical evidence for their causing behavior change. For example, feedback is an active ingredient in learning for which the evidence is incontrovertible. We cannot see the effects of feedback on the brain in real time, but we can explore the mechanisms by which it works by examining the effects of qualitative and quantitative variations in feedback on behavior. We can approach a mechanistic understanding of feedback by discovering, for example, that knowledge of performance and knowledge of results are differentially effective on different kinds of tasks,12,13 or that external feedback enhances some kinds of learning and inhibits others.14,15 Therefore, in the subsequent discussion, we focus more on ingredients than on mechanisms; however, it is clearly beyond our scope to review all literature relevant to either of these. What does it mean to say that ingredients, like targets, are measurable? Ingredients may be measured as the result of selections and decisions on the part of the clinician. Ingredients are delivered through drugs and devices, training in tasks and strategies, and other, specific clinician behaviors that are meant to effect changes in patient functioning. As shown in figure 1, therapists work from selecting a measurable target, to hypothesizing the mechanisms that will change the target, to selecting clusters of ingredients (ways of manipulating tissues, stimulating organ systems, transmitting information, or delivering structured experiences) that engage the hypothesized mechanism of action. The efficacy of treatment can only be determined if the target is measurable, meaning that the intended change, or lack thereof, can be detected. Ingredients must also be measurable if they are to be systematically varied to bring about the optimal change in patient function. Variations in ingredients may be qualitative (eg, when a therapist administers a certain type of feedback) or quantitative (eg, when a therapist changes a *dosing parameter, eg, the frequency or schedule of feedback). Ingredients may be visible and measurable in a single *session of treatment, or they may be reflected in systematic variations over a *course of treatment involving multiple sessions. It will be obvious to clinicians that the linear relations conveyed in figure 1 represent a vast simplification of the actual process that unfolds during a treatment episode. During that process, the interpersonal interaction between the treater and the recipient of treatment adds many ingredients or modifies the delivery and the effect of ingredients already in the equation. In real time, clinicians modify their decisions continually according to the reactions and characteristics of the person treated. By presenting the elements of treatment theory as unidirectional in figure 1, we do not intend to imply that patients are passive recipients of treatment. Indeed, the fact that this is seldom the case www.archives-pmr.org
S35 adds to the richness and complexity of rehabilitation, and contributes numerous challenges to any attempt to classify its treatments.16 Similarly, the use of the term “target” to represent the specific aspect of functioning intended for change is not meant to imply that people treated in rehabilitation should be objectified or that clinicians can or should disregard the whole person. In introducing this scheme, which will inevitably strike some readers as overly reductionistic, the intent is not to explain everything that happens in rehabilitation, but to foster discussion of the utility of the tripartite structure to organize one’s thinking about treatments and the way they act to produce positive changes in human functioning. Some of the issues surrounding the tension between mechanistic models and clinical reasoning are addressed in another article in this supplement as well.7
Target Domains and Treatment Groupings of Rehabilitation In developing the framework of the RTT, we have identified 4 superordinate *treatment groupings that differ from one another in all aspects of treatment theory. That is, they are addressed to different target domains, work via different mechanisms of action, and are characterized by different *essential ingredients. These and other attributes of the 4 groupings are subsequently discussed and are summarized in table 1 with respect to their typical targets, mechanisms, ingredients, and dosing parameters. Table 1 also includes examples of familiar clinical interventions that rely heavily on each treatment grouping. We have labeled these 4 treatment groupings by the types or domains of targets that they include, in recognition of the target-first thinking that characterizes treatment planning (see fig 1). At first glance, this may seem to resemble classifying treatments by goals or deficit areas (mobility, memory, dressing, etc), a practice we previously identified as insufficient. However, our definition of the target of treatment refers to a more specific aspect of functioning than is captured by terms such as these. In addition, these 4 groupings represent domains of targets, which we expect to respond to coherent classes of treatment ingredients, guided by coherent classes of treatment theories. That is, we conceptualize these groupings as containing many different treatments, which are related by virtue of similarity in their tripartite structure (targets, mechanisms, and ingredients).
Structural tissue properties As the name implies, treatments in this grouping are intended to change the structural qualities of tissue (eg, the size or elasticity of muscle or soft tissue) by applying various forms of energy according to a prescribed schedule. As shown in table 1, many of these treatments serve to elongate or enlarge tissues. Examples include the various methods of stretching tendons (eg, serial casting and passive range of motion) and wound healing treatments. Another example in the rehabilitation literature is a bladder training intervention in which increased fluid intake stretches the bladder muscle and helps to restore muscle tone. This, coupled with a voiding schedule, has been used to restore continence.17 Massage to reduce tissue adhesions also fits in this grouping. Although a structural change is the target of this treatment, range of motion and/or pain may be expected to be improved indirectly, via enablement theory.
S36 Table 1
T. Hart et al Treatment groupings by target domains in a preliminary taxonomy of rehabilitation treatments
Attributes of Groupings
Structural Tissue Properties
Typical targets
Organ Functions
Skilled Performances
Size (eg, length) Shape Flexibility
Output Efficiency Capacity (eg, percent of normative function) Response dynamics of system
Mechanisms of action
Tissue remodeling processes (macroor microscopic)
Speed/efficiency of performance Quality (compared with standard of completeness, independence, appropriateness to context, etc) Automaticity Learning: mix of implicit and explicit mechanisms
Essential ingredient
Application of energy to tissues
Up- or down-regulation Nonvolitional learning mechanisms (habituation, classical conditioning) Substitution of function Change in functional Facilitation of performance output of organ (system) on part of recipient
Active ingredients (examples)
Type(s) of energy Methods to enhance applied (thermal, effort on part of mechanical, recipient (eg, electromagnetic, etc) motivational interventions)
Typical dosing parameters
Amount, intensity, schedule of energy applied Progression in physical demands on tissue to maximize change Clinical examples Tendon lengthening (ranging, casting) Wound healing
Progression in demand on system to maintain optimal challenge level
Muscle strengthening Cardiovascular endurance exercise Deep brain stimulation Tilt table (up-regulate baroreceptor system) Transcranial direct current stimulation Prosthetic limbs
Within this treatment grouping, decisions about ingredients that may be empirically pitted against one another in the search for the best treatments for a given type of problem and patient involve consideration of several factors. These include the type(s) of energy that may be used, other attributes of devices that may supply the energy, dosing parameters (eg, the amount and schedule of application), and the type and speed of treatment *progression. Progression refers to algorithms by which demands on tissue properties are gradually increased. For example, accelerating the tissue stretching demands of serial casting by using 1 to 4 days versus 5 to 7 days between cast placements results in faster improvements in range of motion and less pain and fewer pressure ulcers.18 It will be seen that progression is a critical parameter for other treatment groupings as
Cognitive/Affective Representations Amount, completeness, accuracy of knowledge Changes in emotional reactions, emotional nuances of cognition (attitudes, beliefs, etc)
Semantic/affective information processing Semantic memory
Facilitation of acquisition of information on part of the recipient Set manipulations (instructions, Attributes of information rationales, motivational aids) presented (modality, Coaching guidance/cues organization, complexity, etc) during performance Methods of facilitating Feedback, reinforcement, acquisition (didactic, response to error socratic, prompting, Provision of strategies modeling, persuasion, etc) Methods to promote Methods to enhance generalization comprehension/retention/use Progression in demands Amount of information on performance to maintain per unit time optimal challenge level Amount/spacing of Schedules of practice repetition and rehearsal (intensity, number/ duration, patterning) Training in gait, activities of daily living, use of memory orthotics, alternative responses to anger triggers, etc Dexterity/motor control exercises Mental rehearsal of physical movements
