The Effects of Hearing Loss on Balance: A Critical Review

Abstract

Recent epidemiological findings of associations between hearing loss (HL) and poorer mobility and higher falls risk have increased the demand for ecologically valid experimental research to determine the potential mechanisms underlying human hearing-balance relationships. This review provides an overview of the laboratory-based approaches to studying human balance, identifies crucial factors that should be considered to improve the ecological validity of hearing-balance research, and provides a critical review of the scientific literature to date on the effects of HL on balance. Most present studies can be subdivided into those that examine balance changes due to the effects of (1) auditory suppression in individuals with normal hearing, (2) HL with and without hearing aids, and (3) cochlear implants in children and adults. To allow for meaningful comparisons, we based our in-depth critical review on studies that met minimum criteria of having at least one objective kinetic or kinematic measure of standing balance during a two-legged stance with feet side-by-side, for at stance duration of at least 30 sec. With this minimum criterion in place, we found mixed evidence that hearing suppression, HL, or hearing devices affects postural stability, especially when other sensory information is available and/or reliable, and task demands are relatively low. However, hearing may become more important when multiple sensory systems become unreliable, task demands, or cognitive impairments are greater, or when sounds provide important auditory cues to assist with orientation or provide early detection of an impending balance disturbance. However, more research is clearly needed, because there is a wide range of technical and experimental differences and limitations observed across the present literature. To address these gaps, we have provided a number of recommendations and suggested priorities for future research to provide the ecologically valid, reliable, and reproducible evidence needed to uncover any potential relationships between HL, balance, and falls.

INTRODUCTION

When considering the effects of hearing and HL on typical everyday activities, they are most often studied in the context of communication and social interactions. However, hearing also helps us to localize, monitor, and respond to objects and events in our environment and to perceive our own movements through space (^{Campos et al. 2018}). These localizing, monitoring, and reactive abilities are, therefore, particularly consequential to important everyday behaviors involving balance and mobility. As such, when defining the scope of functional outcomes associated with “hearing in the real world” (^{Keidser et al. 2020}), it is important to explore the effects of hearing and HL on mobility-related outcomes. Increasing interest in this research has also been spurred by emerging and converging epidemiological studies demonstrating associations between HL and poor balance (^{Viljanen et al. 2009; Chen et al. 2015}), mobility problems (^{Chen et al. 2015; Mikkola et al. 2015; Polku et al. 2015}), and falls (^{Lin & Ferrucci, 2012; Jiam et al. 2016; Gopinath et al. 2016; Agmon et al. 2017}), particularly in older adults (see also ^{Campos et al. 2018} for a review). For instance, ^{Lin and Ferrucci (2012)} analyzed data from the National Health and Nutritional Examination Survey and demonstrated that individuals with HL were significantly more likely to have recently experienced a fall. Higher degrees of HL were associated with increased odds of having fallen (1.4x increased odds of falling for every 10 dB increase in HL). Notably, the majority of studies demonstrating these associations between HL and poor mobility/falls have been epidemiological studies on aging (e.g., the National Health and Nutritional Examination Survey; Baltimore Longitudinal Study on Aging; Health, Aging and Body Composition Study; Finnish Twin Study on Aging), with far fewer empirical studies focused on experimentally examining the underlying cause(s) of these associations.

Several nonexclusive and nonexhaustive hypotheses have been offered to explain these correlational findings, particularly in the context of aging (^{Viljanen et al. 2009; Lin & Ferrucci, 2012; Jiam et al. 2016; Campos et al. 2018}). First, it is likely that HL causes problems with binaural sound processing and spatial hearing, thereby reducing one’s ability to perceive/monitor the environment and one’s position within the environment. Second, HL taxes cognitive resources (^{Pichora-Fuller et al. 2016}), which could thereby limit the resources available to support balance and safe mobility. Third, there may be paralleled pathologies/declines in the auditory and vestibular systems due to their similar mechanisms of functioning and shared inputs to the brain (^{Zuniga et al. 2012}).

Although there have been some attempts to account for these alternative explanations within epidemiological studies (e.g., by controlling for known vestibular disorders), the nature and directions of these associations cannot be systematically determined using correlational analyses. In contrast, laboratory-based studies provide the opportunity to consider specific subcomponents of balance, mobility, and falls risk, to control the independent factors of interest, and to precisely measure particular outcomes. Hypothesized associations between HL and falls risk can be tested systematically in the laboratory through the independent manipulation of sensory, cognitive, and motor tasks within controlled settings. Laboratory-based studies focused on the effects of HL on mobility and falls risk are few, but increasing, with most research in this area focused specifically on balance. The studies can be divided into two general areas as follows: (1) suppression of sound in normal hearing (NH) subjects and (2) the effects of HL, with and without, (a) hearing aids (HA) and (b) cochlear implants (CIs). Notably, the most commonly used reliable measures of balance currently reported in the literature are those that measure basic static standing balance. While careful measures of standing balance can be sensitive and meaningful when characterizing balance control and identifying balance impairments, they are unfortunately limited from the perspective of ecological validity. Therefore, this article provides a critical overview of what is currently known about the effects of hearing on balance, with a particular focus on standing balance, and provides recommendations regarding important factors that must be considered when interpreting these outcomes from the perspective of understanding real-world hearing-balance-falls associations.

There are many factors that should be considered when developing a framework to evaluate the effects of hearing on balance, but here we have categorized these factors into those associated with the characteristics of the Individual (e.g., age, hearing status, etiologies of HL, emotional, and cognitive status), the Environment (e.g., acoustics, nature of sound cues, visual inputs, and device use), and the Task (i.e., balance-related task types and task demands). In this article, we start by providing a general overview of what is involved in balance control, the research methods that are typically used to study balance, and the measures that are used to describe balance-related outcomes. We then explore some of the factors that should be considered by hearing researchers who are interested in studying balance, including those associated with the characteristics of the individual, environment, and task. Next, we provide a review and critical summary of the scientific literature to date, focusing specifically on laboratory-based studies examining the effects of hearing, HL, and hearing devices on static balance, including a discussion of the gaps and limitations. Last, we provide several recommendations and suggest priorities for future research in the area of hearing and balance and we consider the implications of this research for application and practice. Throughout the article, we reflect on the importance of striving for ecological validity in laboratory studies to ensure that findings are reflective of real-life, hearing-related function, activity, or participation. With respect to the specific goals of striving for ecological validity in hearing research (^{Keidser et al. 2020}), the associations between hearing and balance are directly relevant to (a) better understanding the role of hearing specifically for balance/mobility in everyday life, (b) supporting the development of improved hearing-related interventions by highlighting considerations specific to balance/mobility, and (c) improving methods of predicting the ability of people with various hearing abilities/status to accomplish real-world tasks associated with balance/mobility. Overall, better understanding the causal mechanisms and time course underlying the associations between hearing, balance, mobility, and falls through controlled, yet ecologically valid laboratory studies could ultimately help to inform recommendations and applications, and shift today’s healthcare paradigms from being reactive to becoming more preventative (^{Campos & Launer, 2020}).

