Assessment of reliability in isokinetic testing Among Adolescent Basketball Players

Humberto M. Carvalho1, Manuel J. Coelho E Silva1, Enio Ricardo Vaz Ronque2, Rui S. Gonçalves3, Renaat M. Philippaerts4, Robert M. Malina5, 6 1Faculty of Sport Science and Physical Education, University of Coimbra, Portugal, 2Department of Physical Education, Universidade Estadual de Londrina, Brazil, 3College of Health Technology, Polytechnic Institute of Coimbra, Portugal, 4Faculty of Medicine and Health Sciences, Department of Movement and Sports Sciences, Ghent University, Belgium, 5Department of Kinesiology, University of Texas at Austin, Texas, USA, 6Department of Health and Physical Education, Tarleton State University, Stephenville, Texas, USA


introduction
Isokinetic dynamometry is generally considered the best protocol of dynamic strength measurements (1).However, the literature suggests inconsistency in the application of isokinetic measurement protocols, which may hinder data interpretation (2).Valid isokinetic data require reliable assessment techniques for the peak moment of force, mean peak moment of force, and angle at peak moment of force.It is suggested that coefficients of variation (CV) should not be greater than ±6.1%, and intraclass correlations (ICC) should be greater than 0.88 (3).Most of the available information, however, is derived from studies on adult males and females.
Minimal measurement error (reliability) of performance of a test refers to the consistency or reproducibility of performance when someone performs the test repeatedly (4).A test with poor reliability is unsuitable for tracking changes in performance between trials, and it lacks precision for the assessment of performance in a single trial (5).The magnitude of measurement error in isokinetic testing in general and in particular with young athletes has not been systematically addressed.young athletes are a highly select group with regard to skill, performance, size, and physique; variation in maturity status is an additional factor affecting growth in size and function especially during adolescence (6,7).
Studies on the reproducibility of the isokinetic peak moment of force assessment in children and adolescents are limited, especially on the eccentric action mode (8).Several reports using a test-retest design with youth suggest that a familiarization session may reduce learning effects (9) and that familiarization may be more important in protocols which include eccentric actions (10).Learning effects may interfere with statistical estimates of reliability (11).
The purpose of this study was to investigate the reproducibility of day-to-day testing of isokinetic concentric and eccentric muscular actions in youth basketball players aged 14 to 16 years.Measurement error in isokinetic testing is hypothesized.Interrelationships of age, biological maturation, training experience, body size, lower-body morphology, and initial strength performance with an average within-subject variation between two test sessions of different muscular actions were also examined.

