Effects of robot-assisted gait training on cardiopulmonary function and lower extremity strength in individuals with spinal cord injury: A systematic review and meta-analysis

Abstract

Context

Robot-assisted gait training (RAGT) has been increasingly adopted in many rehabilitation facilities for walking function and activity in individuals with spinal cord injury (SCI). However, the effectiveness of RAGT on lower extremity strength and cardiopulmonary function, especially static pulmonary function, have not been clearly outlined.

Objective

Determine the effect of RAGT on cardiopulmonary function and lower extremity strength in SCI survivors.

Methods

Eight databases were systematically searched for randomized controlled trials comparing RAGT with conventional physical therapy or other non-robotic therapies for survivors with SCI. Study selection required lower extremity strength decline after SCI at baseline. The overall effects of RAGT were calculated using a meta-analytic method. Begg’s test was used to assess the risk of publication bias.

Results

The pooled analysis demonstrated that RAGT may have a positive effect for individuals with SCI on lower extremity strength enhancing (n = 408; standardized mean difference [SMD]  =  0.81; 95% confidence interval [CI] = 0.14–1.48) and cardiopulmonary endurance(n = 104; standardized mean difference [SMD]  =  2.24; 95% confidence interval [CI] = 0.28–4.19). However, no significant effect was established on static pulmonary function. No publication bias was observed according to the Begg’s test.

Conclusions

RAGT may be a useful technique for improving lower limb strength and cardiovascular endurance in SCI survivors. The usefulness of RAGT in enhancing static pulmonary function was not demonstrated by the study. However, these results should be interpreted with caution, given the low number of selected studies and subjects. Clinical studies with large sample sizes will be necessary in the future.

Introduction

Spinal cord injury (SCI) is the partial or total loss of sensory and motor function below the level of injury due to trauma, disease, or congenital factors which hampers patients’ motor skills, living conditions, and ability to work. Recently, SCI incidence has ballooned to around 20–140 cases every one million people globally each year (1). Along with the severe sensory and motor function loss below the injury plane, SCI is also associated with cardiopulmonary dysfunction (2). The vicious cycle of limitation in mobility state and reduction in cardiopulmonary function (CPF) is frequently observed in people with SCI, while prolonged immobilization and reduced mobility states can lead to poor CPF. This limits social integration and increases the risk of developing other comorbidities including heart disease and pneumonia.

A new medical technique, Robot-assisted gait training (RAGT) is a promising and rapidly expanding medical intervention with the potential to provide automatic and repetitive training to enhance functional restitution and motor control for SCI patients. This type of precise care ultimately lessens the manual burden patients need to deal with during rehabilitation (3). However, there exists no definite evidence that RAGT definitely improves CPF in patients with SCI: some studies have shown that RAGT improves metabolic responses, such as heart rate and VO₂ max in patients with spinal cord injury (4). Benson and Xiang revealed a positive effect of RAGT on static pulmonary function (SPF) (5,6), although other studies report otherwise (7,8). With the recent popularization and application of RAGT in clinical practice, various reviews and meta-analyses have confirmed the positive effect of RAGT on walking function in patients with SCI (9–12). Few studies however focused on the effect of RAGT on LES (13,14), while no meta-analysis has synthesized enough evidence to improve CPF. Additionally, most studies included only English language articles. Some also include articles which were non randomized controlled trials (RCTs). However, the quality of the studies included in the meta-analysis often affects the estimation of the intervention efficacy (15). As non-RCTs are generally of lower methodological quality than RCTs. The inclusion of non-RCTs in meta-analyses may affect the reliability of the pooled effect size.

Therefore, the aim of our meta-analysis was to examine the effectiveness of RAGT in improving CPF and LES among individuals with SCI.

Methods

Data sources and searches

A Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (16) was used to perform the meta-analysis (See Supplementary Material 1). The protocol was previously registered at International Platform of Registered Systematic Review and Meta-analysis Protocols (INPLASY), INPLASY202220066.