Patient/caregiver education Adjustment counseling/ psychotherapy Referrals to community services How to (without practice)
well, and underlies the important clinical principle that a systematic, but not overwhelming, increase in demand is a powerful agent of functional change. Further subdivision in this category will most likely be according to the types of energy delivered to tissues and, hence, the types of tissue changes that are likely to result. For example, mechanical energy may result in elongation or enlargement of tissues, whereas heat energy may affect tissue flexibility or elasticity. In some real-world situations, multiple forms of energy are combined or used sequentially to produce the maximum clinical effect. For example, heat could be used to make tissue more pliable, followed by mechanical energy to stretch it out. In our scheme these would count as separate treatments, each with its own target, essential ingredient, and mechanism of action. www.archives-pmr.org
Toward theory-driven classification
Organ functions This grouping includes treatments in which the functions of organs or organ systems are modified, often by systematic stimulation in order to increase or decrease system output, or to bring function to a desired level through natural adaptive processes. The latter may include up- or down-regulation, and habituation to repetitive stimuli. For example, a tilt table may be used to upregulate the baroreceptor system; systematic exposure to sudden noises may reduce startle responses through habituation. Certain forms of massage may also fall into this grouping because massage may alter the sensitivity of pain receptors. This example is notable because massage was also described in the structural tissue properties grouping. The inclusion of similar treatments in 2 groupings illustrates that classification is indeed based on the specific treatment theory hypothesized to account for treatment effects. That is, massage might be hypothesized to reduce pain directly by altering the output of pain receptors, a mechanism belonging to the organ functions group, or the hypothesized mechanism may exert indirect effects by altering the structural properties of soft tissues that cause pain (structural tissue properties group). This is but one example of a case in which empirical research would be needed to clarify the mechanism by which pain is relieved. It may be that massage-related ingredients invoke 2 different mechanisms simultaneously in every case, or it may be that different ingredients or dosing parameters in massage (eg, amount or location of pressure) can shift the balance toward one or the other mechanism that may prove more effective for a given clinical situation. The changes in organ function produced by treatments in this grouping may be permanent or short-lived. For example, both psychostimulant medication and electrical stimulation of the brain from outside the skull (eg, transcranial direct current stimulation) invoke temporary changes in brain function. A different type of temporary modification in brain function is exemplified by immersion in a virtual reality snow world for reduction of pain during wound treatments (eg, in patients with severe burns).19 Here, the presumed mechanism is to capture the attention system by stimulation strong enough to divert it, at least partially, from the pain signal. Another example of a treatment capitalizing on a similar mechanism is playing soothing music through headphones to patients with dyspnea while they are walking.20 The music is hypothesized to divert attention from unpleasant and anxiety-provoking bodily stimuli (eg, shortness of breath), allowing patients to walk with less discomfort. Other treatments in this grouping include those whose mechanism involves substituting for the normative function of an organ (system) by artificial means, presumably for longer-lasting change. Cochlear implants, prosthetic limbs, and deep brain stimulation for treatment of Parkinson’s disease, which provides an artificial stimulus for neurotransmitter release, are some examples of this (see table 1). The musculoskeletal and cardiovascular systems are each the focus of change in many common rehabilitation treatments, such as progressive resistance exercise for muscle strengthening and aerobic exercise for cardiovascular endurance. In aerobic exercise, the aim is to improve the efficiency of oxygen use by gradually pushing the system’s capacity via a systematic increase in demand. This serves to maintain a steady degree of cardiovascular challenge as the function of the system improves. The same principle is in play for progressive resistance exercises for strengthening specific muscle groups. As the
www.archives-pmr.org
S37 muscles adapt to the demand and their force production capacity increases, the weight and/or number of repetitions must also be increased to induce further improvements in strength. Thus, as for the structural tissue properties grouping, dosing parameters that characterize the nature of treatment progression are important for many treatments in the organ functions category (see table 1). This grouping includes both *volitional treatments and *nonvolitional treatments, that is, some of these treatments do not require any active participation (beyond a minimal level of cooperation) to exert their effects on the recipient. Medications and brain stimulation fall into this category. However, others (eg, exercise-based treatments) do require effort on the part of the recipient. Their effects are, therefore, likely to be influenced by active ingredients that are directed toward motivation, as shown in table 1. Such ingredients are not essential for this group because it is the organ stimulation at a given level of challenge that induces the desired change, and this can be achieved by both active and passive means. For example, muscles can be strengthened through volitional exercise or by electrical stimulation, although the effects on the muscle are not identical with these 2 methods. Although learning is not required for all treatments in this grouping, it is common to pair them with learning-based treatments in clinical practice. For example, a therapist may provide a home program by which a patient can carry out muscle strengthening exercises. Learning the routine is not essential for the mechanism of action to work on the target of muscle strengthening. As long as progressive resistance is applied, the muscle will become stronger whether the patient does the program at home or under the continual instruction of the therapist. To help ensure that the patient can continue the program outside of therapy, the therapist might administer other, volitional learning treatments (which we situate in different treatment groupings, which will be subsequently described) to affect other targets beyond muscle strengthening. These might include patients’ understanding of the importance of the muscle-related treatment, their ability to remember to perform it independently, and their ability to self-monitor progress. Although volitional learning is not involved in this treatment grouping, some passive types of learning are inherent in certain treatments in this grouping where the brain is the organ expected to change. For example, rehabilitation treatments may include systematically pairing (or uncoupling) stimuli to alter the circumstances leading to an automatic (unconditioned) response. A familiar example is the case in which a patient develops fear responses to clinicians and/or treatment settings that are associated with pain. In addition to trying to reduce the pain itself, either by addressing the cause or by diminishing the sensation, the rehabilitation team can also work to attenuate the fear response by systematically changing the clinicians or settings that are linked to that experience (counterconditioning). If learning is defined as relatively long-lasting behavior change gained through experience,21 we might also consider sensory habituation, which can be used for vestibular treatments in addition to noise accommodation, as previously described, as a primitive form of learning. All of the learning processes in this treatment grouping are nonvolitional. They are considered primitive in that they are shared by many nonhuman species and require no verbal mediation. Although wakefulness is required for nonvolitional learning, it is thought to occur without the need for conscious awareness of the learning episodes or the stimulus-response pairings involved, although this remains controversial.22
S38 At first glance, these examples of organ function treatments may appear too disparate to belong in 1 grouping. The commonality of these treatments, reflected in their essential ingredient (see table 1), is that they alter the functional output of the organ in question. This alteration may occur either by changing the dynamic relation between inputs and outputs of an organ or system, or by substituting for its functional output. In the example of aerobic exercise, input-output dynamics are changed in that the same level of physical demand (input) results in less heart rate increase (output) after exercise training than before. After habituation or classical conditioning, the same stimulus (input) may produce less autonomic stimulation or an attenuated response (output). As previously noted, substitution (by artificial means) for the normative output of a damaged system occurs with treatments such as prosthetic limbs and deep brain stimulation. Further subdivision within the organ functions grouping may involve first specifying the bodily system selected for change, and then the methods for enacting the desired functions. However, principles that may apply to general mechanisms for enhancing change (eg, the progression of functional challenges or patterning of stimuli to effect optimal adaptation) may cut across organ systems.
