Gait & Posture 31 (2010) 407–414
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Review
Bilateral balance impairments after lateral ankle trauma: A systematic review and meta-analysis Erik A. Wikstrom a,*, Sagar Naik b, Neha Lodha b, James H. Cauraugh b a b
Biodynamics Research Laboratory, Kinesiology Department, University of North Carolina at Charlotte, 9201 University City Blvd, Charlotte, NC 28223, United States Applied Physiology and Kinesiology Department, University of Florida, United States
A R T I C L E I N F O
A B S T R A C T
Article history: Received 6 March 2009 Received in revised form 9 December 2009 Accepted 5 February 2010
Research indicates that balance is impaired in the involved limb following an ankle injury. However, bilateral balance impairments are a viable reason for previous non-significant findings between involved and uninvolved limbs. The purpose of this investigation was to conduct a meta-analysis on studies reporting the effects of lateral ankle trauma on balance of the involved and uninvolved limb after acute ankle injury and chronic ankle instability. Twelve studies qualified for inclusion and assessed static balance for both the involved and uninvolved limbs post-injury and a control group. Meta-analyses calculated standardized mean difference effects and explored moderating variables for the involved and uninvolved limbs relative to controls. A significant cumulative effect size (ES = 0.448, p < 0.00001) indicated that balance of the involved limb is impaired after a history of ankle injury. Moderator variable analysis revealed that both acute (ES = 0.529, p < 0.0002) and chronic (ES = 0.338, p < 0.001) lateral ankle trauma negatively affected balance. Analysis of the uninvolved limb also revealed postural stability impairments (ES = 0.275, p < 0.003). Additional, moderator analysis showed a significant acute effect (ES = 0.564, p < 0.0001), but failed to find significance for individuals with chronic ankle instability (ES = 0.070, p = 0.552). These findings provide strong evidence that balance is bilaterally impaired after an acute lateral ankle sprain. However, these findings suggest that bilateral balance deficits are not present in patients with chronic ankle instability. Based on these findings, the uninvolved limb should not be used as a reference for ‘‘normal balance’’ following an acute lateral ankle sprain. Further, patients with acute lateral ankle sprains should undergo balance training on both limbs. ß 2010 Elsevier B.V. All rights reserved.
Keywords: Ankle sprain Chronic ankle instability Bilateral postural control Systematic review
Lateral ankle sprains are a common consequence of physical activity and sports performance, and frequently residual symptoms arise (e.g. ligamentous laxity, impaired postural control and complaints of the ankle giving way) [1]. Indeed, up to 74% of the individuals who suffer an acute lateral ankle sprain develop residual symptoms defined as chronic ankle instability (CAI) [2]. Unfortunately, the underlying neurophysiological mechanism for CAI remains unclear; however, evidence suggests that both feedback and feed-forward alterations in neuromuscular control may play a role in developing CAI [3,4]. A neurophysiological marker of centrally mediated changes, such as feed-forward alterations, is the presence of bilateral postural control impairments. Specifically, bilateral postural control impairments involve measuring postural control during a single leg stance on both a patient’s involved and uninvolved limb and then comparing the values to performances by a healthy control group. Two recent publications provided robust evidence
* Corresponding author. Tel.: +1 704 687 3764; fax: +1 704 687 3350. E-mail address:
[email protected] (E.A. Wikstrom). 0966-6362/$ – see front matter ß 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2010.02.004
associating an acute lateral ankle sprain with impaired postural control in the involved limb [5,6]. In addition, three recent metaanalyses revealed convincing evidence that postural control is impaired in the involved limb of patients with CAI [6–8]. Moreover, while formulating the discussion of their meta-analysis Wikstrom et al. [6] indicated that bilateral balance deficits were a viable explanation for the lack of significant findings between the involved and uninvolved limbs in previous investigations. Based on the above converging arguments, we asked an important research question: are postural control impairments of the uninvolved limb a function of acute ankle injury or CAI? to answer this question we conducted separate meta-analyses on existing postural control assessments of the involved and uninvolved limbs relative to an uninjured control group. Providing systematic evidence of the presence of bilateral postural control deficits is fundamentally attractive because health care providers frequently use the uninjured limb as a criterion for normal postural control. However, if bilateral balance deficits exist, then using the uninvolved limb may lead to inappropriate conclusions and a hasty return to activity [6]. Thus, quantitatively determining standardized effect sizes will provide empirical evidence regarding the
E.A. Wikstrom et al. / Gait & Posture 31 (2010) 407–414
408
indicative of the time and/or distance a subject spent away from a central point (e.g. center of balance). In addition, given that many studies collected data on multiple dependent measures, we established guidelines for our outcome measure selection process. These guidelines included (a) directional consistency and (b) positive effects on postural instability. Most importantly, following these guidelines ensured that we were logical and consistent in selecting outcome measures as well as cognizant of minimizing selection bias. We purposefully did not choose outcome measures from the original investigation with the largest effect sizes, although this may have happened by chance. While our selected outcome measures may underestimate some identified reported findings, specifically choosing outcome measures with the largest effect sizes would potentially create a significant bias for our meta-analysis, contrary to our stated intention. Further, selecting outcome measures based on effect sizes increases the probability of making a Type II error. Indeed, while independently confirming the outcome measures across the 12 static studies, we found nearly 90% agreement among the investigative team.