LABORATORY-BASED MEASURES OF BALANCE

Here, we define balance as the ability to remain upright with one’s center of mass over the base of support while standing static or when in motion. Human balance can be assessed within the laboratory using performance-based tests and posturography to measure different elements of balance control. We briefly review each category below.

Performance-Based Measures

Performance-based measurements rely on the subjective assessment by an evaluator of the perceived quality of performance during various balance and mobility tasks. Performance-based assessments can involve single tasks, or more often, a battery of multiple tests designed to challenge different aspects of the balance control system, scored by an evaluator, to provide a composite balance and mobility score. Research on hearing-balance interactions have used single-task assessments such as time to fall during one-legged stance, tandem stance, standing on foam with eyes closed (EC) (^{Rumalla et al. 2015; Stevens et al. 2014; Ibrahim et al. 2019}), or the Timed up and Go test (^{Koh et al. 2015}). Hearing effects on balance have also been investigated using composite scores from a set battery of tests, such as the Berg Balance Scale (^{Lacerda et al. 2012}), Balance Error Scoring System (^{Melo et al. 2015}), Mini-Balance Evaluation-Scoring Test (^{Shayman et al. 2017}), Bruininks-Oseretsky Test of Motor Proficiency, second edition (^{Cushing et al. 2008}), and Clinical Test of Sensory Interaction on Balance (^{Kluenter et al. 2010}). While performance-based measures are effective in identifying gross abnormalities in balance performance, the ability to detect subtle changes in balance is lacking and may be confounded by a number of limitations. First, performance-based measures are susceptible to inter-rater variability because they rely primarily on visual assessments and score assignments indicative of deviations from “normal” behavior. Second, tests that measure speeded responses (e.g., Timed up and Go test), or “time to failure”, provide little insight into actual balance performance, and may, in fact, encourage abnormal or risky balance strategies to achieve a better score or outcome (i.e., speed/accuracy trade-off). Last, the scores themselves do not provide insight into the actual mechanisms or basic elements of balance control that may be altered.

Posturography

To address some of the limitations of performance-based measures, researchers use posturography to record objective measures of specific components of human balance control. For instance, unlike performance measures, posturography is not reliant on the observations of a human observer but uses sensors (e.g., force or inertial sensors) to directly measure balance behavior. For example, body-worn balance sensors have been used to record trunk stability in patients with CI during a series of clinical balance and mobility tasks to calculate a composite score such as the Balance Control-Index (^{Stieger et al. 2018}) or % fall-risk score (^{Louza et al. 2019}). These composite scores are then used to compare balance performance in patients with CI against age and sex-specific normative values.

Posturography can also be used to provide more detailed measures in the time or frequency domains for a single dynamic or static postural task. Dynamic posturography measures postural responses to physical perturbations to stance (i.e., disruptions to balance) as a way of eliciting and measuring postural adjustments. These perturbations can be generated through voluntary movements (e.g., arm raises and leg lifts) that require postural adjustments to counteract the anticipated effect on the center of mass (COM), or can be generated externally using an unstable or motorized support-surface, or by applying forces to one or more parts of the body, to elicit postural reactions. In contrast, static posturography provides measures of upright standing in the absence of any physical perturbations. Both static and dynamic posturography provide unique insights into the nature of balance changes, and also have inherent limitations that need to be recognized and accounted for (^{Visser et al. 2008}).

Dynamic Posturography •

Dynamic balance responses are designed to more closely mimic the conditions that precede a fall and trigger a postural response during real-life interactions. Support-surface rotations mimic unstable tilting surfaces that may be encountered when standing on a boat or a moving dock, whereas support-surface translations mimic linear surface displacements such as those experienced on a bus or subway. Other perturbations apply forces to the upper-body through controlled pushes or pulls to a body-worn harness or frame, or by releasing a cable holding the participant in place while they maintain a forward lean (^{Bloem et al. 2003}). Dynamic posturography systems are generally expensive and require special technical support and sophisticated analyses, which limits their widespread utility for clinical and research applications. As such, there are relatively few studies that have examined the effects of hearing on dynamic balance to date (^{Schwab et al. 2010; Greters et al. 2017; Kowalewski et al. 2018}).

Static Posturography •

Static balance is the most widely studied given the greater simplicity of testing methods compared with dynamic posturography. Static posturography can be conducted using a forceplate to record ground reaction forces and moments under the feet to calculate Center of Pressure (COP). Whole-body sway can also be measured to estimate the movements of the COM using either inertial sensors placed on the hip or trunk to record position, velocity, or acceleration of sway, or using motion capture systems to estimate a calculated COM position based on movements of multiple markers placed on different anatomical landmarks (^{Visser et al. 2008}). Participants are typically asked to stand upright, without any purposeful movement for a defined period of time. Differences in body sway characteristics are then interpreted as changes in balance control in response to sensory, motor and/or environmental conditions or differences in balance control between groups. One of the criticisms of static balance tasks is that it is highly constrained, and thus the ecological validity may be limited in terms of its ability to mimic balance conditions that lead to instability or falls during real-life interactions. Specifically, it is rare for humans to stand in one spot for an extended period of time, without any significant weight shifts; consequently, falls under these task conditions are relatively uncommon (^{Robinovitch et al. 2013}). However, by creating this highly controlled condition, it provides an important tool to gain insight into changes happening within the balance control system and can help to define the contributions of different components of postural control.

Because static posturography provides a more targeted assessment of balance compared with other mobility-related tasks and is the most commonly used within a controlled laboratory-based setting, it will be the focus of the remainder of the article.

METHODOLOGICAL FACTORS TO CONSIDER FOR BALANCE AND HEARING RESEARCH

Posture Measurement Duration

While static posturography has utility as a tool for assessing balance control, it is highly sensitive to a number of factors that can confound the outcome measures. One of the key factors that vary greatly across studies is the length of time that the sway behavior is recorded (sampled). The reason that sample duration is a crucial factor lies in the nature of the COP/COM signal during quiet stance. The COP signal has a distribution of frequencies that range from below 0.01 to 1 Hz, with the strongest power (energy of the signal for a given frequency) concentrated within the very low-frequency ranges (0.1 to 0.05 Hz) (^{Carpenter et al. 2001}). These low-frequency components of the COP signal reflect the large, yet slow, oscillations of the COM. In contrast, higher frequency COP oscillations represent only a small proportion of the overall power of the COP signal and are thought to reflect the neuromuscular output of the CNS needed to control the COM (^{Winter, 1995}). To capture the largest amplitude components that dominate a sway signal, the sample duration must be long enough to ensure that a full cycle can occur. For example, 20 sec would be needed to capture a full cycle for a 0.05-Hz signal, whereas as a 0.01-Hz signal needs >100 sec to complete a full cycle. As a result, the outcome measures calculated to summarize the COM/COP signals in terms of amplitude of sway (e.g., root mean square, area, and total sway path) and frequency of sway (e.g., mean-power-frequency and power) are highly sensitive to the duration of the recording. Summary measures calculated from different segment lengths of the same balance trial will, therefore, yield significantly different values, up until lengths that exceed >60 sec (^{Le Clair & Riach, 1996; Carpenter et al. 2001; van der Kooij et al. 2011}). Shorter duration than that and significant content of the balance behavior will not be captured by most summary measures (^{van der Kooij et al. 2011}), nor will the summary measures be reliable, particularly when the duration becomes shorter than 30 sec (^{Carpenter et al. 2001}). Thus, comparisons between groups using sampling durations of different lengths will be significantly confounded and measures using shorter durations should be considered with caution, especially when trying to discriminate between people with and without a balance impairment (^{Jehu et al. 2018}).