Border Zone Between Structural Tissue Properties and Organ Functions At this point, many readers will recognize a quandary that we have faced repeatedly over several years of categorizing and recategorizing treatments: no matter what treatment groupings we compose, the boundaries between them are permeable. Determining whether the target involves changing tissue properties, versus changing organ functions, is difficult because organs are made of tissues, and because the structure of tissue can be intimately related to its function. For example, a leg amputation removes a body structure and also impairs a body function (weight bearing). Providing a prosthetic limb substitutes for the structure but also enhances the function. Similarly, one may seek to stretch the detrusor muscle of the bladder, which would represent a structural change in the bladder’s architecture. However, the same change also increases the bladder’s storage capacity, that is, changes its function. When organ function is changed without a macroscopic change in structure, it is easier to make the distinction. However, when both structure and function are changed, then we need a clearer method of selecting the most relevant treatment target. This and similar challenges to the future and nature of the RTT are subsequently discussed in this article and in another article in this supplement.7
Treatments Based on Volitional Learning We now consider 2 treatment groupings in which all mechanisms of action involve learning and all treatments are volitional. That is, they require active engagement and effort on the part of the recipient, who might be a patient, caregiver, or employer. The interventions in these 2 categories make up what has been termed experience-based treatment.16,23 As clinicians, we expect patients and caregivers to emerge from the experience of rehabilitation having learned new facts, concepts, skills, attitudes, and habits, and to have adapted their prior learning to the new context of disability. However, few rehabilitation practitioners receive a
T. Hart et al strong foundation in learning theories during their training,24,25 and little rehabilitation research is based on explicit models of learning. In developing the theory-driven RTT, we have considered the implications of theoretical frameworks developed over the last century to explain human learning, such as behavioral learning theory, social learning theory, cognitivism, and theories of motor learning. These macrolevel theories differ in emphasis according to their historical contexts, the phenomena of particular interest to their developers, and other factors, such as whether the theories were also meant to explain learning in nonhuman species. There is no accepted scheme by which the relations of these theories to one another may be readily shown, and no straightforward way to overlay them onto the learning-based treatments used in rehabilitation.26 In keeping with our intention to form initial treatment groupings by the types of targets addressed, we have attempted to categorize the types of targets that are likely to be promoted by volitional learning in rehabilitation, regardless of the learning theories that may have been historically invoked to explain various kinds of changes in functioning or behavior. As subsequently discussed, the attempt to classify learningbased treatments into mutually exclusive groups has been extremely challenging. The 2 very broad groupings subsequently described include components of many complex interventions that will require further organization to sort their active ingredients and dosing parameters into combinations useful for both research and practice.
Skilled performances This grouping includes treatments that have in common, in a nutshell, learning by doing. Throughout a lifetime, humans (and other animals) improve the speed, accuracy, and quality of performances through repetition. This improvement with practice may occur in the course of everyday life, as we become better at daily tasks (eg, reading, parenting, getting along with others, and performing various kinds of work), or we may set out to achieve improvement in music, sports, or other skill sets through conscientious practice or drills, and/or by getting coaching from an expert. Rehabilitation clinicians provide the expert coaching and the concentrated practice that patients need to learn new skills (eg, wheelchair management) or to perform previously learned skills (eg, walking and dressing) with a new body. A list of all the skills trained in rehabilitation would be impossible to compile; most of the therapy activities performed in physical, occupational, and speech therapy sessions would fall into this category. Familiar examples include training in gait and other aspects of mobility, and activities and routines of daily living. Also in this grouping would be social skills training, coaching job interviewing skills, prompting mental rehearsal to improve motor recovery, training the use of compensatory devices to keep appointments, and prompting the role playing of anger management strategies. As shown in table 1, the essential ingredient of this category is facilitation of performance on the part of the recipient. This can refer to any type of performance and any type of facilitation on the part of the therapist, from saying “go ahead and start” to prompt a walk down the hall, to providing elaborate environmental set up, cues, and feedback throughout a multistep task. Although repeated training or practice in a particular skill is often necessary to achieve the improvements desired in rehabilitation, it is not an essential ingredient because we assume that even one performance of a skill will result in some change, albeit a small one. However, www.archives-pmr.org
Toward theory-driven classification schedules of practice (massed vs distributed, etc) represent an important set of dosing parameters for this group of treatments.27 Another important dosing parameter is discernible across treatment episodes, and is also found in interventions targeting structural tissue properties and organ functions. This is the progression of treatment demands to maintain an optimal level of challenge. In the skilled performance domain, optimal improvement may occur when the recipient is systematically pushed beyond, but not too far beyond, his or her capacity. Writing mostly of cognitive development, Vygotsky28 termed this critical range of performance the zone of proximal development, in which learning is stimulated but not overwhelmed by challenge. A trainer or therapist can use a variety of means to keep performance in this zone, emphasizing the dimension along which progress is wanted (speed, quality judged against some standard of accuracy or independence, etc). These progression methods may be explicit, as when the therapist engages the patient in setting goals that represent advances in performance, urges the patient to go faster or to pay attention to additional factors bearing on quality, or becomes more selective about the aspects of performance that are cued or reinforced (shaping). The progression may also be more implicit, such as when the therapist gradually decreases the amount of guidance and support while expecting the patient to do more on his or her own, or sets up the task environment to increase difficulty by, for example, introducing background noise. It is important to note that progression means an increase in the level of challenge that is systematic. By implication, this means that progression is typically planned in advance and is not strictly reactive to patient performance. If a therapist decreases the level of cueing or guidance because the patient has improved and needs less help, this is not progression because the demands on performance are unchanged. However, even if progression is not emphasized or if an asymptote is reached, repeated performance can still lead to improved automaticity. In this case, performance may remain at a steady state of quality or speed but will become less demanding of conscious resources, permitting more attention to be directed to concurrent tasks or, if desired, to further improvement in qualitative aspects of performance. The skilled performances category includes many different types of human activities. The performances that are trained may be relatively smooth and continuous (eg, wheelchair propulsion) or accomplished in discrete steps (eg, wheelchair-to-bed transfer). They may involve new or old learning, as previously noted, and they may involve both physical and mental routines. They may encompass treatment approaches that are considered restorative (eg, training hemispatial attention by forcing a response to stimuli further and further away from midline) or compensatory (eg, teaching the use of a cane in a gait sequence). The materials used in treatment sessions may be meaningful (ie, real-world tasks or portions thereof practiced in context) or meaningless (eg, repetitive motor tasks intended to promote dexterity). We have explored some of these dimensions as possible ways to further subdivide this large category, but thus far none have produced results. Some of the distinctions do not hold up well enough for precise definition. Restorative and compensatory approaches can readily shade into one another depending on the precision of measurement of the desired behavior,29 and old versus new learning may be highly dependent on the personal experiences of each patient. For example, the use of a planner or cell phone schedule to aid memory may be second nature to some patients but entirely unfamiliar to others. Even wheelchair locomotion, which is a new set of skills for most learners, taps into previously acquired navigation www.archives-pmr.org
