Relationship between Nintendo’s Wii balance board derived variables and clinical balance scores in individuals with stroke.

Abstract

Background:

The Wii Balance Board (WBB) is widely studied as a balance testing platform and is reliable in detecting changes in body’s center of pressure (COP). However, the relationship between WBB derived measures and clinical tests of balance is currently unknown.

Research question:

To investigate the association between static and dynamic COP measures from the WBB with balance tests commonly used in chronic stroke.

Methods:

This retrospective study included sixty-nine individuals with stroke who performed the Berg Balance Score (BBS), the Mini-BESTest and WBB assessments as a part of their baseline measurements. The WBB assessments included body weight symmetry and a dynamic target matching task that measured Reaction Time (RT) and Completion Time (CT).

Results:

Body weight symmetry performed with eyes open was significantly different between participants who were classified as high and moderate balance using the Mini-BESTest (p=0.03). A significant negative linear correlation was observed between the BBS and CT (rho = −0.29, p=0.021) and between the Mini-BESTest and RT (rho=−0.246, p=0.05).

Significance:

We provide preliminary but weak evidence supporting the relationship between WBB derived variables in relevance to the BBS and Mini-BESTest. Further research is needed to fully understand the clinical utility of the WBB especially in a larger sample and to generalize these results to stroke survivors at all levels of ability.

Introduction

The Wii Balance Board (WBB), an accessory used for the Wii™ video game consoles (Nintendo, Kyoto, Japan), has been widely investigated as a balance testing platform [1,2]. Laboratory grade force platforms have high reliability and validity to detect subtle center of pressure (COP) changes, are typically considered as gold standard for measuring balance and have been extensively used in balance research [3,4]. However, force platforms are not clinically accessible as they are expensive ($1000-$40k), bulky (weighs ~18 to 150 lbs) and require an initial elaborate setup (multiple embedded force plates with each plate dimension ranging from ~15x23x2 inches to 35x35x6 inches). The WBB is a viable and affordable alternative for assessing balance with precise measurements of sway. The WBB is light (weighs ~7.72lbs), compact (19.69x7.87x19.69 inches), battery operated and consists of four force transducers at each corner of the platform. The WBB is globally accessible due to Nintendo’s video gaming market and is more cost-efficient (under $100) as compared to a force platform. It is Bluetooth enabled making it a portable wireless tool. The WBB can provide anteroposterior and mediolateral sway path measures, quantify CoP displacement and provide estimates of weight symmetry. A systematic review of 21 studies found the WBB to have moderate to excellent reliability and excellent concurrent validity in comparison to the same outcome measures collected on a lab force platform [2]. In addition, static standing balance measured with the WBB has shown to be related to gait velocity but not fall risk [10,11].

Although the WBB has been extensively compared to the force platform and to some gait measures, there is still need to clarify its relationship with clinical tests of balance. The Berg Balance Scale and the Mini-BESTest are commonly used psychometric tests of balance in the stroke population [5,6]. They have shown high inter-rater and intra-rater reliability for individuals with impaired balance [7,8,9]. No studies have examined the association between the WBB measures and clinical tests of balance in individuals with stroke. Understanding the extent of association with commonly used balance scales will provide a more comprehensive picture of the utility of the WBB as a clinical balance assessment tool. Hence, the objective of this study was to examine the relationship between static and dynamic COP measurements derived from the WBB with clinical scores from the BBS and Mini-BESTest in individuals with chronic stroke.

Methods

Study Overview

This study is a secondary data analyses of existing data collected as a part of a clinical trial in the Brain Plasticity Lab at the University of Illinois at Chicago, IL (Clinical trial registration: NCT03492229). The WBB and clinical measures were collected by a trained physical therapist as a part of baseline assessments.

Participants

Sixty-nine individuals with chronic stroke (46 males and 23 females, mean age 58.49 ±9.3 years) with residual hemiparetic gait deficits who underwent balance and WBB assessments were included in this study. The inclusion criteria for the RCT included individuals who had a diagnosis of non-cerebellar stroke with residual hemiparetic gait deficits and the ability to stand with or without assistive device for at least 10 minutes. Individuals with severe neurological, cardiopulmonary, orthopedic or medical conditions, significant cognitive (Mini-Mental State Examination score less than 24) or communication impairments and vision problems were excluded. Written signed informed consents were provided by each subject for participation in the larger clinical trial after they demonstrated a complete understanding of the study protocol. The study was approved by the University of Illinois Institutional Review Board.

