A meta-analysis and systematic review of changes in joint position sense and static standing balance in patients with whiplash-associated disorder

Objective To synthesise and analyse the current evidence regarding changes in joint position sense (JPS) and standing balance in people with whiplash-associated disorder (WAD) taking the presence or absence of dizziness into account. Data sources PubMed, CINAHL Plus, Web of Science, Embase, MEDLINE and APA PsycINFO were searched by two independent reviewers from inception until August 2020 and reference lists of all included studies were also reviewed. Study selection Only cross-sectional studies that measured JPS and/or standing balance between people with WAD vs. healthy controls (HC) or people with WAD complaining of dizziness (WADD) vs. those not complaining of dizziness (WADND) were selected. Data extraction Relevant data were extracted using specific checklists and quality assessment was performed using Downs and Black Scale (modified version). Data synthesis Twenty-six studies were included. For JPS, data were synthesized for absolute error in the primary plane of movement for separate movement directions. For standing balance, data were synthesized for traditional time- and frequency domain sway parameters considering the conditions of eyes open (EO) and eyes closed (EC) separately. For meta-analysis, reduced JPS was observed in people with WAD compared to HC when the head was repositioned to a neutral head position (NHP) from rotation (standardised mean difference [SMD] = 0.43 [95%: 0.24–0.62]) and extension (0.33 [95%CI: 0.08–0.58]) or when the head was moved toward 50° rotation from a NHP (0.50 [0.05–0.96]). Similarly, people with WADD had reduced JPS compared to people with WADND when the head was repositioned to a NHP from rotation (0.52 [0.22–0.82]). Larger sway velocity and amplitude was found in people with WAD compared to HC for both EO (0.62 [0.37–0.88] and 0.78 [0.56–0.99], respectively) and EC (0.69 [0.46–0.91] and 0.80 [0.58–1.02]) conditions. Conclusion The observed changes of JPS and standing balance confirms deficits in sensorimotor control in people with WAD and especially in those with dizziness.

Introduction 'Whiplash associated disorder' (WAD) is a term to describe symptoms associated with a whiplash injury [1], caused by a sudden acceleration-deceleration movement of the neck, most commonly following a motor vehicle accident. The acute symptoms include neck pain as well as dizziness and pain in other body regions. Symptoms of WAD persist even one year after injury in~50% of people [2], indicating delayed recovery in a significant proportion of patients.
After pain, dizziness is one of the most common complaints in persistent WAD; a symptom likely due to altered cervical afferent input to the sensorimotor control system [3]. People with WAD may demonstrate alterations in joint position sense (JPS) and standing balance [3], two domains of sensorimotor control. Plausible causes of sensorimotor dysfunction include damage to mechanoreceptors due to trauma [4], morphological changes of neck muscles [5], pain and inflammation [6] and activation of sympathetic nervous system as a consequence of high levels of stress [7]. Nevertheless, there is conflicting evidence with some studies reporting impaired JPS and balance [8,9] in people with WAD whereas other studies report no change [10,11]. Thus, there is a need for a systematic review to determine whether these aspects of sensorimotor control are impaired in people with WAD.
Thus far, changes in JPS in people with WAD have been considered in one systematic review by de Vries et al. [12] and balance in two systematic reviews by Silva et al. [13] and Ruhe et al. [14]. However, these reviews were limited to qualitative analysis of the findings [12][13][14], and in one case [12] a substantial number of relevant studies, were not included. These limitations, together with the need to consider newly published research since the last literature search performed for JPS (December 2014 [12]) and for balance (November 2010 [13] and January 2011 [14]), justified the need to conduct the current systematic review that uses a quantitative approach to synthesize the available evidence. The present study specifically investigated the role of WAD and dizziness on the afore-mentioned outcomes by comparing people with WAD and healthy controls (HC) and people with WAD presenting with dizziness (WAD D ) versus those not presenting with dizziness (WAD ND ) due to potential greater deficits in sensorimotor control in those complaining of dizziness.