existence of bilateral postural control impairments post lateral ankle trauma. Moreover, a comprehensive structured metaanalysis will contribute evidence-based medicine to better enable clinical decisions about restoring or rehabilitating postural control after both acute lateral ankle sprains and onset of CAI. 1. Materials and methods 1.1. Study selection and inclusion/exclusion criteria We conducted a systematic search of databases using three search engines (1980–2009): (a) PubMed, (b) Cochrane database of systematic reviews and (c) Web of Science. Key words used to guide our search for the relevant articles included: ankle instability, ankle sprains/injury, postural control, lateral ankle trauma, bilateral balance deficit/impairment. In addition, references from retrieved papers were scrutinized to identify any articles that were not found through our search of electronic databases. We identified 41 potential articles [1,3,5–43] that investigated postural deficits induced by history of ankle injury. Criteria for inclusion/exclusion were determined by the leading research question: does a history of acute or chronic ankle injury cause bilateral postural control impairments? for inclusion, the studies needed to meet two specific criteria (a) assessment of static postural control deficit for both involved and uninvolved ankles post-acute and/or chronic injury and (b) betweensubjects group experimental design (i.e., both control and injured group). Investigations classified as acute injury studies were operationally defined as those testing patients who had suffered a lateral ankle sprain within six weeks prior to completing the testing protocol. Studies classified as investigating chronic injuries were operationally defined as those using subjects with recurrent lateral ankle sprains, subjects with chronic ankle instability and/or subjects with functional ankle instability. All three terms were allowed because of inconsistent terminology use in the existing literature. Only studies that used static measures of postural control (i.e. maintenance of quiet stance while standing on a single leg) were included. Moreover, investigating measures of static postural control facilitated comparison with previous systematic reviews investigating postural instability post-ankle injury [5–8,29]. Twenty-nine studies were excluded from our analysis for the following reasons: (a) nine review articles [1,3,5–8,21,29,34], (b) 18 articles missing postural control data of the uninvolved limb or control group [9,11–14,16,18–20,22,26,27,30– 32,36,37,39] and (c) two articles failed to provide enough information (i.e. outcome means, standard deviations, p values) to calculate weighted effect sizes [17,28]. On the other hand, Table 1 provides detailed characteristics of the 12 included studies [10,15,23–25,33,35,38,40–43]. A summary of our search and data extraction procedure follows: (a) two authors (EAW and SN) conducted an exhaustive search for the pertinent articles on bilateral postural control deficits in patients with a history of ankle injury, (b) inclusion/ exclusion decisions for each study in the meta-analysis were determined with unanimous consensus and (c) data were extracted and analysed independently by two authors (NL and SN) and confirmed by the two other authors (EAW and JHC).