S39 skills and perhaps knowledge of general principles about how wheels move to propel a vehicle forward. The treatment targets in the skilled performances grouping may also involve either functions or activities, as classified by the International Classification of Functioning, Disability and Health (ICF).30 We made a thorough attempt to subdivide this grouping on the basis of ICF levels, but it ultimately broke down for the following reasons. Functions, as conceptualized in the ICF, cannot be practiced in isolation because performance always requires some integrated combination of perceptual, cognitive, and motor functions. Thus, even performance-based treatments aimed at improving functions must use activities as the delivery vehicle of practice opportunities.11 As we learned in an empirical study of therapy coding, included in this supplement,31 it is difficult, if not impossible, to make a nonarbitrary distinction between these levels in a treatment context. The ICF itself does not always clearly differentiate the function and activity levels for a given performance domain. For example, gait patterns are classified as a body function, whereas walking, which is similarly defined, is an activity. Moreover, except for a few specific techniques for teaching discrete sequences (eg, chaining and certain kinds of mnemonics), diverse skilled performances tend to respond to the same active ingredients. Some of these are presented in table 1 and subsequently discussed. The active ingredients that have a bearing on skilled performances are almost as varied as the performances themselves. Although some ingredients are measurable in a single session, others might require a course of treatment to play out, as in the example of progression of demands along a particular dimension to improve speed, accuracy, or quality of performance. Another example of ingredients visible during a course of treatment would be systematic variations in the environments or stimulus conditions in which a skill is practiced, with the aim of enhancing generalization. Within a session, therapists use a number of active ingredients that are partly orthogonal to one another, that is, that can be varied more or less independently. As shown in table 1, some ingredients are generally administered prior to a therapeutic activity or task to help the recipient get into set. These include various kinds of explanations about the task, verbal instructions or modeling to help ensure accurate performance, pointers about which aspects to attend to in performance, and rationales as to why the task is important. The latter may help to raise motivation and the level of effort when the task is performed. Other ingredients that may be used to increase effort and engagement include various goalsetting techniques, a set of potentially powerful manipulations which may be underused in rehabilitation.32 Among the variations in goals that may be manipulated is their specificity, the level of challenge imposed, and the degree to which goals are set by the patient, therapist, or negotiated between the 2 people. Many therapists also try to influence motivation by selecting materials and tasks of personal relevance to the recipient, with the implicit hypothesis that these ingredients will enhance effort. Feedback is a nearly ubiquitous ingredient in skilled performance treatments. In fact, it could be argued that even in the absence of external feedback from the clinician, the intrinsic feedback from the performance is a key component of learningby-doing, possibly even part of the mechanism of action. For external feedback, therapists select the type, amount, timing, and nature and quantity of any reward or reinforcement. External feedback varies along a number of dimensions, such as modality (eg, verbal information from therapist vs visual feedback from a
S40 mirror or video), valence (eg, positive feedback vs error correction feedback), and timing relative to the task (concurrent or terminal; knowledge of performance vs knowledge of results). Therapists’ decisions about feedback also include the degree to which they will prompt self-evaluation from the patient, and how errors will be handled. Errors may be prevented to the extent possible and ignored when they occur, as in errorless learning33; or processed with the patient in an effort to enhance self-evaluation and prevent similar errors in the future. Quantitative variations, that is, schedules of feedback and reinforcement, are important dosing parameters. In addition to feedback, therapists use a host of ingredients during a session to manipulate performance, such as verbal or physical guidance (cues), deliberate arrangements of materials to change the challenge level of various task components, and provision of internal and external strategies. The mechanisms of action involved in improving skilled performances are assumed to involve a mix of implicit and explicit learning, which consists of partially dissociable learning systems specialized to process and store different kinds of information. Implicit learning appears to be phylogenetically older, more diffusely represented in the brain, and may be thought of as preverbal. This system stores information slowly and gradually over multiple episodes or trials; it is considered primarily responsible for habit learning, which is extremely durable but is marked by inflexibility (ie, task specificity). In contrast, the explicit learning system represents learning in verbal/symbolic codes, producing fast, even 1-trial, learning.22 This system is localized in hippocampal/medial temporal lobe structures,34,35 and processes memory for events, words, places, and factsdarbitrary associations that we are generally aware of acquiring in specific learning episodes, that we can verbalize when necessary, and that we can readily generalize across situations. The distinctions between implicit and explicit learning also seemed to open up possibilities for subdividing the treatments within the skilled performances grouping. It was tempting to think that therapeutic activities could be sorted into one or the other category on the basis of their emphasis on procedural skills versus factual instruction, or repetitive practice versus consciously processed strategies. However, the implicit-explicit distinction proved to be less of a dichotomy and more of a continuum. Although it is possible to tease these systems apart in laboratory settings, and they are affected differentially by certain kinds of brain damage, in everyday functioning the systems are always operating together.36 It even appears that in the absence of brain injury, different individuals depend on one or the other system to learn a given task, for reasons unknown.37 Moreover, we realized that different mixes of implicit and explicit learning occur depending not only on the task, but also on the stage of learning. This opened up the possibility that a single treatment could cross the line between implicit and explicit at a point that was very difficult to define. For a good everyday example of this dilemma, consider learning a new sport, such as tennis. At first, explicit learning predominates as the learner consciously absorbs verbalizable rules for how to act (eg, how to hold the racquet, how to place the feet) and begins to put them into practice. At this stage, performance may be slow and awkward, and marked by conscious processing of every move as the performance gradually acquires the desired shape. Later, implicit learning predominates as the learner repeatedly practices the (ideally) correct actions to stamp them in as habits. With continued practice, the performance becomes smooth and automatic, allowing more cognitive space for fine
T. Hart et al tuning at a more explicit level. Experienced musicians and athletes, for example, talk of their performances seeming to slow down as they acquire expertise, affording them the opportunity to make on-the-spot performance decisions that previously would have been impossible. In rehabilitation, repeated practice of activities, such as walking, may result in automatization of the skill at a certain level; however, explicit learning may be reinvoked at any point that the focus of training is shifted by the therapist to a less familiar target, such as qualitative improvement in the gait pattern or use of a device that provides less support for weight bearing or balance. The continual interplay of the implicit and explicit systems in this and other examples led us to conclude that there was not a meaningful distinction to be made at the level of treatment groupings. Although we do not believe that a distinction between implicit and explicit learning will support a taxonomic subdivision into mutually exclusive categories, the implicit-explicit continuum might still be of value in further efforts to organize ingredients in clusters or dimensions that will help to foster systematic treatment planning. As in the example of learning tennis, clinicians should consider the stage of skill acquisition in deciding whether to try to promote or emphasize learning by one system or the other. The continuum is also helpful in clinical decisions about how to evaluate progress and how to structure cues. Task and learner characteristics may also play a role. For example, implicit learning may be relatively preserved in patients with severe explicit memory impairments, including frank disorientation.38-40 For these learners, strategies that promote implicit learning (eg, minimizing the probability of error during skill acquisition) often result in superior performance.33,41 For all learners, implicit learning is best tested by evaluating performance rather than the verbal accounts that are typically used to assess explicit learning. Implicit learning may be best tapped for tasks that can be performed under invariant stimulus conditions (eg, fixed sequences), and for repeated practice after a skill has been nearly mastered. When a task is seen to involve primarily implicit learning, performance may be aided, but is sometimes hindered, by verbalizable strategies or instructions.42 Interference may occur if the explicit instructions contradict what the performer is learning implicitly, and there may not be a benefit to urging the learner to verbalize what has been learned. In contrast, explicit knowledge that is discovered by, rather than presented to, the learner does not appear to disrupt implicit learning.36
