Dijkers and the team of scholars aiming to define, classify, and measure specific treatments used within treatment sessions in rehabilitation are to be congratulated for proposing a conceptual framework for a rehabilitation treatment taxonomy (RTT) to describe interventions and evaluate their impact on rehabilitation outcomes.1 Rehabilitation outcomes are influenced by 2 major factors: (1) patient characteristics and (2) treatment processes. The latter can be divided into: (1) factors related to clinical settings including characteristics of caregivers and health care systems and (2) specific treatment activities or interventions. As the authors state, intense efforts have been invested previously to create tools that can accurately measure different dimensions of a patient's quality of life (ie, outcomes measures).2–12 Improvements also have been made in the ability to measure a large number of patient characteristics associated with those outcomes.13–17 Some studies have helped understand how treatment-related factors are associated with outcomes (eg, institutional characteristics such as waiting periods or referral source,14,18,19 clinicians' expertise and educational background,20–22 or clinicians' ability to create a strong working alliance with their patients23–26). However, only a few initiatives have been published in the peer-reviewed literature aimed at creating a treatment taxonomy that has demonstrated utility,15–17,27–29 supporting Dijkers and colleagues' efforts. These initiatives need to be examined carefully to either suggest alternative taxonomies or help refine them. Thus, specific requirements for the intended treatment taxonomy need to be clearly defined to enable such examinations.
The basic concepts presented by the authors include the following requirements. The primary concept presented is to improve specificity of treatment description beyond general descriptions of, for example, treatment time provided by a specific discipline or the functional problem addressed such as gait training. The authors set a goal to specify a link among ingredients, mechanisms of action, and functional targets; they consider these the basis for classifying treatments. An assumption is made that when active ingredients of treatments are identified, the ability to conduct efficacy and comparative effectiveness research or to study clinical practice variations will be improved. Therefore, classifying rehabilitation treatments by active ingredients affecting the target of treatment by a known or hypothesized mechanism of action is proposed. Other requirements included in the proposed conceptual framework for an RTT are mutual exclusivity of treatments distinct from one another by their hypothesized mechanisms of action on rehabilitative (functional) targets, and a need to address factors related to clinical settings mentioned above, specifically related to patient-clinician interactions.
Although these basic RTT concepts seem logical, it is not known whether they can be implemented in order to classify rehabilitation interventions due to several challenges. First, can all or most rehabilitation treatments be identified in a consistent manner on the basis of their known or assumed active ingredients? Second, can each active ingredient be defined in a mutually exclusive way, as presented in the proposed conceptual framework shown in Figure 2 of the article by Dijkers et al? Third, does the proposed conceptual framework make sense when applied to activities and interventions currently used in rehabilitation? If these challenges are met, the proposed RTT should perform well when evaluated using a systematic process. An example of such a process involving 2 steps within the framework of a practice-based evidence study design defined previously30,31 is hereby described.
Step 1 would involve implementing intensive data collection, including data reliability testing.30,31 Measures of RTT accuracy, consistency, and overall scores could be defined and tested by having a group of selected frontline clinicians choose the taxonomic categories associated with a set of written clinical scenarios. Accuracy can be defined as the percentage of interventions that are accurately selected by 90% of the group. Consistency can be defined as the percentage of therapists who accurately select 90% of interventions, mechanisms of action, and targets across all scenarios. An overall RTT score can be calculated by averaging all of the therapists' scores, each one reflecting the number of interventions correctly identified by each therapist. Acceptable levels of accuracy, consistency, and overall RTT scores could be predetermined as 90% or more. If acceptable levels are not achieved, the RTT can be continuously modified and re-evaluated until distinctions that do not make sense in routine rehabilitation described in the scenarios or distinctions that are described in ways that make them difficult to understand and apply in routine rehabilitation scenarios no longer are present in the RTT. Once that is achieved, clinician training is conducted until acceptable levels are met. Only then can any RTT be determined to be stable and valid enough to be implemented within rehabilitation settings. Step 2 would involve field testing in actual clinical practice to evaluate the RTT's usability and ability to predict outcomes.
Because rehabilitation care is multidimensional and relatively complex, with many unknown factors associated with assumed theories,32 it is expected that the conceptual framework proposed by Dijkers et al for an RTT, or any RTT proposed, will need several rounds of evaluation and refinement before being determined to be useful for routine clinical practice. Therefore, it is suggested that the proposed conceptual framework for an RTT include a systematic clinical evaluation process, so its chances of achieving clinical acceptability will be improved.
- © 2014 American Physical Therapy Association
References
Issue highlights
