Development of a clinical static and dynamic standing balance measurement tool appropriate for use in adolescents.

Balance training is a key component of rehabilitation following sports injury. (1-4) It also is quickly gaining recognition as a vital component of injury prevention programs for many athletes, including adolescents. (5-11) Currently, there is no "gold standard" for the measurement of standing balance in the young active population. Accurate measurement of standing balance is essential in assessing the effectiveness of balance training.

Balance can be defined as the ability to maintain the body's center of gravity over its base of support with minimal sway or maximal steadiness. (12,13) There is some evidence to suggest that decreased static unipedal balance is a risk factor for ankle sprain reinjury in soccer. (14,15) In sports, an athlete is usually visually attentive to the game, and the activity is dynamic in nature at the time of injury. Some authors, (16,17) therefore, agree that impaired dynamic unipedal balance may be more critical than static balance in sports.

There is some evidence that static unipedal balance ability does improve following balance training using a wobble board. (1-4,18-20) Most of these studies, however, exclusively examined improvement following ankle injury. Other studies (5-10) have demonstrated that balance training is effective in preventing sport-specific injury. Measurements of balance often are not examined in these prevention studies. Consequently, the effect of these training programs on balance remains unclear.

Factors that may influence balance ability, and thus any measurement of balance, must be considered in examining balance as an outcome measurement in rehabilitation or as a risk factor for injury in sports. (21) Factors to be considered include: leg dominance, fatigue or learning effects, (22-24) age, (25-31) sex, (26) height, (31) weight, (31) foot size, (32) physical activity level and specificity, (33) and previous lower-extremity injury. (2,4,27,34-37)

Numerous techniques have been described to measure standing balance, with varying levels of challenge in different populations. Laboratory balance measures (eg, stabilometry, accelerometry, motion analysis) use equipment that is costly, highly technical, and often not portable. The test-retest reliability of the measurements obtained with this equipment is extremely variable. (10,23,38-46) Some measurement tools have been developed for use in the clinical setting, but many of these tools were developed for use in elderly people and people with neurological impairments. (21,47-62) These tools, however, are arguably not challenging enough for adolescents without neurological impairments.

Adequate test-retest reliability for timed static unipedal balance has been reported in both children and adults, (19,60) To date, attempts to establish adequate reliability of a dynamic balance test using a tilt board or a hop-stabilization test have not been successful. (17,60,63) The use of a foam or narrow support surface for the measurement of balance relies on an observer scoring system that includes observing sway (ie, minimal, moderate, and large), movement strategy (ie, control of balance primarily initiated at the ankle, hip, or trunk), and time. (43,64-66) Variable test-retest reliability (intraclass correlation coefficient [ICC] or r=.05-.83) has been reported. (43,64-66) The use of foam to alter proprioceptive feedback from the support surface and create a more dynamic task may be an appropriate tool for the measurement of timed dynamic unipedal balance in adolescents without neurological impairments.

The goals of our study were: (1) to determine the test-retest reliability of data obtained with a timed static and dynamic unipedal balance test in adolescents without neurological impairments, (2) to investigate limits of static and dynamic balance in these adolescents, and (3) to determine the influence of age, sex, leg dominance, body mass index, foot size (length and width), previous injury, sport participation level, sport participation specificity, and visual feedback on static and dynamic balance ability.

Method

Subjects

The sample was recruited from 15 Calgary Board of Education high schools. We randomly selected the order in which schools were approached to participate using computer generation of random numbers. (67) Four adolescent subjects without neurological impairments (2 male, 2 female) from physical education (PE) program rosters in each grade from 10 to 12 were randomly approached for recruitment. In Alberta, participation in PE is mandatory in grade 10 and elective in grades 11 and 12. If a subject declined participation or dropped out at the time of the baseline assessment, another student (from the same school, grade, and sex) was recruited. Three additional students were tested at baseline because 3 students indicated that they would be absent from school for the 7-day follow-up assessment. School and subject recruitment are summarized in Figure 1.

[FIGURE 1 OMITTED]

Subjects were included if they were between the ages of 14 and 19 years and participated in PE class. Subjects were excluded from the study if they reported a history of a musculoskeletal injury (ie, requiring medical attention and time loss from sporting activity of 1 or more days) in the 6 weeks prior to recruitment, a history of a serious musculoskeletal pathology (eg, fracture, rheumatologic disease, systemic disease, surgery) in the 6 months prior to recruitment, or a serious ongoing medical condition or disability. A description of study subjects and dropout subjects is shown in Table 1.

Procedure

Both the subjects and their parent or guardian completed a written informed consent form. Each subject completed a baseline questionnaire regarding previous history of injury and sports participation. Next, the primary examiner (CAE) measured height (in meters), weight (in kilograms), and foot length and width (in centimeters). Then, each subject completed 1 timed static and 2 timed dynamic unipedal balance tests (eyes-closed static=ECS, eyes-open dynamic=EOD, and eyes-closed dynamic=ECD). The static tests were performed barefoot on a gym floor surface. The dynamic tests were performed barefoot on an Airex Balance Pad. * An Airex Balance Pad is a high-density (50-kg/[m.sup.3]), closed-cell foam pad (47 x 39 x 6 cm, 0.7 kg). The order of leg examination (left and right) for each subject was randomly selected for each test. We also randomized the order of testing of all 3 protocols (ECS, EOD, ECD) with blocks of 6. A 30-second rest was provided between protocols. The timed measurements were completed using a stopwatch. For all 3 balance tests, each subject completed 3 trials on each leg. A 15-second rest was allowed between trials. In addition, a 15-second practice session on the foam pad was allowed prior to the start of the dynamic tests so that subjects could gain some familiarity with this support surface.