Eligibility criteria
The PICOS (P: Patient/population; I: Intervention; C: Comparison; O: Outcome; S: Study design) framework, for which we defined P as 'WAD' or 'WAD D ', C as 'healthy controls' or 'WAD ND ', O as 'JPS' or 'balance' and S as 'cross-sectional study design', was utilized to inform the eligibility criteria. The I component was not applicable.
Population. Only studies that measured either 'JPS', 'standing balance' or both in human participants with WAD classified as Grade I, II or III were included. Comparison had to be between people with WAD and HC or between WAD D and WAD ND groups.
Outcome. For JPS, studies were included if they used target matching tasks, i.e. repositioning the head to a neutral or target position, in which the target is achieved by moving the head and neck on the stationary trunk. Studies were excluded if they used trajectory registration tasks that required the participant to follow a visual target, such as Figure of 8 [15,16], The Fly [17,18], or Zigzag [16] tests, as they measure movement sense and not position sense. For standing balance, studies were eligible if they measured postural sway derived from centre of pressure (COP) recordings during bipedal static standing on a force platform. Studies that applied external perturbation using for instance platform translations or rotations were excluded. In case of duplicate publication or publications with similar data, the original version of the article was included in this review.
Study design. Only observational cross-sectional studies, published in English, were included in this review.

Information sources and search strategy
The literature was searched in several electronic databases including PubMed, CINAHL Plus (EBSCOhost), Web of Science, Embase (OVID), MEDLINE (OVID) and APA PsycINFO (OVID) from their inception until August 2020. The search of keywords, across title, abstract and subject headings, and their database-specific variants resulted in a specific search string which can be found in Table 1. In addition to the electronic search, reference lists of selected articles from the original databases were hand searched for additional publications.

Study selection
Two reviewers (MM, DA) independently screened the relevant citations against the pre-determined eligibility criteria. A third reviewer (DF) was consulted in the event of disagreement throughout the stages of the review. Authors were contacted if full texts were unavailable or, if further clarification was required on the study population or outcome.

Data extraction
Two reviewers (MM, DA) extracted data related to demographic and clinical characteristics of patient populations including inclusion/exclusion criteria, sample size, age, height, weight, body mass index, variables matched (or statistically adjusted), WAD grade, time since injury or initiation of symptoms, intensity of pain and pain-related disability. To describe the protocol of the included studies measuring JPS the following data were extracted: measurement method, equipment, start position, movement performed, movement velocity, target position, number of practice and measurement trials and outcome measures. For standing balance, the extracted relevant data included: equipment, test condition, instruction, foot and hand position, duration and number of measurement trials, sampling frequency and outcome measures. Mean and SD or standard error of mean (SEM) of main outcomes were also extracted.

Methodological quality
A modified version of Downs and Black Scale [19], designed for assessing the quality of both randomized and non-randomized studies, was used for quality rating in the present study. The modified version consists of 4 domains (11 items) including quality of reporting (6 items), the generalizability or external validity (1 item), internal validity (3 items) and the adequacy of sample size or study power (1 item) ( Table 2). Two authors (MM, DA) separately assessed the quality of the eligible studies.

Data analysis
For JPS, data were pooled when two or more studies were similar with respect to all of the following aspects: JPS task, type of movement (pure flexion, extension, rotation and side-bending; no complex multi-planar movement) and type of error used for the outcome. For standing balance, when a variety of time-and frequency-domain COP measures were provided in a study, only the traditional time-domain measures (i.e. sway velocity and sway amplitude) were considered for analysis. We reported frequency-domain measures if they were the only outcome measure in the included study. COP data were synthesized in eyes open (EO) and eyes closed (EC) conditions separately.
For meta-analysis, mean, standard deviation (SD) of JPS error and COP parameters and number of subjects were required to be extracted from single groups. To combine mean, SD and number of subjects of two or more subgroups into a single group number, we used the formula recommended by the Cochrane handbook for systematic reviews (https://training. cochrane.org/handbook/current). Where SEM was only reported, SD was calculated. When numerical data was only presented in figures, we used WebPlotDigitizer software (https:// automeris.io/WebPlotDigitizer) to extract the mean and SD or SEM from figure images.
The above variables were entered into Review Manager 5.3 software with the following input parameters: continuous data (data type), inverse variance (statistical method), random effects (analysis model), standardized mean difference (SMD; effect measure), totals and subtotals (totals) and 95% (study confidence interval [CI] and total CI). SMD, CI, p-value and I 2 index were calculated and demonstrated in the forest plot. The effect size represented by SMD was interpreted according to Cohen's suggestion as small, SMD = 0.2; medium, SMD = 0.5; and large, SMD = 0.8 [23]. p-value smaller than 0.05 was considered as significant. The threshold to interpret I 2 , representing the amount of statistical heterogeneity, includes low, I 2 = 25%; moderate, I 2 = 50%; and high, I 2 = 75% [24]. When meta-analysis was not possible, no overall effect was presented in the forest plot. As methodological quality might explain possible heterogeneity among studies, the total quality score was compared using Mann-Whitney U test between studies finding differences in JPS and standing balance between people with WAD and HC compared to those that did not.
3. Are the demographic characteristics of the participants included in the study clearly described?
The question is answered 'yes' if information about age and gender of people with WAD is provided. 4. Are the clinical characteristics of the participants included in the study clearly described?
The question is answered 'yes' if information about injury grade and time since injury/initiation of symptoms (such as pain) and pain or disability level in people with WAD is provided. 5. Is the treatment history of the WAD participants clearly described?
6. Does the study provide estimates of the random variability in the data for the main outcomes?
The question is answered 'yes' if studies have provided quantitative values of the mean and the standard error or standard deviation.

