Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis

Abstract

Background

The congruence or discordance between actual and perceived balance ability has been proposed to be linked to functional outcomes such as falls. However, gaps remain in our ability to quantify discordance, and its relationship to relevant outcomes.

Objective

To investigate a novel quantification of concordance/discordance between balance performance and perception and determine the relationship to falls among people with Parkinson’s disease (PwPD).

Methods

Data from 244 PwPD were aggregated from 5 previously conducted studies. Variables extracted included age, sex, Activities-Specific Balance Confidence scale (ABC; perceived balance), Timed Up and Go (TUG; balance performance), Movement Disorder Society Unified Parkinson Disease Rating Scale (MDS-UPDRS) part III scores, and retrospective falls (6- or 12-month). Data validation between studies was established. Discordance was quantified as the difference between an individual’s predicted ABC, based upon their TUG score, to their reported ABC.

Results

Two methods for calculating discordance were tested: simple linear regression and segmented regression. As there were no differences between the bootstrap distributions of both approaches (P = .520), simple linear regression was utilized for the subsequent logistic regression model. Discordance was the only statistically significant predictor of fall status (OR = 0.98, P = .003), after controlling for age, MDS-UPDRS part III, sex, and TUG. The inclusion of discordance in the logistic regression model boosted the predictive accuracy by 58%.

Conclusions

Discordance between actual and perceived balance was uniquely related to retrospective fall history among PwPD. Clinicians and researchers should consider discordance between actual and perceived balance as a potentially modifiable target to minimize falls.

Keywords

falls mobility perception balance confidence fear of falling Parkinson

Introduction

Actual and perceived balance may interact to facilitate one’s movement, activities, and participation.^1
-4 Recent work suggests that the nature of this relationship may inform how people choose to engage with the tasks and environments they encounter.^1
-3,5
-10 Theoretically, concordance of balance performance and perceived balance or confidence may facilitate an optimal trade-off between safety and engagement in physical activity and community participation.^1,2,11
-13 This could be observed as a continuum where relatively high physical abilities corresponds to high levels of perceived balance or movement confidence, continuing to low physical ability and low perceived balance or movement confidence.^2,13 While concordance is thought to be related to optimal outcomes at the level of the individual, discordance between actual and perceived balance may result in negative outcomes including activity limitation, participation restriction, and falls.^{2,3,5,12,14
-16} This discordance can be observed in at least 2 circumstances (1) having a high balance confidence with poor actual balance ability (“overconfident condition”), and (2) having low balance confidence despite good actual balance ability (“underconfident condition”). The manifestation of concordance or discordance is likely driven by meta-cognitive processes that occur as individuals weigh the perceived demands of the task relative to their perceptions of their own capability.

Previous work has identified these potential conditions of balance discordance, and proposed that they may be predictive of or relevant to functional outcomes such as falls.^2,13 This may manifest as an underconfident pattern in which individuals modify their balance behaviors due to excessive anxiety, fear of falling, or perceived limitations of their balance. Conversely, individuals may demonstrate limited insight into their balance limitations and present with an overconfident pattern in which they fail to appropriately limit their balance behaviors. Both of these manifestations of discordance hold the potential to increase falls.^2,13 However, previously proposed quantification techniques rely on categorization of individuals into bins (over, under, or appropriately confident).^2,13 In reality, the degree of discordance is continuous, and quantifying this outcome as such improves the measurement precision of the construct and, subsequently, our ability to understand and evaluate its relevance and treatment approaches to address it. Further, whether discordance between actual and perceived balance is relevant for functional outcomes, such as falls is poorly understood.^9,10,17
-24

Parkinson’s disease (PD) is a neurodegenerative disorder that results in balance impairments and reduced perceived balance confidence.^{9,10,17,25,26} Falls are a hallmark of PD and are accompanied by progressive reduction in physical activity and community participation as the disease advances.^23,27,28 Due to these factors, PD is an optimal population in which to further investigate the concept of discordance. While current models provide information on the likelihood of a future falls among individuals with PD (eg, age and fall history),^9,10,17
-24 there is less information regarding potentially modifiable physical, psychological, and cognitive characteristics (such as discordance) that relate to falls.^27,29 Identifying modifiable characteristics related to falls and quality of life would provide a deeper understanding of possible contributing factors for falls and could inform treatment strategies targeting fall prevention in these groups.^27,30

In sum, although the concept of discordance between actual and perceived balance has been described previously, gaps remain in (1) our ability to quantify discordance, and (2) its relationship to relevant outcomes such as falls.² Therefore, the purpose of this study was to propose a novel and continuous quantification of concordance and discordance between balance performance and perceived balance, and use this metric to characterize discordance in people with PD. Additionally, this study aimed to determine the relationship between discordance and falls in a fall-prone neurological population (PD). It was anticipated that degree of discordance would be related to a history of falls in individuals with PD.

Methods

Design

This study employed a coordinated analysis, which is a method of harmonizing raw data from individual studies into a single parent data which is then used to perform an integrative analysis to test a specific hypothesis.^31,32 This is a semi-standardized approach, developed to facilitate efficient analysis across multiple data sets. The current study utilized data from 5 previously conducted studies across 4 institutions: Arizona State University (ASU) which conducted 2 of the included studies (ASU1 and ASU2), Saint Louis University (SLU), Oregon Health & Science University (OHSU), and The University of Nevada, Las Vegas (UNLV).^33
-38 Among the 5 studies, data of 244 persons with PD were extracted and compiled to be used as the analytic sample in this secondary analysis. Inclusion criteria for this study were: (1) diagnosis of idiopathic PD and (2) having data for measures included in the primary analyses (Activities-Specific Balance Confidence scale [ABC], Timed Up and Go [TUG], age, sex, and Movement Disorder Society revised Unified Parkinson Disease Rating Scale [MDS-UPDRS] part III scores). Exclusion criteria were: (1) presence of other neurologic disease or injury (ie, cerebrovascular accident, multiple sclerosis, and Alzheimer’s disease), (2) recent (within the prior 6 months) major orthopedic injury or surgery (ie, total joint replacement), and (3) orthopedic conditions that affect the participants walking or balance.