Stance Position

Another key feature that should be controlled and accounted for is the stance position during the standing trial. Commonly tested stance positions include tandem (one foot in front of the other positioned heel-to-toe), feet parallel to each other (either feet together or shoulder-width apart), and one foot off the ground. Stance position has an influential role on balance because the outer borders of the foot define the base of support within which the COM must be maintained to avoid a fall (^{Duarte & Zatsiorsky, 2002}). Anterior-Posterior (AP; forward-backward) and Medial-Lateral (ML—side-to-side) sway are controlled independently, with AP sway controlled primarily through ankle torques and ML sway controlled through a combination of ankle and hip torques (^{Winter et al. 1996}). Thus, changes in stance position, not only modify the base of support, they also affect the relative contribution of torques needed to maintain balance. For example, widening the stance will reduce sway amplitude and increase sway frequency in the ML plane by changing the biomechanical stiffness of the body (^{Kirby et al. 1987}). Likewise, standing in tandem stance increases stability and reduces sway in the AP direction, while also decreasing stability and increasing sway amplitude in the ML direction. Therefore, the stance position needs to be considered when comparing across studies and also when trying to interpret outcome measures that could be affected by the directional effects of the stance width and foot orientation.

Balance is Multisensory

Multiple sources of sensory information are crucial to controlling balance in both static and dynamic conditions. However, there is some redundancy across sensory systems, particularly during static balance. This redundancy is evidenced by the fact that in young healthy individuals the removal or reduced reliability of one sensory input (e.g., closing one’s eyes), usually leads to only minor changes in the amplitude and frequency of sway and is not catastrophic to balance control. Individual sensory inputs are typically weighted by the CNS based on their individual reliabilities (^{Peterka, 2002; Ernst & Bülthoff, 2004}). Where the effects of a sensory loss become most evident, or impactful, is when more than one sensory input is poor (e.g., in the case of age-related hearing and vision loss) in which case the removal of even one sensory input leads to greater changes in balance (^{Woollacott et al. 1986}). Therefore, when considering the effect of reducing/removing one sensory input, such as hearing on balance, the status of the other sensory systems is an important consideration when interpreting the results and when comparing results across studies.

One type of balance test that has been explicitly designed to examine how different sensory inputs may contribute to a balanced deficit is called the “Sensory Organization Test” (SOT). The SOT is based on a protocol designed for the Neurocom balance platform system, which incorporates a forceplate on a moving support surface and visual surround. The rotating platform and visual surround can be fixed in place or moved (via a servo-motor) in the pitch plane in concert with the participant’s estimated AP center of gravity sway. This creates incongruent information that reduces the reliability of the proprioceptive and/or visual inputs. Six different combinations of sensory input are created through manipulations of vision and proprioception as follows: (1) eyes open (EO) with stable surface (2) EC with stable surface, (3) sway referenced vision with stable surface, (4) EO with sway referenced support-surface motion, (5) EC with sway referenced support-surface motion, and (6) sway referenced vision and support-surface motion. The SOT uses an estimate of center of gravity sway angle developed from forceplate measures to calculate an individual’s Equilibrium Score (0 to 100), where a score of 100 represents no sway and 0 indicates sway that exceeds the theoretical limit of stability (12.5°) resulting in a fall. The score can be calculated for each sensory condition (conditions 1 to 6), or combined as a composite score.

The SOT test has been most commonly used to study hearing-balance interactions in patients with CI (^{Buchman et al. 2004; Schwab et al. 2010; Parietti-Winkler et al. 2015; Greters et al. 2017; Murray et al. 2020}). Improved composite equilibrium scores were observed in HL subjects with their CIs turned on compared with off. These improvements were driven by significant changes in conditions 4, 5, and 6 in which proprioceptive input was unreliable (^{Buchman et al. 2004; Schwab et al. 2010}). Subjects with uncorrected HL (pre-CI surgery) had decreased balance performance compared with normal hearing controls based on the composite score, and individual scores for conditions 1, 3, 5, and 6 (^{Parietti-Winkler et al. 2015}). In contrast, ^{McDaniel et al. (2018)}, observed no significant difference in composite, or individual equilibrium scores for any SOT condition for older adults with HL with or without the assistance of bilateral HA. To our knowledge, there have been no studies examining the effects of sound suppression in individuals with normal hearing on balance measured using the SOT.

While SOT research provides insight into how sensory interactions influence balance, the SOT test has inherent limitations that make comparisons with other posturography research challenging (^{Visser et al. 2008}). The limitations include the potential for ceiling effects, particularly for conditions on firm surfaces with EO or EC (^{McDaniel et al. 2018}), reliance on max-min measures that are affected by extreme sway values (^{Parietti-Winkler et al. 2015}), inability to quantify changes in ML directions or the frequency domain, and reliance on short balance trials (20 sec) that restrict the ability to capture the largest amplitude, lowest frequency COP/COM components of sway.

Emotional Factors

The emotional state of the participant can also have a significant effect on static balance control, including fear, anxiety, and confidence. Different sources of anxiety and arousal, for instance, due to environmental threats, first-time exposure to a laboratory setting, or social anxiety caused by the presence of an evaluator, has the capacity to affect the amplitude, frequency, and even mean-position (lean) during static balance (see ^{Adkin & Carpenter, 2018} for review). These emotional influences on balance are relevant to the study of hearing-balance interactions because associations have been found between HL and increased anxiety and fear of falling (^{Viljanen et al. 2012; Arslan et al. 2018}), and the presence of certain sounds may increase or decrease anxiety or arousal (^{Yang et al. 2018}). Different emotional auditory stimuli have also been shown to alter postural control in individuals with normal hearing (^{Chen & Qu, 2017}), with larger COP amplitude changes observed when participants listened to unpleasant auditory stimuli compared with pleasant and neutral auditory stimuli.

