Author Response

We appreciate the interest in our research¹ shown by Merians,² and we are pleased to have the opportunity to comment on some of the important points that she has raised. Merians observes that “it appears that, in addition to comparing the use of the robot as a therapeutic adjunct to task practice, the authors also were comparing particular therapeutic principles, specifically task-based training with a combination of task-based training and impairment-level interventions.” In essence, this project did present an opportunity to assess a robotic device that could be used as a therapeutic adjunct and allow for further study of a task-based training approach. It is important to note that the Hand Mentor is preloaded with patient modules that are designed to reduce spasticity, enhance motor unit recruitment, and encourage controlled active motor function about the wrist. The spasticity module is considered one that addresses a specific impairment. The time spent training in the spasticity module varied across patients and depended primarily on their level of spasticity. The aim of the spasticity module was to reduce resistance to movement during wrist extension motion while the other modules were more focused on enhancing the participant's capacity to recruit the wrist extensors and engage in active control of extensors and flexors about the wrist. During the task-based training portion of each session, participants were encouraged to practice and engage in activities that were important to them. Although such encouragement was not robotically driven, it certainly falls within the domain of task-specific training.

In our study, we reported that hand function and stroke recovery rating improved for participants in both intervention groups, leading Merians to raise several questions and points that we would like to address:

What is the role of impairment-level interventions versus functional training? This question is not easily addressed and forms the cornerstone of much of the applied research currently under way in neurorehabilitation. The question is posed in an “either/or” mode, when, in fact, given the rapidly constricting time for both inpatient and outpatient rehabilitation services, coupled with the paucity of substantive evidence for successful outcomes using impairment-based approaches in the treatment of the upper extremities of patients with either acute or chronic stroke, a functional approach is evolving as more practical and defensible. However, the basis for progressing functional training must entail consideration of impairment levels and how such levels are responsive to either repetition or shaping for retraining to capture a defined function.³

Is the combination important, and if it is important, what is the ideal dosing for each component? We agree that determining ideal dosing is an important issue. This point is of concern for all rehabilitation studies.⁴ Our collective inability to determine the relationship of optimal intensity or dosing to achieve best outcomes has plagued many rehabilitation studies. With respect to an intervention such as constraint-induced movement therapy (CIMT), we did not see a clear relationship between meticulously documented amount of training, using repetitive task practice or shaping, and outcome among patients with subacute or chronic stroke.⁵ The precise dosing necessary to achieve optimal results is unknown at this time. An acknowledged limitation of our study and similar long-term robotic and task practice interventions is the lack of a measure to monitor functional recovery that can be completed on a regular basis during the intervention and that truly reflects a change in function and not simply motor learning to that specific task. We recently provided initial kinematic and kinetic data during the performance of a bimanual dexterity task that may provide appropriate measures to track the recovery process.⁶

Is the total intensity or amount of practice more important than the difference in approach? As we acknowledge in the article, the outcomes that might be seen if patients received 30 hours of repetitive task practice (RTP) alone (without an additional 30 hours of robotic-assisted therapy) remain unknown and are of interest in ongoing studies.

Given the current strong interest in task-based therapy, the possible influence of impairment-level interventions is important to remember. The Stroke Impact Scale (SIS) is a widely used instrument with well-established reliability and validity. Lin et al⁷ recently proposed standards for determining minimal clinically important change on SIS subscales, including hand function. They concluded that the mean change score of a group of patients with stroke on the hand function subscale should reach 17.8 points in order to be considered clinically important. In our study, the estimated change in mean scores of the combined therapy group exceeded this threshold value, both preintervention to postintervention and preintervention to follow-up; the estimated change in mean scores of the RTP-only group did not exceed this threshold value, even though these changes were statistically significant. Merians suggests that this difference may indicate “an essential role for impairment-based interventions.” The improvement in hand function following the combined RTP-Hand Mentor intervention is encouraging and does suggest that an impairment-based robotic system such as the Hand Mentor may have an important role, primarily as an adjunct to functional task practice. The possibility of utilizing technology (ie, robotic or virtual reality) is appealing because of the consistent and controllable nature of these systems. An important unknown is whether the use of these systems provides a cost-effective alternative or adjunct to the current delivery of care model.

