Is it relevant to distinguish scapulothoracic abnormal motion (STAM) type in professional overhead athletes?

Abstract

Background

Scapular dyskinesis, further characterized within the broader concept of scapulothoracic abnormal motion (STAM), is common in both symptomatic and asymptomatic overhead athletes. While STAM type 3 (winging) is a distinct pathological entity, types 1 and 2 remain less well defined and may represent sport-related adaptations rather than disease. The clinical relevance, prevalence, and sport-specific patterns of these conditions are still unclear. The objectives of this study were to: (1) evaluate the prevalence of STAM types 1 and 2 in the dominant and non-dominant shoulders of professional overhead athletes compared with a control population; (2) identify clinical consequences associated with STAM; (3) assess the impact of different overhead sports on STAM; and (4) explore the association between STAM and other shoulder conditions, such as glenohumeral internal rotation deficit (GIRD) and hyperlaxity.

Methods

45 professional in season overhead athletes (15 water polo, 16 basketball, and 14 CrossFit) were included in the study group. Control group consists in 21 volunteers with no or minimal shoulder involvement in sport activity. STAM type was diagnosed using clinical examination and the shoulder flexion resistance test. The shoulder passive range of movement, pain, hyperlaxity, GIRD, American Shoulder and Elbow Surgeons (ASES) score, and the Kerlan-Jobe score were also recorded. A statistical analysis was run to identify any differences in STAM between study and control groups, the clinical role of STAM types, any clinical factor associated to STAM, and the differences between the 3 selected types of overhead sports.

Results

STAM type 1 was found more frequent in the dominant shoulder of athletes than in the control group (q = 0.045). No specific factor associated with scapular dyskinesis was identified in the univariate or multivariate regression analysis. However, STAM type 2 was associated with general joint hyperlaxity and the Beighton score (q = 0.04). A diagnosis of STAM type 1 or type 2 was not associated with higher pain scores, a higher GIRD rate, or lower Kerlan-Jobe or ASES scores.

Conclusion

STAM type 1 was more common in the dominant shoulder of professional overhead athletes. In contrast, STAM type 2 was observed similarly in the dominant and non-dominant shoulders in both the study and control groups and was associated with generalized joint hyperlaxity. STAM types 1 and 2 were not associated with pain, GIRD, ASES score, or Kerlan-Jobe score in this cohort. These findings support epidemiologic differences between STAM subtypes, but do not establish causal mechanisms, distinct biological etiologies, or definite clinical relevance.

Scapular dyskinesis, defined as an abnormal scapular position at rest or during arm elevation, is a frequent condition observed both in asymptomatic individuals and in patients affected by shoulder pain.¹⁷ Muscular imbalance, nerve palsy, scoliosis, acromioclavicular separation or fractures, and thoracic deformity can alter scapulothoracic rhythm, leading to asymmetry between the right and left hemithorax.¹⁸

In 2022, Elhassan et al first introduced the term “scapulothoracic abnormal motion” (STAM),⁸ to encompass, within the same group of pathologies, scapular dyskinesis and winging. The same author recently classified STAM into 7 different types.¹³ The first 3 types of STAM are the more frequently observed, and a specific clinical test, the shoulder flexion resistance test (SFRT), was described to identify a true serratus anterior palsy from dyskinesis.²⁴ STAM type 1 is caused by hyperactivity of the pectoralis minor and upper trapezius resulting in an anterior and internal rotation of the scapula and posterior prominence of the inferior tip. Type 2 is associated with a serratus anterior dysfunction plus a pectoralis minor hyperactivity which cause scapula anterior rotation on the sagittal plane with medial border elevation. Type 3 consists of scapular winging secondary to serratus anterior palsy. While STAM 3—scapula winging—is a specific disease associated with limited anterior elevation and shoulder weakness, STAM types 1 and 2, also referred as scapula dyskinesis, are clinically less well-defined conditions that can be observed in conjunction with shoulder instability, rotator cuff tears, subacromial impingement, shoulder pain, or even in asymptomatic individuals.

Although Kibler et al already described in 2002²² differences between prominence of the inferior scapula border linked to a rotation around a traverse axis (type I) and prominence of the medial border secondary to a rotation around a vertical axis (type II), in the previous literature, to improve interobserver agreement, the diagnosis of dyskinesis has mostly relied on a dichotomous evaluation (yes/no) without detailed differentiation between the types.^9,28,37

Scapula dyskinesis has been reported in 38%-82% of asymptomatic overhead athletes,^2,4,25,30 but no clear association has been established regarding the type of overhead sport, injury risk, or dyskinesis pattern. It remains unclear whether dyskinesis is a sport-specific adaptation, whether preventative interventions are needed, or if STAM type influences shoulder symptoms.^20,21,33

The objectives of this study were: (1) to evaluate the prevalence of STAM types 1 and 2 in the dominant and non-dominant shoulders of professional overhead athletes, compared with a control population; (2) to identify any clinical consequences associated with STAM; (3) to assess the impact of different overhead sports on STAM; and, finally, (4) to explore the association between STAM and other shoulder conditions, such as glenohumeral internal rotation deficit (GIRD) and hyperlaxity.

