Nontraumatic Shoulder Pain Affects Proprioception and Dynamic Stability in Female High School Volleyball Players

Study Design

We designed a cross-sectional study that included a questionnaire covering demographic information and volleyball-specific factors. Additionally, various aspects of shoulder joint function were assessed. The study took place within the gymnasium of the participating teams. This study adheres to Strengthening the Reporting of Observational Studies in Epidemiology checklist and follows ethical guidelines (Institutional Review Board of Saitama Medical University (Approval No. DAI 2022-019) in line with Declaration of Helsinki principles.

Participants

The study’s inclusion criteria targeted female volleyball players aged 15 to 17 years, specifically those within the top 32 of the 139 schools enrolled in the Saitama, Japan. This criterion aimed to standardize the competitive level of players. The exclusion criteria encompassed players who (1) had undergone spinal, upper limb, or lower limb surgery; (2) experienced practice absences due to injuries; and (3) had a history of glenohumeral dislocation. In this study, NSP was defined as a condition that can exist in healthy athletes who do not require activity restriction. Discontinuation criteria comprised suspected pain during the examination or any deterioration in health status. Recruitment for the study took place between March 2023 and July 2023. We visited teams that expressed interest in participating in the study. First, we provided an explanation of the study to the school principals and team coaches. Upon obtaining written consent from them, we then provided the same explanation to the players and their parents. Only those players who, along with their parents, provided written consent were included in the study. To minimize selection bias, we ensured that the study invitation was extended to all eligible players across the selected teams, regardless of their playing position, skill level, or injury status. Additionally, we collected data anonymously to encourage participation from all invited players.

Out of the 32 teams, 5 teams comprising 59 players agreed to participate in the study. All 59 participants met the inclusion criteria, and no one was excluded. No instances of pain or deterioration in health were observed during the measurements, and all 59 participants completed the questionnaire and shoulder joint functional assessment. Teams practiced for approximately 3 hours on weekdays, 6 times weekly, and 4 hours on Sundays.

Demographic Details and Environmental Factors

Demographic details consisted of age (years), height (cm), weight (kg), dominant hand, years of volleyball experience, sleeping time, and number of different sports participated in during elementary school (categorized as 1 sport or less or 2 or more sports). Furthermore, participants’ body mass index (BMI; kg/m²) was calculated (kg/m²). Dominant hand determination was based on the arm used for attack and serve. Participants were queried regarding the presence of dominant-side shoulder pain during attacking or serving within the last 24 hours along with their communication of this to their coach. Players who reported shoulder pain during attack or serve movements provided responses that fell into the following categories as applicable: (1) pain exclusively during the attack or serve, without occurrence during other activities; (2) pain during both the attack or serve and other activities, ceasing after practice; (3) pain persisting until after practice but not extending to the next day; and (4) pain lasting until the next day. The demographic details mentioned above were self-reported through a questionnaire. To eliminate observation bias and ensure the honesty of the players, coaches were situated in a separate room during the completion of the questionnaire. Additionally, to avoid response bias, it was explained to the players in advance that the individual results of the questionnaire would not be seen by their coaches. This is because many players are reluctant to report their pain to their coaches.^18,19 Since we investigated the recent state of pain in this study, recall bias was minimized.

The examined environmental factors encompassed volleyball court position (attackers [outside attacker, middle blocker, opposite] or others [setter or libero]), experience in multiple positions (1/2/≥3), serve type (floater [short: within 3 m from the end line, long: at least 3 m away from the end line, or jump floater] or jump and hybrid), attack form, consciousness of single-leg landing during attack practice, awareness of the ball contacting behind the head during attacks, awareness of frequent mistimed ball hits during attacks, average duration of volleyball practice, weekly practice frequency, and average weekly sleep duration. The above environmental factors were self-reported by questionnaire. Participants’ attack forms were repeatedly recorded on video during attack practice and classified into bow and arrow, circular, or straight swings by Y.M., a volleyball coach certified by the Japan Sport Association. The attack forms were recorded multiple times because the swing used often varies depending on the speed of the ball and the type of attack. We considered the attack form used by each player as the one she used in >80% of her attacks.