External validity
7. Were the participants that were asked to participate in the study representative of the entire population from which they were recruited?
The question is answered 'yes' if the studies have used several recruitment methods (e.g. self-report, hospital, insurance companies, etc.) [20].

Internal validity/bias & confounding
8. Was an attempt made to blind those measuring the main outcomes? 9. Were the main outcome measures, i.e. proprioception and standing balance, reliable? Joint position sense The question is answered 'yes' if JPS was measured at least 6 times [21]. Standing balance The question is answered 'yes' if bipedal standing balance was measured at least 3 times and/or each trial lasted not less than 90 sec. [22]. 10. Were controls matched with WAD participants in important characteristics?
The question is answered 'yes' if appropriate matching on confounders, i.e. age and gender, was performed or if adjustment for these variables is made in the statistical analysis.

Joint position sense
Description of the participants. All of the 16 included studies assessed the difference between people with WAD and HC, among which only 2 studies assessed the difference between WAD D and WAD ND groups [29,35]. The studies investigating the difference between people with WAD and HC recruited a total of 1068 participants to compare the difference between WAD (N = 601; range: 7-102) and HC (N = 467; range: 11-57) ( Table 3). With the exception of one study that measured JPS within 1 month post-injury [34], the majority of studies assessed participants with chronic pain (>3 months). Pain intensity experienced by the WAD group was reported only in 7 studies [10,15,25,26,33,35,37], using 0-10-point pain rating scales, including visual analogue scale (VAS) and numerical rating scale (NRS). Excluding one study that measured pain during the last week [33], other studies reported current pain, with an average of 5.0 (range over studies: 3.4-6.6). Pain-related disability was reported in 8 studies; of the studies using the Neck Disability Index (NDI) [10,[31][32][33][34], the average score was 36.5% indicating moderate disability.
Description of outcome measures. Two types of target-matching tasks are commonly used to measure JPS; head repositioning to a neutral-head-position (HR-NHP) and head repositioning to a specific target remote from the NHP (HR-Target). In the HR-NHP task, the participant's head is actively or passively brought from the 'start position' (a position far from NHP) to NHP, where NHP is the 'target position'. While holding the head in the NHP for a short period, the participant is asked to memorize this target position. The head is then
Among various instruments used to measure JPS, including an electrogoniometer, laser pointer, optoelectronic and Cervical Range of Motion devices, 9 studies used electromagnetic devices to register head and neck movements. The number of repetitions per a specific targetmatching task varied from 2 to 10, with 3 trials reported with greater frequency. With the exception of 1 study that measured HR-NHP in both EO and EC conditions [8], all others measured JPS in an EC condition.
Meta-analysis could not be performed for 6 studies which assessed HR-NHP from flexion due to either averaging JPS error over different movements [10,25] or absence of AE for the primary plane of movement [27,28,32]. The only study [26] that provided AE of JPS from a pure flexion movement found no between-group difference (Fig 3).
For the HR-Target task, pooled results of 2 studies [15,30] that provided data of repositioning towards 30˚of rotation yielded no significant difference between the two groups (pooled SMD = 0.44 [-0.34-1.23], p = 0.27, I 2 = 55%). However, those with WAD showed larger error compared to HC when data was pooled for 2 studies [8,30] that assessed repositioning toward  (Fig 2). One study that assessed JPS when repositioning toward rotation, flexion and extension [10] did not provide AE values for the primary plane of movement and hence was not included for a meta-analysis. No significant difference was found between the two groups in another study when the target was set at 20˚side-bending [30] (Fig 3).