While some HL studies have tried to incorporate measures of subjective balance confidence, such as the Activities-specific Balance Confidence Scale (^{Criter & Honaker 2013; Rumalla et al. 2015}), it should be noted that these generalized scales assess the individual’s confidence in their ability to complete daily tasks of living without falling, and thus do not provide any insight into the emotional state of an individual at the time of the experiment or changes in anxiety between different standing conditions. State-specific measures of anxiety and fear at a particular moment in time can be assessed using self-reported questionnaires and physiological measures of arousal can be recorded noninvasively and reliably using measures of sympathetic nervous system activity (i.e., skin conductance or heart-rate variability) (^{Carpenter et al. 2006; Hauck et al 2008}). These measures can help ensure that emotional state is accounted for when comparing balance performance between groups, or at different time points in a repeated measures design when anxiety and arousal levels may differ with task comfort and familiarity, the presence of the evaluator, and/or the environment. It is recommended that more than one balance trial is used to account for potential first-trial effects (^{Adkin et al. 2000; Adkin & Carpenter 2018}), especially when used for initial baseline assessments to compare against subsequent trials.

Cognitive Factors

Cognitive factors can influence balance in various ways including how attention is focused or distributed during balance tasks and how dual-task demands are managed (e.g., standing while listening or thinking). While often regarded as automatic, maintaining static balance places cognitive demands on the CNS. The amount of cognitive load required to maintain balance varies with age and also when compensating for other sensory, cognitive, or balance deficits. Therefore, when assessing balance, the task requirements have a strong potential to either add or remove attentional demands on the CNS, and this effect may interact with age, sensory, cognitive, or balance abilities. For example, in terms of attentional foci, instructions or conditions that promote an increased focus of attention either toward an individual’s own movements/posture, or away from balance and toward external cues or features of the environment, both (toward and away) have been shown to influence postural control (^{Bonnet, 2016}). This has critical implications for studies on hearing and balance for several reasons. For instance, the acoustics of the environment or auditory cues within the environment have the potential to either support balance control (e.g., if providing orientation cues) or disrupt balance control (e.g., if providing noninformative or distracting cues).

It has also been shown that balance can be affected when an individual has to perform multiple tasks simultaneously (e.g., dual-task paradigms) depending on the magnitude of the additional task demands and the abilities of the individual (^{Lajoie et al. 1993; Teasdale et al. 1993}; Maylor & Wing, 1996; ^{Teasdale & Simoneau, 2001; Woollacott & Shumway-Cook, 2002; Mahboobin et al. 2007; Bernard-Demanze et al. 2009; Bronstein & Pavlou, 2013}). Essentially, balancing while performing different types of tasks (remembering, counting, and tracking) results in a competition for shared cognitive resources, which could result in detrimental effects on one or both tasks. Relevant to the current discussion is that these dual-task costs have also been specifically observed during the simultaneous performance of tasks requiring auditory processing/listening of varying demands while maintaining posture during standing (^{Carr et al. 2019}) and walking (^{Lau et al. 2016; Nieborowska et al. 2019}) tasks. Results have demonstrated that greater dual-task costs are typically observed as the demands of the listening task and/or postural task are increased, but that (for older adults in particular), posture is often prioritized over listening performance. Further, listening under circumstances when auditory input is poor, due to either environmental noise or HL, can result in increased listening effort (^{Pichora-Fuller et al. 2016}), thereby increasing cognitive load and reducing the resources available to support balance control. Balancing while talking can also affect balance measures. For instance, experimental tasks that require articulation (e.g., providing a verbal response) place additional motor-cognitive-attentional demands on the CNS that have the potential to influence balance (Dault et al. 1995). Furthermore, the type and salience of the sound might have different impacts on attention-balance interactions. For example, a distinction should be made between sounds that are used for environmental perception and monitoring and could support balance when informative and congruent (e.g., cues to localization/orientation) versus sounds that are not related to the balance task, but could affect balance (e.g., speech sounds, which draw attention and increase cognitive load), or other incongruent sounds that may be distracting (e.g., pink noise) or disruptive to posture (see also ^{Campos et al. 2018} for a review). Thus, careful attention needs to be paid to the type of sound environments and hearing tasks that are used when comparing studies that assess hearing-balance interactions.

A CRITICAL REVIEW OF STATIC POSTUROGRAPHY RESEARCH

In the next sections, we critically review the present scientific literature on the effect of hearing on static balance control. When reviewing this literature, we highlight, where possible, the potential differences in methodological approaches, with special attention paid to the key factors we have discussed earlier related to the individual, the environment, and the behavior/task, that are likely to have a significant influence on the measured outcomes. We also consider the potential limitations that may confound the interpretation of the results or reduce the ecological validity of the findings. Studies that are included in the final summary are those that involve at least one quantitative kinetic or kinematic measure of postural sway during two-legged, feet side-by-side stance, over a minimum duration of 30 sec. While this criteria-based approach may seem somewhat restrictive, it allows for meaningful comparisons to be drawn across studies that share basic fundamental characteristics and reflect our understanding of the key elements of static postural control.

Effects of Auditory Suppression on Balance in Individuals with Normal Hearing

There is a wide range of studies that have examined how balance changes with the manipulation of different types, sources and directions of sound (see ^{Campos et al. 2018} for review), which provide important insights into the potential mechanisms underlying hearing-balance interactions. Relatively few studies have used auditory suppression in otherwise NH participants to specifically examine how removing sound information affects balance; of these, only four studies involved posturographic measures of standing balance of at least 30 sec duration (see Table 1). ^{Kanegaonkar et al. (2012)} used a WII balance system to examine healthy young adults standing for 30 sec in a clinic room and soundproof audiology booth of similar dimensions with EO and EC, on firm and foam surfaces. In both acoustic settings, the participants performed the standing tasks with and without ear defenders on to further suppress auditory inputs. The results revealed that there was an increase in sway area when auditory information was suppressed with ear defenders compared to without, but only for the EO on foam condition in the clinic room, and EC on foam in the soundproof room. There was also increased sway in the EO conditions without ear defenders when performed in the soundproof room versus the clinic room.

