Cost-Effective Clinical Methods to Evaluate the Efficacy of Gait Training Devices: A Systematic Review of Randomized Controlled Trials

Stroke is a medical condition which results in long term disability such as gait dysfunction occurring due to motor function impairment [1,2]. As per the 2022 report released by the World Stroke Organization, the annual number of new stroke patients exceeds 12.2 million and presently 101 million cases exist globally. In their lifetime, one in four persons over the age of 25 will have a stroke [3]. One major aftereffect of a stroke is gait problems. Around 80% of victims lose their ability to walk due to the decline of one or more of the primary factors that determine mobility, such as balance, gait speed, endurance, and cognitive abilities. Gait training is the process of learning to walk after undergoing an injury or disability [4–6]. Normal human gait is a metabolically efficient process which is executed with minimum energy expenditure. Yet, it is a complex movement that involves the skeleton and muscular system which responds to the stimuli from the brain through the nervous system. Diseases that affect any of the above systems will result in deviation in the gait [7–9]. The disability can be treated by various measures such as rehabilitation medicines, surgical treatments, therapy that targets specific muscles, and activities to train effectively [10].

In conventional gait training (CGT), a team of physiotherapists is involved in manual gait training, where two therapists move the patient's limbs while a third provides support. This approach is time-consuming and laborious. Several clinics have also embraced body weight supported gait training, where the upper body is suspended in a harness; therefore, balance is no longer necessary when the therapist perform manual gait training with patients on the floor or on treadmill. In these techniques, it is impossible to maintain repeatable and consistent gait cycles during the training process.

Assistive technology supports the recovering person to increase their strength to be able to perform normal movements. In all stages of poststroke rehabilitation, there is compelling evidence to support a rehabilitation strategy that is focused on repetitive, regulated, intensive, and task-oriented training [11,12] to improve balance, gait speed, endurance, and patient abilities to perform everyday tasks. Numerous robotic devices have been developed to treat patients with different types of cerebrovascular illness. Robotic gait training devices (RGTDs) have been well established in the past two decades. Clinical trials with these devices have been reported widely in the literature for stroke [13–29] and other neurological impairments for both adults [30–41] and children [42–46]. The outcomes demonstrate the superiority of robot-assisted gait training (RAGT) over CGT techniques [47]. RGTDs also ease the work load of the therapists and increase the confidence in the patients undergoing training [48,49]. However, robotic devices are expensive and are not prominently used in developing countries.

There are many new cost-effective gait training devices [50–63] that are being developed which have prospective market opportunities in developing countries. With simple mechanisms producing the desired output like that of robotic devices, these devices are introduced as an alternative to the RGTDs [64]. It is observed from the literature that several such devices are being designed and developed. Some [51,53,54,56,57,61,62] of the above work is in the design phase, while some [50,52,55,58–60,63] have fabricated prototypes of the device and some have even tested the device on able-bodied individuals and patients. Some [50–52,54–56,59–63] designs have used crank mechanisms with 4, 6, 7, and 8 bars, some [53,57] have used cables, and others [54,58] have used cams to generate the trajectory of normal walking. It is also noticed that some [50,51,54,55,58,59,61–63] devices are designed to have attachment with only the user's ankle and foot, while some [52,53,56,57,60] devices extend support to the shank, thigh, and even pelvis. Some devices [54,62] have fixed cadence and fixed stride of the gait, some [51,52,63] have fixed stride and variable cadence, and some devices [50,53,56,57,59,60] have the ability to permit the varying of both parameters. In these devices, the control is at the device's end. There are other devices [55,58,61] which allow the user to control the cadence and stride. Users who do not have mobility control can be trained with the former type, and the latter can be used by users who have certain control over the limbs. These cost-effective, simplistic devices have not entered the market yet as they are in the development stage. There have been no extensive clinical trials that have been reported with these devices. Clinical trials are necessary to assess a patient's recovery and compare results with the existing methods and devices.

This review focuses on the cost-effective outcome assessment methods that were used to assess the robotic devices that exist in the market. This review will be useful to the inventors and developers of low-cost gait training devices (overall cost of the device is less than 5 lakhs rupees). There are many reviews [65–71] that have discussed gait outcome measures. However, to the best of our knowledge, there have not been any reviews that focus on easy and cost-effective assessment methods for novel gait training device developers who intend to conduct clinical trials. There are a wide range of outcome assessment tests that are available, but only a few are used frequently in studies owing to their reliability, responsiveness, and validity [72]. The review is aimed at helping researchers to choose a cost-efficient outcome assessment method(s) while attempting clinical trials with their devices.