Difference between WAD D and WAD ND groups.
In both studies that assessed the difference between WAD D and WAD ND groups [29,35], JPS was evaluated using HR-NHP from    (Fig 4).
Description of outcome measures. In order to compare the contribution of sensory information into standing balance between people with WAD and HC, various sources of sensory feedback were manipulated including: (1) visual feedback by occluding the eyes in all 12 studies [9,11,25,36,[38][39][40][41][42][43][44][45] or moving the visual surrounds in 3 studies [11,36,44], (2) proprioceptive feedback by alteration of support surface in 4 studies [9,11,36,44], and (3) combination of visual and proprioceptive feedback by alteration of support surface along with either occluding the eyes in 4 studies [9,11,36,44] or moving visual surrounds in 3 studies [11,36,44] (Table 5). This review analysed between-group differences in the two most common conditions of sensory manipulation, i.e., no-manipulation or EO and EC conditions. Traditional time-domain COP measures used in the included studies included sway area, perimeter of sway area, velocity (mean, anteroposterior [AP] and mediolateral [ML]), path length, path length per second, path length per area, acceleration, amplitude and range. The traditional time-domain COP parameters were categorized into sway velocity (velocity, path length and path length per second) and sway amplitude (sway area, perimeter of sway area,  amplitude and range) [46]. We did not include path length per area and acceleration in either amplitude or velocity categories. If more than one COP parameter was reported in a single study, only one parameter representing COP velocity or amplitude was selected for meta-analysis and changes in other parameters were described narratively. This selection was based on prioritizing total sway over sway in specific directions as well as an arbitrary selection of a COP parameter when several parameters provide the same information, such as path length, path length per second and velocity. Two studies [36,44] only used frequency-domain COP measures to quantify standing balance. The measure used in these studies was sway energy, i.e. the amount of energy of the power spectrum of the COP signal. A force platform was most commonly used to measure standing balance [9,11,25,36,39,41,42,44,45], whereas an electronic baropodometer [38] was used in one study and a videobased camera system [40] in another. The instrument in one study was not clearly described [43]. Foot placement during quiet standing on the force platform differed, ranging from feet together [9,36,[40][41][42] to feet apart (comfortable) [9,11,25,36,44,45]. Few studies provided information regarding the instruction given explicitly to the participants and it varied from 'to stand as still as possible' [9,42,44,45] to 'stand relaxed' [38]. In most studies [9, 11, 38, 39, 41,  PLOS ONE 42,44,45], participants were asked to look at a fixed point during standing. The duration of postural assessment ranged from 30 s to 120 s, with 30 s reported more commonly. The number of measurement trials ranged from 1 to 4 trials, but less than 3 trials was used more commonly. Difference between people with WAD and HC. Meta-analysis of COP measures in the EO condition showed that people with WAD demonstrate significantly larger postural sway in terms of velocity (pooled SMD = 0.62 [0.37-0.88], p<0.001, I 2 = 0) and amplitude (pooled SMD = 0.78 [0.56-0.99], p<0.001, I 2 = 0) compared to HC (Fig 5). From time-domain parameters that were not included in a meta-analysis all except one showed no between-group difference (Fig 6). Among the frequency-domain measures, 3 out of 5 parameters showed larger sway energy in people with WAD compared to HC.
Synthesis of COP parameters in the EC condition showed a significant effect for both sway velocity (pooled SMD = 0.69 [0.46-0.91], p<0.001, I 2 = 0) and amplitude (pooled SMD = 0.80 [0.58-1.02], p<0.001, I 2 = 21%) indicating that people with WAD show larger sway compared to HC (Fig 7). Among time-domain parameters that were not present in the meta-analysis, only 3 out of 10 showed significant findings that were consistent with the overall effect obtained from meta-analysis (Fig 8). Within the frequency-domain measures, 4 out of 5 parameters showed larger sway energy in people with WAD compared to HC.
Difference between WAD D and WAD ND groups. In a single study that compared WAD D and WAD ND groups, larger sway energy was observed in WAD D compared to WAD ND in both EO and EC conditions (Fig 9).