Materials and Methods
Experimental Approach to the Problem.Isokinetic dynamometry is a widely recognized component of assessment in athletes (12).As such, reliability of the protocols is important as it provides an indication of the biological and technical variation associated with assessment procedures and conditions (13).
Day-to-day variation in isokinetic assessments was based on repeated tests within a period of 3 to 5 days on the 27 basketball players with at least 48 hours between test sessions.The group of 8 randomly selected athletes from the total sample completed a third testing session to confirm the attainment of optimal performance after the first testing session.Relationships of within-subject variation between two testing sessions with chronological age, duration of formal training, body size, estimated lower-limb volume, and initial strength values (first test session) were examined using correlations.
Subjects.The sample included 27 male adolescent basketball athletes aged 14-16 years at the time of testing, 25 of Portuguese and 2 of African ancestry.From this group, 8 athletes were tested a third time under the same conditions.Participation in the study was voluntary, and each participant's parents (or legal guardians) were provided informed consent following an explanation (in compliance with the Declaration of Helsinki) of the nature of the experiment and any possible risks associated with participation.All procedures were approved by the Scientific Committee of the Faculty of Sport Science and Physical Education of the University of Coimbra.Chronological age was recorded to the nearest 0.1 year by subtracting birth date from the date of the first laboratory assessment.The duration of formal training was documented during an interview.
Anthropometry.Anthropometry was assessed by a single experienced observer following procedures described in Lohman et al. (14).Stature and sitting height were measured with a portable stadiometer (Harpenden model 98.603, Holtain Ltd, Crosswell, United Kingdom) to the nearest 0.1 cm.Leg (subischial) length was estimated as stature minus sitting height.Body weight was measured with a portable scale (Seca model 770, Hanover, MD, United States) to the nearest 0.1 kg.To estimate the total leg volume of the dominant leg (15), the following parameters were recorded: circumferences at the gluteal furrow (the highest possible horizontal circumference), mid thigh (the largest possible mid thigh circumference), minimum circumference above the knee, maximum circumference around the knee (patella level), minimum circumference below the knee, maximum calf circumference and minimum ankle circumference; anterior and posterior skinfolds at the level of mid thigh circumference as well as the medial and lateral skinfolds at the level of maximum calf circumference; and lengths between each circumference from the gluteal furrow to the minimum ankle circumference (e.g., distance between mid thigh and minimum thigh circumferences), sum six partial lengths.Although the protocol is valid, it tended to overestimate total lower-limb volume (approximately 6%) compared estimates based on dual-energy x-ray absorptiometry in 37 young athletes aged 13-22 years.Correlation between estimated volumes ranged between 0.91 and 0.96 (unpublished results).
Based on 18 participants measured twice within one week, intraobserver technical errors of measurement were 0.54 cm for stature, 0.74 cm for sitting height, 0.88 kg for body weights, 0.29 to 0.74 cm for circumferences, and 0.16 to 0.46 cm for lengths.The estimated errors were within the range reported for intra-and interobserver errors for a variety of studies (16).
Maturity Status.Age at peak height velocity (APHV) was estimated with the maturity offset protocol (17) Isokinetic Assessment.The isokinetic assessment was completed within a week period with at least 48-hour interval between sessions.Subjects were instructed not to eat for at least 3 hours and not to drink coffee or beverages containing caffeine for at least 8 hours before each testing session.Subjects wore similar clothing and the same footwear on each testing occasion.No participant was suffering from lower extremity musculoskeletal injury at the time of testing or during 6 months before testing.
Isokinetic concentric and eccentric knee extension and flexion were measured using a calibrated dynamometer (Biodex System 3, Shirley, NY, United States) at angular velocities of 60° s -1 .After the anthropometric assessment, the subject performed a 10-minute cycling warm-up on a Monark cycle ergometer (Monark 814E, Varberg, Sweden) with minimal resistance (basket supported) at 60 rev min -1 and 2-minute static stretching of the hamstring and quadriceps muscles.Athletes were then placed in a seated position, adjusted according to the manufacturer guidelines in a standardized 85° hip flexion from the anatomical position.The lever arm of the dynamometer was aligned with the lateral epicondyle of the knee, and the force pad was placed approximately 3 to 5 cm superior to the medial malleolus with the ankle in a plantigrade position.Range of motion during testing was set using voluntary maximal full extension (0°) to 90° of knee flexion.Cushioning was set using a hard deceleration (according to manufacturer guidelines) and, therefore, 90° constituted the range of motion tested.At the beginning of each session, the participant was asked to relax his leg so that passive determination of the effects of gravity on the limb and lever arm could be accounted for.The participant was instructed to grasp firmly the handgrips at the sides of Biodex System 3 seat during the test procedure.In the concentric action, the participant was instructed to push the lever arm during extension and pull during flexion as hard and fast as possible; in the eccentric action, the subject was instructed to resist the lever arm during extension and flexion as hard as possible.Extension was undertaken first in both muscular action modes.
Each subject performed 5 continuous maximal repetitions.Each set assessed reciprocal muscular action in the dominant leg.The order of testing started with concentric actions, followed by a 120-second rest before the performance of the eccentric actions.Visual feedback of moment versus time was provided during the test, but no verbal feedback was given (18).The maximal knee flexion and extension peak moment of force of the best repetition in both concentric and eccentric muscular actions were retained and expressed in Nm.
Analysis.Descriptive statistics for chronological age, duration of formal training, estimated APHV, body size, and replicate isokinetic measures were computed for the total sample.Differences between test sessions were examined using the paired t test.Coefficients of variation and ICC were calculated (5).Variability in isokinetic dynamic muscle performance between sessions was expressed as the 95% confidence interval (CI).Heteroscedasticity was checked using plots of the log-transformed data.Reliabilities (absolute and percent typical errors) were calculated (5,19).The effect size correlations (ES-r) were estimated using the square root of the ratio of t value squared and the difference between t value squared and degrees of freedom (total amount of variance in the sample) (20).
Relationships of variation in means of the differences between the maximal moment of force in concentric and eccentric muscle actions in the first and second test sessions with chronological age, estimated APHV, duration of formal training, stature, body weight, leg length, leg volume, and initial isokinetic strength (first test session) was examined with Pearson correlations.The magnitude of correlations were interpreted as follows: trivial (r<0.1),weak (0.1<r<0.3), moderate (0.3<r<0.5), strong (0.5<r<0.7), very strong (0.7<r<0.9) and nearly perfect (r>0.9) and perfect (r=1) (21).Positive correlation values indicate an increasing linear relationship and negative correlations indicate a decreasing linear relationship between the variables.Significance was set at P<0.05.Statistical analyses were performed using the SPSS version 17.0 software (SPSS, Chicago, IL).