The dominant leg was identified by asking subjects to kick a ball hard prior to testing. For all trials, the subjects placed their hands on their hips and time started upon elevation of the opposite foot from the floor. Subjects focused on a target placed at eye level, 4 m in front of them. For both the ECS and ECD trials, eyes were closed prior to elevation of the opposite foot. The maximum time allowed for each test was 180 seconds. This maximum time was based on the findings of Hahn et al (33) in which only 1% of their participants (aged 14-24 years) achieved this time on the ECS unipedal balance test. Time was stopped upon loss of balance or opening of eyes in the eyes-closed trials. Balance times were recorded to the nearest 1/100 of a second. Loss of balance included removal of one hand from the hip, touching the foam or floor with the non-weight-bearing foot, movement of the weight-bearing foot from its original position on the floor or foam, or movement of the foam from its original position in the dynamic balance tests. The testing procedure was repeated by the primary examiner (CAE) 7 days later.

Data Analysis

Data analysis was performed using the Stata statistical software package (Release 5.0). (67),([dagger]) Descriptive statistics are used to describe the subjects who

participated in our study. Data were transformed logarithmically if the assumptions of normality and equal variance were not met for statistical tests. We report geometric means, which typically approximate the median and are the best measure of the central tendency when data are skewed, to allow for calculation of a 95% confidence interval (95% CI). (68) The use of an arithmetic mean to describe central tendency is less appropriate if data are skewed. Geometric means and 95% CIs are back-transformed values from the logarithmically transformed data.

For test-retest reliability, all analyses are based on one examiner's measurements at baseline and follow-up. The main outcome measurements included maximum time achieved over 3 trials for each of 2 legs on each of 3 tests (ECS, EOD, and ECD). One of the weaknesses of the ICC in determining reliability is that, as the between-subject variability of a measurement increases, the estimated ICC also increases. (69) Greater between-subject variability clearly does not indicate increased reliability of that measurement. (68) Analysis using ICCs also fails to examine whether the variability of the measurement (and, as a result, the estimated reliability) is independent of the magnitude of the measurement. (70) In addition, use of ICCs fails to use the units of measurement in question. (70,71) It is thus extremely difficult to make decisions regarding clinical relevance of measurement differences. As such, results based on Bland and Altman's methods of agreement (70,72) were examined in this study. We first plotted the individual subject differences between test sessions against the individual mean scores for both test sessions. (70-72) Uniform scatter of points around the mean difference indicates no association between the differences and the magnitude of the measurement. Our data were not uniformly scattered, and we log-transformed them. The final results were then back-transformed and presented as geometric means. We report the ratio of the follow-up to the initial measurement, which is the geometric mean ratio with its 95% limit of agreement. (72) A geometric mean ratio equal to 1 would indicate perfect agreement. The limit of agreement describes the upper and lower limits of the expected ratio, 95% of the time. Intraclass correlation coefficients also are reported. The ICC (3,1) ([double dagger]) and 95% CI, using the method described by Shrout and Fleiss, (73) were calculated to assess test-retest reliability with multiple scores from the same rater. (73-75) Given the underlying assumptions of repeated-measures ANOVA, log-transformed data were used to estimate ICC. Test-retest reliability was examined for each of 3 unipedal balance stances (ECS, EOD, and ECD).

We used t tests and ANOVA (repeated measures and 1-way) to examine the influence of leg dominance, sex, age, potential learning and fatigue, and sport specificity on all 3 balance tests. We used linear regression to investigate the influence of several other factors--leg dominance, order of testing (blocks 1-6), age (in years), sex, body mass index (in kilograms per square meter), foot length (in centimeters), foot width (in centimeters), previous lower-extremity injury within 1 year, sports participation level (estimated hours per week in previous 6-week period), and sports participation specificity--on ECS and ECD balance.

Those subjects who reached the ceiling maximum of 180 seconds on any of the 3 balance tests (ECS, EOD, or ECD) were excluded from the analyses involving those particular balance tests because the difference between measurements would be based on censored time and would bias any estimates based on those values.

Results

There were no important differences between study subjects and dropouts (Tabs. 1 and 2), suggesting that our study results are not subject to selection bias as a result of dropouts. On the ECS test, 4 subjects achieved the maximum time (180 seconds); therefore, 107 subjects remained in the analysis of reliability for ECS balance. On the EOD test, 33 subjects were excluded from the analysis for the same reason. Consequently, 78 subjects remained in the analysis of reliability for EOD balance. All 111 subjects who completed the study were included in the analysis for ECD balance.

No important differences were found between maximum balance times for left and right legs or between sexes. The geometric means (and 95% CIs) as well as medians (and ranges) for all 3 balance tests are summarized in Table 3. Box plots (Figs. 2 and 3) demonstrate the success of log-transformation in meeting the assumptions of normality for all balance measurements (based on a maximum of 6 trials).

[FIGURES 2-3 OMITTED]

The Bland and Altman plot for ECS balance is presented in Figure 4. In this plot, the funnel shape suggests that the differences are clearly greater with measurements of a greater magnitude. Log-transformation was somewhat successful in producing differences unrelated to the magnitude of the measurement (Fig. 5). The funnel shape has been eliminated, but the scatter resembles an oval shape rather than a completely even scatter. This can be seen visually with a more even scatter of points about the mean. The geometric mean ratio is 0.95 (Tab. 4, 95% limits of agreement=0.28-3.2). This means that the 1-week follow-up measurement yielded an ECS maximum on average 5% less than baseline. The limits of agreement indicate that 95% of the time the follow-up measurement should be between 0.28 and 3.2 times that at baseline.

[FIGURES 4-5 OMITTED]

The Bland and Altman plot for EOD balance is presented in Figure 6. Log-transformation was highly successful in producing EOD differences unrelated to the mean (Fig. 7). The geometric mean ratio was 0.88 (Tab. 4, 95% limits of agreement=0.25-3.2). This means that the 1-week follow-up measurement yielded an EOD maximum on average 12% less than baseline. The limits of agreement indicate that 95% of the time the follow-up measurement should be between 0.25 and 3.2 times that at baseline.