Shoulder Joint Function

Acromiohumeral distance (AHD), denoting the distance between the humeral head and the lower acromion end, was measured using an ultrasound imaging device (miruco, NIPPON SIGMAX Co, Ltd; Figure).²⁰ Participants were seated, with their upper limbs resting on a cushion atop their knees, the shoulder joint at 60° abduction, the elbow joint at 90° flexion, and the forearm and wrist joints positioned midway. The shortest humerus-to-acromion distance was measured thrice at the lateral acromion aspect and subsequently stored as an image. Image analysis was carried out using Image J software, with the average AHD value derived from the three measurements. Acromio-humeral distance has shown excellent intrarater reliability (intraclass correlation coefficient [ICC] > 0.90), and this measurement was performed by 1 of the authors (Y.M.).²⁰

View Figure

• Download as Powerpoint
• Download Figure

Passive glenohumeral joint’s ROM (°) was gauged bilaterally for external rotation (ER), internal rotation (IR), horizontal adduction (HAD), and horizontal abduction (HAB), all measured with the participant’s shoulder in 90° abduction (Figure).²¹ A baseline plastic goniometer (12-1000; Fabrication Enterprises Inc; 1° increments) was employed, with ER, IR, and HAD measured in the supine position and HAB measured in the prone position.²¹ During ROM measurements, scapular fixation was maintained. Total rotation motion was calculated as the sum of IR and ER. Glenohumeral IR deficit was defined as a difference in IR ROM of ≥18° and a corresponding loss of total rotation motion of ≥5° when compared bilaterally.^22,23 External rotation deficit was defined as a decrease of 5° or more in ER ROM compared with the nondominant hand side.²² In addition, authors of previous studies have reported good to excellent intrarater reliability for all measures (ICC = 0.85–0.99; variability = 4°–7.5° or less).²¹ The assessments were conducted by 2 experienced physical therapists, each with over 10 years of expertise in the musculoskeletal and sports fields—1 stabilizing the scapula and the other moving the humerus. Although this approach is not typical in clinical practice, it was chosen to ensure precise and consistent measurements (Y.M. and K.S.).²¹

Rotational isometric muscle strength (kgf; IR and ER) was evaluated using a handheld dynamometer (μ-TAS F-1, ANIMA Corp; Figure).²⁴ The measured limb position adopted the belt fixation method based on previous studies, with the shoulder joint at 90° of abduction and the elbow joint at 90° of flexion.²⁴ Muscle strength warm-up routines (50% and 75% of max) preceded maximal measurements. Two 3-second maximum contraction measurements were performed, interspersed with a 10-second rest. A 2-minute break was provided between IR and ER measurements. The dominant hand side underwent maximal contraction assessments, with the mean value adopted. The belt fixation method was used for this measurement, and the intrarater reliability was ICC = 0.90 for IR and ICC = 0.87 for ER.²⁴ One author conducted scapular and elbow fixation, while another author handled the belt fixation (Y.M. and K.S.). The obtained values were divided by body weight and normalized (kgf/kg).

Joint position sense for IR and ER of the shoulder joint was determined bilaterally using a Baseline Digital Inclinometer (12-1057; Fabrication Enterprises Inc), which is capable of measuring angles to an accuracy of 0.1° (Figure).^16,25 Participants sat with the limb positioned at 90° shoulder abduction, 90° elbow flexion, and neutral forearm rotation, with both feet flat on the floor. They sat on a chair without a backrest to avoid tactile feedback from the back or scapular region, ensuring a neutral pelvis and spinal position as much as possible. They wore sleeveless shirts to minimize tactile feedback from clothing. The participants’ eyes were closed throughout the assessment to eliminate visual feedback. The examiner maneuvered participants’ arms to 90% of their maximum range, maintained the position for 3 seconds, prompted participants to recall this position, and quantified the angle (passive movement). The digital inclinometer was in contact with the forearm only during the angle measurement phase, applied lightly to avoid providing additional tactile feedback. Subsequently, participants returned to the starting limb position and executed an active movement to replicate the passive movement-induced position, with the angle recorded. This process was repeated 3 times, and the difference between the passive and active movement angles was represented as an absolute value. The average of the 3 absolute values was then calculated. This passive/active protocol using a digital inclinometer exhibited good intrarater reliability (weighted mean ICC = 0.84), and 1 author performed the measurements (Y.M.).²⁵