Materials and methods

We designed a cross-sectional comparative study that included a study group of professional overhead athletes and a control group of age-matched volunteer medical students. A power analysis was performed to determine the study sample size, taking into account a 40% prevalence of scapula dyskinesis in overhead athletes,² setting the test power value to 80% and alpha at 0.05. A total of 42 shoulders were needed in each group. The study was powered for the primary comparison between the main groups. Subgroup analyses within the study group were performed as secondary exploratory analyses and were not separately powered.

Study group

The study group consisted in 45 healthy, in-season professional overhead athletes. We included 3 distinct types of overhead sports based on shoulder mechanics: (1) overhead throwing in abduction-external rotation (ABER) with dominant arm (water polo), (2) overhead throwing across various arm positions (basketball), and (3) overhead loading and lifting with resistance, using both arms (CrossFit). Fifteen water polo players (first Italian league), 16 basketball players (first Italian league), and 14 CrossFit athletes (competing at national and international levels) were included in the study. To reduce bias related to sex and tissue elasticity differences, only males were included in both study and control groups. Indeed, generalized joint hyperlaxity shows variable sex differences across studies and we decided to not include female subjects to not introduce a bias on the hyperlaxity variable. Athletes affected by any upper extremity injury which caused an activity stop in the last 2 months before the examination were excluded.

Control group

The control group consisted of 21 healthy, non-smoking volunteer medical students matched for age to the study group. Of these, 15 participated in sports with minimal upper limb involvement (9 soccer, 4 fitness, and 2 running) with 1 to 2 training session per week, and 6 did not participate in sports.

Clinical examination

Two expert shoulder and elbow surgeons (R.d.G. and A.G.) trained to STAM diagnosis using video presentations and multiple clinical patients evaluation recorded demographic data (age, dominance, sport start age, training frequency, and previous shoulder and elbow injuries) and clinical data (Beighton score, shoulder hyperlaxity with sulcus, Gagey and Coudane-Walch tests, range of motion, O'Brien test, pain assessment in internal and external rotation and during sport, Kerlan-Jobe, and American Shoulder and Elbow Surgeons [ASES] scores) of the participants. Examiners were blinded to dominance, but not to sport type. Range of motion (external rotation with the arm at the side, external rotation with the shoulder abducted at 90° [ER2], internal rotation with the shoulder abducted at 90°) was measured with a smartphone inclinometer in standing position.²⁹ Pain was measured both as continuous data with visual analog scale (VAS) and as discrete data (yes/no) when the arm was in ABER position. Tenderness at pectoralis minor insertion was assessed by palpation. The presence of scapula dyskinesis and STAM types were evaluated asking participants to perform 5 rounds of active shoulder flexion and abduction while holding a 2-kg weight as already described.¹⁹ The SFRT maneuver (Fig. 1) was also used to exclude patients with serratus anterior paralysis (type 3 STAM) from the study²⁴ and to assess STAM type. Each shoulder was classified as normal (symmetric stable scapula movement during arm elevation without tenderness at pectoralis minor insertion), STAM type 1 (anterior rotation of the scapula on the axial plane with prominence of the inferior angle, negative SFRT at 30-60-100° and tenderness at pectoralis minor insertion), or STAM type 2 (prominence of the medial border and an SFRT positive at 30-60° and negative at 100°). GIRD was defined as loss of internal rotation (>25° compared to the contralateral side), which exceeded the external rotation gain.³⁹ The 2 examiners independently assessed each participant, and in case of disagreement, the senior author (A.C.) was involved in discussion until a consensus was reached.

Figure 1.

Scapulothoracic abnormal motion (STAM) and shoulder flexion resistance test (SFRT). (A) STAM type 2 during anterior elevation of the non-dominant arm of a cross-fit athlete. (B) SFRT positive both at 30° and 60°. (C) STAM type 1 in the right dominant shoulder of control group patient shows hyperactivation of pectoralis minor.

Patient and public involvement

Athletes and medical students were first involved at the earliest conceptualization stage of this research. Before the study commenced, key stakeholders, including representatives from the sport society, team doctors, and the athletes themselves, were involved in dedicated briefing sessions to explain the study objectives and methods. The same detailed explanation and engagement process was undertaken separately with the volunteer medical students. While the core study design and outcomes were not modified directly by participants, these briefing were used to better explain the use of self-reported outcome scales and to fulfill informed consent forms. As requested by participants, the frequency of data collection points was increased and the duration of certain assessments was reduced, aiming to minimize participant burden while maintaining scientific rigor. Finally, the individual results of the study were communicated directly to participants, and the study outcomes were shared with the sport society team to improve athlete preparation and facilitate the setting of individualized physiotherapeutic sessions.