Cognitive/affective representations This last grouping includes treatments aimed at changing internal (ie, central nervous system) representations related to cognitions and affects. Cognitions are referred to here as thoughts and ideas, whereas affect is used as a shorthand term to encompass both automatic and more reflective aspects of emotional experience and response. As shown in table 1, typical targets in this treatment grouping include the amount, completeness, and accuracy of the recipient’s fund of knowledge. Didactic-style patient/ family education interventions are familiar examples. However, the targets in this group are not confined to dry facts, but include the personal relevance or emotional valence of the information. Such factors are at play when the therapist works to change the recipient’s attitudes (propensities to act), beliefs,43 or emotional responses to particular thoughts or events. Thus, this treatment grouping includes many elements of therapies aimed at www.archives-pmr.org
Toward theory-driven classification adjustment to disability. Also included are counseling or psychotherapy interventions aimed at changes in mood states, improvement in insight and self-awareness, and altered interpretations of perceptions and events. For the moment, both cognitive and affective representations are included without attempting to separate them in different subgroups because of the way they are intertwined in normative function. We cannot say, for example, that a particular treatment aims to change cognition directly, with subsequent effects on affect hypothesized to occur via enablement theory. At the state of the science, we cannot pinpoint the causal or temporal relations among internal events that might be termed emotion, cognition, belief, and attribution; in many cases, we cannot specify whether these events precede or follow measurable actions.44 The essential ingredient in this treatment grouping is the facilitation of acquisition of information on the part of the recipient. This information could be novel to the recipient, or it could mean a reinterpretation of information that is already known, as a result of reevaluation of oneself or one’s environment.45 Thus, the common factor is that the recipient gains salient information in cognitive and/or affective form, through some means of facilitation that is set up by the therapist. In some treatments, most of the information is selected and provided by the therapist with the recipient listening and/or viewing materials and asking questions. As previously noted, familiar clinical examples include education sessions in which the therapist informs the recipient about the disability and/or its treatment. Other treatments in this grouping, however, are facilitated by the clinician, and the relevant information is self-discovered by the recipient. For example, the therapist may ask the patient to try several different strategies for energy conservation or fatigue management in the natural environment, with the results to be reviewed in subsequent sessions. As in cognitive behavior therapy, the therapist may also assign the patient to do experiments outside of treatment sessions to test the validity of beliefs about oneself (cognitive reappraisal); or to improve the level of self-awareness about some aspect of the disability. The mechanisms of action for treatments in this category are difficult to specify beyond the rather tautological statement that changes take place via semantic and emotional learning and information processing. Further work to impose theoretical order in this category may be assisted by efforts in related fields concerned with changes in motivation, attitude, and behavior. Psychotherapy outcome research would seem an ideal source of inspiration for conceptualizing the mechanisms that drive changes in this grouping. Recently, however, noting Mark Twain’s observation that everyone talks about the weather but nobody does anything about it, former American Psychological Association President Alan Kazdin remarked that “mechanism is the weather of psychotherapy research.”46(p3) For the same reason that cognitive, affective, and behavioral processes cannot be neatly subdivided in this treatment grouping, the causes and effects within complex treatment programs (eg, psychotherapy) are difficult to tease apart. Nonetheless, there has been progress in identifying specific ingredients associated with specific changes in recipient functioning. For example, in motivational interviewing, a nonconfrontational method used primarily in addictions counseling to explore ambivalence and promote change, measurable recipient behaviors (instances of change talk) have been linked to later changes in addiction behavior. Thus, therapist-supplied ingredients that lead to increased change talk are considered key in the effectiveness of this method.47 www.archives-pmr.org
S41 In another related field, health psychology, Michie,6,48 Abraham,49 Schultz,50 and colleagues have developed taxonomies of therapist behaviors that are used to promote change, such as providing information on the links between health and behavior, prompting identification of barriers, and using various kinds of persuasion. (Both taxonomies also include interventions for facilitating practice of recipient skills, which would fall into the skilled performances grouping of the current scheme.) Michie’s and Abraham’s work, in particular, reflects an attempt to link operationally defined therapist behaviors to empirically supported theories explaining how health-related behaviors may be changed and maintained over time. Self-management interventions for chronic health conditions, such as the programs developed by Lorig and Holman,51 also include a mix of information, selfregulation training, and other skill building. These programs have been studied with respect to their possible mechanisms of action in bringing about a variety of health outcomes. For example, changes in self-efficacy, that is, one’s confidence in the ability to solve health-related problems, may explain positive changes in symptom report apart from actual changes in behavior. Related work from within the field of rehabilitation is exemplified by Siemonsma et al,52,53 who examined cognitive restructuring and both mental and physical experimentation as key components of a treatment theory underlying an intervention for low back pain. The active ingredients in the cognitive/affective representations grouping are those that foster changes in cognition, emotion, and/or propensity for behavior via acquisition of information. As shown in table 1, clinicians presenting information to recipients should consider the attributes of the information itself (length, complexity, etc) and its modality (written, oral, visual, etc). In modern rehabilitation, modality could additionally refer to presentation over the Internet or telephone versus face-to-face. The therapist has, at his or her disposal, many other active ingredients that would be hypothesized to increase the likelihood that the recipient will understand, remember, and use information. The therapist may organize the material in various ways, such as chunking it into manageable packets, or providing a scaffold of basic knowledge on which new information may be overlaid. The therapist may also prompt the recipient to engage in active processing of the material, hypothesizing that this will result in better retention. Active processing may be accomplished through asking the recipient to rehearse or paraphrase the material, to recount a related personal experience, or to ask questions. Other active ingredients include prompting the use of memory aids, such as taking notes, to enhance later access to information. For treatments that are based in self-discovery of knowledge, the therapist may select ingredients from a number of modes of prompting such discovery. These could include asking the recipient to think about and perhaps reinterpret past experiences or concepts of oneself, or providing behavioral homework assignments that involve seeking particular knowledge, to be reviewed in a later session. For treatments that aim to change attitudes, beliefs, or feelings, various methods of exploration and persuasion may be counted as active ingredients, along with provision of normative information and appeals to fear. The therapist providing treatments in this grouping may also select among ingredients sometimes associated with therapist style. The therapist can be more or less directive or persuasive depending on the nature of the treatment target and other factors (eg, recipient characteristics). Depending on the material to be presented or facilitated, the therapist may deliver a preset
S42 curriculum, or flex as they discover the boundaries of the recipient’s knowledge, feelings, and attitudes and resistance to or readiness for change. This is analogous to a therapist who coaches a skilled performance by providing guidance according to the results of a moment-by-moment assessment of the performer’s strengths and weaknesses relative to the task at hand. The dosing parameters in this category are somewhat different from those that affect skilled performances. Changing cognitive and affective representations are certainly enhanced by repetition, not in the sense of practice but in terms of repeated exposure to, and rehearsal of, information. Dosing in this category may also be varied by the density of the material (amount per unit time of exposure). Although the therapist may progress this type of learning by presenting easy before complex material, approaching neutral before emotionally laden material, or assigning material that builds on information received earlier, there is no real analogy to the concept of progression in which the therapist’s systematic increase in demands actually accounts for some of the change in structural tissue properties, organ functions, and skilled performances.