[FIGURES 6-7 OMITTED]

The Bland and Altman plot for ECD balance is presented in Figure 8. As with the EOD test, log-transformation was highly successful in producing ECD differences unrelated to the mean (Fig. 9). The geometric mean ratio was 1.05 (Tab. 4, 95% limits of agreement=0.48-2.29). This means that the 1-week follow-up measurement yielded an ECD maximum on average 5% greater than baseline. The limits of agreement indicate that 95% of the time the final measurement should be between 0.48 and 2.29 times that at baseline.

[FIGURES 8-9 OMITTED]

For all 3 balance tests (ECS, EOD, and ECD), ICCs (3,1) and 95% CIs were calculated on log-transformed maximums for baseline and follow-up assessments. The results of all analyses examining reliability are summarized in Table 4. Based on the 95% limits of agreement, examples of baseline measurements for ECS, EOD, and ECD balance and expected measurements 1 week later are given in Table 5.

We found no evidence that learning or fatigue based on order of testing the 3 protocols influenced our findings or that there were any differences among trials 1, 2, and 3 for ECS or EOD balance. There was an apparent learning effect over 3 trials for ECD balance, however (repeated-measures ANOVA: F=4.69; df=2,244; P=.01).

The final regression model examining factors influencing ECS balance was: logECS=3.309-0.514(Injury), [r.sup.2]=0.1, where logECS denotes the log-transformed ECS balance maximum at baseline and Injury indicates previous lower-extremity injury within 1 year (0=no injury and 1=injury). The coefficient associated with previous injury indicates that static balance in adolescents with a 1-year history of lower-extremity injury was less than that of those without a previous history of lower-extremity injury (-0.514, 95% CI= -0.899 to -0.13). Based on multiple linear regression analysis, no associations were found between EOD balance and other potentially influencing factors.

The final regression model examining factors influencing ECD balance was: logECD=l.699-0.193(Injury), [r.sup.2]=0.03, where logECD=log-transformed ECD balance maximum at baseline. The coefficient associated with previous injury was less than 0, indicating that dynamic balance at baseline in adolescents with a 1-year history of lower-extremity injury was less than that of those without a previous history (-0.193, 95% CI=-0.376 to -0.01). Log-transformed ECS and ECD balance from these regressions were back-transformed to estimate the predicted ECS balance and ECD balance in adolescents with a 1-year history of lower-extremity injury compared with those with no history of lower-extremity injury (Tab. 6).

We also examined the effect of sport specificity on our results. Subjects were grouped by their number one sport for estimated hours spent per week beyond PE class in the past year. Based on 23 different sports reported, no differences were found between groups for log-transformed ECS or ECD balance.

Discussion

This study is the first of its kind to examine timed dynamic balance measurements in adolescents without neurological impairments, using an Airex Balance Pad for a support surface. The Airex Balance Pad is readily available, easy to clean, easy to transport, and resilient to deformation. As such, it is extremely useful in a sport setting or school setting. We determined that timed ECD unipedal balance is an appropriate and adequately reliable clinical measure of standing balance in adolescents without neurological impairments.

Other studies (26,33,62) have examined timed ECS balance in subjects without neurological impairments. Hahn et al (33) found that 1.8% of their sample (competitive athletes aged 14-24 years) reached the maximum time of 180 seconds, which is similar to the findings in our study (1.6%). In our study, the geometric mean for ECS balance based on maximum time attained over 6 trials (3 on each leg) was 25.43 seconds (Tab. 3, 95% CI=22.06-29.31). This finding is in accordance with other research, which demonstrated a mean of 29 seconds based on maximum time achieved over 2 trials. (33) Ekdahl et al (26) demonstrated a mean time of 44 seconds in people 20 to 29 years of age, based on maximum times achieved over 3 trials. In a study by Bohannon et al, (25) the maximum time set was 30 seconds. Only 25% of subjects who were between the ages of 20 and 29 years failed to achieve the 30-second maximum. In our study, 55% of the subjects did not achieve 30 seconds at the baseline assessment. The differences among studies may be related to age and the potential inappropriate use of an arithmetic mean if data were skewed.

Common guidelines for the interpretation of reliability based on ICCs are: <.4=poor, .4 to <.75=moderate, .75 to <.9=good, and [greater than or equal to] .9=excellent. (73,74) Variable test-retest reliability of a timed ECS balance test has been demonstrated previously using the ICC alone. Atwater et al (60) examined children 4 to 6 years of age (ICC=.59-.77), Balogun et al (18) examined young adults without neurological impairments (ICC=.96), and Bohannon et al (61) examined adults following stroke (ICC=.44-.75). Differences among these studies included age and disability. The present study demonstrated adequate test-retest reliability for ECS balance, based on ICC alone, consistent with these other studies (ICC=.69, 95% CI=0.57-0.78). Based on the wide limits of agreement, however, caution is needed in interpretation of ECS balance for people in clinical practice (Tab. 4). For example, an adolescent without neurological impairments may demonstrate "average" balance on the ECS balance test one week (ECS=26 seconds), and his or her expected performance on the ECS balance test the following week might be anywhere between 7 and 85 seconds. As such, the use of ECS balance as an outcome measure of change on an individual level is limited, although it may be appropriate for measuring change at a group level.

Riemann et al 43 demonstrated no difference (F=1.08; df=1,10; P=-.358) in repeated testing of dynamic balance on a foam surface using an error scoring system measured 1 day apart. In our study, dynamic balance measurements (EOD and ECD) appeared to have moderate and poor reliability based on the ICCs alone (Tab. 4, EOD: ICC=.59, 95% CI=0.43-0.71; ECD: ICC=.46, 95% CI=0.31-0.59). However, reliability based on limits of agreement demonstrated sufficiently narrow limits of agreement for ECD balance to be considered in clinical practice. For example, an adolescent without neurological impairments may demonstrate "average" balance on the ECD balance test one week (ECD=5 seconds) and his or her expected performance on the ECS balance test the following week, with no intervention, could be anywhere between 2 and 11 seconds. The reliability of these measurements appears to be greater for relatively low balance ability and decreases as balance ability improves (Tab. 5). Using the limits of agreement presented in Table 5, a clinician may determine the expected change in ECD balance over a 1-week interval. Improvement would be assessed based on change beyond the upper limit of agreement. In the case of ECD balance, improvement would be a follow-up performance of greater than 2.28 times baseline ability.

