This cross-sectional study was part of a larger project involving preseason assessment and lower limb injury risk profile identification of 4 different sport-clubs from the city of Belo Horizonte, Brazil and was approved by the Universidade Federal de Minas Gerais Research Ethics Committee (report number 0493.0.203.000-09). All volleyball, basketball, and soccer youth, competitive level athletes participating in the project were recruited and signed a consent form. All included athletes were regularly practicing sports and did not have pain or history of surgery to the lower limbs and trunk in the past 6 months. Failure to properly follow test instructions and/or having pain during the testing session were exclusion criteria. For those athletes under 18 years-old, the parents signed the consent form after the athlete's assent.

Before the test, each athlete jogged for 5-minutes for warm-up. Two trained examiners (physical therapists with 5 years of clinical practice) followed a standardized procedure to assess hip ER isometric torque of all participants.14 For the assessment, participants were lying in prone, with arms positioned along the body on the treatment table. The knee was positioned at 90° of flexion and hip at 0° (neutral) rotation; this position was monitored by an inclinometer (Starrett®) positioned below the tibial tuberosity. Both legs were randomly tested for each participant.

The examiner positioned the hand-held dynamometer (Microfet2®) using a rigid strap to eliminate measurement bias (examiner force application during the test). The dynamometer's strap was anchored to a metal rod placed perpendicular to the table (Fig. 1). Another stabilization strap was fixed around the pelvic region of the athlete to avoid pelvic movement. The dynamometer was positioned five centimeters proximal to the medial malleolus and the rigid strap limited hip external rotation range of motion (ROM).

Fig. 1.

Participant's position for hip external rotation (ER) isometric torque assessment.

(0.09MB).

We allowed the athletes to produce one practice isometric contraction by pulling against the rigid bar to which the dynamometer was attached with progressively increasing strength, for familiarization purposes. Subsequently, the athlete completed the test following the examiner's standard verbal command to contract the hip ER, gradually increasing strength, until the command to stop the test after 5 seconds of contraction.17 The examiner closely observed the athlete's lower limb during force production. If compensations were observed before the 5-second mark, the test was interrupted and repeated after a 15-second interval. Compensations included pelvic elevation and/or rotation, trunk rotation, extension and/or adduction of the tested lower limb and isometric flexion of the hip of the opposite lower limb.

Participants performed three isometric contractions with each leg, with 30 second interval in between repetitions. The mean value of the 3 repetitions for each leg was multiplied by the respective lever arm to obtain the magnitude of hip ER isometric torque. The lever arm was measured with a ruler from the dynamometer that was positioned 5 cm proximal to the medial malleolus, to the femur medial condyle. Torque was normalized by body mass (Nm/kg). To describe the expected asymmetry between dominant (D) and non-dominant (ND) limbs, a symmetry index was computed according to the following formula: (ND/D)*100.18 An index of 100%, indicates perfect symmetry. Indices <100 and >100 indicate stronger dominant and non-dominant leg, respectively.18 Lower limb dominance was defined through the following question: ‘‘Which leg would you use to kick a ball as hard as possible?’’

A pilot study with 6 participants (4 women and 2 men, healthy university students, with a mean age of 21.6 years, mean body mass of 62.2 kg, and mean height of 1.64 m) was conducted to assess intra- and inter-examiner reliability of the torque assessment procedure. The results showed excellent reliability (intraclass correlation coefficient intra-examiner = 0.98 and inter-examiner = 0.90), a standard error of measurement (SEM) of 0.021 Nm/kg, and a minimal detectable difference (MDD) of 0.05 Nm/kg.

All torque data are reported in Nm per unit of body weight, thus, considering the height (limb length) of the individuals and adjusted for weight. To achieve the primary objective of the study, box plots were used to provide a visual representation of the computed descriptive data (mean and 95% Confidence Interval (CI)) of body-mass normalized hip ER isometric torque (stratified by limb) and Limb Symmetry Index.

Linear mixed effect models were estimated to test hypothesis related to the secondary objectives of this study. This approach allows for the inclusion of random effects to capture and control for individual differences across participants not accounted for by fixed effects. Additionally, linear mixed effect models are robust to imbalances in group sizes. Results, however, should be interpreted as ANOVA results.

