TEST-RETEST RELIABILITY OF THE CLOSED KINETIC CHAIN UPPER EXTREMITY STABILITY TEST (CKCUEST) IN ADOLESCENTS

Abstract

Background

The Closed Kinetic Chain Upper Extremity Stability Test (CKCUEST) has been proposed as an option to assess upper limb function and stability; however, there are few studies that support the use of this test in adolescents.

Purpose

The purpose of the present study was to investigate the intersession reliability and agreement of three CKCUEST scores in adolescents and establish clinimetric values for this test.

Study Design

Test-retest reliability

Methods

Twenty-five healthy adolescents of both sexes were evaluated. The subjects performed two CKCUEST with an interval of one week between the tests. An intraclass correlation coefficient (ICC_3,3) two-way mixed model with a 95% interval of confidence was utilized to determine intersession reliability. A Bland-Altman graph was plotted to analyze the agreement between assessments. The presence of systematic error was evaluated by a one-sample t test. The difference between the evaluation and reevaluation was observed using a paired-sample t test. The level of significance was set at 0.05. Standard error of measurements and minimum detectable changes were calculated.

Results

The intersession reliability of the average touches score, normalized score, and power score were 0.68, 0.68 and 0.87, the standard error of measurement were 2.17, 1.35 and 6.49, and the minimal detectable change was 6.01, 3.74 and 17.98, respectively. The presence of systematic error (p < 0.014), the significant difference between the measurements (p < 0.05), and the analysis of the Bland-Altman graph infer that CKCUEST is a discordant test with moderate to excellent reliability when used with adolescents.

Conclusion

The CKCUEST is a measurement with moderate to excellent reliability for adolescents.

Level of Evidence

2b

INTRODUCTION

Evaluation tests of the upper limb are widely used in clinical and sports practices to provide important information about functional performance.^1,2,3 Specifically, dynamic tests, whether in an open kinetic chain (pull-up, throwing test, and shot putting) or closed kinetic chain (one-arm hop test, upper quarter Y-Balance-Test and the Closed Kinetic Chain Upper Extremity Stability Test) enable not only the identification of possible deficits in strength and muscular power but also to evaluate proprioception and motor control.^3,4,5,6 These evaluation tests of upper limb performance complement the analysis of the individual segments and provide quantitative data on progress and the effectiveness of rehabilitation programs.^1,3,5,7

The Closed Kinetic Chain Upper Extremity Stability Test (CKCUEST) is one option for examining upper limb stability. The test is easy to administer, cost effective, easily understood, and has been validated by the peak torque of internal/external shoulder rotation (isokinetic dynamometer), and maximum grip strength (hand dynamometer), through Pearson correlation coefficients with the average values of the CKCUEST.^7,8 This test consists of counting how many times the subject performs alternating touches on the opposite hand in a closed kinetic chain position (push-up) over 15 seconds, with three trials and allows three potential scoring outcome measures: average touches (the average of the three trials), a normalized score, and power score. The normalized score is obtained by dividing the number of touches by subject height. The power score is obtained by multiplying the average number of touches by 68% of subject's body weight (kg) divided by 15.^1,3

The use of the CKCUEST as an assessment of upper limb function is increasing both by rehabilitation professionals and by researchers. Authors have determined normative values for the test,^7,9 cutoff scores to be used for shoulder injury prediction,¹⁰ and have compared results with results of other tests that aim to evaluate upper limb function.¹¹ In addition, the CKCUEST has also been used to identify risk factors for shoulder pain,¹² determine the effectiveness of different types of intervention,^13,14,15 and to determine their predictive ability for performance of a sport specific task.⁵

Additionally, the literature has provided clinimetric values that reinforce the practical relevance of the CKCUEST, allowing reliable use of data collected both in clinical and sports environment and for research. Some researchers^1,3,8 have found excellent levels of interday reliability—assessed by an intraclass correlation coefficient (ICC)—for athletes over the age of 18 (ICC = 0.92),^1,3 adults with shoulder impingement syndrome (ICC ≥ 0.82),³ and for healthy adults (ICC = 0.97)⁸. Although the above studies provide excellent levels of reliability for the CKCUEST, their results cannot be applied to other populations, especially adolescents, who, for the purpose of this research, are defined as athletes between 15 and 19 years old.

The adolescent population has increased vulnerability to musculoskeletal injury,¹⁶ therefore the use of functional tests can contribute to the assessment of this population. However, the reliability of the CKCUEST in adolescents has not yet been established. The primary purpose of the present study was to examine the reliability CKCUEST on a sample of adolescents, and a secondary purpose was to establish clinimetric values for this test.