One of the weaknesses of the ICC in determining reliability is that, as the between-subject variability of a measurement increases, the estimated ICC also increases. (68) Greater between-subject variability clearly does not indicate increased reliability of that measurement. (68) In our study, between-subject variability of the ECD measurement was very small (Tab. 3, range=2.38-19.63, with only 3 subjects exceeding 12 seconds). This may have contributed to a poor ICC for this test.

In the EOD test, 18% and 24% of subjects achieved the maximum of 180 seconds in each of 2 test sessions. This ceiling of 180 seconds limits our ability to examine changes in EOD balance over time. As such, EOD balance is not the most suitable clinical examination for dynamic balance in this population, unless the maximum time is extended and reliability is further examined.

There are other limitations that may have contributed to the moderate test-retest reliability found in our study. One week between assessments allows time for potential practicing of balance. Testing was performed on the same weekday and time of day for both assessments; however, it is possible that physical activities extraneous to our study may have affected balance ability at one session and not another. Adolescents also may be influenced by boredom during the testing session, peer pressure, or limited attention span, which may influence the reliability of these measurements.

Diminishing visual feedback with eyes closed consistently results in decreased postural stability in comparison with eyes-open conditions. (2,25-27,33,62,69,76-78) The results of our study are consistent with these findings. Based on geometric means, reported balance ability in eyes-open conditions exceeds that of eye-closed conditions. Other studies (4,25,27,33,62,76,78) also have failed to demonstrate a difference in balance ability between the dominant and nondominant legs in subjects without neurological impairments. In addition, we found no association between balance and age in our study. This finding is consistent with the results of the study by Hahn et al. (33) Most of the studies demonstrating that ECS balance decreases with age examined this relationship over a wider age range (ie, adolescent to elderly, child to adolescent). (25,26,62,79) Consistent with other research, (26,29,33,78,80,81) our results also failed to demonstrate any relationship between balance ability and sex in adolescents and adults. Some studies (26,79,82,83) have demonstrated a relationship between sex and balance ability in elderly people and in young children, with female subjects demonstrating better balance ability than male subjects.

In theory, factors that lower the center of gravity (ie, decreased height) and increase the base of support (ie, foot size) will increase postural stability. (13) Odenrick and Sandstedt (31) found both height and weight to be predictors of increased postural sway in boys and girls aged 5 to 15 years. Habib and Westcott (32) found that increased foot length was associated with greater balance ability in children. The age of their subjects differed considerably from that of the subjects in our study. Consistent with our study, however, Peeters et al (79) and Ekdahl et al (26) demonstrated that height and weight bad no direct influence on balance.

Our study failed to demonstrate an association between either hours of sport participation or sport specificity and balance ability. Ekdabl et al (26) also failed to demonstrate an association between postural stability and leisure activities. Hahn et al (33) demonstrated that timed unipedal balance was not associated with type of sport, but was positively associated with hours per week of basketball and number of years of basketball and was negatively associated with hours of swimming.

We found no learning or fatigue effects over repeated trials on the same day for ECS and EOD balance. For ECD balance, however, there was evidence of a learning effect over 3 trials. Further examination of a potential learning effect beyond 3 trials, by increasing the number of repetitions, is recommended for future study. This may be related to the increased difficulty associated with ECD balance. Other studies (22,24,26,84) have demonstrated learning effects over more than 3 repeated trials using laboratory measurements. In our study, there was no evidence of fatigue based on 6 possible orders of ECS, EOD, and ECD tests.

Previous lower-extremity injury (1-year history) appeared to decrease both ECS and ECD balance in our study. Previous injury had no effect on EOD balance, suggesting that vision may have compensated for the effects of previous injury. The difference seems to be more significant based on ECS balance than ECD balance. This is likely related to the low intersubject variability in ECD balance measurement. Differences in balance measurements between previously injured and uninjured athletes also are consistently reported in the literature. (2,14,19,34,37,35,36,85-87)

One of the major strengths of our study is the random recruitment of schools and subjects, which increases the generalizability of the study results. Given that grade 11 and 12 PE is elective in Alberta, however, the generalizability is potentially limited to a more active and healthy population of adolescents. The geometric means provided in our study will give the clinician a relative comparison with a random sample of adolescents without neurological impairments. Other strengths of our study include the high rate of consent to participate (>98%). In addition, the dropout rate was extremely low (<10%). Both of these factors limit selection bias. This study also confirms the need to consider alternate and more appropriate statistical methods, in addition to the commonly used ICC, in the assessment of reliability of outcome measurements in sports medicine.

Future research may include examination of the test-retest reliability of ECS and ECD balance at a shorter interval than 7 days. In addition, EOD balance should be examined without a set ceiling maximum of 180 seconds. Concurrent validity of these timed balance measurements also may be examined by comparison with laboratory stabilometry techniques. Previous lower-extremity injury should always be considered in future research examining timed balance measurements in adolescents.

Conclusion

Timed ECS and ECD balance (using an Airex Balance Pad for base of support), with a 180-second maximum for each test, are appropriate clinical balance measures for use in adolescents. Reliability for both ECS and ECD balance appear to be adequate based on ICCs. Based on wide limits of agreement using Bland and Altman methods, however, caution is needed when examining ECS balance at an individual level in clinical practice. Bland and Altman (70,72) methods of agreement demonstrate sufficient reliability for ECD balance only. Test-retest reliability was moderate for EOD test; however, more than 24% of subjects achieved the maximum of 180 seconds on this test. Consequently, EOD balance using this ceiling maximum is inappropriate for use as a clinical balance measurement in this population. In future research examining balance, it is critical to consider previous lower-extremity injury as a key factor influencing balance.