TABLE 1. - Articles reviewed on the effects of suppression of normal hearing

Article	Kanegaonkar et al. (2012)	Gago et al. (2015)	Vitkovic et al. (2016)	Maheu et al. (2017)
Normal Hearing Group (sample number, age [mean/range])	21 [23–44 yrs]	24 [72 yrs]	50 [29 yrs]	14 [33 yrs]
Foot position	Feet together (Romberg)	Feet together	Feet 10 cm apart	Feet shoulder width apart
Vision: eyes open (EO) or closed (EC)	EO and EC	EO and EC	EO and EC	EO and EC
Surface type	Firm and Foam	Firm	Firm and Foam	Firm and Foam
Stance Instructions	NR	stand quietly	stand as still as possible	stand and count back from 1000
# trials per condition	1	1	1	4
Balance measures	COP area	RR length and radius displ., range (AP and ML)	COP length	COP area, Vel
Recording device	WII	IMU [COM]	WII	Forceplate
Sample duration (sec)	30	30	60	60
Sound suppression type	Ear defenders	Ear defenders	Earplugs	Earplugs and hearing protectors
Sound/noise type	Ambient	Ambient (125–2000 Hz)	Ambient/white (continuous/moving)	Pink noise (100–4000 Hz)
Sound pressure level (dB)	NR	36	60–70 (white)	NR (level of comfort)
Sound source description	N/A	N/A	1 speaker (continuous) or 8 speakers in 180° semicircle (moving); 1 m away	1 speaker, 1 m behind
Room acoustics	Normal and Soundproof	Normal	Normal and Soundtreated	Normal
Test of hearing impairment	History—hearing loss excl.	History—auditory deficits excl. bedside exam (finger rub by ears)	>20 dB HL in both ears for 0.5, 1, 2 and 4 kHz	Mean hearing thresh. for 0.25, 0.5, 1, 2, 3, 4, 6, 8 kHz
Test of vestibular impairment	History—balance disorders excl.	History—vestibular disorders excl.	History-vHit, caloric, oVEMP, cVEMP	Direct measures—vHit, oVEMP, cVEMP
Balance changes with suppression?	Yes	Yes	No	No

Measures include WII balance system (WII); center of pressure (COP); Anterior-Posterior (AP); Medial-Lateral (ML), Displacement (displ.) velocity (vel); total path length (Length); sway area (Area); Romberg Ratio (RR); Center of Mass (COM); Prior Documented Medical History (History); video head impulse test (vHit); ocular vestibular evoked myogenic potentials (oVEMP); cervical vestibular evoked myogenic potentials (cVEMP); decibels hearing loss (dB HL); excluded (Excl.); threshold (thresh.); Variable not reported = NR.

^{Gago et al. (2015)} examined the effects of wearing ear defenders in an older adult population, standing with feet together for 30 sec with EO and EC on firm surfaces in a “quiet clinical hospital room laboratory.” Summary measures of body sway recorded with an inertial sensor placed at the estimated level of the COM were used to calculate ratios indicating the effects of auditory suppression (i.e., no sound: sound) for EO and EC separately. The results demonstrated larger sway responses with auditory suppression compared with when sound cues were available (indicated by positive auditory index scores), particularly in EO conditions, and larger auditory index scores observed in AP compared with ML directions.

^{Vitkovic et al. (2016)} examined the effects of auditory suppression in young adults with NH standing as still as possible for 60 sec on a WII balance system, with feet 10 cm apart, EO and EC, on firm and foam surfaces while COP path length was measured. Each participant completed four sound conditions, which included standing in a normal ambient room, a soundproof room wearing earplugs, and when presented with continuous, moving white noise. The results of this study showed no significant changes in COP path length between the auditory suppression condition compared with any of the noise conditions (ambient and moving white noise).

Likewise, ^{Maheu et al. (2017)} showed no significant change in COP sway area or velocity when individuals stood for 60 sec with or without NH (suppressed with ear-plugs and ear defenders), with either EO or EC, or on firm or foam surfaces. However, the results did reveal greater changes between EO/EC conditions when hearing was suppressed, compared with NH, suggesting that the removal of sound may impact normal sensory re-weighting. This study was distinct from the others described earlier, by having participants perform a cognitive task (count backward) during stance, including an additional sound source (white noise from a speaker), and performing clinical measurements of vestibular function at the time of the experiment.

There are other notable differences across the studies in terms of methodologies and outcome measures that are known to have significant influences on posturographic measures, including foot position, stance instructions, recording devices, and outcome measures. Only the studies that used 30 sec standing trials found any effect of auditory suppression within a stance condition; no differences with sound suppression were observed in the studies that used longer sample durations. There was significant variation across studies in terms of the sound environments in which the balance task was performed, methods used to assess hearing impairment, as well as the age of the participants. On the basis of the wide variation in protocols and measures, it is not unreasonable to expect highly inconsistent evidence for and against any significant effect of removing NH on static balance control.

Effects of Hearing Loss on Balance With and Without Hearing Aids

HA are used by individuals with HL to help amplify and process sounds. Most signal processing implemented within even sophisticated HA is mainly designed to improve speech intelligibility (particularly in noisy environments) with much less known about how they affect balance. Amplifying sound cues that are used to support self-orientation could help to support balance, but HA are also known to interfere with sound localization abilities, which could result in fewer benefits. If HA reduce listening effort during balance-related tasks, this could also help to preserve cognitive resources and support balance-related outcomes. The nature of the effects of HA on balance is likely determined by many factors including the features of the device, the acoustical conditions, and the nature of the task. Of the laboratory-based studies examining the effects of HL with and without HA, only three studies used objective sway-based measures of sufficient duration (>30 sec) to be deemed reliable (see Table 2).

TABLE 2. - Articles reviewed on the effects of hearing loss with and without hearing aids

Article	Vitkovic et al. (2016)	Negahban et al. (2017)	Maheu et al. (2019)
Normal hearing group (sample number, age [mean])	50 [29 yrs]	N/A	14 [31 yrs]
Hearing loss (HL) group (sample number; age [mean])	19 with HA [65 yrs]; 9 no HA [67 yrs]	22 with HA [67 yrs]; 25 no HA [67 yrs]	18 with HA [38 yrs]
Hearing loss type (side; [etiology])	Bi- and Uni- [SN]	Bi-[U]	Bi- [Con/SN]
Foot position	feet 10 cm apart	feet together	feet shoulder width apart
Vision: eyes open (EO) or closed (EC)	EO and EC	EO and EC	EO and EC
Surface type	Firm and foam	Firm and foam	Firm and Foam
Stance instructions	Still as possible	Still as possible	While counting from 1000
# trials per condition	1	3	3
Balance measures	COP length	COP area, Vel. [Mean, St. Dev]	COP area, Vel
Recording device	WII	Forceplate	Forceplate
Sample duration (sec)	60	30	60
Sound/noise type	Ambient/none (earplugs)/white (continuous/moving)	Ambient sound	Pink (100–4000 Hz)
Sound pressure level (dB)	60–70 (white)	NR	56–65
Sound source description	1 speaker (continuous) or 8 speakers in 180° semicircle (moving); 1 m away	N/A	1 speaker, 1 m in back
Room acoustics	Normal and sound-treated	NR	NR
Test of hearing impairment	>20 dB HL in both ears for 0.5, 1, 2, 4 kHz	Mean air conduction thresh. at 0.5, 1, 2 and 4 kHz	Otoscopy, tympanometry, tonal audiometry (bone and air conduction) at 250 Hz to 8 kHz by octave bands plus 3 kHz and 6 kHz.
Test of vestibular impairment	History-vHit, caloric, oVEMP, cVEMP	NR	Direct Measures—vHit, oVEMP, cVEMP
Balance changes HA-On vs. Off?	Yes [with sound]	Yes	Yes (only with VL)
Balance changes HL vs. NH?	NR	N/A	Yes (only with VL)