Measures

Data extracted included information regarding demographic characteristics (age, sex), PD severity (MDS-UPDRS part III, Hoehn and Yahr), falls experienced in the prior 6 months or year, physical function ability (TUG), and perceived physical function (ABC).

PD Severity

The MDS-UPDRS part III, Hoehn and Yahr, were used to characterized PD severity. The MDS-UPDRS part III is a motor examination that consists of 18 items with higher scores indicating more severe PD-related motor impairment.³⁹ The test-retest reliability of the MDS-UPDRS part III is high with an intra-class correlation coefficient (ICC) of .89.⁴⁰

Falls

Data on falls was obtained by asking participants to recall the number of falls they had experienced in the previous 6 months (OHSU and both ASU data sets) or year (UNLV and SLU data sets; see Supplemental Material 1). Falls were defined as any unintentional lowering to the ground. From the falls data a fall status variable was created with zero falls indicative of non-faller status and 1 or more falls indicative of faller status.

Balance Performance

The TUG is a test of balance and functional mobility and was used to quantify balance performance during the performance of basic physical functions. Individuals were asked to stand, walk, turn around and return to the chair where they began the test, with shorter times indicating better balance and functional mobility. The TUG is a commonly used clinical measure and has excellent reliability among individuals with PD (ICC = .85).⁴⁰ Among the 5 studies, this test was performed with some variability, notably 2 studies (SLU and UNLV) instructed participants to walk quickly where the other 3 studies (ASU1, ASU2, and OHSU) instructed participants to walk at a comfortable pace. Additionally, while 4 studies (ASU1, ASU2, SLU, and UNLV) utilized the standard walking distance of 3 m, 1 study (OHSU) utilized an extended distance of 7 m. To address these inconsistencies in administration of the TUG, data from each study were normalized to each sample’s relative mean and standard deviation prior, that is, each sample’s TUG were centered to their mean and divided by their standard deviation prior to being integrated with the data from the other studies. Additionally, given the variability in TUG procedures the data were reviewed to determine variations in the relationship between TUG and ABC, and ensure the data were consistent among each study for use in the coordinated analysis (see 2.3.1 and 3.2). Of note, the TUG was completed with participants ON PD medications in 4 studies, however, in 1 study (OHSU) participants completed the TUG in the OFF PD medication state, following a 12 hour or withdrawal of their PD medications.

Perceived Balance

The ABC is a 16-item self-report measure of confidence with balance and mobility and was used to describe individuals’ perceptions of their balance. Individuals are asked to rate each item from 0 (no confidence) to 100 (complete confidence). The ABC is commonly used clinically and it has excellent reliability among individuals with PD (ICC = .96).⁴¹

Statistical Analysis

The 2 primary goals of this study were to (1) propose a novel and continuous quantification of actual and perceived balance discordance, and (2) determine whether discordance was related to falls. A 3-phase statistical approach was taken to achieve these aims, including (1) validating the use of data across studies, (2) estimating TUG discordance with linear and non-linear formulations, and (3) relating discordance to falls. A vignette including all the code that generated the results and visualizations presented in this paper is included in Supplemental Material 2.

Validation of Between Site TUG to ABC Relationship

The first step in this coordinated analysis, given that the TUG was administered in a variety of ways across sites (see Balance Performance Section for details), was to establish that the relationship between TUG and ABC between studies was consistent. To validly combine data across sites we must ensure that variation in TUG administration does not lead to variation in the general TUG to ABC relationship, that is, presence of a TUG administration to ABC interaction whereby the estimated slope between TUG and ABC changes as a function of the type of TUG administered. To establish that the TUG to ABC relationship was consistent between studies, multivariable linear regression was performed with raw ABC as the dependent variable, normalized TUG by Group interaction as the primary independent variable, and control variables of age, sex (male/female), and MDS-UPDRS part III score. To establish that there was no significant difference of the TUG to ABC relationship across studies, a pairwise comparison of the linear trends was performed using a Tukey method of P-value adjustment for multiple comparisons. If no statistically significant differences between the TUG to ABC relationship between any sites was observed, combination of normalized TUG across sites was unlikely to impact interpretability of results when combining TUG scores.

Coordinated Analysis to Estimate ABC to TUG Discordance with Linear and Non-Linear Formulations

Within this coordinated analysis, our estimation of individual discordance was quantified as the difference between an individual’s TUG-predicted ABC (ie, the ABC value predicted by the fitted TUG coefficient while controlling for age, sex, and MDS-UPDRS part III scores) to their actual ABC. For this measure of discordance, a positive value would suggest perceived confidence is greater than predicted confidence, based on TUG, (“over-confident”), and a negative value would suggest lower than predicted confidence (“under-confident”, see Figure 1 for a conceptual plot of the measure). However, it is unclear if the relationship between TUG and ABC is linear or non-linear (ie, if a constant change in TUG leads to a constant change in ABC). Understanding the linearity of this relationship would inform which type of model to calculate discordance would best relate to individual falls. Therefore, we chose a segmented regression model with optimal knot placement (ie, where the change in regression slope will occur dependent on the value of the primary independent variable (TUG), must first be determined). To determine optimal knot placement, we iteratively fit segmented regression models with incrementally different knot values across the total TUG range and recorded their AIC value to determine the knot of best fit. The identified optimal knot would then be carried forward to the model comparison stage.

