Introduction
Swimming demands repetitive and intense shoulder rotation movements. Elite swimmers may cover up to 14,000 meters/week, performing around 16,000 rotations weekly, placing significant strain on joint structures. The shoulder complex consists of four joints and several muscles, and its integrity ensures stability and movement precision. Pain or injury-related changes can compromise motor control and the kinematics of this joint complex, predisposing it to dysfunction. Consequently, pain (defined as an “unpleasant sensory and emotional experience”) in shoulder complex is reported by 40-91% of athletes, and it is an early sign of overload or incipient injury. This high incidence underscores the need for specific preventive strategies for this population.
Early and multidimensional assessments are essential to detect shoulder dysfunction and guide effective prevention and rehabilitation strategies. In this context, Feijen et al. developed a prognostic model for shoulder pain in young swimmers, identifying the acute-chronic workload ratio as the strongest predictor (4.3 times higher risk). Souza et al. had already associated competitive swimming with shoulder pain and signs of impingement and rotator cuff injuries, while Kennedy et al. linked range of motion, laxity, and muscle imbalances to pain, emphasizing age- and level-specific assessments. Finally, Porter and collegues showed that an acute increase in supraspinatus tendon thickness after training predicts pain at 3 and 6 months, reinforcing the importance of individualized monitoring. However, these studies did not investigate measures of shoulder sensorimotor control.
The stability of this joint depends on the interaction between static and dynamic components, mediated by the sensorimotor system,, and dysfunction in this joint is associated with proprioceptive and neuromuscular control alterations., For assessing shoulder sensorimotor control in competitive swimmers, the Upper Quarter Y Balance Test (UQ-YBT) is a valid and reliable tool, encompassing strength, stability, and mobility. Its results are influenced by factors such as maturation and competitive level., Recently, Albuquerque et al. integrated a force platform assessment into the UQ-YBT, previously used in static conditions., In young swimmers, performance was minimally affected by laterality, while center of pressure displacements was influenced by task demands. Importantly, performance and sensorimotor control on the UQ-YBT were not affected by sex or the presence of shoulder pain at the time of assessment.
The present study aimed to expand on previous findings by prospectively investigating the association between shoulder sensorimotor measures, clinical scores, and functional scores, and the subsequent occurrence of pain in youth swimmers. Considering the multifactorial nature of pain, we sought to identify, among the applied clinical and functional tests, a set of factors that could provide insights into the early identification of shoulder injury risk.
Methods
Ethical considerations
The study was approved by the Institutional Research Ethics Committee prior to its execution (process no. 4.000.902 on April 30, 2020). All parents or guardians of the participants signed an informed consent form after receiving an explanation of the study, and before any procedures were performed.
Study Design
This prospective cohort study was conducted from March to October 2023. The flowchart shown in Figure 1 describes the research protocol. First, athletes were asked to complete an anamnesis and two questionnaires. Subsequently, the Upper Quarter Y Balance Test (UQ-YBT) was performed over a force platform, where participants had to reach in three directions (medial, inferolateral, and superolateral) with each upper limb. Finally, six months later, a follow-up questionnaire was administered to assess shoulder symptoms.
Setting and Participants
The study sample comprises swimming athletes from a local elite sports club, aged 13 to 17 years, who train 5 to 7 times weekly and swim between 3000m and 8000m daily. Exclusion criteria include a history of incapacitating musculoskeletal disorders, neurological conditions, major trauma, or prior surgery on the upper limbs, trunk, or lumbar spine, as well as the current use of anti-inflammatory medication.
Baseline assessment of clinical and functional measures
An interview was taken to collect athletes’ sociodemographic, anthropometric, and sports-related information.
Shoulder instability symptoms were assessed using the Brazilian Portuguese version of the Western Ontario Shoulder Instability Index (WOSI),, which consists of four domains related to: physical symptoms; sports, recreation, and work; lifestyle; and emotional factors. The WOSI contains 21 items, with responses given on a 100-mm visual analogue scale, ranging from "no complaints" (0) to "severe complaints" (100). The items are summed into four scores, with a total score ranging from 0 to 2100, where 0 indicates no limitations and 2100 corresponds to extreme limitations.