Border Zone Between Skilled Performances and Cognitive/Affective Representations Just as it is difficult to sort treatments cleanly into those aimed at structural changes to tissue versus functional changes to organs, the volitional learning categories present some boundary difficulties. Early on in considering these 2 groupings, we recognized that a very large number of clinical interventions consist of admixtures of skill training and practice, along with the provision of information of various kinds. For example, clinicians constantly provide explanations, rationales, and feedback before, during, and after skilled performances. How extensive would these need to be to qualify as separate treatments in the cognitive/affective representations category? How would we recognize them as such? It was previously mentioned that skilled performances may shift from a more automatic to a more conscious mode at any time that explicit information necessitates a reconsideration of strategy or approach. Does this mean that when a therapist talks to a patient about the patient’s gait pattern while walking on a treadmill that the therapist is using a cognitive/affective representation treatment, and the moment the therapist walks away the patient is engaged in a skilled performances treatment? For the sake of simplicity, we would prefer to say no, that the verbal guidance is an ingredient added to the skilled performances treatment, but a moment’s reflection will reveal many examples where the boundaries are muddy. If the patient suddenly refuses to walk and the therapist needs to provide a rationale for walking, persuasion to walk, or a goal-setting manipulation to engage motivation, is this a different treatment with a different target (motivation) or a motivational ingredient overlaid on a treatment whose target is smoothness of gait? In this instance we lean toward calling it a different treatment, but clearly there needs to be some rationaledpreferably not an arbitrary onedfor justifying this and many other similar decisions. Another example of an unclear boundary between these 2 groupings is reflected in the homework dilemma. Many times, as previously noted, therapists ask recipients to practice skills learned in therapy or to complete various assignments between sessions. The dilemma in classifying homework is that the
T. Hart et al change in the target selected by the therapist in assigning the homework, and its associated mechanism of action, are out of the therapist’s control. The recipient is in charge of whether and how the homework gets done. The therapist may provide ingredients (eg, instructions, log sheets, promise of rewards) to try to influence when, where, and how often the homework is accomplished. As long as the therapist is not supervising the performance, however, all of these ingredients are aimed to targets within the cognitive/affective treatment grouping (knowledge, motivation, etc). However, if the homework itself addresses targets in the skilled performances grouping, is it important to capture that fact to adequately characterize the treatment administered by this therapist? We think yes, but the precise method for doing so is not clear cut. This is but one example of issues that would need to be resolved if the general ideas for treatment classification expressed here are to be translated into concrete schemes for coding and measuring the components of actual treatments.11
Shape of Things to Come Clearly, with these 4 treatment groupings, we now have only the rudiments of a taxonomy. We have alluded to some of the challenges that arise at the boundaries of the current groups. We have also noted the difficulties encountered in finding logical subdivisions with which to move forward in creating the remainder of the RTT, especially in the categories related to learning. What is the best format for this taxonomy, and how should we proceed to achieve it? Before beginning this work, we envisioned that the RTT would be a hierarchical tree structure, of the type Linnaeus used to classify living things. This type of taxonomy shows relations from general categories, of which there are no exemplars because they are abstractions (eg, there are no generic mammals, only individuals belonging to species of mammals), to the most specific categories which contain the actual items. The relations between superordinate and subordinate classes are consistent throughout the tree and may be expressed by concepts such as contained with, caused by, or shared history with.1 We anticipated that the RTT would also meet the criteria for a scientific hierarchy, like that of Linnaeus, which stipulates that the top categories contain all of the subordinates, that the relations between each 2 levels of the tree are the same, and that each specific item in the taxonomy belongs to 1, and only 1, class. The original proposal for the RTT was that it be developed in part by the creation of rules, a blueprint, for defining treatments and recipients along the lines described elsewhere.11 These rules would be tested by the creation of 2 minor branches of the RTT (ambulation interventions and executive function interventions), which we termed local taxonomies. We expected these branches to rely on substantially different treatment theories. Development of the local taxonomies was expected to highlight remaining conceptual issues needing to be addressed by the blueprint, and to result in the beginnings of a framework that we could replicate for other types of treatments, to fill out the remaining branches of the tree. The expected form of this work is depicted in figure 2. What actually happened is shown in figure 3. In trying to classify treatments according to the tripartite structure of their underlying theories, we recognized that treatments focused on ambulation or on executive function would not all belong together, as they might in a more traditional scheme focused on www.archives-pmr.org
Toward theory-driven classification
Fig 2
Proposed development of an RTT based on local taxonomies.
deficits. For example, some of the treatments used to improve both ambulation and executive function entail skilled performances, which all respond to many of the same active ingredients and would, therefore, reside together in the taxonomy. Because the treatments involved in ambulation and those involved in executive function would each end up scattered on different branches of the RTT, the local taxonomies would not allow for construction of discrete parts of the tree. We made other attempts at creating a tree structure that bypassed the idea of local taxonomies, but these ran aground because it was difficult to apply consistent criteria for subdividing the groups. Some of our early subdivisions were based on presumed mechanisms of action, as when we identified groupings based on functional regulation and stimulus-response links (all of which are now together in the organ functions grouping). However, others were primarily based on types of targets, as when we ended up with a grouping called extrinsic changes. This included environmental modifications, devices, prostheses, and memory aids in an unwieldy grouping of dubious utility. Still others were based on no identifiable portion of the tripartite
Fig 3 What thwarted the attempt to develop the RTT using local taxonomies.
www.archives-pmr.org
S43 structure, as when we created an initial split between treatments that require volition on the part of the recipient and those that do not. This had a strong intuitive appeal but could not be sustained on any grounds that would also support the remainder of the taxonomy. It may well be that living things are among the few entities that fit reasonably well into a Linnaean hierarchical tree structure, which a taxonomy expert has termed a pipe dream for the goal of organizing knowledge or information.1 For the RTT, we may need to consider, instead of trees, multidimensional matrices for organizing treatments along clusters of ingredients in addition to target classes. Perhaps concept maps could represent key relations in a less linear, more intuitive way, or we may need to create multiple groupings (eg, clades), which show interrelations among items with common features without encompassing an entire universe of items in one view.54 Our challenge is to incorporate input from the field, one of the primary motivations for this series of articles, to develop an RTT that is simple and useful, consistent yet expandable, to allow for classification of future treatments. This may be a pipe dream, but it is one worthy of pursuit.
Keywords Classification; Rehabilitation; Therapeutics
Corresponding author Tessa Hart, PhD, Moss Rehabilitation Research Institute, 50 Township Line Rd, Elkins Park, PA 19027. E-mail address: [email protected].
References 1. Lambe P. Organising knowledge: taxonomies, knowledge and organisational effectiveness. Oxford: Chandos Publishing; 2007. 2. Dijkers MP, Hart T, Tsaouside T, Whyte J, Zanca JM. Treatment taxonomy for rehabilitation: past, present and prospects. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S6-16. 3. DeJong G, Horn S, Gassaway J, Slavin M, Dijkers M. Toward a taxonomy of rehabilitation interventions: using an inductive approach to examine the “black box” of rehabilitation. Arch Phys Med Rehabil 2004;85:678-86. 4. Bode RK, Heinemann AW, Semik P, Mallinson T. Patterns of therapy activities across length of stay and impairment levels: peering inside the black box of inpatient stroke rehabilitation. Arch Phys Med Rehabil 2004;85:1901-8. 5. Horn SD, DeJong G, Ryser DK, Veazie PJ, Teraoka J. Another look at observational studies in rehabilitation research: going beyond the holy grail of the randomized controlled trial. Arch Phys Med Rehabil 2005;86(12 Suppl 2):S8-15. 6. Michie S, Prestwich A. Are interventions theory-based? Development of a theory coding scheme. Health Psychol 2010;29:1-8. 7. Dijkers MP, Hart T, Whyte J, Zanca JM, Packel A, Tsaousides T. Rehabilitation treatment taxonomy: implications and continuations. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S45-54. 8. Keith RA. The role of treatment theory. In: Fuhrer MJ, editor. Assessing medical rehabilitation practices: the promise of outcomes research. Baltimore: Paul H. Brookes Publishing; 1997. p 257-74. 9. Whyte J. Contributions of treatment theory and enablement theory to rehabilitation research theory and practice. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S17-23.