Articles on randomized controlled trials (RCTs) on RGTDs that treated stroke patients were identified by searching different databases such as SCOPUS, Web of Science, PubMed, and Cochrane Library until March 2024. The key search words were (effectiveness OR effects OR efficacy) AND robot AND assisted AND randomized AND controlled AND trial AND gait AND training AND stroke. The review was conducted in preferred reporting items for systematic reviews and meta-analyses (PRISMA) style [73].

The studies were included if (a) a RCT was involved in the study, (b) a RGTD was used with the control group as CGT methods, (c) the participants of the study were diagnosed with stroke, (d) the authors used standard outcome measures to assess the device comparing with CGT methods, (e) the article was written in English language or translated into it, and (f) it is an original research article (review articles were used as a secondary source to identify additional articles). The studies are excluded if (a) the device intervention was combined with any other neuromuscular treatments, (b) the study was a protocol design with no trials conducted, (c) studies on prosthetics, wearable devices, or implants, and (d) the combination of experiment group and controlled group was considered in the study. To overcome the risk of bias, the search was conducted over several repeated days, and the order of keywords was changed. The search was performed by one author under the supervision of another author to prevent potential bias and omission of articles.

A total of 489 articles were found using electronic searches of different databases, such as SCOPUS (91), Web of Science (93), PubMed (141), and Cochrane Library (164) from origination to Mar. 31, 2024. After duplicates were eliminated, 135 articles were examined. Of those, 83 articles were disqualified for not meeting our inclusion criteria. A more thorough study was conducted on 52 articles. Considering our exclusion criteria, 35 articles were eliminated from the study. Our systematic review's inclusion criteria resulted in including 17 articles in total. The comprehensive study selection was precisely presented in a PRISMA style flow diagram (Fig. 1).

The characteristics of the studies that were part of the analysis are shown in Table 1. On the whole, 17 RCTs were analyzed [13–29], which consisted of 705 stroke survivors in both RAGT and CGT group. Out of which 359 (51%) patients underwent RAGT and 346 (49%) patients were given CGT. Sample sizes ranging from 20 to 109 were observed in the studies, and the average age varied from 52.05 to 72. The time elapsed since stroke was observed to be different in different studies, 29.4% of the studies considered stroke duration less than 1 yr, 11.76% less than 6 months, 11.76% less than 3 months, 17.64% less than 2 months, 5.8% less than 1 month, and 11.76% if the studies have not reported the duration of stroke.

The duration of the intervention was decided as per the concerned physician's opinion. It ranged from 2 weeks to 10 weeks. The intervention duration was 10 weeks for 5.8% of the studies [23], 8 weeks for 17.64% of studies [16,24,27], 6 weeks for 11.76% of studies [20,28], 5 weeks for 5.8% of studies [13], 4 weeks for 29.42% of studies [14,17,21,26,29], 3 weeks for 17.64% of studies [15,18,22], and 2 weeks for 11.76% of studies [19,25]. Each session ranged from 30 min to 180 min. The time duration was used differently in all the studies. Some studies incorporated physiotherapy as a part of the training and also included rest periods. Total number of sessions for a week varied from two sessions to seven sessions. Two sessions for 5.8% of the studies, three sessions for 5.8% of the studies, four sessions for 5.8% of the studies, five sessions for 58.82% of the studies [13–15,17,20,21,26,27,29,30], seven sessions for 11.76% of the studies [16,19], and two studies did not consider weekly sessions instead preferred observing overall number of sessions (10 (18) and 24 (24) sessions) (Fig. 2).

Patients who received RAGT were supplemented with standard physiotherapy. In this review, studies that used RGTDs that provide body weight support were included. There were different types of RGTDs that were used. The 35.29% of studies [13,22,23,25,27,28] used Lokomat, 11.76% [16,17] used GEAR, and 52.94% of studies used different RGTDs such as Gait-Assistance Robot [14], Morning Walk [15], SUBAR [18], A3 [19], Welwalk [20], MRG P100 HIWIN [21], Autoambulator [24], Walkbot [26], and Exowalk [29]. These devices incorporated exoskeleton like orthosis to house the knee and the ankle, which are in turn operated by linear actuators. Some devices also have sensors to measure joint torques and potentiometers to measure angular positions. The process of human gait movement occurs in all three planes: sagittal, transverse, and frontal planes [74]. These devices demonstrate how rehabilitation can be simplified by limiting motion actuation to the sagittal plane.

