Walking adaptability therapy after stroke: study protocol for a randomized controlled trial

In total, 40 persons who had a stroke will be recruited from the inpatient and outpatient population of rehabilitation center Reade (Amsterdam, The Netherlands) to participate in this study. Inclusion criteria are first-ever stroke ≥3 months ago, walking or balance deficits confirmed by a physician, clinical diagnosis of hemiparesis, age ≥18 years, general walking ability as indicated by a Functional Ambulation Category score ≥3 [23], and the ability to understand and execute simple instructions. Exclusion criteria are orthopedic and other neurological disorders that affect walking (e.g., Parkinson’s disease), other treatments that could influence the effects of the interventions (e.g., recent Botulin toxin treatment of the lower extremity), contra-indication to physical activity (e.g., heart failure, severe osteoporosis), moderate or severe cognitive impairments as indicated by a Mini-Mental State Examination [24] score below 21, or severe uncorrected visual deficits. Persons with stroke who are eligible for participation will be informed about the study by their rehabilitation specialist, both orally and in writing. All participants will provide a written informed consent.

The proposed study is a single-center, parallel group randomized controlled trial with pre-intervention, post-intervention, retention, and follow-up tests to determine the relative efficacy of the interventions: treadmill-based C-Mill therapy and the overground FALLS program. After giving informed consent, participants will be randomly assigned to one of the two interventions using an automated, custom-made minimization algorithm written in MATLAB. The minimization procedure is based on time after stroke, age and Functional Ambulation Category score to balance groups for these stratification factors. The research assistant will enter the data for randomization in the algorithm and the participant will subsequently be informed about the resulting group allocation before the pre-intervention tests. Subsequently, the assessor will schedule the participants for the assigned 5 week intervention program. Pre-intervention tests (T0) to characterize groups and obtain baseline values of primary and secondary outcome measures will be performed one week prior to the intervention program. Within one week after completing the intervention, post-intervention tests (T1) will be performed. The same tests will be conducted 5 weeks (retention tests, T2) and 12 months (follow-up tests, T3) after completing the intervention. All assessments will be performed at the rehabilitation center. Because of the nature of the intervention studied, therapists and participants cannot be blinded to group allocation. The assessor will also not be blinded to group allocation, because of pragmatic constraints related to the planning of assessments and therapy sessions. Figure 2 shows a flow chart of the procedures that participants will undergo at T0, T1, T2 and T3.

C-Mill therapy is a structured treadmill training program with a specific emphasis on practicing walking adaptability (as detailed in Table 1 and Additional file 1), using gait-dependent augmented-reality content projected on the instrumented treadmill surface to elicit step adjustments [1, 13‐15, 25‐28]. Figure 3 shows various exercises of C-Mill therapy, including exercises to practice avoidance of projected visual obstacles (Fig. 3a), exercises to practice accurate foot placement on a step-to-step basis by walking to a regular or irregular sequence of visual stepping targets (Fig. 3b), exercises to practice acceleration and deceleration by maintaining position within a projected walking area that moves along the treadmill (Fig. 3c), and a functional and interactive walking adaptability game (Fig. 3d). C-Mill therapy is a patient-tailored type of training in that the therapist can adjust the difficulty of the different exercises by manipulating content parameters as the obstacle size and available response time for obstacle negotiation, the variation in the sequence of stepping targets, and the degree of acceleration and deceleration of the moving walking area. As progressive training has previously been shown to have beneficial effects [29‐31], therapists are instructed to increase the difficulty of C-Mill exercises as tolerated by the participant, by either changing content parameters or increasing the belt speed, as described in the pre-defined training protocol (Table 1). To assist therapists in progressively scaling the C-Mill therapy sessions, the participant’s perceived fear and difficulty levels during the sessions will be assessed at a scale from 0 (no fear or not difficult) to 10 (much fear or very difficult), as well as their rating of perceived exertion using the Borg scale (range 6–20, [32]). Furthermore, the pre-defined protocol will guide therapists to vary C-Mill exercises, both in terms of content and the order in which the exercises will be performed, inspired by recent insights into motor learning showing superior transfer and retention effects with variability in practice [16]. C-Mill therapy will be performed in groups of two persons with stroke supervised by one therapist. Therapy sessions will last 1.5 hours each, divided in exercise blocks of 3–8 min, during which the participants alternately train and rest (Table 1).