To determine the participant's preferred upper limb, we used the Edinburgh Handedness Inventory. This questionnaire was administered in an interview format, where each subject indicates their preferred hand for each of the listed manual activities by selecting the right or left column for their response.
Baseline assessment of Upper Quarter Y Balance Test
First, the upper limb length was measured with a tape measure (precision of 0.5 cm), with participants standing, arms abducted to 90°, elbows extended, forearms in neutral position, and thumbs pointing upward. The measurement was taken from the spinous process of the 7th cervical vertebra to the tip of the middle finger.
Next, the athletes were positioned on a force platform in a push-up posture, with the shoulders and wrists at 90 degrees of flexion, elbows and knees extended, trunk straight, and feet hip-width apart. The head remained in a neutral position, in line with the trunk. In this position, body weight is distributed between the athlete’s feet and the supporting hands (randomly assigned as the reaching or supporting limb), which stay in contact with the platform.
After a familiarization period with the procedure, athletes performed the UQ-YBT adapted for the force platform, using a wooden ruler (5 cm in height) positioned in each of the three directions. This test challenges shoulder mobility and stability, requiring the participant to bear weight on the supporting limb on the platform, while reaching with the contralateral limb in the medial, inferolateral, and superolateral directions. Each direction was tested three times in random order, with reach distance and posturographic data recorded. UQ-YBT reach distances were averaged and normalized by upper limb length.
Baseline assessment of postural sway
Center of pressure (COP) coordinates were acquired at 30 Hz using a force platform (Wii Balance Board, Nintendo Co Ltd, Japan). Data was transmitted via Bluetooth to a laptop running a custom LabVIEW program (National Instruments, USA). The platform was calibrated according to the manufacturer's guidelines. The COP signal was resampled to 100 Hz using the SWARII algorithm, converted into resultant distance (square root of the sum of the squared displacements in the lateral and anteroposterior directions), and expressed as path length (summed distances between consecutive points) for each direction. Data processing was performed offline in the Python 0.5.4 environment.
Outcome assessment
After six months, a follow-up phone call was made to assess sports-related shoulder symptoms. The questionnaire evaluated the occurrence of pain in the previous six months, including: specific conditions or activities associated with it; affected body parts; duration of activity suspension; whether treatment was sought; and whether the athlete resumed their usual activities. Athletes were classified as positive (=1) or negative (=0) for pain based solely on reported shoulder symptoms (with/without pain symptoms, respectively).
Statistical analysis
Missing data (approximately 6% of total) from functional assessments (UQ-YBT performance and force platform variables) were addressed using Multiple Imputation by Chained Equations (MICE). Normality assessment indicated that 60% of the variables showed Gaussian distribution (Shapiro-Wilk test, P>0.062).
Group comparisons (positive vs. negative shoulder pain symptoms) used Welch's t-tests for unequal sample sizes, with Cohen's d effect sizes being computed (trivial: <0.2; small: 0.2–0.5; moderate: 0.5–0.8; large: >0.8). False Discovery Rate (FDR) correction maintained false positives below 5%.
Variables with large effect sizes (d>0.8) entered univariate logistic regression to assess predictive value for pain status. Predictors included age, WOSI subscores (physical symptoms, sports/recreation/work, emotion, and total score), with previous shoulder injury/pain as covariate. Continuous predictors were Z-score normalized (mean=0; SD=1) to allow for direct comparison of effect sizes across different measurement scales. All logistic regression models were fitted using maximum likelihood estimation, with results expressed as odds ratios (OR) with corresponding 95% confidence intervals.
All analyses were performed in Python 3.11.7, using the pingouin (version 0.5.5) and statsmodels (version 0.14.0) packages. Statistical significance was set at 5%.