results
Characteristics of the sample are presented in Table 1.With the exception of two participants, maturity offset values were positive indicating that the sample was beyond the age at maximum growth rate in height during the adolescent spurt.Table 2 shows the differences between means of the two test sessions, and CVs and ICCs of the isokinetic strength indicators.
The difference (mean of mean differences as a Values of the mean peak moment of force for the second test session were higher than for the first session in all muscular actions with an exception of eccentric knee extension (Table 2).Effect sizes of within-subject variation were only moderate.The CV for concentric knee extension was 8.1% (95% CI, 6.3 to 11.2), while CVs for other muscular actions ranged from 15.1% to 17.4%.The retest correlations ranged from 0.72 to 0.89.
The pairwise CVs between test sessions 2 and 3 for the subsample of 8 athletes were 4.9% and 6.0% for concentric knee extension and flexion, respectively, and 3.9% and 5.3% for eccentric knee flexion and extension, respectively (Table 3).The corresponding ICCs ranged from 0.95 to 0.99.
Correlations between the variance in withinsubject variation between the two test sessions and age, estimated APHV, duration of formal training, body size, lower-body morphology (length and volume), and initial strength are presented in Table 4. Variance in within-subject variation for concentric knee extension was not associated with age, duration of formal training, body size, lower-body morphology, and initial strength.The mean differences between pairwise eccentric knee flexion knee were negatively correlated (P<0.05) with session 1 values of eccentric knee flexion and knee extension.Within-subject differences in the mean differences between pairwise concentric knee flexion were correlated with training experience (P<0.05) and initial values of concentric knee flexion (P<0.01).Withinsubject eccentric knee extension variation was significantly (P<0.05)correlated with chronological age, estimated APHV, stature, body weight, and leg volume, but the correlations were only moderate.
Studies of the reproducibility of the isokinetic peak moments of force measurements are limited, specifically for eccentric actions.Retest correlations for the concentric peak moment of force of knee extensors and flexors in boys aged 6-8 years tested in 3 sessions with 3-day rest between sessions, using a velocity of 100° s -1 , ranged from 0.85 to 0.95 (9).Among pubertal (pubic hair, stage 3) soccer players aged 13 years, ICCs for the concentric and eccentric knee extensors and flexors of the dominant leg ranged from 0.85 (eccentric) to 0.98 (concentric) (10).The coefficients of variation of 4% and 9.9% for the eccentric peak moment of force were documented in 11-year-old boys (22).It was suggested that reliability could have been improved by a familiarization session.Test-retest correlations in the present study after familiarization were consistent with the range of values reported for concentric actions but higher than ICCs reported for eccentric muscular actions.Nevertheless, the reliability estimates from the familiarization session to the second test were lower than values reported in children, and exceeded the limit of values proposed for good reliability in isokinetic strength assessment (2,4).
Few studies have analyzed possible learning effects during day-to-day testing of isokinetic strength.No differences in the concentric peak moment of force in the knee extensors and flexors at 60° s -1 were noted in adult men and women (3), but comparable data for eccentric knee muscular actions in general and specifically for younger subjects or athletes are lacking.Differences between the first and second test sessions and the reliability estimates for the familiarization session (n=8) suggested the presence of systematic bias and/or random error in both concentric and eccentric muscular actions when testing inexperienced young athletes.The reproducibility in isokinetic strength might be related to biological variation.
Several factors affecting reliability have been noted in the literature (4,23).At the speed used in the present study, isokinetic dynamometry is a valid and reliable testing tool (24).All measurements were made by a single experienced observer.Reliability estimates of the observer for the used protocol, based on repeated measurements in 13 subjects (university students, age 21.1±3.1 years), were as follows: concentric knee extension, CV=4.1% and ICC=0.98;concentric knee flexion, CV=4.1% and ICC=0.98;eccentric knee extension, CV=5.2% and ICC=0.97; and eccentric knee flexion, CV=4.8% and ICC=0.97.In order to control potential effects of time-of-day, replicate data were obtained at the same time of day within ±30 minutes (25).Therefore, it is possible to assume that within-subject variation in the adopted protocol of isokinetic strength assessment is mainly related to subject-specific factors.