Etiology types include Congenital (Con), Sensorineural (SN); Unknown (U); Bilateral (Bi-); Uni-lateral (Uni-). Measures include: WII balance system (WII); center of pressure (COP); velocity (Vel); total path length (length); sway area (Area); SD (SD); Prior Documented Medical History (History); video head impulse test (vHit); ocular vestibular evoked myogenic potentials (oVEMP); cervical vestibular evoked myogenic potentials (cVEMP); decibels hearing loss (dB HL); excluded (Excl.); threshold (thresh.); Variable not reported = NR

^{Vitkovic et al. (2016)} measured quiet stance in individuals with NH and individuals with HL with and without HA. Participants were instructed to stand as still as possible on a WII for 60 sec, with feet 10 cm apart, EO and EC, on firm and foam surfaces, under four sound conditions, which were then subgrouped post-hoc into sound and no-sound categories. Of the HL participants, those with HA showed decreased total COP sway path in the “sound” compared with the “no-sound” tasks. Note, comparisons between HL and NH participants were not performed due to significant group differences in age, identified post-hoc, that limited further analysis.

^{Negahban et al. (2017)} investigated balance in HL participants with and without HA, using similar stance conditions and instructions as ^{Vitkovic et al. (2016)}, albeit using a forceplate and a shorter duration (30 sec). They observed no significant change in measures of COP sway area but did report a significant decrease in the SD of AP and ML COP velocity when HL participants stood with EO on foam and when their HA were On versus Off. One specific limitation of Negahban et al. was that the HA group was always tested with their HA On first, which introduces the potential for an order effect to confound the data.

^{Maheu et al. (2019)} compared NH participants with and without hearing protection (earplugs and earmuffs) with HL participants with their HA turned On and Off. Participants stood for 60 sec, with feet shoulder width apart, and EO and EC, on firm and foam surfaces. Two key distinctions for this study were the addition of a cognitive task that required participants to count backward from 1000 during each balance task, and the subgrouping of HL participants on the basis of clinically confirmed vestibular loss. The results showed decreased COP sway area in HL participants with vestibular loss compared with NH, and a significant improvement in participants with HL with vestibular loss with their HA-On versus HA-Off. However, no significant difference in postural sway measures was observed in HL participants without HL, with HA-On versus HA-Off.

When examining the evidence as a whole, there does not appear to be strong converging evidence either to support or refute an effect of HA use on static balance control, particularly when other sources of sensory information (vision, proprioception, and vestibular) are available and reliable. The inconsistent results from the three studies reviewed here may be attributed to the large variation in methodologies used by the studies, including the type of measurement devices, cognitive load requirement, sample duration, summary measures used, sound conditions, instructions, and use of age-matched controls between HL and NH groups, which significantly hampers the ability to make any meaningful comparisons across studies. We also must keep in mind the potential for individual differences in these studies. For example, the etiology of the HL (congenital, late onset, conductive, and sensorineural), and degree of hearing and vestibular loss are critical factors that may affect balance, yet the methods used to confirm the degree of hearing and vestibular function in subjects varied widely. Likewise, age and health status of the individual all have important implications for any comparisons between groups and individuals, yet for the studies reviewed, the age groups differed greatly, and health status is rarely measured and reported.

Effects of Cochlear Implants on Balance

CIs are most typically used by individuals who are deaf or who have severe sensorineural HL. Like HA, CIs improve access to sound but do not restore all sound cues/inputs in a way that leads to normal speech understanding or sound localization. Unlike HA, CIs are surgically implanted directly within the cochlea and because of this, there is often consideration given to whether this procedure itself can negatively affect balance by disrupting the vestibular organs, which are in close spatial proximity. Balance is, therefore, sometimes measured preimplantation and postimplantation. CIs are also often the most common intervention to address severe HL that occurs congenitally or early during development, with many etiologies of early/congenital hearing impairments also affecting vestibular functioning (i.e., vestibulocochlear loss; ^{Cushing et al. 2008}). Therefore, CI users are a unique subgroup when considering the hearing-balance connection for a number of reasons. While a large host of studies have assessed static balance measures in adult and adolescent HL patients with CI, there are only five that have used objective sway-based balance measures of stance with a sample duration of >30 sec (see Table 3). Given developmental differences in standing balance due to age, we will review the literature on adult and adolescent CI effects on balance separately.

TABLE 3. - Articles reviewed on the effects of hearing loss with and without cochlear implants

	Adults			Children/adolescents
Article	Shayman et al. (2018)	Oikawa et al. (2018)	Miwa et al. (2019)	Suarez et al. (2007)	Huang et al. (2011)
Normal hearing group (sample number, age [mean/range])	N/A	8 [22/20–24 yrs]	N/A	22 [8–11 yrs]	24 [15 yrs]
Cochlear implant (CI) group (sample number; age [mean/range])	13 [63/23–84 yrs]	8 [44/20–61 yrs]	9 [56/35–79 yrs]	13 [8–13 yrs]; (5 VL)	24 [15/13–17 yrs]
HL etiology	NR	IP (7); Men (1)	IP(5); SO(2); OM(2)	NR	EVA (3); MD (1); CH (2); Men (5); CVI (1); U (12)
CI surgery Info (side; [time from surgery])	Bi-CI [0.5–14 yrs]; n = 10; Uni-CI [0.5–20 yrs]; n = 3	Uni-[3–21 yrs]	Uni- [<1 mo prior/4–94 mo post]	Uni-[>6 mo]	Uni-[>5yr/mean = 8.5 yr]
Foot position	feet together (Romberg)	feet together	NR	NR	NR
Vision: eyes open (EO) or closed (EC)	EC	EO and EC	EO and EC	EO (Firm) and EC (Foam)	EO and EC
Surface type	Firm/foam (subject specific)	Firm	Firm	Firm and Foam	Firm and Foam
Stance instructions	NR	NR	NR	NR	stable as possible
# Trials per condition	3	1	1	1	1
Balance measures	RMS accel, vel, length	COP length, area, mean disp	COP RR-area, RR-length	COP vel, area	COP vel, area
Recording device	IMU (head, lumbar)	Forceplate	Forceplate	Forceplate	Forceplate
Sample duration (sec)	30	60	60	80	60
Sound/noise type	White (0–4000 Hz)	Ambient (125–16,000 Hz) White (70–17000 Hz)	Background	NR	NR
Sound pressure level (dB)	60–70	Ambient ≤ 15/White = 70	50	NR	NR
Sound source	1 speaker, 1 m front	1 speaker, 1 m front	N/A	N/A	N/A
Room acoustics	NR	Anechoic	NR	NR	NR
Test of hearing impairment	Unaided hearing level thresh. < 45 dB in better ear	Aided hearing level thresh. between 23–33 dB HL	Pure-tone audiometry, speech perception, DPOAE	Age-specific audiological testing (details NR)	NR
Test of vestibular impairment	NR	NR	Direct measures-caloric, cVEMP, stabilometry	History and direct measure-passive whole-body (electronystagmography)	History -balance problems, dizziness (NH only-excl.)
Balance changes CI-On vs. Off?	Yes	N/A	Yes	No	No
Balance changes CI vs. NH?	N/A	Yes	N/A	NR	Yes