The FALLS program [9] is an overground therapy program aimed at reducing the number of falls in people with stroke by practicing walking adaptability, among other aspects (as detailed in Table 2 and Additional file 2). Figure 4 shows various exercises of this pre-defined FALLS program, including exercises to practice obstacle avoidance (Fig. 4a), exercises to practice foot placement while walking over uneven terrain (Fig. 4b), tandem walking (Fig. 4c), and slalom walking (Fig. 4d). These exercises must also be performed while cognitive and motor dual-tasks are imposed, as well as under visual constraints. In addition, the program incorporates exercises to simulate walking in a crowded environment and to practice falling techniques (one session per week). The FALLS program was originally performed in groups of six persons with stroke, with two or three therapists per group in therapy sessions lasting 2 hours each [9]. Following design considerations for this study (as detailed in the next section), the FALLS program will be performed in groups of four to six persons with two or three therapists per group, with sessions lasting 1.5 hours, including rest.

Both interventions are matched for therapy duration (90 min), frequency (twice weekly) and therapist attention (mean participant-to-therapist ratio, 2:1). The amount of walking practice per session (defined as the number of steps performed during therapy sessions) will be compared between the two interventions and treated as the process measure. Therefore, the number of steps taken during C-Mill therapy sessions will be registered using the treadmill’s inbuilt step counter, while an observer will count the number of steps taken during the FALLS program offline in a random selection of FALLS program sessions using video recordings of the sessions in question.

Finally, after completing the last session of the intervention, participants will be asked to fill in a purpose-designed questionnaire to register perceived discomfort during and after therapy sessions, as well as their experience with the therapy, to compare the feasibility of the interventions from a participant’s perspective.

After group allocation, pre-intervention tests will be performed to assess the baseline values of primary and secondary outcome measures and to collect participant characteristics (sex, age, height, body mass, medication use, co-morbidities, side and location of the lesion, current living situation, daily functioning and the use of assistive devices). The primary outcome measure in this study will be walking speed. Walking speed will be assessed using the 10 Meter Walking Test [33], which has been shown to be a reliable and robust means for measuring walking speed [34].

The secondary outcome measures are inspired by the targeted-stepping and obstacle-avoidance results of Hollands et al. [35] and Van Ooijen et al. [15], underscoring the importance of task-specificity and context-specificity in walking adaptability assessments. Van Ooijen et al. [15] showed enhanced obstacle-avoidance success rates at lower attentional costs after a period of C-Mill walking adaptability therapy, while Hollands et al. [35] showed that measures of targeted stepping were clinically meaningful components in the recovery of functional mobility after stroke. Therefore, the 10 Meter Walking Test will also be performed in combination with context (10 Meter Walking Test with three obstacles, a tandem walking path and three stepping targets) (Fig. 5), a cognitive dual-task (10 Meter Walking Test while counting backwards in steps of three [36]) and both context and dual-task (10 Meter Walking Test with three obstacles, a tandem walking path, and three stepping targets, and while counting backwards in steps of three). Walking adaptability will also be assessed using the Interactive Walkway (Technology4Science, Vrije Universiteit Amsterdam, The Netherlands), a walkway instrumented with multiple Microsoft Kinect for Windows sensors and a projector to present visual context, such as obstacles and stepping targets in a gait-dependent manner (Fig. 6). The walking adaptability evaluation with the Interactive Walkway includes targeted-stepping assessments, obstacle-avoidance assessments, and obstacle-avoidance assessments while counting backwards in steps of three. The obstacle-avoidance assessment of the Interactive Walkway differs from the 10 Meter Walking Test with context in that the Interactive Walkway obstacles can be suddenly presented in a gait-dependent manner, that is, the obstacle suddenly appears at the location where the participant would place his or her foot without adjusting gait. Hence, a step adjustment is always required to avoid the obstacle successfully. Moreover, this step adjustment needs to be performed under high time-pressure demands, which is especially difficult for persons after stroke [37]. The difference between the stepping targets within the 10 Meter Walking Test with context and the Interactive Walkway targeted-stepping assessment is that the Interactive Walkway targets are presented in regular and irregular sequences of visual stepping targets based on participants’ self-selected step length. In this way, it is possible to evaluate foot placement errors on a step-to-step basis for each participant. The 10 Meter Walking Test scores and Interactive Walkway assessment scores will be given in seconds required to complete each test, as well as in the number of errors made during the obstacle crossings, targeted stepping, and tandem walking. The cognitive dual-task, a serial-3 subtraction task, will be analyzed by counting the number of subtractions, as well as the number of mistakes made (dual-task performance [DTP]). Subtraction-task performance while walking will be normalized to subtraction-task performance while sitting (i.e., single-task control condition). These walking adaptability evaluation tools are expected to be sensitive and specific for finding improvements after walking adaptability interventions.