Figure 1.

Conceptual graph representing the calculation of discordance between a measure of capability (x-axis) to a measure of confidence (y-axis). The black line represents the predicted value of confidence given a specific value of capability. Individual discordance would be calculated as the residual difference between the actual value of confidence versus the predicted value of confidence. The arrows that connect the black fitted line to the actual value of confidence represent the varying magnitude of discordance given this approach, that is longer arrows represent greater discordance than short arrows. Furthermore, arrows that are coded as red represent negative discordance (individuals with perceived confidence greater than their predicted confidence based on their balance performance) and blue arrows represent positive discordance (individuals with perceived confidence less than their predicted confidence based on their balance performance).

Model Selection to Identify Method of Discordance Calculation

To determine which approach (simple linear vs segmented regression) would provide the strongest association for individual falls, a bootstrapping approach for each model was performed. Specifically, the entire dataset was iteratively resampled 1000 times to generate random training and test sets to first generate the TUG-specific discordance metric using the training set (ie, either linear or segmented regression model with ABC score as the dependent variable, TUG as the primary dependent variable, and age, sex, and MDS-UPDRS part III score as control variables). Then, using the fitted model from the training set estimated TUG-specific discordance (predicted ABC – actual ABC) on the test set was calculated and then used as the primary independent variable in a logistic regression model with Falls (Faller vs non-Faller) as the dependent variable and age, sex, and MDS-UPDRS part III scores as control variables. For each bootstrap iteration, a predicted versus actual confusion matrix for falls was generated and the area under the curve for each model was stored for each model type (linear regression discordance AUC versus segmented regression discordance AUC). This bootstrapping approach allows for a distribution of fall predicted accuracy between discordance type to be compared using an independent t-test. Results of the t-test would allow us to infer which model for deriving discordance (linear vs segmented) had the better association with falls.

Analysis of Discordance to Falls in PD

If there was no significant difference between the discordance metrics, then the simple linear model of discordance would be used as the simpler of the 2 models. Once the discordance metric of best fit was determined, then logistic regression on the entire dataset would be used to examine the relationship between the validated measure of discordance to individual falls (Faller vs non-Faller) with discordance as the primary independent variable and control variables of age, sex, and MDS-UPDRS part III score.

Results

Participant Characteristics

A total of 244 people with PD were used in this analysis across 5 studies with a mean age of 69.5 ± 8.1 years, with 103 (42.2%) identifying as female. The average ABC score among all participants was 77.6 ± 19.0, with an average MDS-UPDRS part III score of 35.9 ± 14.3, and 97 (40%) identified as having a recent fall (Table 1). Between sites there were significant differences in ABC score (P < .001), distribution of sex (P < .001), MDS-UPDRS part III score (P < .001), and number of people with a recent fall (P = .04; Table 2). Between group differences in TUG by study could not be calculated given the difference in TUG method used by site. Overall, inclusion of each study demonstrates the variation in balance confidence, gender distribution, disease severity, and faller status across sites.

Table 1.

Descriptive Characteristics for the Combined Sample.

Characteristic	N = 244
ABC (scale points)	81.00 [12.00-100.00] IQR: 26.93
TUG (seconds)	17.89 (8.74)
Group
ASU1	30 (12%)
ASU2	16 (6.6%)
OHSU	130 (53%)
SLU	31 (13%)
UNLV	37 (15%)
Number of falls	0.00 [0.00-30.00] IQR: 1.00
Age (years)	69.51 (8.10)
Gender
Female	103 (42%)
Male	141 (58%)
UPDRS part III (scale points)	36.00 [3.00-82.00] IQR: 18.00
Hoehn & Yahr (scale points)	2.00 [0.00-4.00] IQR: 0.00
Fallers	97 (40%)

Note. Median [Min–Max] IQR: interquartile range; Mean (SD); n (%).

Table 2.

Descriptive Characteristics by Data Set.

Characteristic	ASU1	ASU2	OHSU	SLU	UNLV	P-value
Characteristic	N = 30	N = 16	N = 130	N = 31	N = 37	P-value
ABC (scale points)	75.00 [18.38-91.56] IQR: 14.06	91.25 [37.50-98.75] IQR: 14.64	84.00 [25.00-100.00] IQR: 23.00	94.38 [25.68-100.00] IQR: 18.38	66.30 [12.00-100.00] IQR: 38.50	<.001
TUG (seconds)	13.71 (4.12)	10.39 (1.95)	23.32 (7.61)	8.03 (2.89)	13.66 (6.17)	<.001
Number of falls	1.00 [0.00-20.00] IQR: 2.75	0.00 [0.00-30.00] IQR: 0.00	0.00 [0.00-11.00] IQR: 1.00	0.00 [0.00-6.00] IQR: 1.00	0.00 [0.00-30.00] IQR: 1.00	.056
Age (years)	70.71 (7.18)	71.68 (5.92)	68.81 (7.94)	67.71 (7.94)	71.57 (9.79)	.2
Gender						<.001
Female	10 (33%)	5 (31%)	46 (35%)	14 (45%)	28 (76%)
Male	20 (67%)	11 (69%)	84 (65%)	17 (55%)	9 (24%)
UPDRS part III (scale points)	30.50 [9.00-56.00]IQR: 18.00	42.00 [27.00-68.00]IQR: 11.00	40.00 [13.00-82.00]IQR: 15.00	24.00 [6.00-71.00]IQR: 15.00	31.00 [3.00-72.00]IQR: 24.00	<.001
Hoehn and Yahr (scale points)	2.00 [1.00-3.00]IQR: 1.00	2.00 [2.00-4.00]IQR: 1.00	2.00 [1.00-4.00]IQR: 0.00	1.00 [1.00-3.00]IQR: 1.00	2.50 [1.00-3.00]IQR: 2.00	<.001
Fallers	18 (60%)	3 (19%)	51 (39%)	14 (45%)	11 (30%)	.040

Note. Median [Min–Max] IQR: interquartile range; Mean (SD); n (%). One-way analysis of means (not assuming equal variances); Kruskal–Wallis rank sum test; Pearson’s Chi-squared test.