1.3. Data synthesis and analysis Consistent with Rosenthal’s guidelines [45], synthesis and analysis functions were applied to the current studies. Synthesis function describes the characteristics of each study in terms of the effect size. Analysis function determines the overall effect size by computing the weighted effect sizes [45]. We computed the standardized mean difference for individual effect sizes of the 12 bilateral control deficit studies [44]. Standardized mean difference is a robust and conventional method of analysis that computes the mean effect size for each study by calculating the difference between normalized means [44,46,47]. Two separate meta-analyses were conducted. First, we systematically calculated postural control deficits in the involved limbs of acute and chronic ankle injury participants with balance performances relative to healthy control subjects. Second and more importantly, we determined postural control impairments in uninvolved limbs of acute and chronic ankle injury participants with balance outcomes relative to healthy control subjects. Classifying the 12 studies as acute or chronic ankle injury cases allowed us to conduct an additional analysis for evaluating the contribution of ankle injury history as a potential moderator variable on individual effect sizes. These additional analyses allowed us to determine if balance was impaired on the involved and uninvolved limb following acute lateral ankle sprains and the development of CAI independently. The current meta-analysis was conducted using comprehensive meta-analysis software (Biostat. Inc., Version 2), and effect sizes were interpreted (small = .10–.24, medium = .25–.39, large >.40) according to Cohen [48]. 1.4. Heterogeneity Variation in the outcome measures across studies was determined by computing I2. This technique measures inconsistency across study outcomes due to heterogeneity rather than chance alone [45–47]. Additionally, I2 evaluates variation as a percentage with no consideration to the number of studies included in the meta-analysis. Ideally, lower values of I2 ( 50%). Further heterogeneity information was obtained from calculating traditional meta-analysis Q values on the included 12 studies.
1.2. Outcome measures Specific measures of postural control varied across studies. To determine the effect of ankle injury on postural control deficits of involved and uninvolved ankles, common outcome measures across studies were selected and standardized before analysis. In line with recommendations from the Cochrane Handbook of Systematic Reviews [44], standardized effect sizes were based on one outcome measure per study. The three outcome measures most frequently reported in our 12 bilateral balance impairment articles were: (a) center of pressure area, (b) center of pressure velocity and (c) stability/postural sway index, operationally defined as a measure
1.5. Publication bias and fail–safe analysis To evaluate the presence of publication bias [49] in our sample of studies, we adopted two meta-analytic techniques: (a) funnel plot and (b) fail–safe analysis. A
Table 1 Characteristics of the 12 ankle injury studies used in the present meta-analyses (i.e., bilateral postural control deficits). Study
Control ankle
Injured ankle
Ankle trauma type
Quality score
19.7 1.4 26.0 6.0 20.4 1.5
Acute Acute Acute
72.7 36.4 27.3
22.9 3.2 21.0 3.1 19.7 1.3 23.7 6.6 20.3 1.3 19.2 1.3 25.1 3.9 21.9 3.1 30.0 11.0
Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic Chronic
20.0 25.0 27.8 27.8 47.4 25.0 21.1 22.2 36.8
Total (N)
Mean age (years)
Total (N)
Mean age (years)
Evans et al. [15] Perron et al. [33] Rose et al. [35]
– 36 18
– 30.0 7.0 21.2 1.6
28 34 19
Bernier et al. [10] Hale et al. [42] Hertel & Olmsted-Kramer [23] Hiller et al. [24] Hubbard et al. [25] Lee et al. [41] Mitchell et al. [40] Rozzi et al. [43] Santos & Liu [38]
9 19 9 20 30 8 19 13 16
26.2 2.3 21.1 3.1 22.7 2.6 24.5 9.9 21.3 3.8 19.6 1.6 26.5 3.1 21.2 2.5 31.0 11.0
9 16 15 19 30 8 19 13 19
Table 2 Summary statistics for the 12 studies included in each of two meta-analyses: (a) I: involved limb versus the control limb and (b) II: uninvolved limb versus control limb. The top three studies are acute ankle studies and the bottom nine are chronic ankle studies. SMD = standardized mean difference, CI = confidence interval, COP = center of pressure, BSS = biodex stability system, SD = standard deviation, ML = mediolateral, AP = anterioposterior. Study
Calculation of primary outcome measure
I: Involved limb
II: Uninvolved limb
SMD
SMD
95% CI
95% CI
Average excursion velocity: 15 s stance with eyes open Time taken to complete a dynamic limit-of-stability protocol on the BSS SD of the time and distance away from balance center: 20 s stance with eyes open SD of the time and distance away from balance center: 10 s stance with eyes open