S44 10. Keith RA, Lipsey M. The role of theory in rehabilitation assessment, treatment, and outcomes. In: Glueckauf R, Sechrest L, Bond G, McDonel E, editors. Improving assessment in rehabilitation and health. Newbury Park: Sage; 1993. p 33-58. 11. Whyte J, Dijkers MP, Hart T, et al. Development of a theory-driven rehabilitation treatment taxonomy: conceptual issues. Arch Phys Med Rehabil 2014;95(1 Suppl 1):S24-32. 12. Cirstea MC, Levin MF. Improvement of arm movement patterns and endpoint control depends on type of feedback during practice in stroke survivors. Neurorehabil Neural Repair 2007;21:398-411. 13. Peled A, Karniel A. Knowledge of performance is insufficient for implicit visuomotor rotation adaptation. J Mot Behav 2012;44: 185-94. 14. Boyd LA, Winstein CJ. Impact of explicit information on implicit motor-sequence learning following middle cerebral artery stroke. Phys Ther 2003;83:976-89. 15. Wulf G, Chiviacowsky S, Schiller E, Avila LT. Frequent externalfocus feedback enhances motor learning. Front Psychol 2010;1:190. 16. Hart T. Treatment definition in complex rehabilitation interventions. Neuropsychol Rehabil 2009;19:824-40. 17. Glenn J. Restorative nursing bladder training program: recommending a strategy. Rehabil Nurs 2003;28:15-22. 18. Pohl M, Ruckriem S, Mehrholz J, Ritschel C, Strik H, Pause MR. Effectiveness of serial casting in patients with severe cerebral spasticity: a comparison study. Arch Phys Med Rehabil 2002;83:784-90. 19. Patterson DR, Jensen MP, Wiechman SA, Sharar SR. Virtual reality hypnosis for pain associated with recovery from physical trauma. Int J Clin Exp Hypn 2010;58:288-300. 20. Brooks D, Sidani S, Graydon J, McBride S, Hall L, Weinacht K. Evaluating the effects of music on dyspnea during exercise in individuals with chronic obstructive pulmonary disease: a pilot study. Rehabil Nurs 2003;28:192-6. 21. Ormrod JE. Human learning. 3rd ed. Upper Saddle River: PrenticeHall; 1999. 22. Shanks DR. Learning: from association to cognition. Annu Rev Psychol 2010;61:273-301. 23. Hart T, Fann J, Novack T. The dilemma of the control condition in experience-based cognitive and behavioral treatment research. Neuropsychol Rehabil 2008;18:1-21. 24. Fuhrer MJ, Keith RA. Facilitating patient learning during medical rehabilitation. Am J Phys Med Rehabil 1998;77:557-61. 25. Ehlhardt LA, Sohlberg MM, Kennedy M, et al. Evidence-based practice guidelines for instructing individuals with neurogenic memory impairments: what have we learned in the past 20 years? Neuropsychol Rehabil 2008;18:300-42. 26. Hart T, Powell JM. Principles of learning in TBI rehabilitation. J Head Trauma Rehabil 2011;26:179-81. 27. Schweighofer N, Lee JY, Goh HT, et al. Mechanisms of the contextual interference effect in individuals poststroke. J Neurophysiol 2011;106:2632-41. 28. Vygotsky L. Mind and society: the development of higher psychological processes. Cambridge: Harvard Univ Pr; 1978. 29. Massie C, Malcolm MP, Greene D, Thaut M. The effects of constraint-induced therapy on kinematic outcomes and compensatory movement patterns: an exploratory study. Arch Phys Med Rehabil 2009;90:571-9. 30. World Health Organization. International Classification of Functioning, Disability and Health (ICF). Geneva: World Health Organization; 2001. 31. Hart T, Ferraro M, Myers R, Ellis CA. Opening the black box: lessons learned from an interdisciplinary inquiry into the learning-based contents of brain injury rehabilitation. Arch Phys Med Rehabil 2014; 95(1 Suppl 1):S66-73. 32. Hart T, Evans J. Self-regulation and goal theories in brain injury rehabilitation. J Head Trauma Rehabil 2006;21:142-55.
T. Hart et al 33. Clare L, Jones RS. Errorless learning in the rehabilitation of memory impairment: a critical review. Neuropsychol Rev 2008;18:1-23. 34. Poldrack RA, Packard MG. Competition among multiple memory systems: converging evidence from animal and human brain studies. Neuropsychologia 2003;41:245-51. 35. Murre JM. Implicit and explicit memory in amnesia: some explanations and predictions by the tracelink model. Memory 1997;5:21332. 36. Sun R, Slusarz P, Terry C. The interaction of the explicit and the implicit in skill learning: a dual-process approach. Psychol Rev 2005; 112:159-92. 37. Willingham DB, Nissen MJ, Bullemer P. On the development of procedural knowledge. J Exp Psychol Learn Mem Cogn 1989;15: 1047-60. 38. Ewert J, Levin HS, Watson MG, Kalisky Z. Procedural memory during posttraumatic amnesia in survivors of severe closed head injury: implications for rehabilitation. Arch Neurol 1989;46: 911-6. 39. Ptak R, Gurbrod K, Schnider A. Association learning in the acute confusional state. J Neurol Neurosurg Psychiatry 1998;65:390-2. 40. Cavaco S, Anderson SW, Allen JS, Castro-Caldas A, Damasio H. The scope of preserved procedural memory in amnesia. Brain 2004;127: 1853-67. 41. Ptak R, der Linden MV, Schnider A. Cognitive rehabilitation of episodic memory disorders: from theory to practice. Front Hum Neurosci 2010;4. 42. Vidoni ED, Boyd LA. Achieving enlightenment: what do we know about the implicit learning system and its interaction with explicit knowledge? J Neurol Phys Ther 2007;31:145-54. 43. Meeus M, Nijs J, Van Oosterwijck J, Van Alsenoy V, Truijen S. Pain physiology education improves pain beliefs in patients with chronic fatigue syndrome compared with pacing and self-management education: a double-blind randomized controlled trial. Arch Phys Med Rehabil 2010;91:1153-9. 44. Lowe R, Ziemke T. The feeling of action tendencies: on the emotional regulation of goal-directed behavior. Front Psychol 2011; 2:346. 45. Prochaska JO, DiClemente CC. The transtheoretical approach. In: Norcross JC, Goldfried MR, editors. Handbook of psychotherapy integration. New York: Oxford Univ Pr; 2005. p 147-71. 46. Kazdin AE. Mediators and mechanisms of change in psychotherapy research. Annu Rev Clin Psychol 2007;3:1-27. 47. Miller WR, Rose GS. Toward a theory of motivational interviewing. Am Psychol 2009;64:527-37. 48. Abraham C, Michie S. A taxonomy of behavior change techniques used in interventions. Health Psychol 2008;27:379-87. 49. Michie S, van Stralen MM, West R. The behaviour change wheel: a new method for characterising and designing behaviour change interventions. Implement Sci 2011;6:42. 50. Schultz R, Czaja SJ, McKay JR, Ory MG, Belle SH. Intervention taxonomy (ITAX): describing essential features of interventions. Am J Health Behav 2010;34:811-21. 51. Lorig K, Holman H. Self-management education: history, definition, outcomes, and mechanisms. Ann Behav Med 2003;26:1-7. 52. Siemonsma PC, Schroder CD, Dekker JH, Lettinga AT. The benefits of theory for clinical practice: cognitive treatment for chronic low back pain patients as an illustrative example. Disabil Rehabil 2008;30:1309-17. 53. Siemonsma PC, Schroder CD, Roorda LD, Lettinga AT. Benefits of treatment theory in the design of explanatory trials: cognitive treatment of illness perception in chronic low back pain rehabilitation as an illustrative example. J Rehabil Med 2010;42:111-6. 54. Blanchard DL. The ABC’s of animal taxonomy. The Cold Blooded News (Newsletter of the Colorado Herpetological Society). Available at: http:// www.dlblanc.com/coloherp/cb-news/archive/taxonomy/Taxonomy.php. Accessed November 19, 2013.