The authors declare no known conflicts of interest. Specifically, they are not paid consultants nor shareholders in Fitter International Inc and have no vested interest in their products (Airex Balance Pads) or sales.

This article was received February 19, 2004, and was accepted November 9, 2004.

References

(1) Gauffin H, Tropp H, Odenrick P. Effect of ankle disk training on postural control in patients with functional instability of the ankle joint. Int J Sports Med. 1988;9:141-144.

(2) Goldie PA, Evans OM, Bach TM. Postural control following inversion injuries of the ankle. Arch Phys Med Rehabil. 1994;75:969-975.

(3) Rozzi S, Lephart SM, Sterner R, Kuligowski L. Balance training for persons with functionally unstable ankles. J Orthop Sports Phys Ther. 1999;29:478-486.

(4) Tropp H, Ekstrand J, Gillquist J. Factors affecting stabilometry recordings of single limb stance. Am J Sports Med. 1984;12:185-188.

(5) Bahr R, Lian O. A twofold reduction in the incidence of acute ankle sprains in volleyball after the introduction of an injury prevention program: a prospective cohort study. Scand J Med Sci Sports. 1997;7: 172-177.

(6) Caraffa A, Cerulli G, Projetti M, et al. Prevention of anterior cruciate ligament injuries in soccer: a prospective controlled study of proprioceptive training. Knee Surg Sports Traumatol Arthrosc. 1996;4:19-21.

(7) Holme E, Magnusson SP, Becher K, et al. The effect of supervised rehabilitation on strength, postural sway, position sense and re-injury risk after acute ankle ligament sprain. Scand J Med Sci Sports. 1999;9: 104-109.

(8) Tropp H, Askling C, Gillquist J. Prevention of ankle sprains. Am J Sports Med. 1985;13:259-62.

(9) Wedderkopp N, Kaltoft M, Lundgaard B, et al. Prevention of injuries in young female players in European team handball: a prospective intervention study. Scand J Med Sci Sports. 1999;9:41-47.

(10) Wester JU, Jespersen SM, Nielsen KD, Neumann L. Wobble board training after partial sprains of the lateral ligaments of the ankle: a prospective randomized study. J Orthop Sports Phys Ther. 1996;23: 332-336.

(11) Laskowski ER, Newcomer-Aney K, Smith J. Refining rehabilitation with proprioceptive training: expediting return to play. Phys Sportsmed. 1997;25:89-102.

(12) Horak FB. Clinical measurement of postural control in adults. Phys Ther. 1987;67:1881-1885.

(13) Shumway-Cook A, Woollacott MH. Motor Control: Theory and Practical Applications. Baltimore, Md: Williams & Wilkins; 1995.

(14) Tropp H, Ekstrand J, Gillquist J. Stabilometry in functional instability of the ankle and its value in predicting injury. Med Sci Sports Exerc. 1984; 16:64-66.

(15) McGuine TA, Greene JJ, Best T, Leverson G. Balance as a predictor of ankle injuries in high school basketball players. Clin J Sport Med. 2000; 10:239-244.

(16) Lephart SM, Pincivero DM, Rozzi SL. Proprioception of the ankle and knee. Sports Med. 1998;25:149-155.

(17) Riemann BL, Caggiano NA, Lephart SM. Examination of a clinical method of assessing postural control during a functional performance task. The Journal of Sport Rehabilitation. 1999;8:171-183.

(18) Balogun JA, Adesinasi CO, Marzouk DK. The effects of a wobble board exercise training program on static balance performance and strength of lower extremity muscles. Physiother Can. 1992;44:23-30.

(19) Hoffman M, Payne VG. The effects of proprioceptive ankle disk training on healthy subjects. J Orthop Sports Phys Ther. 1995;21:90-93.

(20) Mattacola CG, Wills Lloyd J. Effects of a 6-week strength and proprioception training program on measures of dynamic balance: a single-case design. J Athl Train. 1997;32:127-135.

(21) Emery CA. Is there a clinical standing balance measurement appropriate for use in sports medicine? A review of the literature. J Sci Med Sport. 2003;6:492-504.

(22) Holliday PJ, Fernie GR. Changes in the measurement of postural sway resulting from repeated testing. Agressologie. 1979;20:225-228.

(23) Pincivero D, Lephart SM, Henry T. Learning effects and reliability of the Biodex Stability System. J Athl Train. 1995;30:S35.

(24) Nawoczenski DA, Sharp WB, Maiers DJ, et al. Reliability of performance measurements obtained using the Stability Testing and Rehabilitation Station (STARStation). Phys Ther. 1991;71:706-714.

(25) Bohannon RW, Larkin PA, Cook AC, et al. Decrease in timed balance test scores with aging. Phys Ther: 1984;64:1067-1070.

(26) Ekdahl C, Jarnlo GB, Andersson SI. Standing balance in healthy subjects: evaluation of a quantitative test battery on a force platform. Scand J Rehabil Med. 1989;21:187-195.

(27) Goldie PA, Evans OM, Bach TM. Steadiness in one-legged stance: development of a reliable force-platform testing procedure. Arch Phys Med Rehabil. 1992;73:348-354.

(28) Baloh RW, Fife TD, Zwerling L, et al. Comparison of static and dynamic posturography in young and older normal people. J Am Geriatr Soc. 1994;42:405-412.

(29) Colledge NR, Cantley P, Peaston I, et al. Ageing and balance: the measurement of spontaneous sway by posturography. Gerontology. 1994; 40:273-278.

(30) Kolleger H, Baumgartner C, Wober C, et al. Spontaneous body sway as a function of sex, age, and vision: posturographic study in 30 healthy adults. Eur Neurol. 1992;32:253-259.

(31) Odenrick P, Sandstedt P. Development of postural sway in the normal child. Hum Neurobiol. 1984;3:241-244.

(32) Habib Z, Westcott S. Assessment of anthropometric factors on balance tests in children. Pediatric Physical Therapy. 1998;10:101-109.