Etiology types include ideopathic (IP), Meningitis (Men), Sudden Onset (SO), Otitis Media (OM), Enlarged Vestibular Aqueduct (EVA), Mondini Dysplasia (MD), Cochlear Hypoplasia (CH), Congenital Viral Infection (CVI), Unknown (U), Vestibular Loss (VL). Measures include Inertial motion sensors (IMU); center of pressure (COP); root mean square (RMS); acceleration (Accel); velocity (Vel); total sway path length (length); displacement (disp); Romberg Ratio (RR); sway area (Area); normal hearing group (NH); Prior Documented Medical History (History); Distortion Product Otoacoustic Emission (DPOAE) test; cervical vestibular-evoked myogenic potentials (cVEMP); decibels hearing loss (dB HL); excluded (Excl.); threshold (thresh.); Variable not reported = NR.

Cochlear Implant Use in Adults •^{Shayman et al. (2018)} compared balance in 13 patients with bilateral (n = 10) and unilateral (n = 3) CI postsurgery, with the implants turned On and Off, using inertial sensors mounted on the head and trunk while standing with feet-together, with EC for 30 sec. Results showed a significant decrease in estimated mean AP velocity and root mean square acceleration of the head with the implants turned On compared with Off; however, no changes were observed at the lumbar region, suggesting that effects on whole-body postural stability are minimal (^{Shayman et al. 2018; Zhong & Yost 2013}). One major limitation of the study was that the stance conditions were inconsistent across participants. Specifically, the stance condition (EC on firm or foam) for each participant was selected on the basis of the most challenging task that they could complete for the duration of the trial. This creates a highly variable measure of balance that cannot be compared with results from other studies and cannot be used to help understand the interactions between hearing and other sensory inputs within this study.

^{Oikawa et al. (2018)} recorded forceplate measures from eight NH participants and eight patients with unilateral CI turned On, during 60 sec quiet stance trials with EO and EC on a firm surface, with and without sound (white noise from a speaker), in an anechoic chamber. Results showed no significant change in COP path length or COP area between groups or sound conditions. A three-way interaction was observed for mean shift in ML COP displacement, with a shift toward the CI side with EC compared with EO in the no-sound condition.

^{Miwa et al. (2019)} recorded forceplate measures from nine patients with unilateral CI at <1 mo before their surgery and >4 mo postsurgery with CIs turned On and Off. Participants stood for 60 sec on a forceplate with EO and EC, on a firm surface. The Romberg Ratio (EC/EO) was calculated for COP measures of sway area and total sway length. Results revealed Romberg Ratio length and Romberg Ratio area significantly increased with CI Off versus CI On (i.e., greater difference in EC compared with EO sway), and no differences between CI On compared with preoperative control measures. One major limitation of this study is the constant order of testing (CI Off followed by CI On) and lack of a control group for the presurgery versus postsurgery comparison to account for potential order effects of the repeated measures. Also, the Romberg Ratio demonstrates how sound restoration influences the ability to adjust balance to different visual conditions, but does not provide insight into the actual balance changes for a given standing condition.

There are a number of limitations shared across the aforementioned three studies on CI in adults. Most notably these studies (^{Shayman et al. 2018; Oikawa et al. 2018; Miwa et al. 2019}) were all based on relatively small sample sizes for patients with CI (n = 13, 8, 9), which included individuals with diverse HL etiologies and a very wide range of ages that averaged 63, 44, 56 years, respectively.

Cochlear Implant Use in Children •^{Suarez et al. (2007)} examined 36 children with HL, of whom thirteen had unilateral CI surgery >6 mo before testing and 22 age-matched children with NH. Forceplate measures of 80 sec quiet stance were compared between NH and HL participants with EO on a firm surface and EC on a foam surface. The 13 children with CIs were also tested with their CIs turned On and Off during the EC on foam task only. The results showed no significant difference in COP sway velocity or elliptical area for HL patients with CI On versus Off. However, it should be noted that within the sample of 13 children tested with CI On versus Off, 5 patients had clinically confirmed hypoactivity of vestibular function measured 6 mo postsurgery.

^{Huang et al. (2011)} studied balance in CI patients after a relatively long period of postsurgery recovery (>5 yr) compared with NH controls. As a result, participants included in the study by ^{Huang et al. (2011)} were older than those included in the study by ^{Suarez et al. (2007)} (means of 14.6 and 10.2 yr, respectively). Participants were instructed to stand as still as possible for 60 sec on firm and foam surfaces with EO and EC. The results showed a significantly higher sway velocity in the CI On group compared with NH controls. However, there were no significant differences between CI On and CI Off across all tasks and measures.

There are a number of limitations to the studies as a whole for CI influences on static balance control in children and adults. First, direct clinical assessments of vestibular function were only performed presurgery and postsurgery in two studies (^{Suarez et al. 2007; Miwa et al. 2019}), and postsurgery in ^{Huang et al. (2011)}. This is of particular relevance to CI studies, as the effects of CI may be mediated by the concurrent electrical stimulation of residual, or in some cases fully intact, vestibular hair cells (^{Buchman et al. 2004}). Furthermore, most studies had very limited sample sizes, with only one study involving >13 patients with CI. Like the HA studies, there is wide variation in methodologies used including the use of control groups, age of participants, differences in sound sources, summary COP measures, stance positions, standing conditions, HL etiology, and time following surgery. Clinical methods used to confirm the extent of hearing impairment in patients with CI and NH subjects differed greatly across studies, or were reported with limited detail. Therefore, direct comparisons across studies are not possible and meaningful interpretations are difficult. Taken as a whole, the literature suggests there is little evidence for significant differences between NH and patients with CI, or any improvement in children with CI On- versus Off; however, there is significant opportunity and need for further controlled and well-designed studies on this topic.