PLOS ONE
For JPS and standing balance, clinical characteristics of people with WAD (injury grade and time since injury or initiation of symptoms and pain or disability level) was reported in a half and a third of studies, respectively. Information regarding treatment history was rarely reported (2/16 and 3/12). With a few exceptions where quantitative values were presented in

PLOS ONE
figures, the remaining studies (15/16 and 8/12) detailed estimates of random variability, i.e. mean and SD or SEM.
Only one of the included studies investigating standing balance and very few (4/16) examining JPS recruited participants using multiple recruitment methods such as self-report, hospital or insurance companies-a factor that limits the generalizability of the findings.
Few studies in both groups of studies blinded the assessors of outcome measures to the participants (4/16 and 2/12). For JPS, less than a third of the included studies (5/16) used at least 6 trials to calculate JPS error. In studies of balance, 3/12 studies reported recording of COP at least 3 times and/or more than 90 sec (per measurement trial), whilst poor reliability was obtained for other studies. People with WAD and HC were comparable (matched or statistically adjusted) in terms of age and sex in approximately half of the studies measuring JPS (9/16) and standing balance (6/12). Sample size was justified in a very limited number of studies (2/16; 4/12).

PLOS ONE
The overall quality was similar between 5 studies examining JPS [8,15,30,35,36] that found at least one significant difference between people with WAD and HC and 6 studies [25,26,29,31,34,37] that did not find any difference between the two groups (Mann-Whitney U = 11.5; p = 0.51). Inspection of individual methodological quality items showed that studies finding an effect obtained lower scores for item #4 (description of clinical characteristics; 1/5 studies reporting a difference vs. 4/6 studies reporting no difference) and item #7 (use of several recruitment methods; 1/5 vs. 3/6 respectively), but higher scores were noted for item #10 (matching groups; 4/5 vs. 3/6). Only one study among the studies which found a difference between WAD and HC used a reliable measure JPS whereas none of studies among those which found no difference used a reliable JPS measure. Even though a similar proportion of studies in both groups described or justified power or sample size, the median sample size of the studies finding no difference (n = 83) was nearly twice that of the studies that did report a difference (n = 43). Specifically, 3 studies finding no difference tested at least 100 participants [29,34,37], whilst only one study [35] which did report a difference, utilized sample of 100 Standing balance Juul-Kristensen (2013) [

PLOS ONE
participants. Due to poor reporting of subject clinical characteristics, we could not verify the role of factors such as WAD grade, pain/symptom duration/intensity or disability level in explaining differences between studies. All studies testing standing balance apart from 1 [45], found at least one significant difference in either the EO or EC conditions between people with WAD and HC, making the comparison of methodological quality meaningless.

Discussion
This review assessed whether JPS and standing balance are altered in people with WAD compared to asymptomatic individuals and whether the extent of change is greater in those presenting with the symptom of dizziness. Results of the meta-analysis showed moderately larger JPS error in people with WAD compared to HC when the head was repositioned to a NHP from extension and rotation or when the head was moved toward 50˚rotation from a NHP. Similarly, WAD D group performed markedly worse compared to WAD ND group when the head was repositioned to a NHP from rotation. Meta-analysis of standing balance studies showed markedly larger and faster sway in people with WAD compared to HC during conditions with the EO or EC. In a single study that took the presence or absence of dizziness into account [44], WAD D group showed larger sway energy compared to WAD ND group.