0.391 0.562 0.886 0.394
0.007 0.085 0.211 0.539
0.776 1.040 1.562 1.327
0.391 0.718 0.794 0.062
0.007 0.234 0.125 –0.862
0.776 1.201 1.464 0.986
Average excursion velocity: 15 s stance with eyes open
0.568
0.111
1.246
0.117
0.549
0.782
Hertel & OlmstedKramer [23] Hiller et al. [24] Hubbard et al. [25] Lee et al. [41]
COP Velocity Balance Test duration: level 4 Sway index: single leg stance Sway index: single leg stance: ml tilt COP velocity: rehabilitation group COP velocity: ML direction
Average ML excursion velocity: 10 s stance with eyes open
0.614
0.230
1.459
0.313
0.518
1.144
ML Ankle movement: foot flat COP area Radius of COP distribution
0.773 0.042 1.784
0.122 0.464 0.625
1.423 0.549 2.942
0.239 0.000 0.023
0.869 0.506 1.003
0.391 0.506 0.957
Mitchell et al. [40] Rozzi et al. [43] Santos & Liu [38]
Postural sway: ML direction Sway index: level 2 Balance control
SD of talar movement in ML direction: 5 s stance with eyes open Area of COP excursion (ML AP): 10 s stance with eyes open Difference between each COP X–Y coordinate and mean COP X–Y coordinate: 10 s stance with eyes open Sum of the absolute values of all +x and x coordinates: 35 s stance with eyes open Variance in BSS platform tilt from level in all motions: 30 s stance with eyes open Area of COP excursion (ML AP): 15 s stance with eyes open
0.233 0.356 0.509
0.871 0.419 0.166
0.405 1.131 1.185
0.223 0.208 0.112
0.415 0.562 0.553
0.861 0.979 0.778
Hale et al. [42]
Meta-analysis
SMD: overall
SE
95% CI
I2 (%)
Fail–Safe (N)
I: Involved limb II: Uninvolved limb
0.448 0.275
0.091 0.090
0.269–0.626 0.098–0.452
27.58 1.00
72 11
E.A. Wikstrom et al. / Gait & Posture 31 (2010) 407–414
Evans et al. [15] Perron et al. [33] Rose et al. [35] Bernier et al. [10]
Primary outcome measure
409
410
E.A. Wikstrom et al. / Gait & Posture 31 (2010) 407–414
Fig. 1. The left panel displays a forest plot of studies analysed in the involved (injured) limb versus control limb meta-analysis. Numbers in the column on the far right identify the three acute and nine chronic ankle injury studies. The tick marks in the center of each line represent individual effect sizes and the 95% confidence interval are the distal ends of each line. The two open diamond shapes represent cumulative effect sizes and confidence intervals of the studies listed above (i.e., acute and chronic studies). The black diamond at the bottom of the forest plot refers to the overall effect size of all studies in the meta-analysis and its horizontal tips indicate the associated confidence interval. Because none of the diamond shapes intercept the vertical line of no effect (0.00), each represents significant effect sizes. The right panel shows a funnel plot with relatively no publication bias in the 12 acute and chronic studies comparing the involved (injured) limb and control limb. The observed studies (open circles) were plotted as a function standardized mean difference on the x-axis and standard error on the y-axis. The three black circles on the left side of the funnel (negative effects) were added to balance the open circles in symmetry representing no publication bias.
funnel plot is a scatter plot with treatment effect size assigned to the x-axis and standard error (study size) on the y-axis. A symmetrical distribution of studies around the individual effect sizes indicates little or no publication bias [49]. A fail–safe analysis determines the number of additional non-significant studies required to reverse the overall effect size [50]. This is a classic meta-analytic technique for countering the tendency of submitting and accepting publications because the findings are in a certain direction. 1.6. Quality assessment Finally, each study was rated using the quality assessment instrument and methods described by Arnold et al. [7]. This instrument calculates the quantity of threats to construct, internal and external validity present in each included investigation that are specifically associated with CAI research. A higher score is an indicator of a higher quality investigation with fewer threats to validity. As discussed by Arnold et al. [7] this particular scale was used because traditional quality scales for randomized control trials penalize for not randomizing and data for this meta-analysis were extracted from cross-sectional studies or before randomization occurred. Quality scores of the six [23–25,40–42] investigations included in the current investigation and by Arnold et al. [7] were compared to ensure accurate quality ratings. The results of this comparison indicated that the current quality scores were significantly correlated (r = 0.892) with those of Arnold et al. [7]. Table 1 displays the quality of the 12 bilateral postural control studies.