Factors related to subjects, particularly specific issues in pediatric populations, and assessment conditions that may influence dynamometric measurements of extremity muscles have been addressed in the literature (2,8,26).Potential factors include variation in chronological age per se, body size (stature and body weight), limb dominance, presence of impairment, and athletic background; variation in biological maturity status is an additional consideration.Body size, in particular body weight, apparently influences the magnitude of isokinetic moment of force production in adolescent populations (22).Muscle mass has also been indicated as a relevant covariate in the interpretation of isokinetic strength in youth (27), but intrasubject variation does not appear to be related to this measure.It is possible that individual differences in maturity status may influence strength through its effect on body size and muscle mass.Within a given age group of adolescent boys, those advanced in maturation are, on average, taller and heavier and have a larger muscle mass than boys later in maturation (7).To our knowledge, no study has apparently addressed the possible influences of chronological age per se, late adolescent growth in muscle mass (peak rate of growth in muscle mass occurs on average after peak height velocity) and accumulated effects of training on reliability estimates in isokinetic dynamometry.
Age, estimated maturity status, training experience, and body size were the variables correlating with mean within-subject variation between two sessions of isokinetic strength of the hamstrings muscles assessed at 60° s -1 (concentric knee flexion and eccentric knee extension) and in particular for eccentric contractions (Table 4).The results highlight the importance of age, maturation, and body size in the consideration of the reliability of assessments of knee flexion muscles in adolescent basketball players.As for the assessment of quadriceps muscles (concentric knee extension and eccentric knee flexion), within-subject variation in the concentric action was smaller than in the other muscular actions and closer to reported values in literature.Moreover, within-subject variation in eccentric actions was negatively related to performance values in the familiarization session in both eccentric knee actions.Therefore, higher reliability estimates between the familiarization session and second test Humberto M. Carvalho, Manuel J. Coelho E Silva, Enio Ricardo Vaz Ronque, et al. may be related to higher values of strength in this muscle group.
The issue of maturity-associated variation in isokinetic strength performance has not received much attention (8).Some evidence indicated an increase in concentric knee extension and flexion with age in male and female adolescent athletes but without statistically controlling for interindividual variability in growth and maturation in the analysis (28)(29)(30).This is relevant because short-term power outputs improve with age during adolescence independently of pubertal status (7).It is likely that maturity exerts an influence on isokinetic strength performance via body size and muscle mass.Some limited data for male adolescent soccer players showed an independent maturity effect on isokinetic indicators in the knee joint, albeit the limitations of the use of selfreported sexual maturation (31).Overall, the one interested in assessing lower-body isokinetic strength should consider biological inter-individual variation and fitness level as factors that significantly affect measurement reliability.On the other hand, learning effects, i.e., systematic bias, seem to diminish after familiarization session, allowing for more reliable data in adolescent basketball players.

Conclusions
A familiarization session may improve the reliability of assessment when using concentric and eccentric knee isokinetic strength testing at 60° s -1 and improve the evaluation of the effectiveness of strength training programs or clinical assessments of adolescent basketball players.Intraindividual and interindividual differences in chronological age, maturity status, and accumulated experience are potential factors that should be considered when using knee isokinetic strength testing at 60° s -1 , in particular for the assessment of the hamstrings muscular actions.

Table 2 .
Reliability Estimates of Concentric and Eccentric Knee Flexion and Extension Peak Torque at 60° s -1 Measured in Two Sessions in Adolescent Basketball Players (n=27)

Table 3 .
Reliability Estimates in the Subsample of Adolescent Basketball Players (n=8) Tested on a Third Occasion Ext, extension; flex, flexion; con, concentric; ecc, eccentric; PHV, peak height velocity.

Table 4 .
Correlations Between Variance in Within-Subject Variation Between Two Trials of Isokinetic Muscular Actions at 60° s -1 With Age, Maturity Status, years of Formal Training, Body Size, Lower-limb Morphology, and Initial Strength in Adolescent Basketball Players