Validation of Between Study TUG to ABC Relationship

Results that compared significant deviation in the TUG to ABC relationship by group were non-significant (adjusted P > .05 for all group-by-TUG pairwise comparisons, Figure 2) which suggests that, when normalized, there is no difference in the relationship between the different type of TUG variation to individual ABC score when controlling for individual age, sex, and MDS-UPDRS part III score.

Figure 2.

A comparison of the distribution and relationship between TUG when normalized to the individual site mean both when (A) combined and (B) stratified by group. A represents the combined relationship between normalized TUG to the raw ABC scores with the distribution of each measure on the x and y axes, respectively. B represents the individual site relationships between normalized TUG and raw ABC scores with the individual distributions for each sites TUG and ABC scores on the x and y axes respectively.

Selection of Discordance Measure

The optimal knot for our segmented regression approach based on knot placement that yielded the lowest AIC was 1.469 (Supplemental Figure 1, segmented model fit to complete dataset can be seen in Figure 3B). The segmented regression analysis then used in the bootstrapping approach used this knot. Comparison between bootstrap distributions (simple linear vs segmented regression) of AUC values revealed no significant difference (P = .52, mean linear = 0.692 ± = 0.035, mean segmented = 0.691 ± −0.035, Supplemental Figure 2). Therefore, the value of discordance derived from the simple linear regression model was used (simple linear regression model fit to complete dataset can be seen in Figure 3A).

Figure 3.

Visual representation of the calculated discordance between the different regression methods. (A) Simple Linear Regression. (B) Segmented Regression with an optimal knot at 1.47. Arrows that are coded blue represent positive discordance values that equate to individuals with less perceived confidence than their model predicted confidence. Arrows that are coded red represent negative discordance values that equate to individuals with greater perceived confidence than their model predicted confidence.

Relationship to Discordance to Faller Status

Results of the logistic regression model with faller status as the dependent variable (Faller vs non-Faller) demonstrated that discordance was the only statistically significant measure (OR = 0.98, 95% confidence interval [CI] [0.96, 0.99], P = .003, Figure 4). This result indicates that a 1 unit increase in discordance (less confidence per ability) is associated with a 2% increase in probability of being classified as a faller. Neither TUG (OR = 1.34, 95% CI [0.98, 1.86], P = .07), Age (OR = 1.01, 95% CI [0.98, 1.05], P = .52), MDS-UPDRS part III score (OR = 1.01, 95% CI [0.99, 1.3], P = .25), or sex (OR_male = 1.22, 95% CI [0.70, 2.16], P = .48) were associated with faller status. The overall model AUC was .68, with a sensitivity (prediction of faller) of 35%, a specificity (identification of non-faller) of 82%, and an overall accuracy of 63%. Within the context that the no-information criterion is 40% (ie, the predictive accuracy of a naïve classifier (Faller ~ 1)) the inclusion of discordance in the logistic regression model boosted the predictive accuracy by 58% (63/40).

Figure 4.

The binomial relationship between probability of being a faller or a non-faller to magnitude of simple linear regression derived discordance. Density plots on the superior x-axis visualize the distribution of the fall types given the value of discordance with greater confidence (higher perceived balance confidence compared to predicted balance confidence) associated with a greater probability of being a non-faller. The horizontal dashed line represents the no-information criterion which is equal to 40% given that only 97 out of 244 participants in the study were classified as Fallers.

Discussion

It is well established that deficits in actual and perceived balance relate to negative outcomes such as falls.^17,42
-44 Results of the current project extend this work, showing that the accuracy of individual’s perceived balance (ie, discordance between actual and perceived ability) also relates to fall risk. Specifically, using a novel, continuous quantification of discordance between actual and perceived balance ability, we show that under-confidence is related to retrospective falls, even after accounting for balance performance alone. Although additional work will be necessary to understand the directionality of this relationship, this work highlights an important and under-studied aspect of falls and fall prevention.

Previous work has introduced the idea of discordance between actual and perceived balance, and its potential importance.^2,13 However, while this work provided important theoretical groundwork, methods proposed to quantify discordance in a sensitive and quantitative way were limited. Landers and Nilsson² and Delbaere et al¹³ both suggested categorization of individuals as inappropriate or maladaptive (discordantly over or under confident) or appropriately adaptive (concordant). However, this approach loses the continuous nature of the relationship between perceived and actual ability and is not sensitive to subtle changes in concordance. Therefore, a goal of the current analysis was to propose a quantitative approach to characterize this construct. To achieve this, we first needed to establish a line of best fit relating perceived with actual balance, or a concordance line. Theoretically, an individual at any point on this line would have an appropriate weight between perceived and actual balance and, thus, would be concordant. The results of this study indicate that a simple linear relationship best fit the data, rather than alternative non-linear approaches. Discordance was then operationalized as the vertical distance from the concordance line or the difference in ABC from predicted ABC based on the concordance line. This relies on the assumption that there is a relatively similar number and degree of individuals that manifest as under- and over-confident and that the individuals closest to the fitted line exhibit a concordant relationship. It is possible that in certain populations, including PD, other factors (eg, disease severity, psychological contributors, impaired cognition) or measurement tool limitations (eg, floor and ceiling effects) may shift the mean away from “true” concordance. However, this study successfully developed a quantitative method for describing discordance and showed that discordance as uniquely related to falls, providing initial evidence for the construct validity of this measurement approach.

