Overground Gait Training for Individuals with Chronic Stroke: A Cochrane Systematic Review

Abstract

Background and Purpose: 

Overground gait training—observation and cueing of patient’s walking pattern along with related exercises—forms a major part of rehabilitation services for individuals with chronic stroke in almost every setting. This report of a Cochrane systematic review assessed the effects of overground gait training on walking ability for individuals with mobility deficits subsequent to chronic stroke.

Data Sources: 

We searched the Cochrane Stroke Group Trials Register, Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE, EMBASE, CINAHL, Science Citation Index Expanded, http://www.clinicaltrials.gov, among other databases through spring 2008. We also searched reference lists and contacted authors and trial investigators.

Review Methods: 

Only randomized controlled trials comparing overground physical therapy gait training with a placebo or no treatment condition were included.

Results: 

Nine studies involving 499 participants were included. We found no evidence of a benefit on walking function, the primary variable, based on three studies (n = 269). Some performance variables did show significant effects. Walking speed increased by 0.07 m/sec [95% confidence interval (CI), 0.05-0.10] based on seven studies (n = 396), Timed Up and Go test improved by 1.81 seconds (95% CI, −2.29 to −1.33) based on three studies (n = 118), and six-minute walk test increased by 26.06 m (95% CI, 7.14-44.97) based on four studies (n = 181).

Conclusions: 

We found insufficient evidence to determine whether overground physical therapy gait training benefits walking function in individuals with chronic stroke, although limited evidence suggests potential benefits for some performance variables. High-quality randomized controlled trials are needed to replicate and extend these findings.

INTRODUCTION

Stroke is a leading cause of serious, long-term disability in the United States, with about 700,000 new or recurrent incidents each year.¹ Almost two thirds of the immediate survivors of stroke have initial mobility deficits,^2,3 and six months after a stroke, >30% of individuals still cannot walk independently.^3–5 However, the relatively high rate of walking independence may mask substantial mobility deficits. One year after a stroke, one half of community-dwelling individuals with stroke (and thus those with relatively good recovery) could not complete a six-minute walk test (6MWT), and those who could were only able to walk 40% of their predicted normal distance.⁶ This limitation in walking function is frequently responsible for long-term disability and handicap because recovery of gait is central to regaining independent living for individuals after a stroke.⁷ Although intensive rehabilitation services, including physical therapy, can aid significantly in recovery within the first three to six months,^1,8,9 relatively few studies have addressed whether continuing services for individuals with chronic stroke (more than six months since stroke) will lead to further recovery in mobility.

Within physical therapy services for stroke, gait training forms the major component of intervention, at least for individuals with acute stroke (less than six months).¹⁰ Gait training refers to a wide range of physical therapy interventions, all aimed at improving the functional activity of walking. Overground gait training, perhaps the predominant form, includes a physical therapist’s observation and manipulation of the individual’s gait over a regular floor surface and is often accompanied by practice walking on common surfaces such as flat ground, ramps or stairs, and exercises specifically designed to improve gait. Overground gait training is performed in virtually every setting from home care to small outpatient facilities to large rehabilitation units and reflects a basic element of physical therapist training.

Recently, efforts aimed at improving rehabilitation of walking function have shifted toward increased amounts of gait practice by the use of mechanical devices or fitness-oriented exercise protocols. A systematic review that considered treadmill training with or without body weight support for people with stroke did not find significant and consistent benefits compared with other gait training methods.¹¹ A systematic review that considered fitness programs for individuals with stroke¹² found limited evidence that cardiorespiratory training improves functional ambulation category and maximum walking speed, at least for a high-intensity treadmill training that progressed quickly through different levels of exercise. In contrast, cardiorespiratory training done within the context of usual care or in combination with strength training did not improve functional mobility. Interestingly, no systematic review has specifically addressed whether the less technologically demanding intervention of overground gait training is effective at improving mobility in individuals with stroke. Although there is an overwhelming clinical consensus that overground gait training is needed during the acute stage of recovery for those patients who cannot walk independently,¹³ there has been little discussion of whether overground gait training would be beneficial for individuals with chronic stroke and continuing mobility deficits. A recent systematic review on the impact of physical therapy after stroke¹⁴ found strong evidence that patients benefit from exercises in which functional tasks are directly and intensively trained. Despite this, the efficacy of overground gait training, per se, for individuals with stroke has not been assessed previously by a systematic review.