Clinical Assessments:

Berg Balance Scale (BBS)

The BBS is a static and dynamic balance test consisting of 14 functional items rated on a 4-point ordinal scale of 0 (lowest functional level) – 4 (highest functional level) with a maximum score of 56 [12]. The sub-components of BBS assess balance in various positions used in everyday life (standing, sitting, turning etc.) challenging participants to maintain postures of varying difficulty. It has a high inter-rater (ICC=0.97) and a high intra-rater (ICC=0.98) reliability [7]. Participant were divided into two groups based on their score, moderate (21–40, n = 4) and high (41–56, n = 55), for secondary analyses.

Mini-BESTest

The Mini-BESTest is a balance test targeting six different dynamic balance control systems of the human body [13]. It has 14 test items with a 2-point score range of 0 (lowest- unable to perform) to 2 (highest- normal performance) with a maximum achievable score of 28. The various components of the Mini-BESTest assess anticipatory transitions, dynamic gait stability, sensory orientation and reactive postural control. It also has a high inter-rater (ICC=0.98) and a high intra-rater (ICC=0.98) reliability [8]. It has an excellent concurrent validity with the BBS [14]. Participants were divided into two groups based on their score, moderate (≤17.5, n = 29) and high (> 17.5, n = 40) for secondary analyses.

Wii Balance Board

The Wii Balance Board (WBB) was interfaced with a customized WeHab software (WeHab, University of Notre Dame, IN http://sites.nd.edu/aaron-striegel/2020/01/26/wehab-code-downloads/) to record mean COP data during quiet standing and weight shifting [15,16] The WBB has shown to have good reliability and validity to assess standing balance (ICC 0.66–0.94)[17,18].

The following three tasks were performed to assess static and dynamic balance:

1. Body Weight Symmetry (BWS): Lateral body weight distribution during standing with eyes open and closed was measured. Each trial consisted of quiet standing for a maximum of 30 seconds. Two trials each were conducted for the eyes open (EO) and eyes closed (EC) conditions. Body weight distribution on the paretic and non-paretic limb was computed as the ratio of the absolute value of the average COP position of the paretic limb over the nonparetic limb. A value of 1 indicates symmetrical weight distribution over both legs, whereas a value <1 indicates greater weight distribution on the non-paretic side and value of >1 indicates greater weight distribution on the paretic side [19,20].

2. Dynamic Stability: A dynamic weight shifting task was used to test the participant’s ability to shift their COP in various target directions without losing balance (Figure 1). As each target appeared on the screen, the participants were instructed to move their COP as quickly as possible towards the target and maintain their COP for 5 secs within each target. Reaction time (RT) was calculated as the time in seconds between the appearance of the target and the first movement made by the participant (i.e., when COP deviated more than five standard deviations from the mean). Completion time (CT), total time to complete the task, was calculated as the time in seconds from the appearance of the target until the participants successfully placed their COP in the target for a continuous 5 secs. The targets appeared in a random order and two practice trials were provided before testing [19,20].

Figure 1.

Visual schematic of the Wii Balance Board’s dynamic weight shifting test. The dark gray circles depict the target regions. The small light grey circle depicts the participant’s center of pressure (COP). The participant was instructed to place his/her COP within one of the 5 targets in the order of its appearance (as shown by the numbers 1–8). Each target circle stayed until the patient was able to place his/her COP within the circle for 5 s continuously. Reaction time (RT), was the time between the appearance of the target and the patient’s first COP movement, and the Completion Time (CT), was the time from the appearance of the cursor until the participant successfully placed the cursor in the target for continuous 5 s.

Statistical Analyses

Statistical analyses were performed using Windows SPSS 24.0 (SPSS Inc, Chicago, IL) with significance level of 0.05 for all tests. Normality of all variables was examined using the Shapiro-Wilk’s test of normality. As none of the variables were normally distributed, Spearman’s rho correlation analyses were conducted to examine the strength of the relationship between the clinical tests and the WBB variables (clinical scores vs BWS-EO and BWS-EC ratios, BBS vs RT, BBS vs CT, MBT vs RT, MBT vs CT). The strength of correlation coefficient was defined as negligible (below 0.3), low (0.3 to 0.5), moderate (0.5 to 0.7) and high (0.7 and above) [21]. Mann-Whitney U tests were done to compare BWS-EO, BWS-EC and WBB variables between individuals who were categorized as moderate and high balance based on the Mini-BESTest scores. Because of the uneven sample size in each group of the BBS, we did not perform further analyses. We performed k-means clustering analyses to examine for trends within the WBB data. k-means clustering is an exploratory non-hierarchical partitioning method that divides data observations into k pre-defined mutually exclusive clusters based on their similar characteristics within each cluster, while also keeping clusters far apart as possible. After the primary analysis, we decided to accept the 2-cluster solution as this resulted in a more distinct clusters compared to the 3 or 4 cluster solutions. Post-hoc analyses via Mann Whitney U tests were done to see whether there was a difference in clinical scores between the two clusters.