Two models were estimated. Model 1, computed with data from male athletes, tested for significant differences in hip ER torque between sports, between dominant and non-dominant limbs, and for significant sports modality × leg dominance interaction effect. Model 2, computed with data from volleyball athletes, tested for differences in hip ER torque between males and females and for significant sex × leg dominance interaction effect. Finally, simple Analyses of Variance were performed with data from male and volleyball athletes, respectively, to determine if torque asymmetry differed (a) among sports (Model 3), and (b) between males and females (Model 4). Appropriate follow-up pair-wise comparisons with Tukey correction were conducted, as needed, on the least square means estimated by each of the models.

Fig. 2 illustrates the typical distributions of (body-mass normalized) hip ER isometric torque (stratified by limb) and between-legs asymmetry index. Fig. 2A and B present data of male participants stratified by sports, and Fig. 2C and D present data of volleyball athletes stratified by sex. Distributions of hip ER torque scores (Fig. 2A and C) look quite symmetric for males of all three sports. The distribution for females looks right skewed, suggesting a higher prevalence of lower torque values than would be expected from a strictly normal distribution. Skewness and kurtosis scores for both dominant and non-dominant legs were, however, within acceptable ranges (<2) for all distributions. The distribution of asymmetry indices (Fig. 2B and D) looks right skewed for basketball athletes and symmetric for soccer and volleyball athletes. Again, the skewness and kurtosis scores (<2) did not suggest excessive deviations from normal in any of the distributions. Extreme outliers for torque and asymmetry index (i.e., values that were more than 3 times the interquartile range above the mean) were identified and excluded from the dataset prior to further analysis. In total, we excluded one observation of torque obtained from the dominant leg and one obtained from the non-dominant leg. We have also excluded 7 observations of Limb Symmetry Index.

Fig. 2.

Distributions of (body-mass normalized) hip external rotator isometric torque for dominant and non-dominant legs (A and C) and for between limb torque symmetry index (B and D). A and B: data from male athletes in the study sample stratified by sport. C and D: data from volleyball athletes in the study sample stratified by sex. (*) indicates extreme outliers that were removed from data set.

(0.13MB).

Table 1 provides the normative data for (a) hip ER torque for dominant and non-dominant limbs and (b) Limb Symmetry Index, stratified by sports and sex. Notably, intervals obtained for male athletes are smaller than 0.05 Nm/kg (the MDD of the torque variable) and only slightly greater than that for female athletes. This suggests that the sample size was sufficient to provide estimates of hip ER torque with small margin of errors, which is critical if these estimates are to be useful to guide decision making of sport physical therapists and trainers.

Model 1 (male data) showed that hip ER torque significantly differed among male athletes of different sports, p < 0.001, and between dominant and non-dominant legs, p < 0.001. The model also returned a significant sports × leg dominance interaction effect, p < 0.001.

Pair-wise comparisons of least squares means were performed to follow-up on the interaction effect. Results showed that soccer athletes produced significantly higher torques than both volleyball and basketball athletes regardless of limb dominance, all p values < 0.0001. There was no significant difference between volleyball and basketball players regardless of limb. Differences observed between dominant and non-dominant limbs depended on sports and were only statistically significant for basketball (p < 0.01) and soccer athletes (p < 0.01).

Model 2 (volleyball data) returned a significant effect of sex, p = 0.001, with male volleyball athletes showing higher torque values than female volleyball athletes. There was no significant effect of dominance, p = 0.11, nor sex × leg dominance interaction effect, p = 0.08.

Model 3 (male data) showed that Limb Symmetry Index differed statistically among sports. Pair-wise comparisons of least squares means returned significant differences in Limb Symmetry Index between basketball athletes and both volleyball and soccer athletes (p-values < 0.01), with basketball athletes showing higher asymmetry. No significant difference was observed between soccer and volleyball athletes.

Model 4 (volleyball data) returned a significant effect of sex, p = 0.005, with greater asymmetry observed for males than females.