The Upper Quarter Y-Balance Test (UQYBT), using the FMS Professional Y-Balance Test Kit (Functional Movement Systems), assessed dynamic stability, joint mobility, and proprioception asymmetry within the upper extremity and trunk (Figure).²⁶ The test commenced from a push-up position, wherein participants extended 1 hand in 3 directions (medial, superolateral, and inferolateral), avoiding ground contact until all 3 directions were attempted. Each direction was attempted 3 times, the average of which, divided by upper limb length (LL), was normalized (average distance in 3 directions/LL × 100 = %LL). The total composite score was calculated by dividing the sum of the averages across the 3 directions by 3 times the participant’s upper extremity length, subsequently multiplied by 100. Reach distances were measured to the nearest 0.5-cm increment. Upper extremity length was measured from the C7 spinous process to the tip of the middle finger in the 90° abduction limb position of the shoulder joint. The intrarater reliability of the UQYBT was ICC > 0.99, and 1 examiner performed the measurement (N.S.).²⁶

The Seated Medicine Ball Throw (SMBT) test, designed to measure upper extremity power, was evaluated with the participant in a seated position against a wall, holding a 2-kg medicine ball in front of her chest with both hands and propelling it as far as possible (Figure).²⁷ A measuring tape, affixed to the wall, gauged the throwing distance, with the horizontal distance from the wall to the ball’s front subtracted. For measurement normalization, body weight factored in, resulting in division by 0.35 to the power of body weight.²⁸ The intrarater reliability of the SMBT was ICC = 0.98, and 1 author performed the measurement (S.H.).²⁷

The evaluators were blinded to whether the participants were experiencing shoulder pain during volleyball play at the time of conducting the assessments. This blinding ensured that the evaluators’ measurements were not influenced by knowledge of the participants’ pain status.

Sample Size

The sample size was determined based on logistic regression analysis, considering 2 predictors (ER ROM and imbalance in isometric muscle strength of IR and ER) which are critical factors influencing shoulder pain, as indicated in previous studies.^10–12 The dependent variable event rate was estimated to be 60%.^1,2,12 Using the formula proposed by Peduzzi et al, the required sample size was calculated as follows: N = (10 × 2)/0.6 ≈ 33 per group.²⁹ Furthermore, considering additional factors such as joint position sense, dynamic stability, and power, the number of predictors increases to 5. Thus, the required sample size would be N = (10 × 5)/0.6 ≈ 83 per group. Therefore, considering at least 2 predictors, a minimum sample size of 33 participants per group is necessary.

Data Analysis

Participants were subgrouped according to the presence of shoulder pain during the attack or serve. Data normality was assessed using the Shapiro-Wilk test. Continuous variables were analyzed with t tests or Mann-Whitney U tests and categorical data with χ² tests. Logistic regression analysis with increasing variables (likelihood ratio) was used to examine the relationship between shoulder pain and specific factors. IBM SPSS Statistics for Windows (version 29.0) was used for all statistical analyses, with a significance level of P = .05. Clopper-Pearson 95% confidence intervals (CIs) for the results of binomial variables were calculated using R (version 4.4.1). A post hoc power analysis was performed using G*Power 3.1.9.7 to ensure the robustness of our findings. Furthermore, the minimal clinically important difference (MCID) was calculated for each variable where applicable, to provide context for the clinical significance of the findings. The MCID was determined using a distribution-based approach, specifically by calculating a portion of the standard deviation combined with effect size (ES; Cohen d). This approach was chosen to ensure that the MCID reflects clinically meaningful changes based on the variability observed within the study population.

Prevalence and Characteristics of Shoulder Pain

A total of 32 participants (54.2%; 95% CI = 40.6, 67.6) reported shoulder pain during attacking and serving, leaving 27, without pain. Among those reporting pain, 68.8% (95% CI = 48.9, 82.9) did not inform their coach. Pain duration responses were pain exclusively during the attack/serve without occurrence during other activities (n = 14, 43.75%; 95% CI = 27.4, 60.8), pain during both the attack/serve and other activities, ceasing after practice (n = 5, 15.625%; 95% CI = 5.3, 32.8), pain persisting until after practice but not extending to the next day (n = 7, 21.875%; 95% CI = 9.3, 39.9), and pain lasting until the next day (n = 6, 18.75%; 95% CI = 7.2, 36.4).