Results

We included data from 69 participants for the clinical tests and BWS measures. For the RT and CT measures, a sample of 59 was used because of missing data due to technical problems with the WeHab software during data collection. Demographic data are presented in Table 1.

Table 1.

Participant characteristics. Values are in Median (Interquartile Range). BBS: Berg Balance Score.

Demographics, stroke (n=69)	Median (IQR)
Age (years)	56 (52–67)
Gender (male/female), n	46/23
Time since stroke (months)	8 (6–13)
Hemiplegic side(left/right), n	41/28
Leg dominance (left/right), n	13/56
BBS (/56)	49 (46–53)
Mini-BESTest (/28)	19 (15–22)
Reaction Time (seconds)	1.42 (1.20–2.19)
Completion Time (seconds)	7.25 (6.43–11.91)

Static COP Distribution

We observed a trend towards greater weight bearing on the non-paretic side for both the EO and EC conditions, which was not statistically significant. No significant difference was found between BWS between EO and EC conditions. BWS was not significantly correlated with the clinical scores (see Supplementary material for scatter plots). Only BWS-EO was significantly different between the two groups of Mini-BESTest (U = 456, P = 0.03), with those with moderate balance showing lesser symmetry than those with high balance (see Table in Supplementary material).

Dynamic COP

Correlations between WBB variables and clinical tests are shown in Figure 2. A significant but negligible negative linear correlation was observed between BBS and CT (rho = −0.29, P = 0.021) and between Mini-BESTest and RT (rho = −0.246, P = 0.05) (Figure 2). No other correlations were statistically significant.

Figure 2:

Scatter plots showing the relationship between clinical tests and WBB derived temporal variables. Each scatter plot also indicates the two clusters identified using the k-means cluster analyses with cluster 1 represented with black circles and cluster 2 with grey circles. Note the uneven size of the two clusters. Correlations between Berg Balance Score and Completion Time (rho = −0.296, P = 0.021) and Mini-BESTest and Reaction Time (rho = −0.246, P =0.05) were significant. * indicates P < 0.05.

Cluster Analyses

The k-means clustering divided our data into two uneven non-overlapping groups based on similarity in WBB variables. Table 2 summarizes the cluster data for all the WBB derived variables. A k-means cluster analysis for BWS-EO revealed two clusters: cluster 1 (n=59) aggregated those with greater weight on the non-paretic limb side and cluster 2 (n=10) included those with greater weight on the paretic side. Similarly, BWS-EC showed two clusters: cluster 1(n=57) with greater weight on the non-paretic limb side and cluster 2 (n=12) with greater weight on the paretic side. Post hoc tests did not reveal a significant difference between the two clusters for the clinical scores. K-means cluster analysis for RT divided participants into cluster 1 (n=11) including those with higher reaction time and cluster 2 (n=48) with a lower reaction time (Figure 2). Similarly, the CT clusters were divided with cluster 1 (n=8) having participants with higher completion time and cluster 2 (n=51) with a lower completion time (Figure 2). After identification of the two clusters, post hoc tests did not reveal a significant difference for the clinical scores between the two identified clusters.

Table 2.

K-means cluster analysis for Body Weight Symmetry ratios and Wii Balance Board derived temporal variables. No significant differences were noted between the clusters for any of the clinical variables. BWS-EO: Body Weight Symmetry in Eyes Open condition, BWS-EC: Body Weight Symmetry in Eyes Closed condition, BBS: Berg Balance Score.

Variable	Clusters	No. of	Cluster center/	Mini-BESTest	BBS
		participants	Median	Median (IQR)	Median (IQT)
BWS-EO	1	59	0.90	20(16–22)	50(46–53)
	2	10	1.33	16(12.5–23)	50(43–55)
BWS-EC	1	57	0.94	20(15–22)	49(45–53)
	2	12	0.83	22(15.25–23)	52.5(47.25–55)
Reaction Time (s)	1	11	3.82	16(13–21)	47 (42–54)
	2	48	1.32	20(317–22)	50(47–53)
Completion Time (s)	1	8	20.5	16 (12.25–19.5)	44.5(42.5–46.75)
	2	51	6.95	20(16–22)	51(47–54)

Discussion

The purpose of this study was to explore the relationship between static and dynamic balance measured with the WBB in relationship to clinical balance assessments in post stroke individuals. We found that body weight symmetry measured with the WBB was different between the individuals with moderate and high balance, categorized using scores on the Mini-BESTest. Time to complete the target matching task on the WBB was significantly correlated to the BBS, and WBB derived reaction time was correlated to the Mini-BESTest; but with both relationships showing negligible correlations. These results suggest that there is a potential for the WBB to capture differential aspects of balance that are measured clinically, however a comprehensive picture may not be elicited with the just the WBB.