The second goal of this study was to investigate the relationship of discordance to a functionally relevant outcome such as falls. We observed that discordance uniquely and more strongly related to fall status (faller vs non-faller) than other included factors, including MDS-UPDRS, ABC, TUG, age, or sex. Additionally, inclusion of discordance substantially improved the predictive accuracy of the model to correctly classify participants according to fall status. This illustrates the potential of quantifying and considering the interplay between perceived and actual balance abilities. This study found that as discordance increased (less confidence per ability) the odds of being classified as a faller also increased. In other words, participants who were under confident were more likely to be classified as fallers. This finding is somewhat surprising, given previous notions that over confident individuals may be a greatest risk for falls given their lack of activity curtailment to match their diminished balance abilities or their lack of insight to their balance deficits.² However, this finding is consistent with those of Delbaere et al¹³ who found that among older adults with similarly elevated physiologic fall risk, those with lower perceived fall risk were less likely to experience a fall. While the factors that contribute to and help provide meaning to this finding will require further investigation, it is possible that this finding may be related to altered attentional strategies for balance control. One possible explanation of this finding is that under confident individuals may excessively attend to their movement particularly in balance challenged situations which may interrupt automatic processes which are potentially more efficient and effective at maintaining balance control and safety in these situations. Notably, prior work has found an increased propensity for conscious control of movement production as PD progresses.⁴⁵ Follow-up work will need to determine whether discordance is related to outcomes such as mis-prioritization of attentional resources or cognitive dysfunction and robust longitudinal studies are needed to understand the relationship between balance, balance confidence, falls, physical activity restriction, and the development of discordance. It is possible that changes in constructs such as balance confidence, self-efficacy, and fear of falling may exert an important and distinct influence on discordance and subsequently on mitigating fall risk. Additionally, future studies should consider the progression and development of discordance, notably the sequence of worsening of balance and balance confidence, relative to falls and PD progression.

This study utilized a coordinated data analysis approach in which data across 5 studies and 4 sites were combined to accomplish the aims of this project. Meta-analytic approaches of this sort align with contemporary open science approaches and the efficient use of scientific resources. In the context of this project, coordinated data analysis allowed for the combination of data to increase the breadth/range of the data; this was advantageous as understanding the value of discordance is best accomplished by having a variety of data points in both the under and overconfident regions. Additionally, this study combined sets of diverse data which utilized a variety of administrations of the TUG. Despite these differences in approach, the analytic method revealed that the relationships were consistent among the data sets and this allowed for inferences to be generalized across the variations in the mobility-related postural control task of the TUG. Additionally, as the results of this study were consistent, despite data stemming from multiple sites and methods of TUG administration, this underscores the robustness of the findings related to the aims of this study.

Discordance as a measure that relates both perceived and actual balance abilities, may have clinical utility for fall reduction. Current approaches, utilizing physical or behavioral change interventions to improve physical or perceived balance and mobility outcomes, only partially mitigate falls and increase physical activity.^7,46,47 Therefore, discordance may be an additional intervention target to both optimize physical activity and reduce falls, by addressing both the perceptual and actual balance and their alignment.^2,13 Optimization of intervention for these purposes would require a multifaceted approach that includes physical and behavioral change methods, and potentially others, to be used together in a coordinated fashion.^7,47
-49 Assuming this to be the case, optimal outcomes are likely to be achieved more efficiently and more frequently when applying this theoretical approach to the clinical management of disorders that impact physical abilities, including but not limited to PD.

This study has several limitations which should be considered when interpreting its findings. First, as noted, this study utilized retrospective data from 5 prior studies, across 4 sites. Although the coordinated data analytic approach employed in this study provided a robust method to, at least partially, address the heterogeneity across these studies, it should be considered in the interpretation of results. Additionally, across these data sets, differing TUG procedures were utilized. Despite analyzing the congruence of relationships between TUG and ABC across all data sets, it is possible that differing TUG protocols may have subtly affected the relationships with ABC, which would consequently affect the calculation of discordance. However, analytic approaches were undertaken to confirm that the discordant relationships were not significantly different across studies. Further, while not ideal for internal validity, the variety of TUG procedures does allow for more generalizable interpretation of findings. The use of ABC as the variable of perceived balance is prone to a ceiling effect which may make estimation of concordance among those with a maximum measurement of balance confidence limiting (ie, a perceived balance measure without a ceiling effect would enable better estimation of discordance among individuals with high balance confidence). Previous research has shown that measures that are not prone to either ceiling or floor effects provide the most robust relationships.^50,51 The use of the TUG as a measure of balance performance is also a limitation as it is a measure of both balance and functional mobility, and may have a different relationship to balance confidence than a measure that is confined exclusively to the domain of balance, such as the MiniBESTest. Future research should investigate if the discordance relationship is improved by inclusion of more aligned performance and perceived ability measurement tools (ie, MiniBESTest and ABC). Although the line of best fit for quantifying discordance was determined to be derived from simple linear regression, it should not be assumed this is true across all cases and will generalize to a more impaired population. Due to measurable skew in both the ABC and TUG measures, representing a sample with generally higher balance confidence and faster TUG times on average, among a sample that is more impaired a non-linear line may serve as a more sensitive measure of discordance. Future research should aim to reproduce the findings presented here among a sample that is more normally distributed across the measures of perceived and actual balance. TUG testing was completed in different PD medication states between studies. Given that no differences in the relationship between TUG and ABC were observed between studies, that the TUG data was normalized, and that the anticipated effect of lack of PD medications would be a universal slowing of gait, we do not anticipate that this had an influence on the findings of the current study. However, it is difficult to determine to effect of medication on discordance with confidence due to the differences in medication status and TUG administrations between studies. Additionally, the reporting period for retrospective falls varied between studies from 6 months to 1 year. Considering the varying falls reporting period and PD severities between studies it is possible that the results of this study could be affected by different degrees of PD progression. Additionally, it is possible that the retrospective nature and differences in reporting period could have influenced the findings regarding the relationship between discordance and falls, however, falls data was used in the analyses as dichotomous variable (faller vs non-faller) which may limit this influence. Further, as falls were obtained via retrospective report this data is prone to recall bias. Future studies should employ prospective approaches for capturing falls-related data. Additionally, future studies should investigate factors that influence the development and manifestation of discordance, such as cognition, mood, and gait and balance limitations.