2.1.2. Publication bias and fail–safe analysis Fig. 1 (right panel) displays a funnel plot to examine the symmetry in the studies. Even though publication bias appears to be minimal, an ideal symmetry is shown by adding three circles on the left side of the funnel (black circles). In addition, the fail–safe analysis indicated a large number of studies (N = 72) required to reduce this standardized mean difference to a null effect. 2.1.3. Moderator variable analysis: type of lateral ankle sprain Analyzing type of lateral ankle sprain as a potential moderator variable revealed a significant medium size for the overall mixed effects model = 0.487 (SE = 0.104, Z = 4.66, p < 0.00001; 95% confidence interval lower limit = 0.282 and upper limit = 0.692). Further, fixed effect model analysis of both types of lateral ankle sprains indicated two significant medium effect sizes (Fig. 1: left panel): (a) acute = 0.529, SE = 0.140, Z = 3.79, p < 0.0002, lower limit = 0.255 and upper limit = 0.803 and (b) chronic = 0.388, SE = 0.120, Z = 3.23, p < 0.001, lower limit = 0.153 and upper limit = 0.623.
2. Results 2.1. Postural deficits meta-analysis I: involved limb versus control limb The fixed effect model meta-analysis of the involved and control limbs indicated a standardized mean difference = 0.448 (SE = 0.091, p < 0.00001) with confidence interval limits of 0.269 and 0.626. This medium effect size is shown in Table 2 and Fig. 1 (left panel) displays a forest plot of the weighted effect sizes for each study. 2.1.1. Heterogeneity Variability calculations on the involved limb and control limb comparisons indicated an I2 = 27.58%. Again, a low heterogeneity value reveals a relatively high amount of consistency in the studies. Consequently, we selected a fixed effect model for analysis. In addition, variability results according to the Q test indicated a homogeneous group of studies (Q = 15.20).
2.2. Postural deficits meta-analysis II: uninvolved limb versus control limb A fixed effect model meta-analysis revealed a standardized mean difference equal to 0.275 (SE = 0.0902, Z = 3.05, p < 0.003). As the last three columns of Table 2 shows, this significant summary effect had a 95% confidence interval of 0.0984–0.452. This standardized mean difference is a relatively small effect found in the static postural control deficit outcomes comparing the uninvolved limb to a control limb. A forest plot (left panel) displays the individual effect sizes in Fig. 2. 2.2.1. Heterogeneity Variability calculations on the 12 studies revealed an I2 = 1%; Q value = 10.65. Such low heterogeneity values indicate that a relatively high amount of consistency in these bilateral balance control studies. Given that heterogeneity is considerably less than 50%, a fixed effect model analysis was reported.
E.A. Wikstrom et al. / Gait & Posture 31 (2010) 407–414
411
Fig. 2. The left panel shows a forest plot of studies analysed in the uninvolved (uninjured) limb versus control limb meta-analysis. Numbers in the column on the far right identify the three acute and nine chronic ankle injury studies. Significant summary effect sizes were found for the acute studies (open diamond = 0.564) and overall effect size (black diamond = 0.275). The right panel displays a funnel plot with relatively no publication bias in the 12 acute and chronic studies comparing the uninvolved (uninjured) limb and control limb. The observed studies, open circles, were plotted as a function of the standardized mean difference on the x-axis and standard error on the y-axis. The two diamonds on x-axis are identical: observed and imputed.
2.2.2. Publication bias and fail–safe analysis Examining the funnel plot (right panel) in Fig. 2 clearly reveals a relatively symmetrical distribution of studies. Thus, we cautiously conclude that publication bias is minimal. Further, the fail–safe analysis determined that 11 null effect findings were necessary to lower the standardized mean difference effect to an insignificant level.
current results clearly indicate postural control impairments on the involved limb despite a myriad of postural control methods and measures, a diverse sample of acute ankle sprain severities, multiple operational definitions and inclusion criteria for subjects with CAI. While the previous investigations [5–8] demonstrate strong evidence showing deficits in the involved limb, none of the studies determined if bilateral postural control deficits existed.