www.archives-pmr.org
Toward theory-driven classification
Supplemental Appendix S1 Glossary of Terms Active ingredients e Attributes of a treatment, selected or delivered by the clinician, that play a role in the treatment’s effects on the target of treatment. These include the essential ingredients associated with the treatment’s known or hypothesized mechanism of action, and any other ingredients that moderate the treatment’s effect(s) but may be common to multiple treatments. Aim(s) (of treatment) e Aspect(s) of the patient’s or other recipient’s functioning or personal factors that is predicted to change indirectly (via mechanisms specified in enablement/disablement theory) as a result of the treatment-induced change in the treatment target. A single treatment may have multiple aims, and there may be a chain of treatment aims (eg, exercises leading to increased strength leading to improved ambulation leading to greater community participation) with at least 1 (in the case of treatments delivered to other recipients) involving the patient’s functioning. Although highly relevant to the ultimate clinical value of a treatment, these distal treatment aims are not relevant to the definition or classification of the treatment. (As previously noted, when the target of treatment is clinically and functionally significant in its own right, we avoid calling it a treatment aim to avoid confusion about direct vs indirect effects of treatment.) Course of treatment e Series of treatment sessions or therapistrecipient contacts that are pursuing change in a specific treatment target, sometimes with some form of treatment progression within or between contacts. Dosing parameters e Quantitative variations in the strength, intensity, frequency, and/or quantity of treatment ingredients; these are often expressed as an amount of time during which the recipient is continuously exposed to the ingredient, the number of times a discrete ingredient is administered, or the magnitude of the ingredient on a quantifiable scale. Enablement/disablement theory e Conceptual system that specifies how change in one aspect of a patient’s functioning (eg, at the level of an International Classification of Functioning, Disability and Health component: body structure, body functioning, activity/activity limitation, participation/participation restriction, personal factor, or environment) will translate into changes in another aspect, specifically a characteristic classified elsewhere in the framework being used. Essential ingredients (of a treatment) e Active ingredients, selected or delivered by the clinician, that define a particular treatment and distinguish it from other treatments. The essential ingredients are those that are specified by the corresponding treatment theory, and are hypothesized or known to be necessary for the treatment’s effects on the treatment target. Inactive ingredients (of a treatment) e Attributes of a treatment that do not define or moderate the impact of the treatment on the target. Ingredients may be presumed to be inactive because they are not addressed by a treatment theory (eg, the building in which the treatment is conducted) or have been empirically determined to be inactive. Ingredients
e
See Treatment ingredients
Mechanism of action e Process by which the treatment’s essential ingredients induce change in the target of treatment. A treatment theory should specify how the essential ingredients engage mechanisms of action to bring about desired treatment effects, that is,
www.archives-pmr.org
S44.e1 specification of the mechanism of action explains how the essential ingredients alter the treatment target within the framework of the treatment theory. Similarly, additional mechanisms of action specify how other active ingredients moderate the effects of the treatment. Nonvolitional treatments e Treatments whose hypothesized mechanisms of action require no effort on the part of the recipient (other than cooperation/nonresistance). Unlike volitional treatments (subsequently defined), the recipient of nonvolitional treatment is always the patient/client undergoing rehabilitation, not a third party (eg, caregiver). Patient/client e Person with a disability or at risk of a disability who is the intended beneficiary of treatment. Progression e Clinician’s deliberate, systematic alteration of the treatment to maintain, over time, the degree of challenge to the body system/behavior(s) selected for change. Progression is often triggered by improvements in the target of treatment; therefore, the pace of progression (within a single treatment contact or a course of treatment) typically depends on the pace of change in the treatment target. The form that treatment progression takes (and hence the nature of the challenge that is being maintained) is often specified by the treatment theory; therefore, multiple treatment sessions may need to be observed to distinguish between one treatment and another. Recipient (of treatment) e Individual whose function/behavior is intended to be changed directly as a result of treatment. In most cases this is the person with a disability (patient/client recipient), but in some instances a caregiver or employer may be the other recipient who is changed by the intervention (eg, to provide care or intervention to the patient/client or to create a more supportive environment for the patient/client). Rehabilitation treatment taxonomy e System of classifying rehabilitation treatments based on a principle or set of principles that allows for distinctions between treatments, which have practical and/or theoretical utility. Session (of treatment) e Individual episode of treatment (typically minutes to a few hours), which may be repeated during a course of treatment. Target of treatment/treatment target e Aspect of the recipient’s functioning, or personal factor, that is predicted to be directly changed by the treatment’s mechanism of action. Specification of the target of treatment in a theory in terms of an International Classification of Functioning, Disability and Health variable(s) helps to define the scope of the treatment/treatment theory. (The aim of treatment refers to changes in functioning obtained in indirect waysdsee Aim(s) (of treatment), above. Although there are instances where the target of treatment is functionally important in its own right, without reference to distal enablement effects, we nevertheless reserve the term treatment aim only for the functionally relevant clinical effects that are distal to the treatment target.) Taxonomy e System of classification or categorization based on characteristics that have important pragmatic or theoretical implications. Treatment e Action taken by a health professional, in the context of contact with a treatment recipient, to alter the functioning of an individual with a disability or at risk of a disability. Treatment is defined broadly to include provision of information, devices, and referrals, specific active experiences, and passive interventions.
S44.e2 Treatment grouping e Broad class of treatments that is similar in essential ingredients (eg, forms of energy) and is able to act on a class of similar treatment targets (eg, tissue properties). Treatment ingredients e Observable (and, therefore, in principle, measurable) actions, chemicals, devices, or forms of energy that are selected or delivered by the clinician. See also Active ingredients, Dosing parameters, Essential ingredients, and Inactive ingredients.
T. Hart et al Treatment theory e Conceptual system that predicts the effects of specific forms of treatment on their targets, specifying the law(s) that expresses the relations between essential ingredients and treatment target changes. (This is similar to Mechanism of action but may be broader and more inclusive.) Volitional treatments e Treatments where a hypothesized mechanism(s) of action requires some effort on the part of the treatment recipient.
www.archives-pmr.org