(33) Hahn T, Foldspang A, Vestergaard E, Ingermann-Hansen T. One-legged standing balance and sports activity. Scand J Med Sci Sports. 1999;9:15-18.

(34) Friden T, Zatterstrom R, Lindstrand A, Moritz U. A stabilometric technique for the evaluation of lower limb instabilities. Am J Sports Med. 1989;17:118-122.

(35) Leanderson J, Wykman A, Eriksson E. Ankle sprain and postural sway in basketball players. Knee Surg Sports Traumatol Arthrosc. 1993;1: 203-205.

(36) Leanderson J, Eriksson E, Nilsson C, Wykman A. Proprioception in classical ballet dancers: a prospective study on the influence of an ankle sprain on proprioception in the ankle joint. Am J Sports Med. 1996;24:370-374.

(37) Zatterstrom R, Friden T, Lindstrand A, Moritz U. The effect of physiotherapy on standing balance in chronic anterior cruciate ligament insufficiency. Am J Sports Med. 1994;22:531-536.

(38) Brouwer B, Culham EG, Liston RA, Grant T. Normal variability of postural measures: implications for the reliability of relative balance performance outcomes. Scand J Rehabil Med. 1998;30:131-137.

(39) Dickstein R, Dvir Z. Quantitative evaluation of stance balance performance in the clinic using a novel measurement device. Physiother Can. 1993;45:102-108.

(40) Kamen G, Patten C, Du CD, Sison S. An accelerometry-based system for the assessment of balance and postural sway. Gerontology. 1998;44: 40-45.

(41) Mattacola CG, Lebsack DA, Perrin DH. Intertester reliability of measuring postural sway using the Chattecx Balance System. J Athl Train. 1995;30:237-242.

(42) Schmitz R, Arnold B. Inter-tester and intra-tester reliability of a dynamic balance protocol using the Biodex Stability System. The Journal of Sport and Rehabilitation. 1998;7:95-101.

(43) Riemann BL, Guskiewicz KM, Shields EW. Relationship between clinical and forceplate measures of postural stability. The Journal of Sport and. Rehabilitation. 1999;8:71-82.

(44) Birmingham TB. Test-retest reliability of lower extremity functional instability measures. Clin J Sport Med. 2000;10:264-268.

(45) Kejonen P, Kauranen K. Reliability and validity of standing balance measurements with a motion analysis system. Physiotherapy. 2002;88: 25-32.

(46) Baker CP, Newstead AH, Mossberg KA, Nicodemus CL. Reliability of static standing balance in nondisabled children: comparison of two methods of measurement. Pediatr Rehabil. 1998;2:15-20.

(47) Berg K, Wood-Dauphinee S, Williams JI, Gayton D. Measuring balance in the elderly: preliminary development of an instrument. Physiother Can. 1989;41:304.

(48) Berg KO, Wood-Dauphinee SL, Williams JI, Maki B. Measuring balance in the elderly: Validation of an instrument. Can J Public Health. 1992;83(suppl 2):S7-S11.

(49) Berg K, Wood-Dauphinee S, Williams JI. The Balance Scale: reliability assessment with elderly residents and patients with an acute stroke. Scand J Rehabil Med. 1995;27:27-36.

(50) Bonanni M, Newton R. Test-retest reliability of the Fukuda Stepping Test. Physiother Res Int. 1998;3:58-68.

(51) Brauer S, Burns Y, Galley P. Lateral reach: a clinical measure of medio-lateral postural stability. Physiother Res Int. 1999:4:81-88.

(52) Duncan PW, Weiner DK, Chandler J, Studenski S. Functional reach: a new clinical measure of balance. J Gerontol. 1990;45:M192-M197.

(53) Duncan PW, Studenski S, Chandler J, Prescott B. Functional reach: predictive validity in a sample of elderly male veterans. J Gerontol. 1992;47:M93-M98.

(54) Mackenzie M. A simplified measure of balance by functional reach. Physiother Res Int. 1999:4:233-236.

(55) Podsiadlo D, Richardson S. The Timed "Up & Go": a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;39: 142-148.

(56) Shumway-Cook, Horak FB. Assessing the influence of sensory interaction of balance: suggestion from the field. Phys Ther. 1986;66: 1548-1550.

(57) Tinetti ME. Performance-oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc. 1986;39:142-148.

(58) Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med. 1988;319: 1701-1707.

(59) Whitney SL, Poole JL, Cass SP. A review of balance instruments for older adults. Am J Occup Ther. 1998;52:666-671.

(60) Atwater SW, Crowe TK, Deitz JC, Richardson PK. Interrater and test-retest reliability of two pediatric balance tests. Phys Ther. 1990;70: 79-87.

(61) Bohannon RW, Walsh S, Joseph MC. Ordinal and timed balance measurements: reliability and validity in patients with stroke. Clin Rehabil. 1993;7:9-13.

(62) Stones MJ, Kozma A. Balance and age in the sighted and blind. Arch Phys Med Rehabil. 1987;68:85-89.

(63) Broadstone BJ, Westcott SL, Deitz JC. Test-retest reliability of two tiltboard tests in children. Phys Ther. 1993;73:618-625.

(64) Crowe TK, Deitz JC, Richardson PK, Atwater SW. Inter-rater reliability of the pediatric clinical test of sensory interaction for balance. Physical and Occupational Therapy in Pediatrics. 1991;10:1-4.

(65) Deitz JC, Richardson P, Atwater SW, et al. Performance of normal children on the Clinical Test of Sensory Interaction for Balance. Occupational Therapy Journal of Research. 1991; 11:336-356.

(66) Horneij E, Holmstrom E, Hemborg B, et al. Inter-rater reliability and between-day repeatability of eight physical performance tests. Advancements in Physical Therapy. 2002;4:146-160.

(67) Stata Statistical ,Software. Release 5.0 [computer program]. College Station, Tex: Stata Corp; 1997.

(68) Ahman DG. Practical Statistics for Medical Research. London, United Kingdom: Chapman & Hall; 1991.