Conclusion

This study utilized a coordinated analysis approach to develop a novel method for the quantification of agreement between actual and perceived balance referring to this on a spectrum from concordant to discordant. Discordance was subsequently found to be uniquely related to falls among individuals with PD. Clinicians and researchers should consider discordance between actual and perceived balance in conditions that impact physical abilities, such as, but not limited to PD as a potentially modifiable target to minimize falls and fall risk while optimizing physical engagement.

Supplemental Material

sj-pdf-1-nnr-10.1177_15459683241300456 – Supplemental material for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis

Supplemental material, sj-pdf-1-nnr-10.1177_15459683241300456 for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis by Jason K. Longhurst, Andrew Hooyman, Merrill R. Landers, Martina Mancini, Erika Franzén, Breiffni Leavy, Hanna Johansson and Daniel Peterson in Neurorehabilitation and Neural Repair

Supplemental Material

sj-pdf-2-nnr-10.1177_15459683241300456 – Supplemental material for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis

Supplemental material, sj-pdf-2-nnr-10.1177_15459683241300456 for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis by Jason K. Longhurst, Andrew Hooyman, Merrill R. Landers, Martina Mancini, Erika Franzén, Breiffni Leavy, Hanna Johansson and Daniel Peterson in Neurorehabilitation and Neural Repair

Supplemental Material

sj-pdf-3-nnr-10.1177_15459683241300456 – Supplemental material for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis

Supplemental material, sj-pdf-3-nnr-10.1177_15459683241300456 for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis by Jason K. Longhurst, Andrew Hooyman, Merrill R. Landers, Martina Mancini, Erika Franzén, Breiffni Leavy, Hanna Johansson and Daniel Peterson in Neurorehabilitation and Neural Repair

Supplemental Material

sj-pdf-4-nnr-10.1177_15459683241300456 – Supplemental material for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis

Supplemental material, sj-pdf-4-nnr-10.1177_15459683241300456 for Discordance Between Balance Ability and Perception Is Associated With Falls in Parkinson’s Disease: A Coordinated Analysis by Jason K. Longhurst, Andrew Hooyman, Merrill R. Landers, Martina Mancini, Erika Franzén, Breiffni Leavy, Hanna Johansson and Daniel Peterson in Neurorehabilitation and Neural Repair

Footnotes

Author Contributions

Jason K Longhurst: Conceptualization; Data curation; Funding acquisition; Methodology; Project administration; Resources; Writing—original draft; Writing—review & editing. Andrew Hooyman: Data curation; Formal analysis; Methodology; Project administration; Resources; Software; Visualization; Writing—review & editing. Merrill R Landers: Methodology; Resources; Writing—review & editing. Martina Mancini: Funding acquisition; Methodology; Resources; Writing—review & editing. Erika Franzén: Methodology; Writing—review & editing. Breffni Leavy: Methodology; Writing—review & editing. Hanna Johansson: Methodology; Writing— review & editing. Daniel Peterson: Conceptualization; Data curation; Funding acquisition; Methodology; Project administration; Resources; Writing—review & editing.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The authors acknowledge the support of the Saint Louis University Applied Health Research Council, the Irma Ruebling Endowed Research Fund, the Michael J. Fox Foundation (under Grant #008373), Department of Veterans Affairs Grant No. I01 RX001075, NIH, NIA Grant No. R01AG006457, and NIH grant No. R25HD105583-03.

ORCID iDs

Jason K. Longhurst

Merrill R. Landers

Martina Mancini

Erika Franzén

Hanna Johansson

Daniel Peterson

Supplementary material for this article is available on the Neurorehabilitation & Neural Repair website along with the online version of this article.

References

Ellmers

Wilson

Kal

Young

WR.

The perceived control model of falling: developing a unified framework to understand and assess maladaptive fear of falling. Age Ageing. 2023;52(7):afad093. doi:10.1093/ageing/afad093

Landers

Nilsson

MH.

A theoretical framework for addressing fear of falling avoidance behavior in Parkinson’s disease. Physiother Theory Pract. 2023;39(5):895-911. doi:10.1080/09593985.2022.2029655

Deshpande

Metter

Lauretani

Bandinelli

Guralnik

Ferrucci

Activity restriction induced by fear of falling and objective and subjective measures of physical function: a prospective cohort study. J Am Geriatr Soc. 2008;56(4):615-620. doi:10.1111/j.1532-5415.2007.01639.x

Sales

Levinger

Polman

Relationships between self perceptions and physical activity behaviour, fear of falling, and physical function among older adults. Eur Rev Aging Phys Act. 2017;14:17. doi:10.1186/s11556-017-0185-3

Landers

Jacobson

Matsunami

McCarl

Regis

Longhurst

JK.