Results
Participants
Thirty-four young athletes were originally enrolled, but two dropped out before the study's completion due to leaving the sport. The study included 32 competitive swimmers (18 female, 14 male) aged (mean and range) 14 (13-16) years, with a body mass of 59.2 (44.0-80.0) kg, height of 169 (158-187) cm, and BMI of 20.6 (15. 8-23.8) kg/m². Participants had been swimming for 7 (3-13) years and reported an average training volume of 7.1 (6.0-9.5) km/week. The sample was predominantly right-handed (N=30, 93.8%), with an equal distribution of swimmers with and without a history of previous shoulder injury/pain (N=16 in both groups). Regarding swimming specialization, the cohort consisted of 17 sprinters, 14 long-distance swimmers, and 1 middle-distance swimmer. The most common stroke was freestyle (N=14), followed by butterfly (N=8), breaststroke (N=5), backstroke (N=4), and individual medley (N=1).
Shoulder pain symptoms after 6 months
Of the total participants, 9 (28%) experienced shoulder pain in the last six months, while 23 (72%) reported no shoulder pain. All affected athletes reported pain during training, with the following distribution: bilateral shoulder pain (N=6), right shoulder pain (N=2), and left shoulder pain (N=1). Regarding physical therapy, seven sought treatments, while two did not. All athletes with shoulder pain symptoms returned to or maintained their regular sports activities without prolonged interruption (at least one day off due to pain was taken). Of the 9 athletes with positive pain symptoms, 6 (67%) had already reported shoulder injury/pain in the previous 6 months; these overlapping was accounted in logistic modeling.
Group comparisons
Welch's independent sample t-test revealed no significant differences between groups regarding height, weight, BMI, practice time, or swimming volume (P>0.104; Table 1). The only exception was the age, where the no pain symptoms group was older than the positive group. In addition, age showed large effect size, warranting its inclusion in the subsequent regression model.
After FDR correction, close-to-threshold P-values and large effect sizes were found for several WOSI’s subscales (all P=0.061;Fig. 2): physical symptoms domain (d=0.936;Fig. 2A); the sports/recreation/work domain (d=1.017;Fig. 2B); the emotion domain (d=1.026;Fig. 2C); and the total WOSI score (d=0.959;Fig. 2D). All showed a tendency for higher scores for those with positive pain symptoms.
For the UQ-YBT, no significant differences were observed in composite scores or normalized reach distances (P>0.962; Table 1). Similarly, path length measurements showed no significant differences between groups (P>0.348; Table 1). The computed effect sizes were all below d=0.583.
In general, the most pronounced differences between groups were in age and shoulder function (WOSI), while dynamic balance and functional measures were largely similar between groups.
Predictive regression modeling
Logistic regression analysis (Table 2), considering previous shoulder pain as covariate, revealed significant associations between shoulder-related outcomes and demographic and clinical measures. Along with age, the WOSI sports and emotion subscale, together with close-to-threshold effect for WOSI physical function and total score, emerge as significant predictors. Overall, higher disability scores increased the odds of the outcome (shoulder pain symptoms) while increased age reduces its odds.
| Odds ratio | |||||
|---|---|---|---|---|---|
| Predictor variable | Beta | P-value* | OR | CI95%, lower | CI95%, upper |
| Age | -1.672 | 0.012 | 0.188 | 0.051 | 0.698 |
| WOSI, physical | 0.893 | 0.061 | 2.444 | 0.960 | 6.220 |
| WOSI, sports | 0.943 | 0.030 | 2.569 | 1.095 | 6.027 |
| WOSI, emotion | 0.901 | 0.044 | 2.461 | 1.025 | 5.912 |
| WOSI, total | 0.892 | 0.050 | 2.441 | 0.999 | 5.968 |
Discussion
In this prospective study of 32 young swimmers, we investigated clinical and functional factors associated with shoulder pain development. Results showed that athletes reporting shoulder pain after six months already exhibited higher baseline WOSI scores, indicating greater dysfunction and pain perception, while functional measures (UQ-YBT and postural displacements) did not differ significantly between groups. Regression analysis revealed that higher WOSI sub-scores, along with younger age, increased pain risk, suggesting clinical assessments have greater predictive value than functional tests for this outcome in young competitive swimmers.