(69) Bland JM, Altman DG. Measurement error and correlation coefficients. BMJ. 1996;313:41-42.

(70) Bland JM, Altman DG. Measuring agreement in method comparison studies. Statistical Methods in Medical Research. 1999;8:135-160.

(71) Atkinson G, Nevill AM. Statistical methods for assessing measurement error (reliability) in variables relevant to sports medicine. Sports Med. 1998;26:217-238.

(72) Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986; 1:307-310.

(73) Shrout PE, Fleiss JL. Intra-class correlations: uses in assessing rater reliability. Psychol Bull. 1979;86:420-428.

(74) Fleiss JL. The Design and Analysis of Clinical Experiments. New York, NY: John Wiley & Sons Inc; 1986:1-32.

(75) Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. 2nd ed. Upper Saddle River, NJ: Prentice Hall; 2000.

(76) Goldie PA, Bach TM, Evans OM. Force platform measures for evaluating postural control: reliability and validity. Arch Phys Med Rehabil. 1989;70:510-517.

(77) Lee DN, Lishman JR. Visual proprioceptive control of stance. Journal of Human Movement Studies. 1975;1:87-95.

(78) Stribley RF, Albers JW, Tourtellotte WW, Cockrell JL. A quantitative study of stance in normal subjects. Arch Phys Med Rehabil. 1974;55: 74-80.

(79) Peeters H, Breslau E, Mol J, Caberg H. Analysis of posturographic measurements on children. Med Biol Eng Comput. 1984;22:317-321.

(80) Black Ft, Wall C III, Rockette HE Jr, Kitch R. Normal subject postural sway during the Romberg test. Am J Otolaryngol. 1982;3: 309-318.

(81) Brocklehurst JC, Robertson D, James-Groom P. Clinical correlates of sway in old age: sensory modalities. Age Ageing. 1982;11:1-10.

(82) Mechling RW. Objective assessment of postural balance through use of the variable resistance balance board. Phys Ther. 1986;66: 685-688.

(83) Overstall PW, Exton-Smith AN, Imms FJ, Johnson AL. Falls in the elderly related to postural imbalance. Br Med J. 1977;1:261-264.

(84) Ageberg E, Zatterstrom R, Moritz U. Stabilometry and one-leg hop test have high test-retest reliability. Scand J Med Sci Sports. 1998;8: 198-202.

(85) Tropp H, Odenrick P. Postural control in single-limb stance. J Orthop Res. 1988;6:833-839.

(86) Tropp H, Odenrick P, Gillquist J. Stabilometry recordings in functional and mechanical instability of the ankle joint. Int J Sports Med. 1985;6:180 -182.

(87) Hoffman M, Schrader J, Koceja D. An investigation of postural control in postoperative anterior cruciate ligament reconstruction patients. J Athl Train. 1999;34:130-136.

* Alcan Airex AG, Speciality Foams, CH-5643 Sins, Switzerland.

([dagger]) StataCorp LP, 4905 Lakeway Dr, College Station, TX 77845.

([double dagger]) ICC (3,1) is based on Shrom and Fleiss model 3 (based on repeated-measure analysis of variance [ANOVA] where tested raters are the only raters of interest), form 1 (single measurement is the unit of analysis rather than a mean). Calculations are based on a repeated-measures ANOVA with one fixed effect for one rater.

CA Emery, BSc(PT), MSc(Epidemiology), PhD, is an Assistant Professor and Head Physiotherapist, Sport Medicine Centre, Faculty of Kinesiology, University of Calgary, 2500 University Dr NW, Calgary, Alberta, Canada T2N 4E4 ([e-mail]). Address correspondence to Dr Emery.

JD Cassidy, PhD, DrMedSc, is Professor, Public Health Sciences, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada, and Senior Scientist, Division of Outcomes and Population Health, Toronto Western Research Institute, University Health Network.

TP Klassen, MD, MSc, FRCPC, is Director, Alberta Research Centre for Child Health Evidence, and Professor and Chair, Department of Pediatrics, Faculty of Medicine, University of Alberta, Edmonton, Alberta, Canada.

RJ Rosychuk, PhD, is Assistant Professor, Department of Pediatrics, Faculty of Medicine, University of Alberta.

BH Rowe, MD, MSc, CCFP(EM), FCCP, is Professor and Research Director, Department of Emergency Medicine, Faculty of Medicine, University of Alberta.

All authors provided concept/idea/research design. Dr Emery and Dr Rowe provided writing. Dr Emery provided data collection, and Dr Emery, Dr Cassidy, and Dr Rosychuk provided data analysis. Dr Emery provided project management. Dr Cassidy, Dr Kiassen, Dr Rosychuk, and Dr Rowe provided consultation (including review of manuscript before submission). The authors acknowledge the Calgary Board of Education high school principals and physical education teachers. Without their support, this research would not have been possible. They are especially grateful to the many high school students who consented to participate in this study. They acknowledge the financial support of the Department of Pediatrics, University of Alberta.

Dr Emery was supported by the Canadian Institutes of Health Research (CIHR) in partnership with the Physiotherapy Foundation of Canada, the Alberta Heritage Foundation for Medical Research (AHFMR), the CIHR Institute of Musculoskeletal Health and Arthritis, Bone and Joint Health Training Program, and the Walter Johns Graduate Scholarship Fund, University of Alberta. Dr Cassidy was supported by AHFMR as a Health Scholar. Dr Rosychuk was supported by the AHFMR as Population Health Investigator, and Dr Rowe was supported by the CIHR as a Canada Research Chair.

Ethics approval was granted by both the University of Alberta Health Research Ethics Board and Calgary Board of Education.