A vicious cycle of fear of falling avoidance behavior in Parkinson’s disease: a path analysis. Clin Park Relat Disord. 2021;4:100089. doi:10.1016/j.prdoa.2021.100089

Baltes

Herber

Meyer

Stephan

Fear of falling from the perspective of affected persons: a systematic review and qualitative meta-summary using Sandelowski and Barroso’s method. Int J Older People Nurs. 2023;18(1):e12520. doi:10.1111/opn.12520

Rider

Longhurst

Lekhak

Young

Landers

Fear of falling avoidance behavior assessment and intervention in Parkinson’s disease: a scoping review. Res Rev Parkinsonism. 2022;12:1-17. doi:10.2147/jprls.S350890

Valentine

Simpson

Worsfold

Fisher

A structural equation modelling approach to the complex path from postural stability to morale in elderly people with fear of falling. Disabil Rehabil. 2011;33(4):352-359. doi:10.3109/09638288.2010.491575

Lee

Altmann

McFarland

Hass

CJ.

The relationship between balance confidence and control in individuals with Parkinson’s disease. Parkinsonism Relat Disord. 2016;26:24-28. doi:10.1016/j.parkreldis.2016.02.015

10.

Curtze

Nutt

Carlson-Kuhta

Mancini

Horak

FB.

Objective gait and balance impairments relate to balance confidence and perceived mobility in people with Parkinson disease. Phys Ther. 2016;96(11):1734-1743. doi:10.2522/ptj.20150662

11.

Nilsson

Jonasson

Zijlstra

GAR

. Predictive factors of fall-related activity avoidance in people with Parkinson disease: a longitudinal study with a 3-year follow-up. J Neurol Phys Ther. 2020;44(3):188-194. doi:10.1097/NPT.0000000000000316

12.

Steckhan

Fleig

Wurm

Wolff

Schwarzer

Warner

LM.

Fear of falling carries over into overprotection in old age: a cross-lagged panel analysis. Int J Aging Hum Dev. 2024;98(4):436-451. doi:10.1177/00914150231196819

13.

Delbaere

JCT

Brodaty

Sachdev

Lord

SR.

Determinants of disparities between perceived and physiological risk of falling among elderly people: cohort study. BMJ. 2010;341:c4165. doi:10.1136/bmj.c4165

14.

Deshpande

Metter

Bandinelli

Lauretani

Windham

Ferrucci

Psychological, physical, and sensory correlates of fear of falling and consequent activity restriction in the elderly: the InCHIANTI study. Am J Phys Med Rehabil. 2008;87(5):354-362. doi:10.1097/PHM.0b013e31815e6e9b

15.

Steckhan

GMA

Fleig

Schwarzer

Warner

. Perceived physical functioning and gait speed as mediators in the association between fear of falling and quality of life in old age. J Appl Gerontol. 2022;41(2):421-429. doi:10.1177/0733464820979188

16.

Pliner

Hezel

Lord

SR.

Visual field dependence, sensorimotor function and falls in community dwelling older people. Braz J Motor Behav. 2024;18.

17.

Mak

Pang

MY.

Balance confidence and functional mobility are independently associated with falls in people with Parkinson’s disease. J Neurol. 2009;256(5):742-749. doi:10.1007/s00415-009-5007-8

18.

Landers

Lopker

Newman

Gourlie

Sorensen

Vong

A cross-sectional analysis of the characteristics of individuals with Parkinson disease who avoid activities and participation due to fear of falling. J Neurol Phys Ther. 2017;41(1):31-42. doi:10.1097/NPT.0000000000000162

19.

French

Moore

Pohlig

Reisman

Self-efficacy mediates the relationship between balance/walking performance, activity, and participation after stroke. Top Stroke Rehabil. 2016;23(2):77-83. doi:10.1080/10749357.2015.1110306

20.

French

Miller

Pohlig

Reisman

DS.

Depressive symptoms moderate the relationship among physical capacity, balance self-efficacy, and participation in people after stroke. Phys Ther. 2021;101(12):pzab224. doi:10.1093/ptj/pzab224

21.

Beauchet

Fantino

Allali

Muir

Montero-Odasso

Annweiler

Timed Up and Go test and risk of falls in older adults: a systematic review. J Nutr Health Aging. 2011;15(10):933-938. doi:10.1007/s12603-011-0062-0

22.

Beck Jepsen

Robinson

Ogliari

, et al. Predicting falls in older adults: an umbrella review of instruments assessing gait, balance, and functional mobility. BMC Geriatr. 2022;22(1):615. doi:10.1186/s12877-022-03271-5

23.

Duncan

Earhart

GM.

Measuring participation in individuals with Parkinson disease: relationships with disease severity, quality of life, and mobility. Disabil Rehabil. 2011;33(15-16):1440-1446. doi:10.3109/09638288.2010.533245

24.

Mak

Pang

MY.

Fear of falling is independently associated with recurrent falls in patients with Parkinson's disease: a 1-year prospective study. J Neurol. 2009;256(10):1689-1695. doi:10.1007/s00415-009-5184-5

25.

Crouse

Phillips

Jahanshahi

Moustafa

AA.

Postural instability and falls in Parkinson’s disease. Rev Neurosci. 2016;27(5):549-555. doi:10.1515/revneuro-2016-0002

26.

Balestrino

Schapira

AHV

. Parkinson disease. Eur J Neurol. 2020;27(1):27-42. doi:10.1111/ene.14108

27.

Fasano

Canning

Hausdorff

Lord

Rochester

Falls in Parkinson’s disease: a complex and evolving picture. Mov Disord. 2017;32(11):1524-1536. doi:10.1002/mds.27195

28.

Nilsson

Iwarsson

Thordardottir

Haak

Barriers and facilitators for participation in people with Parkinson’s disease. J Parkinsons Dis. 2015;5(4):983-992. doi:10.3233/jpd-150631

29.