Studies published up to December 31, 2022 were searched in journals such as PubMed/Medline, Embase, Web of Science, Physiotherapy Evidence Database (PEDro), China National Knowledge Internet (CNKI), China Science and Technology Journal Database (VIP), Wanfang Data. The following search terms were used to search for RAGT: “RAGT”, “robot*”, “robotic-assisted gait training”, “lower limb robot”, “lower limb exoskeleton robot”, “lokomat” and “exoskeleton”. Meanwhile, the following search terms were used for CPF: “cardiopulmonary function”, “pulmonary”, “endurance”, “maximal oxygen consumption (VO2max)”, “peak expiratory flow (PEF)”, “forced expiratory volume in first second (FEV1)”, “forced vital capacity (FVC)”, “maximal voluntary ventilation (MVV)”. Other keywords included “spinal cord injury” and “lower extremity strength”. We combined multiple terms about RAGT with “OR”, multiple terms about CPF with “OR” and then combined the four parts with “AND”. We made the search algorithm by taking PubMed as an example, and this form shows in Supplementary Material 2. We also manually searched the references of reviews and included aticles to identify additional relevant studies. All potentially eligible studies will be considered with no language limitation.

Study Selection: Studies were chosed if their content met the following: (1) patients had been diagnosed with spinal cord injury; (2) randomized controlled trials (RCTs) comparing RAGT with other training methods in SCI; (3) must have reported at least one outcome among CPF and LES at baseline and end of treatment. As for CPF, we try to look for some outcome measures such as VO2max or VO2peak, anaerobic threshold (AT), metabolic equivalent of energy (MET), and maximum heart rate (HRpeak). As for LES, we mainly look for outcome indicator lower extremity motor score (LEMS). Because LEMS is usually used to evaluate LES in patients with SCI. However, they were outright excluded if the following were noticed: (1) compared the effects of two different types of robots; (2) were designed as a non RCT trial such as a self-crossover control trial or an uncontrolled trial; (3) were without accessible data and contacts of the authors; (4) were case reports or pre–post design studies. article references and sources were evaluated for eligibility as well.

Quality Assessment (Risk of Bias): The Physiotherapy Evidence Database (PEDro) scale was used to assess the quality of the studies, measuring methodological quality of randomized controlled trials (17). The PEDro scale is an 11-item scale with a maximum score of 10. Studies with a rating of at least four were considered to be of moderate to high quality (18). Two independent reviewers evaluated the quality of each study, and any discrepancies were discussed with a third reviewer.

Data Extraction and Analysis: The following data were extracted from the selected studies: study design, intervention, number of participants, characteristics of participants, outcome measures (mean, standard deviation) for CPF, as well as LES of treatment group and control group. Two independent reviewers searched for relevant articles and extracted information from the studies. A third reviewer participated in discussions and focused on discrepancies. The reviewers were among the authors, however the author responsible for statistical analysis did not participate in the literature screening to avoid manipulation of data.

Outcome measures assessing the effect of RAGT on SCI were as follows: CPF and LES. The Cochrane Q and I² tests were used to assess inter-study heterogeneity (19),A value of p < 0.1 within the Cochrane Q test was considered to represent heterogeneity between studies and a value of I²> 50% meant moderate to high heterogeneity (20)^.

A random effect model was used to properly account for the added uncertainty brought on by variation between studies. Because there were many CPF evaluation scales, we employed the standardized mean difference (SMD) to report treatment effect. A predetermined sensitivity analysis was also performed by excluding studies one by one to explore the effect of each selected study on overall estimates (21). If the pooled effect remained unaltered after each study was respectively removed, results were then considered relatively credible. Begg’s test (22), a regular statistical test for detecting publication bias, was used to identify the publication bias. All P values were calculated using a two-sided test where P < 0.05 was considered statistically significant. STATA 16.0 (StataCorp LP, College Station, TX) was then used to perform subsequent data analysis.