Group Comparisons: Demographics, Environmental Factors, and Shoulder Function

The results of the between-groups comparisons are shown in Table 1. The number of different sports participated in during elementary school was more frequently reported as 1 sport or less in the pain group (93.8%) compared with the no-pain group (70.4%). These sports included not only volleyball but also swimming, soccer, basketball, karate, and others. Regarding attack form, a higher proportion of the pain group used a bow and arrow swing (pain group: 65.6%; no-pain group: 37.0%). Circular swings were not used in either group, and straight arm swings were more common in the no-pain group (pain group: 34.4%; no-pain group: 63.0%). No significant differences were found between the groups in terms of age, height, weight, BMI, dominant hand, volleyball experience, sleep hours, court position, serve type, and attacking awareness.

Table 1. Group Comparison of Demographics, Environmental Factors, and Shoulder Function^a

View Fullscreen

For shoulder joint function, the pain group had a greater joint positional sense error for IR (pain group: median 3.6° [25%–75%: 2.5–5.3]; no-pain group: 2.5° [1.6–3.6]; P = .03, ES = 0.61). However, the observed difference did not meet the MCID threshold of 1.47°, suggesting that, while statistically significant, the difference may not be clinically meaningful. The pain group also had a lower UQYBT composite score (pain group: 86.6 ± 8.9; no-pain group: 91.2 ± 5.8; P = .03, ES = 0.61; 95% CI = −8.63, 0.61). The MCID for the UQYBT composite score was 4.85 points, indicating that the observed difference exceeds the threshold for clinical significance. Additionally, the pain group showed significantly lower UQYBT medial-%LL (pain group: 98.2 ± 9.7; no-pain group: 105.0 ± 6.7; P = .003, ES = 0.82; 95% CI = −8.93, −1.18) and UQYBT superolateral-%LL (pain group: 71.4 ± 10.7; no-pain group: 77.8 ± 7.8; P = .01, ES = 0.68; 95% CI = −8.8, −1.52). The MCID values for UQYBT medial-%LL and UQYBT superolateral-%LL were 7.44%LL and 6.76%LL, respectively, supporting the clinical relevance of these findings.

Logistic Regression Analysis of Shoulder Pain Factors

A logistic regression explored the relationship between specific factors and shoulder pain. Volleyball elements (attack form) and shoulder joint function (IR joint position sense, UQYBT-medial [%LL], UQYBT-superolateral [%LL]) were used as independent variables. Shoulder pain was the dependent variable, focusing on items with significant between-groups differences. Results are in Table 2. Significant factors associated with shoulder pain were UQYBT-Medial (P  = .008) and joint position sense-IR (P  = .031). Odds ratios for shoulder pain were 0.89 (95% CI = 0.82, 0.97) for UQYBT and 1.41 (95% CI = 1.03, 1.93) for joint position sense-IR. The Hosmer-Lemeshow test for this model indicated compatibility with P  = .27, and the percentage of correct classifications was 67.8%. As the final logistic regression model included 2 independent variables, the initial sample size requirement of 33 participants was met, indicating that the sample size was sufficient for this analysis.

Table 2. Factors Associated With Current Shoulder Pain in Female High School Volleyball Players^a

View Fullscreen

Limitations

Despite the insights gained in this study, certain limitations warrant consideration. The cross-sectional design restricts the establishment of causal relationships between variables. Authors of longitudinal studies could provide a more comprehensive understanding of how changes in shoulder joint function and volleyball-specific elements contribute to the onset and progression of shoulder pain over time. Additionally, in this study, we focused exclusively on female high school volleyball players, limiting the generalizability of the findings to other populations. Our findings do not apply to male players, as gender differences in shoulder ROM exist.⁶ Additional research is needed to understand the effect of shoulder pain on shoulder function and performance tests in males. However, our findings may be more directly applicable to women’s teams with practice times and frequencies like those observed in the present study. Furthermore, the glenohumeral ROM assessments were conducted by 2 physical therapists, 1 stabilizing the scapula and the other moving the humerus. While this method ensured precise measurements, it is not typical in clinical practice, where 1 clinician usually performs both tasks. This may affect the generalizability of the findings. Similarly, in the joint position sense assessments, while participants’ eyes were closed to eliminate visual feedback, a blindfold was not used, and the digital inclinometer was applied lightly to the forearm during measurement. Additionally, the lack of backrest in the chair may have required participants to actively stabilize their trunk, potentially introducing variability. These factors should be considered when interpreting the joint position sense results. Finally, we did not obtain detailed information on the number of attacks or serves performed per practice or per competition. This limitation prevents us from assessing the cumulative load of spikes and serves on the development of shoulder pain. Future researchers should investigate how these cumulative loads contribute to shoulder pain to develop more targeted preventive measures.