Secondary outcome measures are drawn from a comprehensive set of common clinical measures to determine walking ability, balance, and other walking-related constructs, including Timed Up-and-Go test [38] and Functional Ambulation Category [23]. The obstacle-avoidance subtask of the modified Emory Functional Ambulation Profile will be performed [39], a conventional clinical test closely related to the construct of walking adaptability. The modified Emory Functional Ambulation Profile is reliable and valid for use in people with stroke [40]. Balance will be assessed using the Berg Balance Scale, which provides a psychometrically sound measure of balance impairment for use in post-stroke assessment [40, 41]. Executive function will be assessed using the valid and reliable Trail Making Test [42]. Balance confidence will be assessed with the Activities-specific Balance Confidence scale, a questionnaire measuring balance confidence in performing specific activities, which has good test-retest reliability and validity [43, 44]. Self-reported limitations in walking will be assessed using the Walking Questionnaire [45], which targets experienced limitations in indoor and outdoor walking relative to pre-stroke walking limitations. Finally, the Nottingham Extended Activities of Daily Living scale [46‐48] will be used to assess activities of daily living. Table 3 provides an overview of the tests that will be performed at T0, T1, T2 and T3.

Finally, the number of steps taken per therapy session will be recorded, since we expect that the amount of walking practice per session (defined as the number of steps performed during therapy sessions) will be higher for treadmill-based C-Mill therapy than for the overground FALLS program. This expectation will be tested by comparing this process measure between the two intervention groups.

The primary outcome measure in this study will be walking speed. Previous clinical trials in people with stroke by Yang et al. [49] and Jaffe et al. [50] showed greater improvements in walking speed after treadmill training in a complex and challenging virtual reality environment than after, respectively, conventional treadmill training and overground obstacle-avoidance training [49, 50]. Unfortunately, effect sizes and required sample sizes for a controlled clinical trial with multiple comparisons cannot be estimated from the results of these studies, but both reported significant between-group differences in walking speed with small sample sizes of 9 to 10 participants in each intervention group. The study of Yang et al. [49] allows for a sample size calculation for post-hoc analyses for significant group effects on walking speed with independent t tests. Based on those results, we aim for a relative, clinically relevant, improvement in walking speed of 0.50 km/h (∆) with a common standard deviation (SD) of 0.47 km/h, which results in a sample size of 14 participants in each group to achieve 80 % power with a two-tailed α of 0.05, i.e., following[51]. Considering a drop out of 10–25 %, we chose to increase our sample to 20 participants in each intervention group to be on the safe side for establishing the relative efficacy of the two interventions in terms of improvements in walking speed.

Primary and secondary longitudinal outcome measures that are normally distributed will be analyzed using repeated-measures ANOVA with the between-subject factor group (two levels: C-Mill therapy and the FALLS program) and the within-subject factor time (four levels: pre-intervention [T0], post-intervention [T1], retention [T2], and follow-up [T3] tests). Post-hoc analysis using independent t tests between groups per time level will be performed in case of significant interaction effects. For ordinal or non-normal distributed variables, we will use Mann–Whitney U tests and Friedman tests to evaluate possible main effects of group and time, respectively. To analyze possible interactions between groups and times, we will apply Kruskal–Wallis tests to change scores (i.e., relative to the previous time level) at T1, T2, and T3. When significant, Mann–Whitney U post-hoc tests will be performed to identify between-group differences in change scores per time level. Significant effects are assumed for P < 0.05. Data will be analyzed as randomized. Missing data will be imputed using the data from the last available measurement.

Recruitment commenced in 2013 and is ongoing. Results of this study are expected in 2017.