Results

A total of 1440 relevant studies were identified using the previously described search strategy. Eleven studies (2,6,23–31) met the inclusion criteria after a thorough screening process (Fig. 1). Two articles were published in Chinese language-based journals (27,28). Four selected studies (23,25,27,31) compared RAGT with overground training (OGT). Six selected studies (2,6,24,26,28,29) compared RAGT with conventional physical treatment (CPT) . One selected study (30) compared RAGT with aquatic therapy (AT). The intervention of these studies included locomotor training (Lokomat) (10 trials) (2,23–31)^, exoskeleton-assisted walking (EAW) (one trial) (6) . The characteristics of selected studies on design, participants, intervention, outcome, details of treatment protocol and evaluation of study quality are presented in Supplementary Material 2. Four studies (2,24,30,31) measured VO₂ max as outcomes, and two studies (2,6) measured PEF, FEV₁, FVC, MVV and LEMS as outcomes . The duration of the intervention ranged from four weeks to three months and an evaluation of study quality is also shown in Supplementary Material 2. All selected studies scored 5 or higher on the PEDro scale.

Figure 1.

Flow chart of study selection.

Lower extremity strength

All included studies (n = 408 patients) assessed LES with lower extremity motor score (LEMS) (2,6,23–31). The pooled analysis demonstrated significant effect of RAGT on LES increasing after treatment in individuals with SCI (SMD  = 0.81; 95% confidence interval [CI] = 0.14–1.48) (Fig. 2). In lower-limb robots, ten studies (2,23,31) used Lokomat and only one (6) used EAW. The results of subgroup analysis favored Lokomat over controls for LES increasing (SMD = 0.88; 95% confidence interval [CI] = 0.16–1.60). No significant effect was immediately detected on the effect of EAW versus control group on LES after treatment (SMD = 0.07; 95% confidence interval [CI] = −0.85–0.99) (Fig. 2A). Eight studies (2,23,26,28,30,31) performed RAGT sessions over 6 weeks where results showed significant effect (SMD = 1.07; 95% confidence interval [CI] = 0.23–1.91). Three studies (6,27,29) performed RAGT sessions for less than six weeks, with results showing no significant effect (SMD = 0.08; 95% confidence interval [CI] = −0.51–0.68) (Fig. 2B). Meanwhile, four (23,25,27,31), six (2,6,24,26,28,29), and one (30) study, compared the effects of RAGT with OGT, CPT, and AT on LES, respectively. Subgroup analysis shows that RAGT was more effective in improving LES than CPT (SMD = 1.21; 95% confidence interval [CI] = 0.09–2.33). However, it was not better than OGT and AT (SMD = 0.46; 95% confidence interval [CI] = −0.43–1.35)/ (SMD = 0.04; 95% confidence interval [CI] = −0.65–0.72) (Fig. 2C). Additionally, no significant effects were immediately detected for Paraplegia nor Tetraplegia effects versus the control group on LES after treatment (SMD = 1.37; 95% confidence interval [CI] = −0.11–2.84) and (SMD = 0.52; 95% confidence interval [CI] = −0.19–1.23) (Fig. 2D). Despite the statistically obvious variation between studies, the RAGT groups outperformed the CPT control group.

Figure 2.

The effect of RAGT compared with control on LES by pooling data from 11 trials (n = 408): A (Robot device type); B (Intervention time); C (Control group training method); D (Lesion level).

Cardiopulmonary function

Four studies (n = 104 participants) (2,24,30,31) reported data concerning CPF. Pooled analysis demonstrated that RAGT led to better cardiopulmonary endurance compared to control group (SMD = 2.24; 95% confidence interval [CI] = 0.28–4.19) (Fig. 3). Despite the statistically obvious variation between studies, the RAGT groups showed a beneficial effect over other control groups.

Figure 3.

The effect of RAGT compared with control on CPF by pooling data from 4 trials (n = 104).

Two studies (n = 34 participants) (2,6) used PEF, FEV1, FVC and MVV to measure static pulmonary function. No significant effect of RAGT were noticed on PEF in individuals where SCI was observed. (SMD = 0.34; 95% confidence interval [CI] = −0.34–1.02). Moreover, no significant difference was observed on FEV1 (SMD = 0.62; 95% confidence interval [CI] = −0.19–1.43), FVC (SMD = 0.42; 95% confidence interval [CI] = −0.40–1.24) and MVV (SMD = 0.07; 95% confidence interval [CI] = −0.60–0.75) after treatment (Table 1).