The studies that conducted RCTs were chosen in this review. Randomization reduces the chance of a possible bias and is proven to be a great tool to examine cause-effect relationships between interventions and outcomes [75]. In RCTs, the population is carefully selected and recruited to either the control group or the experimental group by random allocation method. There are various randomization methods that were adopted in the studies considered. These methods include randomization tables, randomization software, and block randomization. Although randomization wipes out bias during the initial allotment, blinding is necessary for the entire process of the trial [76]. Concealment during allocation, assessment, and evaluation is imperative to eliminate misleading and faulty results [77]. In conventional RCTs, participants are also blinded. In gait training trials with CGT devices as control group, blinding the participants is not feasible. However, blinding the assessors and evaluators is possible, and it is followed in the studies reported. There are few studies [23,24,26,30] that have not blinded the assessors and evaluators.

The outcome measure can vary depending on the clinic or research institute. Some have the facility to conduct sophisticated outcome assessing tests and analysis, such as metabolic analyzers, which measure the metabolic cost of performing activities [78], and gait analysis, which uses high-end motion capture systems that measure the spatiotemporal parameters in patients [18–20]. Other institutes assess recovery using standard outcome assessment methods (e.g., methods mentioned in Table 2).

These methods require basic materials such as a bed, a chair, a ruler, and a stopwatch. The tests can be performed using these commonly available low-cost materials (Table 2). The most commonly used method was FAC, which was used in nine studies, 10 MWT was used in eight studies, BBS, 6 MWT, TUG, and FIM were used in six studies each, RMI was used in five studies, FMA, MI-L, BI, MAS, and TPOMA were used in three studies each, GA, 2 MWT, and 3 MWT was used in two studies, and 8 MWT, EQ-5D, SF-36 PF, mEFAP, MM, HWAP-AS, NIHSS, SAS, MI-LM, SIAS-L/E, SF-8; GRC, MI-LM, STS, SLS, PASS, AROM, MMT; PROM, 5x-STS; FI; SAEs were used in one study each. Table 3 and Fig. 3 give visual pictorial representation of the most attested and valid outcome assessment methods used in the articles reviewed.

Intensive, regulated, repetitive, and consistent training is proven to stimulate motor skills and thereby promote the restoration of gait in a patient with asymmetric gait. CGT methods are not repeatable, inconsistent, require more staff, and are expensive. The outcomes of therapists-based gait training depend on the personal skills of therapists, and the patients undergoing treatment do not benefit from standardized and uniform therapy [79]. Gait training is an intricate process; if the rehabilitation is not correctly performed, the rehabilitation outcome turns more into a compensating gait than normal gait [80]. RGTDs show promise in terms of repeatability and consistency in the training process when compared to the traditional gait training methods. However, RGTDs are very expensive and are unaffordable in developing countries. Mechanical gait training devices have the advantage of achieving repeatability and consistency and at the same time being cost-efficient.

The developers of RGTDs have tested the efficiency of their devices using RCT with conventional therapist-based gait training as the control group. The trials and methodologies of 17 studies have been discussed in this review. These devices are expensive for large scale use in the developing countries. Simple, cost-effective, electromechanical devices are being designed and developed, which are discussed in the Introduction section of this article. These devices have not reported clinical trials. Even though clinical trials can be expensive, it is imperative to ensure the safety of potential users. A controlled study is required to effectively monitor the effectiveness of these devices. The outcome assessment methods used in the trials with RGTDs are cost-effective. These methods can be adopted to conduct trials with the newer cost-effective devices that are being developed. The assessment methods that were used in three studies or more are chosen to be more reliable and explained briefly in Table 4. The materials required for each of the assessment methods are graphically compared in Table 3.

The most commonly reported outcome assessment methods from the studies are FAC, 10 MWT, BBS, 6 MWT, TUG, and FIM. The outcome measures discussed in this review have the advantage of being easily available, can be easily administered and interpreted, and are cost-effective [108,109]. This is contrary to the complex and expensive laboratory setting that involves detailed analysis of kinetic and kinematic variables [110]. Gait assessment must be performed with reliable, responsive, predictive, and concurrently valid methods [81]. The validated results from various studies [13–29] give compelling evidence that the methods satisfy the requirement. FAC, which is the most widely used outcome assessment method, has shown excellent reliability and responsiveness. It is an expeditious visual measurement of gait which is easily interpreted [109]. Studies indicate FAC scores correlate with step length and walking velocity [111–113]. The timed ambulation tests, 10 MWT, 6 MWT, and TUG, showed remarkable test-retest and interobserver reliability. Owing to its easier implementation and shorter time requirement of the 10 MWT, it is the most effective timed ambulation test [114]. It is also observed that the 10 MWT has excellent intra and inter-rater reliability [115]. BBS demonstrates remarkable consistency and reliability in the monitoring of the static and dynamic sitting and standing balance. The clinical feasibility of BBS is high as it requires minimal equipment and can be done in a short time span [116]. FIM is meant to reflect on a patient's typical performance instead of their peak performance. The reliability and validity of FIM are rated to be good with a fine inter-rater reliability. It is generally compared with BI and is found to be more responsive than BI in stroke patients [117].