This article provides an abbreviated report of a systematic review recently published in the Cochrane Database of Systematic Reviews.¹⁵ The purpose of this systematic review was to assess the effects of overground physical therapy gait training on walking function for individuals with chronic stroke and mobility deficits. We measured effectiveness based on (1) multidimensional measures of walking function; (2) performance measures such as overground walking speed, the Timed Up and Go test (TUG), and the 6MWT; and (3) adverse events and death or disability.

METHODS

Types of Studies

The randomized controlled trials (RCTs) were included if they compared overground physical therapy gait training with no intervention or a control intervention (ie, other rehabilitative techniques that do not include gait training).

Types of Participants

The population of interest was adults (older than18 years) who had a stroke at least six months before inclusion in the study and who had mobility deficits; that is, they used an assistive device, exhibited an abnormal gait pattern, or had slowed walking speed. Participants must have had the cognitive ability to follow directions and participate in physical therapy treatments. There were no restrictions on the participants’ living environment or the setting in which care was received.

Types of Interventions

Overground gait training was the intervention of interest. For the purposes of this review, we defined overground gait training as treatment that consisted of at least one of the following.

1. Real-time cueing of the patient’s gait by the use of manual, verbal, positional, or rhythmic cueing techniques
2. Practice of the walking pattern overground
3. Pregait activities such as step-up and step-down exercises, dynamic balance training, weight-bearing exercises to strengthen the lower extremities, and other exercises that require standing and weight-shifting.

We included only studies in which the main intervention was overground gait training as defined earlier. We included interventions that combined overground gait training with other rehabilitation techniques if the main intervention focused on either full-gait activities (items 1 and 2 from the above definition) or on pregait activities (item 3).

Interventions that do not include consistent face-to-face interactions between the therapist and patient (such as home exercise programs) or the primary goal of which was not to improve gait (such as progressive resistance training using exercises for isolated muscle groups) do not fit within this definition. In addition, gait training interventions that focus on the use of treadmills, complex technical equipment, such as body weight-supported treadmill training, functional electrical stimulation, biofeedback based on electromyography, or joint-position measurement, or virtual reality systems do not fit within this definition. The rationale for this restrictive definition is to focus on the types of gait training that require direct interaction with a physical therapist and are available in a typical community-based physical therapy facility or during home-care visits by a physical therapist. We excluded studies that only used treadmill-based training, only used technologically demanding forms of gait training such as body weight-supported treadmill-based training, only used progressive resistance training without gait-oriented exercises, or only included home-care programs.

Types of Outcome Measures

The primary outcome measure was walking function. Walking function was assessed by any of various multidimensional, ordinal scales that evaluate walking function and that are validated for the use with individuals with stroke. Suitable scales included the following.

• Rivermead Mobility Index^16,17
• Stroke Rehabilitation Assessment of Movement¹⁸

Barthel Index^19,20 was included as an unanticipated secondary measure of mobility function that provided information on the participant’s ability (or inability) to function independently because it was available in each of the studies that also provided one of the designated primary measures. Several other secondary outcome measures assessed independent walking function with objective, quantitative variables.

• Walking speed measured over a short distance (≤10 m)^21,22
• TUG^20,23,24
• 6MWT^21,25

For inclusion in the analysis, outcome measures must have been recorded before the intervention and immediately after the intervention. We examined follow-up data for the follow-up point closest to three months after the intervention.