Table 1.
Comparison of Study and Dropout Subjects

                                 Subjects Who
                                 Completed Study
                                 (n=111)

                                 Mean (95% CI (a)) or
                                 Proportion (95% CI) * or
Covariate                        Frequency (%) ([dagger])

Age (y)                          16.59 (16.4, 16.78) *
Sex                              Male: 56 (50.45%) ([dagger])
                                 Female: 55 (49.55%) ([dagger])
Grade                            Grade 10: 39 (35.14%) ([dagger])
                                 Grade 11: 36 (32.43%) ([dagger])
                                 Grade 12: 36 (32.43%) ([dagger])
Previous injury (lower           15/111=13.51%
  extremity)                     (7.77, 21.31) *
Previous injury (all)            25/111 =22.52%
                                 (15.14, 31.43) *
Height (m)                       1.70 (1.68, 1.72)
Weight (kg)                      68.05 (65.41, 70.68)
Body mass index (kg/[m.sup.2])   23.42 (22.57, 24.28)
Foot length (cm)                 25.24 (24.87, 25.6)
Foot width (cm)                  9.64 (9.51, 9.77)
Sports participation             9.93 (7.98, 11.89)
  previous 6 wk (hr/wk)

                                 Dropout Subjects
                                 (n=12)

                                 Mean (95% CI) or
                                 Proportion (95% CI) * or
Covariate                        Frequency (%) ([dagger])

Age (y)                          16.5 (16.05, 16.95) *
Sex                              Male: 5 (41.67%)
                                 Female: 7 (58.33%) ([dagger])
Grade                            Grade 10: 3 (25%) ([dagger])
                                 Grade 11: 5 (41.67%) ([dagger])
                                 Grade 12: 4 (33.33%) ([dagger])
Previous injury (lower           3/12=25%
  extremity)                     (5.49, 57.19) *
Previous injury (all)            4/12=33.33%
                                 (9.92, 65.11) *
Height (m)                       1.69 (1.64, 1.76)
Weight (kg)                      69.17 (55.8, 82.53)
Body mass index (kg/[m.sup.2])   23.95 (19.58, 28.33)
Foot length (cm)                 25.47 (24.28, 26.66)
Foot width (cm)                  9.64 (9.22, 10.05)
Sports participation             10.2 (7.53, 12.98)
  previous 6 wk (hr/wk)

(a) CI=confidence interval.

Table 2.
Geometric Means (95% Confidence Interval [CI]) for Comparison of
Study and Dropout Subjects (a)

                  Subjects Who
                  Completed Study           Dropout Subjects
                  (n=111)                   (n=12)

Covariate         Mean (95% CI)             Mean (95% CI)

ECS balance (s)   25.57 (21.91-29.85) (b)   24.17 (17.5-33.38)
EOD balance (s)   54.59 (46.92-63.54) (c)   52.7 (34.93-79.52) (b)
ECD balance (s)    5.38 (5.02-5.77)          4.75 (3.83-5.91)

(a) ECS=eyes-closed static, EOD=eyes-open dynamic, ECD=eyesclosed
dynamic.

(b) Two subjects were excluded because they achieved a maximum of 180
seconds.

(c) Twenty subjects were excluded because they achieved a maximum of
180 seconds.

Table 3.
Summary of Balance Tests

                              Geometric Mean (s)    Median (s)
Balance Test                  (95% CI) (a)          (Range)

Eyes-closed static balance    25.43 (22.06-29.31)   26.43 (3.84-157.59)
  (n=107)
Eyes-open dynamic balance     54.4 (47.3-62.61)     58.38 (8.64-174.09)
  (n=78)
Eyes-closed dynamic balance    5.32 (4.98-5.68)      4.94 (2.38-19.63)
  (n=111)

(a) Based on back-transformed log-balance at baseline.
CI=confidence interval.

Table 4.
Test-Retest Reliability (by the Same Rater Over a 1-Week Interval)

                              Geometric Mean Ratio
                              (95% Limits of         ICC (3,1)
Balance Test                  Agreement)             (95% CI) (a)

Eyes-closed static balance    0.95 (0.28-3.2)        .69 (0.57-0.78)
  (n=107)
Eyes-open dynamic balance     0.88 (0.25-3.2)        .59 (0.43-0.71)
  (n=78)
Eyes-closed dynamic balance   1.05 (0.48-2.29)       .46 (0.31-0.59)
  (n=111)

(a) ICC=intraclass correlation coefficient, CI=confidence interval.

Table 5.
Examples of Expected 1-Week Follow-up for All 3 Balance Measures
(in Seconds) Based on 95% Limits of Agreement

                                                         Potential
              95% Limits                                 Interval
              of Agreement                 Expected      Length for
              (Bland and     Baseline      Follow-up     Follow-up
Balance       Altman         Measurement   Measurement   Measurement
Test          (70,72))       (s)           (a) (s)       (s)

Eyes-closed   0.28-3.2         3.8          1.06-12.2     11.14
static                        15.4          4.31-49.28    44.97
                              26.4          7.39-84.5     77.11
                              46.4         12.99-148.5   135.51
                             166.2         46.53-180     133.47

Eyes-open     0.25-3.2         8.6          2.15-27.5     25.35
dynamic                       35.7          8.93-114.2   105.27
                              58.4          14.6-180     165.40
                              94.7         23.68-180     156.32
                             174.1         43.53-180     136.47

Eyes-closed   0.48-2.28        2.4          1.15-5.5       4.35
dynamic                        4.0          1.92-9.1       7.18
                               4.9          2.35-11.2      8.85
                               6.4          3.07-14.6     11.53
                              19.6          9.41-44.7     35.29

(a) Calculated by multiplying the baseline measurement by the limits
of agreement.

Table 6.
Predicted Eyes-Closed Static (ECS) and Eyes-Closed Dynamic (ECD)
Balance Times (in Seconds) by Injury History (a)

              Subjects Who          Subjects Who
              Reported No           Reported Previous
              Previous Lower-       Lower-Extremity
              Extremity Injury      Injury

Balance       Geometric Mean        Geometric Mean
Test          (95% CI)              (95% CI)

ECS balance   27.35 (23.58-31.73)   16.36 (10.52-25.43)
ECD balance    5.47 (5.1-5.87)       4.51 (3.66-5.56)

(a) All values are back-transformed from log ECS and ECD.
CI = confidence interval.