The outcome assessment methods should be chosen as per the parameter requirements that are intended to be measured. It is essential to monitor the patient's status during recovery in terms of gait, balance, transfers, and specific goals. Depending on the patient group's disability levels, the outcome measures are chosen. The 3, 8, and 10-m walking tests are widely used for assessing gait speed at short distances, whereas the 2, 3, and 6-min walk test is used to measure endurance over a long duration. Other outcome measures are used to measure patients' independence to perform daily activities. To perform the outcome measures discussed above, one does not require any special training. All the required information that is required to perform these tests is available online.

To the best of our knowledge, there is no review article that provides directions for novel gait training device developers to conduct simple and cost-effective outcome assessing methods to prove the efficacy of their gait training devices. Considering the need for low-cost gait training devices to be deployed in large numbers in the developing countries, it seemed necessary to analyze the existing outcome assessment methods used in research. Herein, a review was conducted with 17 studies which used outcome assessment methods to conduct trials with patients. The main strength of this review is that it narrows down the wide number of tests that are available to a few prospective, easy to use, and cost-effective outcome assessment methods. Moreover, there is enough evidence in the literature that proves these tests as reliable and valid. Despite the strength of this review, there are a few limitations. First, the methods discussed in this review do not assess the qualitative improvement in gait, unlike gait analysis and metabolic measurements. Second, a meta-analysis could not be performed as the studies have a high level of heterogeneity. The last limitation of this review is associated with the selection criteria, which excluded studies that were not conducted on stroke patients. Therefore, studies of other neurological impairments which have similar outcome measures were omitted. These restrictive criteria were chosen to obtain a homogeneous systematic review.

This review can be of use to assist researchers and medical professionals in choosing outcome metrics to assess the efficacy of new cost-effective gait training devices. These outcome assessing methods discussed in this review are quantitative in nature as the patient's ability to perform the tasks is assessed by measuring it in terms of time taken and levels of accomplishment. The assessments have standard scales in which the evaluator gives scores for task based on the quality of accomplishment. These tests require commonly available materials which do not incur huge material costs. The low-cost gait training devices can be an effective alternative to the expensive RGTDs. Many researchers and inventors have designed various novel gait training devices. Performing clinical trials and quantifying and comparing outcome measures can determine the efficacy of these devices. The adoption of standard outcome measures mentioned in this review may facilitate clinical trials and multicenter investigations. This will enable wide scale deployment of these devices in clinics and rehabilitation centers.

The authors do not have any relevant financial or nonfinancial declaration of interest.

No data, models, or code were generated or used for this paper.

Abbreviations

2 MWT =

2-min walk test

3 MWT =

3-m walk test

3 MWT =

3-min walk test

5x-STS =

five times sit-to-stand

6 MWT =

6-min walk test

8 MWT =

8-m walk test

10 MWT =

10-m walking test

AROM =

active range of motion

BI =

Barthel index

BBS =

Berg balance scale

BRS =

Brunnstrom recovery stages

EQ-5D =

EuroQol-five dimension

FAC =

functional ambulation classification

FI =

feasibility of implementation

FIM =

functional independence measure

FMA =

Fugl–Meyer assessment

FMA-LE =

Fugl–Meyer assessment of lower extremity

GA =

Gait analysis

GRC =

global rating of change

HWAP-AID =

Hochzirl walking aids profile—walking aids

HWAP-AS =

the Hochzirl walking aids profile personal assistance

K-MBI =

Korean modified Barthel index

MAS =

modified Ashworth scale

MBI =

modified Barthel index

MBS =

modified Borg scale

mEFAP =

Emory functional ambulation profile

MI-LM =

motricity index-lower

MM =

mobility milestones

MMT =

manual muscle test

NIHSS =

National Institutes of Health Stroke Scale

PASS =

postural assessment scale for stroke

PROM =

passive range of motion

RMI =

Rivermead mobility index

SAEs =

serious adverse events

SAS =

stroke activity scale

SF-8 =

eight-item short form health survey

SF-36 PF =

36-item short-form health survey physical functioning

SIAS-L/E =

stroke impairment assessment set-total lower limb motor score

SLS =

step length symmetry

STS =

swing time symmetry

TPOMA =

Tinetti’s performance-oriented mobility assessment balance and gait subscores

TUG =

timed up and go test