Joint position sense-Difference between people with WAD and HC
People with WAD performed poorer compared to HC when the head was repositioned to a NHP starting from either rotation or extension. Although repositioning to a NHP from flexion was not significant between groups, this could be due to lack of power as this finding is based on a single study [26] with a relatively small sample size. For target-matching tasks locating the target far from the NHP, people with WAD produced larger errors compared to HC but only when the head was repositioned to 50˚of rotation. This result is in contrast to the results of Kritjansson et al. [15] who found that HR-Target is not as good as HR-NHP to differentiate people with WAD and HC, but the limited number of studies investigating HR-Target tasks makes it hard to draw firm conclusions regarding the discriminative ability of various target matching tasks. The results showed no difference between groups when the head was repositioned to 30˚. Assuming that positions farther from the NHP are less practiced or learned and rely more on proprioceptive input, repositioning the head to 50˚of rotation could better differentiate people with WAD versus HC compared to rotation at 30˚. However, this proposition has no support since the difference between groups was evident when the head is repositioned to the most learned position, i.e. NHP. Taking into account the similar magnitude of SMD between 30˚and 50˚, we would suggest that these differing results are likely due to underpowered studies examining the 30˚of rotation target-matching task. Further studies are needed to examine positioning to different points in range.
Even though the overall quality was similar between studies that did find a difference compared to those that did not, a detailed inspection of other possible explanatory factors showed that the sample size was surprisingly larger in the studies showing no difference compared to the studies demonstrating a difference. This may indicate either stricter inclusion criteria leading to more homogenous groups or stricter experimental control resulting in higher sensitivity in the smaller studies. Recruitment of participants with WAD in specific settings as well as tight control of the WAD and HC groups for age and sex in the studies which did report group differences supports homogeneity of the studied population. Finding a between group difference can also be attributed to the use of more reliable measures of JPS, as suggested by de Vries et al. [12]. In their review, de Vries et al. found that all studies finding a difference between people with neck pain and HC, calculated JPS error over at least 6 trials. In contrast, we found no difference in reliability of JPS error between studies that found an effect and those that did not find an effect. This discrepancy could be attributed to the fact that de Vries et al. combined all the groups (traumatic and non-traumatic neck pain) and outcome measures (AE and CE, movement directions) together.

Joint position sense-Difference between WAD D and WAD ND groups
WAD ND group performed worse compared to WAD ND group only when the head was repositioned to a NHP from rotation but not from extension. One plausible explanation is that the rotation movement is accompanied by activation of the vestibular system in addition to proprioceptive system, in contrast to extension movement where there is less stimulation of the vestibular system [35]. Therefore, in rotation, the mismatch between sensory organs, i.e. abnormal proprioceptive and normal vestibular input, may be more pronounced, resulting in larger JPS error in people complaining of dizziness.

Standing balance-Difference between people with WAD and HC
The results of the meta-analysis showed larger and faster sway in both visual conditions, i.e. EO and EC, in people with WAD compared to HC. As additional sensory input provided by visual feedback is supposed to compensate for impaired proprioceptive feedback in conditions with the EO, observation of similar results between EO and EC came as a surprise. For instance, in contrast to WAD, sensorimotor deficits due to anterior cruciate ligament injury [46] are detected in EC, but not EO conditions. Independence of postural sway to visual feedback in people with WAD may highlight the significant impact of impairment of either cervical afferent input, vestibular input or combination of both on standing balance even when other sources of sensory information are still available.
Absence of a significant effect of WAD on COP parameters not included in the meta-analysis could be attributed to lack of statistical power.

Standing balance-Difference between WAD D and WAD ND groups
WAD D group showed poorer standing balance compared to WAD ND group. This was evident as larger sway energy in those people who complained of dizziness. This indicates higher effort needed by the postural control system to overcome the instability induced by postural fluctuations when dizziness is present.