METHODS

Sample

For sample size calculation, the following equations described by Shoukri, Asyali, and Donner¹⁷ in the Gpower 3.1.7 program was used considering the following values: α = 0.05; β = 0.10 (90%); power correlation ratio to null hypothesis (ρH₀) = 0.40; correlation ratio for alternative hypotheses (ρH₁) = 0.80, and a potential loss of 20%. A minimum required sample of 31 subjects was determined.

Subjects aged between 15 and 19 years old, all of whom were healthy, physically active and had no history of injury to the upper limb, were included in this research. Exclusion criteria included refusal to perform the test and the anthropometric measurements, and/or absence from the second evaluation. Because of the departure of six participants, the total sample was composed of 25 adolescents (14 girls and 11 boys), reflecting a final statistical power of 89.64%. Demographic characteristics of all participants are presented in Table 1.

Table 1.

Demographic characteristic mean and (standard deviation) of the total sample and stratified by sex.

Variable	Total (n = 25)	Male (n = 11)	Female (n = 14)
Age (years)	16.92 (1.41)	16.18 (1.40)	17.50 (1.16)
Weight (kg)	59.25 (10.23)	65.95 (10.25)	53.99 (6.66)
Height (m)	1.67 (0.08)	1.74 (0.06)	1.61 (0.04)

All the participants signed a consent agreement and their respective legal guardians signed the consent form. The study was approved by the Research Ethics Committee of the University of Pernambuco – Brazil.

Procedures

After the signed consent forms were collected, information was gathered regarding the age of the subjects, body weight (kg) and height (m) measurements were performed using a digital scale and a portable stadiometer. The first CKCUEST was then performed. Prior to the study, the evaluators responsible for measuring were properly trained. In order to examine test-retest reliability, avoid possible memory effects and adaptation to the test, the second assessments were performed after an interval of seven days. In addition, the subjects were not informed about the scores obtained during the first evaluation at the time of reassessment in order to minimize the motivational effects.

The CKCUEST was carried out following the guidelines as described by Tucci et al.³ During this test, the boys assumed the push-up position and the girls assumed the modified push-up position (with knee support), with both hands positioned on two pieces of tape affixed to the ground at a distance of 91.4 cm apart (Figure 1). The subject remained in the push-up position while they alternatively touched the opposite hand, for a period of 15 seconds. Each subject was allowed to perform the test once sub-maximally in order to familiarize themselves with the task prior to the execution of the three repetitions of the actual test. One evaluator was responsible for counting the number of touches, and the other evaluator timed the test and verbally informed the first evaluator of the beginning and end of the test. Evaluators gave verbal cueing during the test to encourage the maximum effort of the subjects. Three repetitions of the 15-second CKCUEST were performed with an interval of 45 seconds between each test. The average touches score was calculated on the basis of the arithmetic mean number of touches recorded during the three attempts. The normalized score, the score was calculated by dividing the average number of touches by the subject's height in meters. The power score was obtained by the product of the average number of touches and 68% of body weight in kilograms divided by 15.¹⁸

Figure 1.

Positioning for performing the CKCUES-test. Starting position (A) and execution of touch (B).

Statistical Analysis

The statistical procedures were conducted using two computer software packages: Statistical Package for Social Sciences (SPSS) version 20.0 and GraphPad Prism version 5.03. The distribution of the numerical data concerning the age, body weight, height, average score, normalized score, and power score was evaluated using the Shapiro-Wilk test with a significance level of 0.05. In cases where the distribution was symmetrical (scores, body weight, and height) the measures of central tendency and dispersion were presented on average and standard deviation and for cases where the distribution was asymmetrical (age) the data was evaluated in the form of median and interquartile range amplitude.

The comparison between the two scores (first evaluation and re-evaluation) was accomplished using a paired t test. The relative reliability analysis was calculated using the intraclass correlation coefficient (ICC_3,3) two-way mixed model, with the absolute consent method and presentation of confidence interval 95% (CI 95%). ICC values above 0.75 were considered to represent excellent reliability; values between 0.40 and 0.74 represented moderate reliability; and values less than 0.40 indicated poor reliability.¹⁹

In addition, a Bland-Altman chart was plotted for each score from the CKCUEST to verify the absolute agreement between the first and second assessment, from the scatter plot between the difference of the two assessments and the average of the two evaluations. That way, it was possible—accounting for bias—to determine the error limits of agreement, outliers, and trends.^19,20 Systematic error was verified using a one-sample t test based on the average of the difference between testing and retesting and accepting the significance level of less than 0.05.

The Standard Error of Measurement (SEM) was calculated to estimate the variance of each score and the Minimal Detectable Change (MDC) was also evaluated to determine the threshold value for measurement error. Both were calculated by the respective equations:

$$SEM95\%=SD*\sqrt{\left(1-IC{C}_{test-retest}\right)},$$

for which the SD is the standard deviation of the mean of the first evaluation (test).

$$MDC=1.96*SE{M}_{95\%}*\sqrt{2},$$

for which the constant 1.96 represents the z-score associated with the 95% confidence level.