IMPLICATIONS AND RECOMMENDATIONS FOR APPLICATION AND PRACTICE

Given the growing body of epidemiological evidence demonstrating significant associations between HL and balance instability (^{Viljanen et al. 2009; Chen et al. 2015}), and falls risk (^{Lin & Ferrucci, 2012; Jiam et al. 2016; Gopinath et al. 2016; Agmon et al. 2017}), there is widespread interest in research designed to confirm direct causal links, and elucidate the potential mechanisms underlying hearing-balance interactions. The present review has identified two different research approaches that have been undertaken to examine the effects of HL on balance control. The first approach involves within-subject designs comparing NH participants with and without artificial sound suppression (^{Kanegaonkar et al. 2012; Gago et al. 2015; Vitkovic et al. 2016; Maheu et al. 2017}); and in HL participants with and without artificial sound amplification through HA (^{Vitkovic et al. 2016; Negahban et al. 2017; Maheu et al. 2019}) or CI (^{Suarez et al. 2007; Huang et al. 2011; Shayman et al. 2018; Miwa et al. 2019}). The results of these within-subject experiments are highly variable; irrespective of the type of hearing manipulations, some studies reported significant effects of reduced hearing on whole-body static postural control (^{Kanegaonkar et al. 2012; Gago et al. 2015; Vitkovic et al. 2016; Miwa et al. 2019}), and others demonstrated no significant effect of reduced hearing on an individual’s balance (^{Suarez et al. 2007; Huang et al. 2011; Vitkovic et al. 2016; Shayman et al. 2018}) (NH group; ^{Maheu et al. 2017}), particularly when controlling for vestibular loss (^{Maheu et al. 2019}). The second approach involves between-subject designs that compare balance in individuals with NH and those with HL (^{Suarez et al. 2007; Huang et al. 2011; Oikawa et al. 2018; Maheu et al. 2019}). The results of these between-subject studies are also inconclusive, with some studies reporting significant differences between groups (^{Huang et al. 2011; Oikawa et al. 2018}) and others finding no differences between groups (^{Suarez et al. 2007}), especially when controlling for concomitant vestibular loss (^{Maheu et al. 2019}).

Studies included in this critical review met the basic minimum criteria we determined to be crucial to ensuring reliable static posturographic measures, which were objective measures of sway using ground reaction forces or body-worn sensors recorded during two-legged, feet side-by-side stance, over a minimum duration of 30 sec. However, despite sharing this common base design, any attempt to reconcile the differences in results and find converging evidence across studies was still hampered by the lack of consistent methodology and poor control for individual, environmental, and task-specific factors known to have significant influences on standing balance. In summary, the evidence is weak that hearing suppression, HL, or hearing devices affect postural stability within the context of basic static balance when other sensory inputs are presumed intact/reliable and task demands (sensory, motor, and cognitive) are reasonably low. It is likely that hearing may become more important when other sensory inputs are poor and/or when task demands are high, but there is little empirical evidence to support this and this should be a goal of future research. This goal is particularly critical when considering the ecological validity of the current approaches/measures, which could be improved by better measurements and/or challenging other sensory, motor, and cognitive processes during experimental static and dynamic balance tasks in individuals with HL.

Future studies are clearly needed that adequately control, or account, for the many critical factors that can potentially confound meaningful interpretations and comparisons of results in a way that can also improve ecological validity; this includes considerations of the individual, environment, and task (^{Keidser et al. 2020}). Individual factors are most relevant for studies comparing balance across groups with different levels of hearing abilities. In particular, the studies must rule out other potential sensorimotor or cognitive comorbidities based on valid clinical assessments (not only self-report), that could otherwise mask/mimic the effects of HL on balance. Within-subject designs help control for individual differences (age, health status, cognitive function, and emotional state), yet still, need to provide important individual participant details to facilitate meaningful comparisons across studies and justify the generalisability of the results. This includes details on the etiology of the HL (congenital, late onset, conductive, and sensorineural) as well as degree of HL (partial/complete and unilateral/bilateral), and clinical evidence of vestibular involvement. Studies examining the effect of HA should report the type of HA signal processing, length of time using the device, frequency of use, the quality of the fit, and cleanliness. Studies comparing the effects of CI should report important features about the surgical procedure (time pre-post surgery, side of implantation, degree of vestibular function presurgery and postsurgery), as well as details about the CI itself (device type and length of time using device).

In terms of environmental factors that can influence balance, the most crucial variable to control is the acoustics of the environment in which the experiments are performed. Studies should be designed with careful consideration of the acoustics of the testing area, as well as the source type, location, and amplitude of the auditory stimuli that may or not be controlled by the experimenter. Particular sound cues may have differential effects on balance, depending on whether they are considered relevant to balance (i.e., orientation cues), relevant to the performance of a secondary tasks, or irrelevant, and the extent to which the sounds may convey emotional meaning and/or require cognitive processing. Given the multisensory nature of balance control, the environment must also be designed to control for visual, vestibular, and proprioceptive cues, that will affect the degree to which these sensory inputs are weighted, and indirectly influence the cognitive load attributed to controlling posture in that particular condition.

In terms of the task, the crucial aspects to control for are the stance duration, stance position, and instructions to the participant. Using a sufficient stance duration is most important, as it will directly affect the magnitude and reliability of most COP/COM measures used to characterize balance. Thus, we would recommend future hearing studies follow established recommendations to have individuals stand quietly for at least 60 sec (^{Carpenter et al. 2001; van der Kooij et al. 2011}). Maintaining a narrow width stance (feet-together or width equal to foot-length) is also recommended to ensure an equal postural challenge in AP and ML directions. Unless cognitive load is being specifically manipulated, studies should limit any instructions that direct the participant’s focus of attention toward their balance (i.e., stand as still as possible), or a secondary task (reaction time, counting, and articulation), to minimize the influence of cognitive load. Measures of balance should ideally be recorded using a forceplate, IMUs, or optical motion sensors to provide accurate sway measures in both the AP and ML direction, and support analyses in both the time and frequency domains. Last, the studies should incorporate multiple trials for each stance condition, and counterbalance the order of conditions to minimize potential learning effects or changes in state anxiety and arousal across repeated trials.

In summary, many theories have been developed to explain the potential causal link between HL, balance, and falls risk. However, the results of the present review suggest that there are sufficient gaps in the literature and limitations to many existing studies, individually, and as a whole, to conclude that this issue is still very much unresolved. Of the hypothesized links between HL and falls previously considered in the literature, the measures of balance reviewed in the present article are likely most relevant for considering whether there is possible comorbid vestibular loss with HL and whether HL is associated with increased cognitive load. The extent to which balance measures can be used to reflect on these two hypotheses; however, depends on the nature and demands of the task (e.g., the extent to which the task challenges the vestibular system or introduces high cognitive demands). Therefore, future studies should be carefully designed to account for the wide range of factors known to influence static balance control, and how these factors may map on to the hypothesized links between HL and falls. Studies must also experimentally control or account for key elements involved in normal and pathological hearing to understand how HL may directly or indirectly affect balance control. Further research is also needed to examine the effect of HL on dynamic balance control, as well as multisensory interactions and multi-tasking challenges, that will provide more realistic conditions needed to obtain more ecologically valid findings and to unravel the complex relationship between HL and balance control. Ultimately, the outcomes of this research could be used to provide interprofessional guidelines regarding how, for instance, teams of health professionals (e.g., audiologists and primary care physicians) and hearing device developers could implement these findings into practice, thereby promoting mobility-related safety for their clients with HL (^{Campos & Launer, 2020}).

ACKNOWLEDGMENTS

Authors thank Rocio Hollman and Kyle Johnson for their assistance with the literature search and data analysis.