2.2.3. Moderator variable analysis: type of lateral ankle sprain Analyzing the 12 studies for a potential moderator variable indicated that a significant overall mixed effects model equal to 0.275 (SE = 0.090, Z = 3.05, p < 0.003; lower limit = 0.0984; upper limit = 0.452). Separately examining the contribution of the three acute and nine chronic lateral ankle sprain studies in a mixed effects analysis revealed a significant effect for acute studies (0.564, SE = 0.140, Z = 4.03, p < 0.0001; lower limit = 0.290; upper limit = 0.839). However, analysis of the postural deficits of the uninvolved limbs for the CAI studies did not find any significance (0.070, SE = 0.1118, Z = 0.60, p = 0.552; lower limit = 0.16; upper limit = 0.301).
3.2. Postural deficits meta-analysis II: uninvolved limb versus control limb
3. Discussion 3.1. Postural deficits meta-analysis I: involved limb versus control limb A recent systematic review by McKeon and Hertel [5] and metaanalysis by Wikstrom et al. [6] provided compelling evidence that the involved limb of acute lateral ankle sprain patients demonstrated quantifiable deficits in postural control. Similarly, three meta-analyses [6–8] reported convincing evidence indicating deficits in postural control of the involved limb of CAI patients. Indeed, Wikstrom et al. [6] conducted a fail–safe analysis of the 25 included investigations and determined that 707 studies would be needed to find no statistically significant differences between injured and control groups to reverse the finding that various types of ankle trauma (acute, CAI) affect postural control. The current meta-analysis of 12 included investigations support these previous reports and indicate that 72 null findings are needed to conclude that various types of ankle trauma (acute, CAI) do not affect postural control. Furthermore, previous findings [5–8] and the
The current comprehensive meta-analysis indicated that postural control deficits were present on the uninvolved limb of patients with an acute lateral ankle sprain but not in individuals with CAI. This meta-analysis provides evidence that bilateral postural control impairments are present following an acute lateral ankle sprain and support the findings of Evans et al. [15] in a true prospective study indicated that following an acute lateral ankle sprain postural control deficits in the uninvolved limb take approximately seven days to be resolved whereas deficits in the involved limb typically take at least four weeks [15]. Further, our findings support the growing breadth and depth of the literature that indicates centrally mediated changes occur after lateral ankle trauma and suggest that these changes may be the underlying cause of CAI [11,15,23,25,42,51–60]. Centrally mediated changes have been previously reported in a variety of neuromuscular control measures. For example, BullockSaxton et al. [51] and Bullock-Saxon [11] reported altered proximal muscle activation patterns following acute lateral ankle sprains. Additionally, Hubbard et al. [25], Nicholas et al. [59], and Friel et al. [56] all reported hip musculature strength deficits in CAI patients while Sedory et al. [60] revealed bilateral hamstring inhibition in CAI patients. Even though the exact mechanisms of these alterations remain unknown, a plausible explanation focuses on altered motor control programs [3]. The injury literature is replete with studies highlighting multiple altered motor control issues in CAI patients: (a) altered hip biomechanics during dynamic balance tests [57], (b) altered ankle biomechanics during gait [55,58] and (c) different landing patterns [52–54]. Additionally, evidence of centrally mediated changes in postural control following CAI development was demonstrated by Hale et al. [42] when
412
E.A. Wikstrom et al. / Gait & Posture 31 (2010) 407–414
rehabilitative exercises performed only on the involved limb resulted in bilateral postural control improvements. Moreover, Hertel and Olmsted-Kramer [23] found bilateral balance deficits post-CAI development using time-to-boundary but not traditional measures of postural control. These data strongly suggest that spinal-level motor control mechanisms are altered in individuals with CAI. However, a plausible alternative concerns supraspinal aspects of motor control. Changed supraspinal aspects of motor control are viable given that in vivo measurements of sensorimotor function, which are impaired in subjects with CAI, require conscious perception of peripheral information [61]. Despite the above-mentioned evidence, our structured metaanalytic findings on cumulative values of individual weighted effect sizes failed to reveal balance deficits in the uninvolved limb of CAI subjects relative to uninjured healthy control subjects. The lack of significant findings may be because the true effect of CAI may have been underestimated in both the original investigations and the current meta-analysis. Indeed, recent scientific evidence has demonstrated that traditional COP based measures, used in many of the included studies, are not as sensitive to CAI related postural deficits as other postural control measures such as timeto-boundary [23]. Further, inconsistent operational definitions of and inclusion criteria for individuals with CAI may have confounded the current results by comparing individuals with a wide range of CAI severity. In addition, the quality of the included studies may be a reason for the lack of statistical significance. All of the included investigation’s quality scores were lower (