RESULTS

The description and comparison of the average touches score, normalized score, and power score in the first and second evaluation, as well as the results of the ICC with their respective 95% confidence intervals, SEM, and MDC for each score of the CKCUEST are presented in Table 2 and Table 3.

Table 2.

Mean and standard deviation scores for test (evaluation) and retest (revaluation) of Closed Kinetic Chain Upper Extremity Stability Test scores (CKCUEST).

CKCUEST score (n = 25)
Average touches		Normalized score		Power score
Test	Retest	Test	Retest	Test	Retest
25.6 (3.8)	28.0 (5.4)*	14.8 (2.4)	16.2 (3,4)*	69.1 (17.99)	75.9 (20.7)*

^*p<0.05, when test was compared to retest.

Table 3.

Values of the intraclass correlation coefficient (ICC), 95% confidence interval (CI 95%), standard error of measurement (SEM) and minimal detectable change (MDC) of the average number of touches, normalized and power scores of CKCUEST.

CKCUEST score (n = 25)
Scores	ICC (CI 95%)	SEM	MDC
Average touches	0.68 (0.26 – 0.86)	2.17	6.01
Normalized score	0.68 (0.27 – 0.86)	1.35	3.74
Power	0.87 (0.64 – 0.95)	6.49	17.98

The agreement analysis between measurements of the three scores—average touches, normalized, and power—can be observed in Figures 2, 3 and 4, respectively. The chart reveals that the bias for the three scores was within the limits of agreement and close to zero, and that outliers were present. The presence of statistical differences in the one-sample t test for the average score (p = 0.012), normalized score (p = 0.014) and power score (p = 0.009) indicates the presence of systematic error.

Figure 2.

Bland-Altman plot for the average CKCUES-test touches.

Figure 3.

Bland-Altman plot for normalized score.

Figure 4.

Bland-Altman plot for power score.

DISCUSSION

The absence of studies that have determined the reliability of the CKCUEST for assessing upper limb functional performance in an adolescent population motivated this study. Despite the relative coefficients demonstrating moderate to excellent reliability values, the examination of agreement still shows fluctuations, being possibly linked to the form of measurement and individual adaptation to the test. In short, the results of this study demonstrate that the CKCUEST has moderate to excellent reliability for evaluation of upper limb stability in adolescents when performed in just one familiarization session.

Previous authors that have assessed the test-retest reliability of the CKCUEST have demonstrated excellent reliability values when three to seven day intervals were allowed between sessions.^1,3,8 Goldbeck & Davies¹ initially developed this test as an alternative for evaluating the function of the upper limb in a closed kinetic chain manner, observing an equivalent ICC of 0.92 for the score when referring to average touches. Tucci et al³ assessed the reliability of three groups of adults: physically active, sedentary and healthy, and sedentary with diagnosis of impingement syndrome in the shoulder (SIS). The authors demonstrated ICCs ranging from 0.85 to 0.96 in the average touches score, 0.87 to 0.96 for the normalized score, and 0.82 to 0.96 for the power score. Recently, Lee & Kim⁸ assessed the reliability of the CKCUEST in 40 male and female adults and also found excellent reliability values for the average touches score (ICC = 0.97). In addition, the authors evaluated the concurrent validity of the CKCUEST and found a strong positive correlation with the values of maximum grip strength (r = 0.78–0.79) and peak torque of the internal and external shoulder rotators (r = 0.87–0.94), measured by hand and isokinetic dynamometer, respectively.

However, it should be emphasized that the aforementioned studies used a relative reliability analysis (ICC), which is subject to overestimated values of reliability. Deceptive reliability values can occur because when measuring a variable group of different subjects, it is expected that measures although different, can get too close to the average, thus producing high values of ICC. ²¹ Thus, the use of methods that assess the variance and data agreement appear as an additional option and have been widely used in studies assessing the reliability of muscle strength measures, such as one-repetition maximum tests.^22-27 In order to avoid erroneous conclusions about the reliability levels, the current study assessed not only the ICC, but also utilized an analysis of the Bland-Altman plots which allowed the authors to better examine models of variance and agreement between measurements, thus aiming to make conclusions about the results more complete.

The reliability values found in the results of the current study are lower than those found in the previous studies, therefore, it is important to highlight the difference between studied populations. Since there are no other studies that assess the CKCUEST in adolescents, a comparison of the values to other adolescents is not possible. Therefore, it is appropriate to explore the possible reasons for the lower reliability values seen in this group of adolescents compared to prior results obtained with adults. Tucci et al³ speculated that the anthropometric characteristics of individuals can influence the results of the tests, especially because the equations used to calculate two of the scores depend on body weight and height variables. In addition, the authors³ suggest that the standardization of the distance between the hands during the test (91.4 cm) may be a factor that contributes to the result, positively or negatively, depending on the individual anthropometrics.