Search Strategy

A computerized literature search was conducted from April 2006 and updated in spring 2008 in consultation with the Cochrane Stroke Group and incorporating their guidelines for intervention-based search strategies. Searching included the Cochrane Stroke Group’s Trials Register (last searched in March 2008), the Cochrane Central Register of Controlled Trials (The Cochrane Library, issue 2, 2008), MEDLINE (1966 to May 2008), EMBASE (1980 to May 2008), CINAHL (1982 to May 2008), AMED (1985 to March 2008), Science Citation Index Expanded (1981 to May 2008), ISI Proceedings (Web of Science, 1982 to May 2006), the Physiotherapy Evidence Database (http://www.pedro.org.au/) (May 2008), REHABDATA (http://www.naric.com/research/rehab/) (1956 to May 2008), http://www.clinicaltrials.gov/ (May 2008), http://www.controlled-trials.com/ (May 2008), and http://www.strokecenter.org/ (May 2008). We also tracked the reference lists from retrieved articles, contacted investigators of identified trials and authors of relevant Cochrane physiotherapy reviews, and used Science Citation Index Cited Reference Search to track relevant papers. Full details of the search strategy are provided within the Cochrane Review.

Search Results

The electronic search using multiple databases yielded 3793 citations. The studies retrieved by the electronic search were distributed to three review authors (R.A.S., Y.S., and E.P.) and obviously irrelevant studies were eliminated (ie, obviously false hits such as “running during a tennis stroke” or studies investigating pediatric subjects). Two review authors (R.A.S. and Y.S. or R.A.S. and E.P.) independently screened the abstracts for the remaining articles, classifying each study as “definitely irrelevant” or “possibly relevant.” Where there was disagreement between review authors, a third review author assessed the abstract and discussion among the three review authors led to consensus. This process led to identification of 83 possibly relevant citations. The three review authors applied the selection criteria by using complete articles for studies classified as possibly relevant. We resolved disagreements by discussion leading to the “included studies.”

Methodological Quality and Data Extraction

Key data were extracted from each included study and were documented using a data coding form. For articles that did not contain sufficient information, study authors were contacted. The 3 review authors independently assessed the methodological quality of the included studies with disagreements resolved by discussion. We evaluated quality, according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions,²⁶ by reporting the score for each of the 11 items contained in the PEDro Scale for rating quality of RCTs,²⁷ and by determining whether validity had been established for each relevant outcome measure from the study. We considered the last category from the Cochrane Handbook (detection bias) in two parts: blinding of evaluators and reporting bias.

Data Synthesis

Meta-analyses were conducted at the end of the treatment period and again for data from the latest time point within the 3 months after the end of the intervention. Thus, meta-analyses were conducted for the following two comparisons on all available variables: (1) overground gait training versus a control condition at the end of the treatment period and (2) overground gait training versus a control condition at the three-month follow-up time point. The Cochrane Collaboration’s Review Manager²⁸ was used for all analyses. Following Review Manager guidelines, the mean values reported here represent change scores from pretest to posttest for studies that provided them, otherwise they reflect values at posttest.

RESULTS

Description of Studies

Nine studies with 499 participants met our inclusion criteria and were analyzed. All the included studies were RCTs, although 2 studies used randomized crossover designs.^29,30 For those 2 studies, only the first phase was analyzed, and hence they were treated as traditional RCTs. Data from a follow-up period of about three months were available and analyzed from five studies (comparison 2).^29–33 Information on study characteristics and study quality are summarized in Table 1.

TABLE 1: Summary of Study Characteristics—Key Characteristics of the Included Studies Are Shown. More Complete Details Are Available in the Full Cochrane Review¹⁵

Comparison 1: Gait Training Versus Control at the End of Treatment

Only three of the nine studies (269 participants) reported data for the primary variable of walking function or for the related variable (Barthel Index). There were no statistically significant differences between the experimental and control groups at the end of treatment as shown in Figures 1 and 2.

FIGURE 1.: Walking function (%). Standard mean differences are shown for each study that contributed to this measure along with the total overall effect calculated from the meta-analysis. For an explanation of the subgroup analyses, see the full Cochrane review.¹⁵

FIGURE 2.: Barthel Index (unitless). Mean difference values are shown for each study that contributed to this measure along with the total overall effect calculated from the meta-analysis. For an explanation of the subgroup analyses, see the full Cochrane review.¹⁵

There were statistically significant differences for the secondary variables of walking speed [mean difference (MD), 0.07 m/sec, 95% confidence interval (CI), 0.05-0.10; Fig. 3], TUG test (MD, −1.81 seconds; 95% CI, −2.29 to −1.33; Fig. 4), and 6MWT (MD, 26.1 m; 95% CI, 7.14-44.97; Fig. 5). For each variable, the gait training group demonstrated better performance than the control group.