Table 1.

The effect of RAGT compared with control on static pulmonary function by pooling data from 2 trials (n = 34).

Primary outcome	Studies including	I2	Results of meta-analysis
			Effect model	SMD (95%CI)	P
PEF	2	0.0%	random	0.34 (−0.34∼1.02)	0.842
FEV1	2	25.5%	random	0.62 (−0.19∼1.43)	0.247
FVC	2	29.7%	random	0.42 (−0.40∼1.24)	0.233
MVV	2	0.0%	random	0.07 ( −0.60∼0.75)	0.841

PEF: peak expiratory flow; FEV1: forced expiratory volume in first second; FVC: forced vital capacity; MVV: maximal voluntary ventilation.

As shown in (Fig. 4), the point estimates after excluding one study at a time were unanimously within the pooled effect of all studies, indicating that these data are relatively credible. The results of Begg’s test and Egger’s test (Fig. 5) both indicate a low risk of publication bias (P > 0.1). Individual P value for each Begg’s and Egger’s test were presented in Table 2.

Figure 4.

Sensitivity analysis of included study. A: lower extremity strength after RAGT. B: Maximal oxygen consumption after RAGT.

Figure 5.

Begg’s funnel plots of the publication bias. A：lower extremity strength after RAGT. B: Maximal oxygen consumption after RAGT. SE: standard error. SMD: standardized mean difference.

Table 2.

Results of Begg's test for each meta-analy.

Begg's test	P
Lower extremity strength	0.35
Cardiopulmonary function	0.75
Egger's test	P
Lower extremity muscle strength	0.75
Cardiopulmonary function	0.124

Discussion

Our study showed that RAGT had a positive improvement on LES and CPF in persons with SCI, but the effect on static pulmonary function is relatively limited.

In the field of gait therapy for those with SCI, RAGT has drawn a lot of interest. This extensive exposure to task-specific repetitive training aids in the promotion of reorganization of the primary motor cortex, and functional outcomes can be improved in patients with neurological conditions like SCI (32). It allows wheelchair users to practice up to 1000 steps through a 30-minute treatment session, as opposed to a peak of only 50–100 steps during a traditional treatment session (33). Our meta-analysis of eleven selected RCTs concluded that RAGT may help enhance muscle strength in lower limbs after SCI. Consistent with previous studies, our data supports the use of RAGT for LES improving in clinical settings (34). However, according to subgroup analysis results, patients must have sufficient training intensity with at least six weeks of training to achieve the expected training effect. Given the significant effect of the combined effect size, this suggests that the results are statistically significant. However, the actual clinical effect needs further study. In addition, we found that some of the included studies had higher significance, such as the studies done by Ana (23), Melike (26) and Shi (28). Compared with other studies, these studies all had a higher sample size (greater than 30), and most patients with spinal cord injury had a shorter course of disease (<6 months), AISA rating of C or D, and training time of more than six weeks. We speculate that the high significant difference in the results of these studies may be related to the above factors.

People with SCI may have decreased cardiopulmonary capacity as a result of their extended limited mobility. The ACSM guidelines for exercise prescription (35) recommend that people with SCI should participate in an aerobic exercise activity of moderate to vigorous exercise intensity (30–60%HRR, 30–60%V̇O2 max, or 3.0–6.0 METs) at least twice per week. However, due to the injuries resulting from paralysis, the modes of exercise available to this population are limited. A recent systematic review by Peters et al. (36) indicated that vigorous training was effective to increase cardiorespiratory fitness (peak power output and oxygen uptake) among people with SCI (36). Hence, RAGT is currently attempted clinically in persons with SCI to enhance their CPF considering the elicitation of a metabolic response during the exercise.