Anticipating that standardization of the hand measurement could change the test execution, Tucci et al²⁸ proposed comparing three variations of hand positioning in the CKCUEST (standard positioning, distance between the acromion processes, and 150% of the acromion-hand distance) and found that biomechanical testing (scapular kinematics, maximum force and time to maximum force) was independent of positioning; however, these findings has not been tested in adolescents - who are in the process of both neuromuscular and anthropometric maturation and development, and therefore the same results may not be observed.

The Bland-Altman graph is a good option to assess the agreement of a test and the present study used a graph to suggest that the CKCUEST in adolescents is a correlated test-retest measure and not very concordant. Confirmation of systematic error may indicate that the volunteer performed the test better or worse on revaluation due to factors such as learning effect or a change in motor behavior or motivation during the test performance.^19,20 This can be assessed by the presence of outliers mainly in the average touches score and normalized score, the majority of the scores in the plots below zero, representing a negative bias, and by a significant increase in the average of the three scores in the second evaluation.

The differences between the first and second evaluations allows speculation that familiarization can change the results of the CKCUEST. This hypothesis is also supported in studies that evaluate the one-repetition maximum test on strength exercises (1-RM), in which it has been seen that more than one session familiarization session on different days is required for load stabilization.²¹ Considering that Lee & Kim⁸ found a strong correlation between CKCUEST and muscle strength measurements, it can be speculated that the variation of this test may be similar to the 1-RM tests. Another factor that could influence the results of the revaluation is the motivational issue since the subjects were encouraged to make as many touches as possible. However, the authors of this study believe that this factor was minimized, since the verbal encouragement were offer by the same tester and did not varied between the two days and the subject did not have access to an amount of touches completed in the first evaluation.

The CKCUEST is a multi-joint movement assessment, which induces instability in the subject and performance can depend on muscular strength levels, balance, coordination, and, consequently, the physical ability of the individual. These factors raise the need to investigate whether only a single session would be sufficient to indicate the actual value of the subject's performance, considering that no previous study has investigated the amount of attempts required for the stabilization of CKCUEST values. The performance requirements of the test itself seemed to present a considerable challenge to the some subjects, requiring processing time for learning and motor control and possibly requiring practice sessions on different days for familiarization.

Regarding the variability of the CKCUEST, a small standard error is present, so it is expected that the average touches scores, normalized scores, and power scores will be slightly different when repeated by the same individual but at different times, an expected error value of 2.17, 1.35, and 6.49 respectively. Only Tucci et al³ previously evaluated the SEM of the CKCUEST. The SEM in their research for average touches ranged from 1.45 to 2.76 touches between groups, from 0.02 to 0.04 touches/height for the normalized score, and from 8.52 to 28.32 touches x 0.68 kg/m for the power score. Although several different populations were studied by Tucci et al,³ Weir²⁹ has suggested that the value of SEM is independent of the population subjected to the test, and instead is a fixed characteristic value of the measure.

Finally, for the minimum detectable change interpretation, it is important to differentiate the values obtained in a CKCUEST that may be related solely to measurement error. Tucci et al³ indicated that the MDC values for their adult populations ranged from 2 to 4 points for the average touches score, from 0.03 to 0.06 for the normalized score (touches/height), and from 9 to 29 touches x 0.68 kg/m for the power score. The results of the present study indicate that in adolescents the minimum difference that is not attributable to variation of measurement is seen with values above 6 touches on the average touches score, 4 in the normalized score, and 18 in the power score.

Some limitations can be described in this study. Failure to stratify the sample by sex and the use of physically healthy adolescents, makes the study restricted regarding extrapolation for other populations, be they healthy or with any pathological conditions. Thus, it is suggested that further studies be carried out dealing with greater sample sizes and different age groups and populations that have different types of dysfunction in the kinetic chain of the upper limb to allow for further development of normative values for the adolescent population.

Another limitation concerns the execution of a single familiarization session. It has been suggested that multiple tests be conducted on different days in order to assess accurately the number of sessions necessary for stabilization of test scores. It is possible that better standardization of the distance between the hands according to the anthropometry of adolescents and the determination of the number of attempts to stabilize the values could contribute to better implementation and completion of the test. Moreover, the form of to equate standardization of the current scores may not be suitable for adolescents. Considering these limitations, the authors believe that future studies should perform allometric analysis in order to get the best CKCUEST standardization system for adolescents.

CONCLUSION

The CKCUEST is a tool that demonstrates questionable reliability levels for use in assessing stability and functional performance of the upper limb of adolescents when subjects were provided only a single familiarization attempt. The results show moderate to excellent reliability, and relative agreement between the three scores derived from the evaluation.