FIGURE 3.: Walking speed (in meters per second). Mean difference values are shown for each study that contributed to this measure along with the total overall effect calculated from the meta-analysis. For an explanation of the subgroup analyses, see the full Cochrane review.¹⁵

FIGURE 4.: Timed Up and Go test (in seconds), Mean difference values are shown for each study that contributed to this measure along with the total overall effect calculated from the meta-analysis. For an explanation of the subgroup analyses, see the full Cochrane review.¹⁵

FIGURE 5.: Six-minute walk test (in meters). Mean difference values are shown for each study that contributed to this measure along with the total overall effect calculated from the meta-analysis. For an explanation of the subgroup analyses, see the full Cochrane review.¹⁵

Comparison 2: Gait Training Versus Control at Three-Month Follow-Up

At the end of the 3-month follow-up period, there was a statistically significant difference between the experimental and control groups for walking function (SMD, 0.34 units; 95% CI, 0.01-0.66), but this was based on only one study with 150 participants. There were no significant differences evident for Barthel Index, walking speed, TUG test, or 6MWT, and data were limited to three or fewer studies for each variable.

DISCUSSION

The primary aim of this review was to evaluate the effect of overground physical therapy gait training on the walking function of individuals with chronic stroke and mobility deficits. The results were mixed, with no significant effect for the primary variable at the end of treatment but small effects for several secondary variables. Our mixed findings may be due to the small number of studies that met the inclusion criteria, the limited number of enrolled participants, and the lack of consistent outcome measures used in the included studies.

Meta-analysis showed no significant effect on the primary variable, walking function, at the end of the treatment period, although a small effect was evident at follow-up. Even that follow-up effect, SMD of 0.34, is considered statistically small,³⁴ and it was based on a single study. The results on the related variable, Barthel Index, showed no significant effects for the intervention or follow-up periods. Hence, there is little evidence of a clinically important benefit of overground gait training on multidimensional measures of walking function for individuals with chronic stroke.

The results for the secondary variables (walking speed, TUG test, and 6MWT) were mixed. Although there were some statistically significant effects immediately after treatment, these did not reach the level of “smallest real difference” as documented by Flansbjer et al³⁵ for individuals with chronic hemiparesis subsequent to stroke; Flansbjer et al reported 95% CIs for smallest real differences of −0.15 to −0.25 m/sec for comfortable walking speed, −3.75 to −2.59 seconds for TUG, and −37.3 to −66.0 m for SMWT. Moreover, there was no evidence of significant effects at three-month follow-up, although walking speed was the only variable reported at follow-up by more than one study. Taken together, there is insufficient evidence to show that walking speed, TUG, and 6MWT will translate into reliable and important clinical improvements.

The results for the secondary variables may suggest a promising area for future work. The results for the TUG test and the 6MWT were generated from research focused on well-documented, therapeutic protocols for overground gait training, most of which compared gait training with methodologically rigorous placebo conditions.^31,36–39 Thus, the studies investigated a specific set of exercises aimed at task-oriented training,^31,37 cardiovascular fitness,³⁶ effective weight transfer and increased duration during the support phase of the affected limb,³³ strength training,³⁸ or dual-task performance.³⁹ These more focused interventions may generate benefits because of larger amounts of practice on a limited set of gait skills. Recent work to develop additional performance variables that can be measured in contextually rich environments, such as the use of a functional obstacle course to measure task-oriented training⁴⁰ and multitasking situations to measure dual-task training,³⁹ may help to clarify the specific effects of particular focused gait protocols.

The primary limitation in conducting this review was the paucity of large, high-quality RCTs on the effectiveness of overground gait training for individuals with chronic stroke. Other limitations include the large variability in the disability status of individuals with chronic stroke and mobility deficits, the slow rate of overall recovery for individuals with chronic stroke, the limited descriptions of the experimental interventions in studies focused on broadly based community physical therapy services that included overground gait training, the blunt nature of the primary variable—walking function, the limited duration of follow-up (generally not more than three months), and the diversity of outcome variables used in the included studies.