Several studies have used COP data from the WBB for assessment of balance in the elderly and individuals with stroke and demonstrated that the WBB can effectively capture changes in COP variability with excellent reliability [22–24]. Asymmetry in body weight distribution is common after stroke, with stroke survivors bearing more weight on their less affected side compared to the more affected. Our data support these results, while providing new information that BWS measures may be sensitive to identifying those with impaired balance. Individuals with poor balance (scores less than 17.5 on the Mini-BESTest) demonstrated greater weight bearing on the less affected side. It is possible that the WBB can be efficiently used to capture impairments in balance as the first line of evidence as an alternative to performing a comprehensive balance assessment. However, it should be noted that in a study performed by Bower et al. (2019) static balance measured with the WBB was not a significant predictor of falls following stroke compared to other dynamic tests of balance such as the Timed Up and Go which better identified fallers from non-fallers [11]. Therefore, the clinical utility of the WBB as a diagnostic tool for fall risk may be currently limited.

An interesting finding was the differential (albeit negligible) association of the two components of the dynamic tasks to the clinical balance tests. Those with better balance on the BBS were able to complete the WBB task in a lesser time. While those with better balance on the Mini-BESTest responded faster during the dynamic tasks. This is not a surprising result as the discriminative ability of the BBS and the Mini-BESTest are shown to be different. As the Mini-BESTest involves more dynamic tasks such as the single leg standing task, it has been shown to discriminate stroke survivors better on walking speed classifications than the BBS [12]. Again, these results suggest that the dynamic assessment using the WBB may weakly capture components of balance and postural adjustments that the clinical tests of balance measure.

In an attempt to further explore the distribution of WBB derived data and to estimate cut-off scores, we performed cluster analyses for the static and dynamic WBB measures. We found that participants could be divided into two distinct groups based on their performance on the WBB tasks, however there was no significant difference in the clinical scores between the two statistically derived clusters. A potential reason for this could be due to the uneven number of participants in the two clusters, possibly due to a predominance of those with higher functional recovery that our cohort of participants included. Given this limitation, we are unable to further extrapolate on the significance of the two identified clusters. Having a larger representation of balance impairments within the subject sample may help establish further quantitative guidelines for identifying subgroups of participants when using the WBB.

We provide preliminary information exploring the use of the WBB for clinical use and the possibility of using it as a clinical balance assessment tool. The WBB is portable, simple to use, affordable, reliable and compatible with customizable software such as WeHab and LabVIEW. This makes it an excellent first-line assessment tool to be used in the clinical or home setting for clinicians aiming to screen for balance assessments before performing a comprehensive balance assessment. In recent years, numerous studies have established the reliability and validity of COP measures using the WBB, further supporting potential clinical implications. However, our results do not fully support the utility of WBB derived variables as replicative of clinical balance assessments. Our data provides some evidence that WBB derived variables are related to clinical scores of balance, however these relationships were negligible and our results need to be interpreted with caution. Future studies could investigate a larger heterogenous sample of individuals with stroke to elucidate the clinical utility and generalizability of the WBB. We studied the overall displacement of WBB derived COP; delineating the antero-posterior and mediolateral sway components of body sway may provide better insight into the relationships being studied. Currently, newer models of WBB have been discontinued by Nintendo but older versions of the WBB are still available to purchase through various ecommerce sites, a caveat clinicians will need to consider before clinical translation of the WBB.

Conclusion

In conclusion, we provide preliminary but weak evidence supporting the clinical utility of static and dynamic balance measured with the WBB in relevance to the BBS and Mini-BESTest. Further research is needed to fully understand the clinical utility of the WBB especially to generalize these results to stroke survivors at all levels of ability.

Highlights.

• Clinical utility of the Wii balance board (WBB) was examined

• Time to complete a task with the WBB was related to the Berg Balance Score

• Time to react to a target with the WBB was related to the Mini-BESTest

• WBB maybe a useful addition to routine clinical assessments for stroke