Sensorimotor control deficits in people with WAD
Damage to any element involved in sensorimotor control can lead to deficits of JPS or standing balance. Nociceptive inputs can induce changes at multiple levels of sensorimotor system including spinal and supra-spinal levels, which may contribute to disturbed sensorimotor deficits [47,48]. Persistence of proprioceptive deficits may occur even after the resolution of acute pain [49]. The contribution of vestibular dysfunction in sensorimotor deficits cannot be ruled out, but the differing results of standing balance between people with WAD and non-traumatic vestibular pathology supports a cervical origin of somatosensory deficits in WAD [36]. People with WAD show larger deficits across the majority of sensory-and surface-type manipulations when compared to people with vestibular disorders and HC [36]. Therefore, the results of the both JPS and standing balance assessments highlight a likely modification of cervical afferent input in people with WAD.
Our findings confirm a greater disturbance of sensorimotor control in people with WAD presenting with dizziness compared to those that do not have dizziness. If dizziness in the WAD D group is mainly attributed to a vestibular disorder, we would expect impaired balance but not necessarily JPS in WAD D compared to WAD ND . Nevertheless, we cannot exclude the presence of vestibular dysfunction, in least in some participants. Further support for a disturbance in cervical afferent input, rather than vestibular, as a cause of sensorimotor deficits in people with WAD originates from findings of altered JPS and standing balance in people with no dizziness, as demonstrated by larger JPS error [29,35] or larger sway energy [44] in those with WADND compared to HC. The symptom of dizziness might be due to, or related to, more severe disturbance of proprioceptive inputs in WAD D group compared to WAD ND group. Poor performance in other domains of sensorimotor control, such as oculomotor control [50], in people with WAD with the symptom of dizziness compared to those without dizziness, further supports the role of altered cervical somatosensory input in developing the symptom of dizziness.
Since some the above-mentioned mechanisms, such as alteration of proprioceptive feedback resulting from pain and/or injury, may have consequences for both JPS and standing balance, we may expect a high correlation between these two measures. However, as demonstrated by Treleaven et al. [51] in people with WAD, a strong linear association between JPS and standing balance does not exist. In other words, poor performance in one domain does not imply poor performance in the other domain. This can be attributed to different neural networks underlying JPS and postural control. For instance, cerebellum activation has been consistently reported in human and animal neuroimaging studies of static postural control [52]. However, findings of neural correlates of proprioception does not support a clear involvement of the cerebellum [53]. Although WAD related impairments may affect both domains, the nature of impairments may differ across people with WAD, and therefore the effects of WAD on JPS and standing balance may differ across individuals. These observations and deductions may suggest that JPS and standing balance are two separate constructs implying the need to assess both domains in clinical practice of people with WAD. Nevertheless, we do not exclude the existence of a relationship, as the moderate effect sizes of both measures JPS and balance measures may hinder detection of such an association in a sample of limited size.
The results of the current review highlight the existence of deficits in JPS and standing balance in people with WAD, especially in those with dizziness. These findings imply the importance of incorporation of assessment and management of these specific domains of sensorimotor control within the clinical management of people with WAD. The interventions should address the causes of altered cervical afferent input, such as pain and inflammation, morphological muscle changes or psychological stress, as well as secondary adaptive changes in the sensorimotor control system, such as altered coordination between cervical, visual and vestibular systems [54].

Limitations
Various tests have been developed to measure JPS, but our review included only studies utilizing a specific test, i.e. target matching tasks by moving the head on a stationary trunk. Due to the uniqueness of different tests, the results of the current review are only applicable to target matching task. However, this is of minor concern as the majority of studies investigating JPS in people with WAD have employed HR-NHP and HR-Target tasks. Furthermore, these tasks have shown higher ability to discriminate between HC and people with WAD compared to other complex JPS tests [15]. We analysed AE as the measure of JPS since it was more frequently reported in the included studies. Preferential analysis of AE may limit extrapolation of study findings to other types of error, i.e. CE and VE. Furthermore, due to heterogeneity of outcomes measures in both JPS and standing stability domains, we analysed some studies using a narrative method which is a less rigorous method compared to meta-analysis. A very limited number of studies addressed the effect of dizziness associated with WAD on JPS and standing balance, which should be considered when interpreting the conclusions.
Few studies reported treatment history in the WAD group which could bias the results because they may have affected standing balance and JPS. Other quality criteria which were not adequately addressed involved describing clinical characteristics, blindness of assessors, reliability of measures, matching the comparison groups and power justification.

Direction for future research
To date, very few studies have measured sensorimotor deficits in the acute phase. With the exception of two studies, one assessing JPS within 1 month post-injury [34] and the other examining standing balance 2.8 months post-injury (on average) [43], all studies in both groups included participants with chronic WAD. Even though the results of these studies indicate disturbance of JPS and standing balance, future studies should be conducted to determine how quickly these deficits occur following the onset of pain/dizziness and to what extent these disturbances tend to increase or decrease over time.

Conclusions
This review found impairment of JPS and standing balance in people with WAD compared to HC. Altered JPS was more pronounced in WAD D group compared to WAD ND group. Even though similar underlying mechanisms are proposed to cause impaired JPS and standing balance, further research is needed to clarify if and how these impairments are related.