Our findings align with previous research. Freijen et al. identified five significant predictors of shoulder pain in swimmers: acute-to-chronic workload ratio, competitive level, posterior shoulder muscle endurance, flexion range of motion, and hand entry error. Souza et al. demonstrated that competitive swimming is associated with higher frequency of shoulder pain and rotator cuff injuries, suggesting excessive training load without adequate control contributes to chronic pain development. Finally, Kennedy et al. and Porter et al. highlighted that anatomical and kinesiological variables – such as range of motion, joint laxity, and supraspinatus tendon thickness – are associated with shoulder pain risk.
Our results corroborate previous findings showing that the WOSI is a robust predictor of subjective disability perception,, with elevated scores correlating with pain presence. Conversely, the UQ-YBT performance was insufficiently sensitive to detect functional deficits related to subclinical pain stages, consistent with studies finding no significant correlation between this test and shoulder injury.
Pain alone does not confirm injury presence. However, its persistence, particularly when accompanied by positive clinical tests, significantly increases musculoskeletal injury risk. According to Hoegh et al., distinguishing between sport-related pain (absence of measurable tissue damage, typically associated with training adaptations) and sport-related injury (objective clinical signs or imaging findings compatible with structural impairment) is essential. In this context, disabling pain measured by WOSI represents a more advanced stage in the pain-injury continuum, associated with greater dysfunction risk and higher structural injury probability. This distinction, evidenced by higher scores across all WOSI domains (physical symptoms, sports/recreation/work, emotional, and total score) in positive shoulder pain athletes, captures aspects functional tests cannot detect, such as perceived instability, shoulder insecurity, and emotional impact.
Corroborating this view, body height, weight, and swimming volume were not significant predictors of pain presence in the studied sample. This suggests that, in homogeneous groups regarding activity level, anthropometric and load factors alone may not explain pain onset in athletes with a history of instability. It is important to note, however, that these variables may act as adjusting predictors or even as confounding factors in more complex models and should therefore be considered in the design of future studies.
Limitations
Study limitations include sample size and assessment interval. A larger sample could reveal differences between symptomatic and asymptomatic groups not observed here. Additionally, longer follow-up would increase shoulder the probability of pain occurrence, making group comparisons more equivalent. However, observed differences, even after multiple comparison correction, showed strong effect sizes, indicating relevant pain and functionality alterations despite sample size asymmetry. Conversely, longer intervals could increase sample loss due to sport dropout (as occurred with two athletes) or contact loss. Despite limitations, results provide important insights into clinical decision-making and future investigations.
Conclusion
WOSI are useful instruments for identifying athletes at higher risk of developing or exacerbating shoulder complex dysfunction. Elevated scores on these scales are associated with subsequent pain development, even when objective functional tests (UQ-YBT and hand-support postural displacement measures) detect no significant alterations. Our findings corroborate current models positioning perceived disability as a more sensitive risk marker in the pain-injury continuum. Future studies should validate these scales' predictive power and establish clinical cut-off points in sports contexts.
Funding
This study was funded by the Carlos Chagas Filho Foundation for Research Support of the State of Rio de Janeiro (FAPERJ; grant number E-26/211.104/2021) and the Coordination for the Improvement of Higher Education Personnel (CAPES; Finance Code 001, grant number 88881.708719/2022-01, and number 88887.708718/2022-00).
Acknowledgments
We would like to thank the CUIDAR Project of the Clube de Regatas do Flamengo, represented by Carlos Alexandre Souto de Assis, for their support during the development of this research.
References
1.
2.
3.
IASP. IASP Revises Its Definition of Pain for the First Time Since 1979. 2020. Accessed May 7, 2025. https://www.iasp-pain.org/wp-content/uploads/2022/04/revised-definition-flysheet_R2-1-1-1.pdf
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.