Canning

Paul

Nieuwboer

Prevention of falls in Parkinson’s disease: a review of fall risk factors and the role of physical interventions. Neurodegener Dis Manag. 2014;4(3):203-221. doi:10.2217/nmt.14.22

30.

Allen

Canning

Almeida

LRS

, et al. Interventions for preventing falls in Parkinson’s disease. Cochrane Database Syst Rev. 2022;6:CD011574. doi:10.1002/14651858.CD011574.pub2

31.

Weston

Graham

Piccinin

. Coordinated data analysis: a new method for the study of personality and health. In: Hill

Allemand

, eds. Personality and Healthy Aging in Adulthood: New Directions and Techniques. Springer International Publishing; 2020:75-92.

32.

Hofer

Piccinin

AM.

Integrative data analysis through coordination of measurement and analysis protocol across independent longitudinal studies. Psychol Methods. 2009;14(2):150-164. doi:10.1037/a0015566

33.

Monaghan

Hooyman

Dibble

Mehta

Peterson

DS.

Stability changes in fall-prone individuals with Parkinson disease following reactive step training. J Neurol Phys Ther. 2024;48(1):46-53. doi:10.1097/npt.0000000000000442

34.

Peterson

Barajas

Denney

Mehta

SH.

Backward protective stepping during dual-task scenarios in people with Parkinson’s disease: a pilot study. Neurorehabil Neural Repair. 2020;34(8):702-710. doi:10.1177/1545968320935814

35.

Walker

Chiu

Y-F

Schwarz

Soudagar-Turkey

Longhurst

JK.

The relationship between cognition, fear of falling avoidance behavior, and backwards gait. Paper presented at: American Physical Therpay Association’s Combined Sections Meeting; 17 February 2024; Boston, MA.

36.

King

Mancini

Smulders

, et al. Cognitively challenging agility boot camp program for freezing of gait in Parkinson disease. Neurorehabil Neural Repair. 2020;34(5):417-427. doi:10.1177/1545968320909331

37.

King

Peterson

Mancini

, et al. Do cognitive measures and brain circuitry predict outcomes of exercise in Parkinson disease: a randomized clinical trial. BMC Neurol. 2015;15:218. doi:10.1186/s12883-015-0474-2

38.

Landers

Haller

Aldaco

, et al. The psychometric properties of the modified fear of falling avoidance behavior questionnaire in Parkinson’s disease and older adults. Arch Physiother. 2024;14(1):11-19. doi:10.33393/aop.2024.2702

39.

Goetz

Tilley

Shaftman

, et al. Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (MDS-UPDRS): scale presentation and clinimetric testing results. Mov Disord. 2008;23(15):2129-2170. doi:10.1002/mds.22340

40.

Steffen

Seney

Test-retest reliability and minimal detectable change on balance and ambulation tests, the 36-item short-form health survey, and the unified Parkinson disease rating scale in people with parkinsonism. Phys Ther. 2008;88(6):733-746. doi:10.2522/ptj.20070214

41.

Lohnes

Earhart

GM.

External validation of abbreviated versions of the activities-specific balance confidence scale in Parkinson’s disease. Mov Disord. 2010;25(4):485-489. doi:10.1002/mds.22924

42.

Nilsson

Drake

Hagell

Assessment of fall-related self-efficacy and activity avoidance in people with Parkinson’s disease. BMC Geriatr. 2010;10:78. doi:10.1186/1471-2318-10-78

43.

Lindholm

Hagell

Hansson

Nilsson

MH.

Factors associated with fear of falling in people with Parkinson’s disease. BMC Neurol. 2014;14:19. doi:10.1186/1471-2377-14-19

44.

Lindh-Rengifo

Jonasson

Mattsson

Ullén

Nilsson

MH.

Predictive factors of concerns about falling in people with Parkinson’s disease: a 3-year longitudinal study. Parkinsons Dis. 2019;2019:4747320. doi:10.1155/2019/4747320

45.

Masters

Pall

MacMahon

Eves

FF.

Duration of Parkinson disease is associated with an increased propensity for “reinvestment”. Neurorehabil Neural Repair. 2007;21(2):123-126. doi:10.1177/1545968306290728

46.

Kendrick

Kumar

Carpenter

, et al. Exercise for reducing fear of falling in older people living in the community. Cochrane Database Syst Rev. 2014;11:CD009848. doi:10.1002/14651858.CD009848.pub2

47.

Zijlstra

van Haastregt

van Rossum

van Eijk

Yardley

Kempen

GI.

Interventions to reduce fear of falling in community-living older people: a systematic review. J Am Geriatr Soc. 2007;55(4):603-615. doi:10.1111/j.1532-5415.2007.01148.x

48.

Whipple

Hamel

Talley

KMC

. Fear of falling among community-dwelling older adults: a scoping review to identify effective evidence-based interventions. Geriatr Nurs. 2018;39(2):170-177. doi:10.1016/j.gerinurse.2017.08.005

49.

Wetherell

Johnson

Chang

, et al. Activity, balance, learning, and exposure (ABLE): a new intervention for fear of falling. Int J Geriatr Psychiatry. 2016;31(7):791-798. doi:10.1002/gps.4393

50.

Hawe

Scott

Dukelow

SP.

Taking proportional out of stroke recovery. Stroke. 2019;50(1):204-211. doi:10.1161/STROKEAHA.118.023006

51.

Hooyman

Malek-Ahmadi

Fauth

Schaefer

SY.

Challenging the relationship of grip strength with cognitive status in older adults. Int J Geriatr Psychiatry. 2021;36(3):433-442. doi:10.1002/gps.5441

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.07 MB

0.10 MB

0.42 MB