Our meta-analysis showed that RAGT can improve CPF in patients with SCI. However, no improvement was fund in static pulmonary functions. This may be related to the fact that patients in RAGT mainly perform lower limb walking training, which is relatively less involved for trunk and respiratory muscles. Undoubtedly, the forced expiratory volume in the first second (FEV1) of both included studies attained the minimal clinically relevant difference value of 0.1 L despite the statistical findings of the meta-analysis on static pulmonary function showing no significant difference (37). Therefore, RAGT may have potential clinical significance for static pulmonary function. Studies such as Xiang's (6) used an exoskeleton robot as a training device which requires the participation of the patient's upper limb.

Results also showed that FVC, FEV1, PEF, and MVV had significantly improved after treatment, which may reflect an increase in respiratory muscle strength and ventilation. Therefore, the extent and plane of the patients’ injuries, as well as whether the robotic training method used involves the upper limb, may determine whether RAGT can enhance the pulmonary function of patients. Additionally, we also tried to extract the indicators that could reflect cardiac function except VO2max from the article. Only one article, though, mentions peak heart rate, and none of the others include data results like exercise endurance times or MET levels.

This meta-analysis was the first to investigate the effects of RAGT on static pulmonary function. In addition, the study conducted updated meta-analyses for LES and CPF, respectively. Furthermore, the outcomes were unaffected by a sensitivity analysis that disregarded each trial individually. The results of Begg’s test indicated a low risk of publication bias among included studies with P values higher than 0.1. Therefore, the results in this study are statistically robust.

However, when analyzing the results of meta-analysis, we should not only consider whether the results have statistical significance according to the combined effect size, but also judge whether the results have clinical significance based on professional knowledge. It has been reported that a minimum clinically significant difference (MCID) of >3 units (38) of LES was necessary to improve walking function, but the MCID in VO2 max was not found. Obviously, the LES statistical effect size (SMDs = 0.81) in our study has not reached the MCID, and it is not clear whether VO2 max has reached the MCID. Even so, considering that RAGT can greatly reduce the physical expenditure of therapists, at the same time is conducive to improving the patient's daily active training amount, improve mood, its clinical application is of certain significance.

Some limitations are also noted in this study herein. Ultimately, eleven studies were included but only a very small number has assessed changes in CPF which limits the statistical power of the results. Moreover, there was a lack of comprehensive assessment indicators for cardiac function in the included studies. Due to the absence of some data, it is difficult to conduct a subgroup analysis based on SCI severity, age, and course of disease in this study. And these factors may have a certain impact on the merger results. In addition, some studies (23,26,28) had significantly higher results than others, which may have led to better results. We only searched a limited number of databases, ignoring some important ones such as SCOPUS, CINHAL and SPORTDiscuss. This may cause us to miss some studies of interest. Obviously, it is necessary to further expand the scope of database search to include as many appropriate literatures as possible in future research.

Finally, most robot types included in our study were Lokomat while EAW was relatively few, hence the results may not be accurate enough for the training comparison of the two devices. Further reviews are suggested to compare effects of different robots in SCI patients with different injury planes and severities. Future studies are therefore suggested to include more indicators that can reflect cardiac function should be considered to comprehensively evaluate changes in cardiac function by RAGT. It may also be a direction worth studying to explore the effect of RAGT on pulmonary function in patients with spinal cord injury.

Conclusions

RAGT had positive effects on LES and CPF for patients with SCI. No significant effect was observed of RAGT on static pulmonary function. However, these results should be interpreted with caution, given the low number of selected studies and subjects. Clinical studies with large sample sizes will be necessary in the future.

Disclaimer statements

Contributors CW, SH and XW were involved in the study selection, data extraction and manuscript writing. PG, ZW and YZ performed the statistical analysis and participated in the data interpretation. BS and YL conceived the project and the study hypothesis, designed the study selection criteria, analyzed data. All authors read and approved the final manuscript. Chunli Wan, Xue Wang and Yuting Zhang are in training.

Funding This study was supported by the Major Scientific Research Project of Wuxi Health and Family Planning Commission [grant number Z202013], Wuxi Health and Family Planning Commission“Wuxi Taihu Talent Plan” [grant number WXTTP2020008], Jiangsu Provincial Commission of Health and Family Planning [grant number LR2021040] and Wuxi Municipal Bureau on Science and Technology.

Conflicts of interest Authors have no conflict of interests to declare.