We are confident that our extensive search strategy across multiple databases identified all relevant trials. We made multiple contacts with investigators to obtain additional information on published studies or updated information on ongoing studies. We also had access to studies registered with the Cochrane Stroke Group and thoroughly reviewed the titles of studies included in the bibliography of relevant articles. We did not perform hand searching of conference proceedings because it has been shown to be ineffective in identifying relevant studies.¹¹

The effectiveness of overground gait training for improving walking function among individuals with chronic stroke is crucial, given the large number of individuals with chronic stroke and mobility deficits.³ The limited available evidence suggests that overground gait training has no significant effect on walking function, although it may provide small, time-limited, benefits for the more unidimensional performance variables of walking speed, the TUG test, and the 6MWT. Despite the lack of statistical significance, the importance of gait training for individuals with chronic stroke may lie in its ability to educate the patient about ways to maintain safety during walking while the patient is encouraged to engage in related exercises and therapeutic approaches to increase strength and cardiovascular fitness, alter coordination or quality of gait, and achieve improved agility and efficiency. If that hypothesis holds true, overground gait training should not be expected to generate substantial changes in specific gait parameters or even in broad measures of walking function by itself. Instead, overground gait training may create the most benefit in combination with other therapies or exercise protocols. This hypothesis is consistent with the finding that gait training is the most common physical therapy intervention provided to individuals with stroke.¹⁰ It is also consistent with other systematic reviews that have considered the benefit of overground gait training in combination with treadmill training or high-technology approaches such as body weight-supported treadmill training,¹¹ or both, or with exercise protocols¹² for individuals with acute or chronic stroke. These combination treatments seem to be more effective than overground gait training alone perhaps because they require larger amounts of practice on a single task than is generally available within overground gait training. In contrast with overground gait training, however, the combination therapies can be difficult to apply in community settings where resources are limited. Every effort was made to define gait training in a consistent and replicable manner that reflects clinical practice. However, some clinicians may disagree with some aspects of the definition; for example, they may believe that pregait activities do not constitute gait training, and by including these activities, we contaminate the results. Although this is plausible, subgroup analyses were conducted to contrast one study that used only pregait activities with others that did only full-gait activities or with those that did a combination. The results of those comparisons were similar to the results reported above.

Despite the plethora of clinically relevant studies on people with stroke and the fact that overground gait training is a mainstay of physical therapy rehabilitation for individuals with chronic stroke, very few studies have investigated its effectiveness as a unique intervention. The literature search for this review suggests that a larger number of studies examining the effectiveness of overground gait training may be available for individuals with acute or subacute stroke. Focusing on the effects in subacute patients may also be fruitful, given the limited data available for individuals with chronic stroke. This review found insufficient evidence to determine whether overground gait training directly benefits broad measures of walking function, although it does suggest that specific training protocols provide limited benefits for more unidimensional measures of gait performance such as walking speed and the 6MWT. Additional evidence is needed to determine whether any type of overground gait training intervention by itself is sufficient to substantially affect broad measures of walking function in individuals with chronic stroke.

CONCLUSIONS

The findings of this study demonstrate a clear need for well-designed randomized controlled studies examining the effectiveness of overground physical therapy gait training for individuals with chronic stroke. Future studies should give a detailed description of a well-defined therapeutic protocol used for the intervention, include large sample sizes so that they can be adequately powered, ensure that randomization procedures are adequate and clearly outlined, ensure that the assessors who administer outcome measures are blind to group assignment, and include outcome variables that measure quality of life, function, and adverse events as well as the more unidimensional performance variables such as walking speed and the 6MWT. Investigators should consider the findings of this review when designing trials and attempt to overcome the limitations of the studies presented.

ACKNOWLEDGMENTS

The authors greatly appreciate the efforts of Jill Lucas-Findley and Mariya Shiyko who reviewed the statistical analysis and provided suggestions on reporting results. Limited amounts of article retrieval and archival help were provided by graduate assistants supported by Long Island University, Brooklyn Campus.