Given that study participants were trained within 3 to 9 months poststroke, a concern during this time frame is how to separate changes in function resulting from the intervention from changes due to spontaneous recovery. In patients with subacute stroke, differentiating change in function resulting from an intervention from changes due to spontaneous recovery is challenging, as we note with respect to the issue of change in quality-of-life ratings over time. As Merians suggests, double-baseline testing, separated by a period of time, may help to control for this issue. The value of implementing additional assessments always must be balanced against possible increased respondent burden and willingness to continue participation. Rather than using double-baseline testing for clinical measures, we collected data for kinematic and kinetic measures of hand function to complement data for clinical measures and provide insight into changes in motor control patterns that may be responsible for improvements in hand function. Nevertheless, we appreciate Merians’ point and recommend a combined approach of double-baseline testing and the utilization of quantitative measures of function that are not necessarily part of the training and are precise enough in nature to detect subtle changes in motor functioning.

Clinicians would like to be able to discriminate between recovery based on compensation and recovery based on true improvement in motor abilities of the hemiparetic limb. Discriminating between recovery based on compensation and recovery based on true improvement in motor abilities of the hemiparetic limb is another research challenge. Quality-of-life measures provide patient perceptions of intervention effects and can be especially useful for judging the acceptability of a therapy. The National Institutes of Health Program Announcement under which our research was funded was titled Increasing Quality of Life in Mobility Disorders. However, we appreciate the difficulty of differentiating between compensatory mechanisms and true recovery. In our earlier work examining the effects of CIMT on the control of grasping forces and torques, we demonstrated that the motor improvements associated with CIMT appear to be related to actual improvements in patients’ ability to initiate and control grasping forces over time (ie, following CIMT, the rate of force production was similar to patterns others have reported in individuals who were healthy).⁸ More recently, for participants from this project, we have provided initial data suggesting that individuals in both groups exhibited a fundamental change in motor control strategy following each intervention from primarily a feedback control of grasping forces and torques to a greater reliance on feedforward control of grasping forces and torques during a functional bimanual dexterity task.⁶ Furthermore, participants were better able to focus their grasping forces (ie, produce smaller shear force) following these 2 interventions.⁶ These changes in specific kinematic and kinetic variables and motor control suggest that recovery was based on true improvement of the hemiparetic limb rather than a simple compensatory activity.

To provide the reader with a context for changes in participants’ functional ability over the course of our study, mean scores on the Fugl-Meyer Motor Assessment (FMA) improved at postintervention by 7 to 8 points in both intervention groups, and this change was maintained at follow-up in both intervention groups. In a companion manuscript in process from this study in which the FMA is an outcome, our data are consistent with this sustained improvement in FMA at the 2-week follow-up evaluation.

Development of robots to assist hand function is particularly challenging. Inconsistent use of the same outcome tests and measures and kinematic analyses hinders interpretation of results and impedes progress of the development and incorporation of these technologies. We agree that development of robots to assist hand function is particularly challenging, which may explain why, to date, few studies have shown meaningful gain in dexterous function following a robotic intervention. The multiple degrees of freedom and extensive number of intrinsic and extrinsic muscles engaged, coupled with the vast amount of sensory information gathered, make the hand part of an exquisite control system. Developing a robotic system that takes all of these variables into account would be daunting and most likely cost-prohibitive. The importance of hand function in daily life is universally appreciated in the rehabilitation and bioengineering communities. The development of effective robotic systems for the hand requires a multidisciplinary approach between these 2 communities. We are encouraged to see more of these types of collaborations throughout North America and Europe. Such efforts will yield effective and practical solutions to improving hand function. As Merians suggests, it is incumbent on clinicians and bioengineers to utilize consistent outcomes across studies. The selection of outcomes should not be based on the “historical value” of a test per se. Outcomes, whether clinical or biomechanical, must be reliable, precise, and quantitative in order for robotic or other technology-driven approaches to be translated into patient care.

• © 2010 American Physical Therapy Association