Statistical analysis

We used the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) reporting guideline¹⁰ to draft this manuscript and the STROBE reporting checklist¹¹ when editing. Continuous data were reported as mean or as numbers (percentages) ± standard deviation. The distribution of variables was tested using the Shapiro-Wilk test. Comparison between study and control groups was performed using Welch's t-test for continuous variables and Fisher's exact test for categorical variables.

Because multiple statistical comparisons were performed, the risk of type I error inflation was addressed by applying the Benjamini-Hochberg procedure to control the false discovery rate. Adjusted q-values were calculated across the P values reported in the comparative tables. Statistical significance was interpreted primarily using the false discovery rate‑adjusted q-values.

Because bilateral shoulder data were analyzed, the potential non-independence of shoulders from the same participant was addressed using generalized estimating equations (GEE). Each participant was treated as a clustering unit, with an exchangeable working correlation structure to account for within-subject correlation between shoulders. Continuous outcomes were analyzed using Gaussian GEE models, and binary outcomes were analyzed using logistic GEE models. Group status and shoulder dominance were included as model covariates. Results from the clustered models were used as a sensitivity analysis to confirm the robustness of the primary findings.

Interobserver reliability for scapula dyskinesis and STAM type diagnosis was assessed using the Cohen's kappa coefficient, and the results were interpreted using the Landis and Koch thresholds (k > 0.61 substantial agreement; 0.41< k < 0.6 moderate agreement; 0.21 < k < 0.4 fair agreement; 0 < k < 0.2 slight agreement). Interobserver reliability for continuous variables, as range of motion, was assessed using a 2-way random-effects intraclass correlation coefficient (ICC) for absolute agreement (ICC: 2.1), with 95% confidence intervals (CIs).

Univariate regression was conducted to investigate potential risk factors for scapula dyskinesis, isolated STAM type 1 and isolated STAM type 2. Because STAM 2 positivity was observed in only 8 participants, we used Firth penalized logistic regression to reduce small-sample bias and avoid instability related to sparse data. Those variables potentially associated with the outcome of interest (ie, with a P value < .05) were then included in a multiregression model. Analysis of variance and Tukey post hoc tests were used to compare the study variables among the different sports and to analyze factors associated with shoulder hyperlaxity and GIRD. No missing data were recorded. The level of significance was set at P < .05. Statistical analyses were performed using Statistical Package for Social Sciences, version 29.0.1.0 (IBM, Armonk, NY, USA).

Results

Demographic and clinical findings for the study and control groups are summarized in Table I. The groups were comparable for age (P = .3) and Beighton score (P = .6), but differed in the number of trainings sessions (P < .001), ASES score (P = .01), and Kerlan-Jobe score (P < .001).

Table I.

Differences between the study and control groups

Variables	Study group	Control group	P value
Age	24.6 ± 0.6	26.1 ± 1	.3
Beighton	2.5 ± 0.4	2.9 ± 0.5	.6
Symptoms age	9.3 ± 0.9		-
N° training	6.5 ± 0.1	1.7 ± 0.4	<.001
ASES	93.5 ± 2	100	.01
Kerlan-Jobe	89.4 ± 2.3	99.7 ± 0.3	<.001

ASES, American Shoulder and Elbow Surgeons.

Statistically significant results are indicated in bold italics.

Several differences were observed when comparing overhead athletes with the control group. In particular, the study group exhibited higher VAS pain score and a higher prevalence of GIRD in the dominant shoulder (Table II). In the clustered GEE analysis accounting for bilateral shoulder observations within participants, the main findings were generally consistent with the unadjusted comparisons. Significant group differences remained for VAS, internal rotation, GIRD, and ER2. No significant group differences were observed for external rotation with the arm at the side, dyskinesis, STAM, or pain in ABER after accounting for clustering. After accounting for bilateral clustering using GEE, the primary interpretation was unchanged: STAM type 1 remained more frequent in the dominant shoulder of athletes, whereas STAM type 2 remained associated with generalized joint hyperlaxity. Nevertheless, the CIs were wide for subtype-specific analyses, reflecting the limited number of STAM subtype events.

Table II.

Clinical findings by study group and arm dominance

Variables	SG dominant arm	CG dominant arm	Difference 95% CI	Unadjusted P value	BH-adjusted q	SG non-dominant arm	CG non-dominant arm	Difference 95% CI	Unadjusted P value	BH-adjusted q
Gagey	14 (31%)	1 (4.5%)	26.6% (5.8 to 41.6)	.014	0.052	11 (24%)	0 (0%)	24.4% (6.4 to 38.7)	.012	0.046
Sulcus	1 (2%)	2 (9%)	−6.9% (−25.7 to 4.5)	.25	0.48	1 (2%)	3 (14%)	−11.4% (−31.2 to 1.5)	.1	0.24
Coudane Walch	4 (9%)	3 (14%)	−4.7% (−25.2 to 10.1)	.67	0.78	2 (4%)	2 (9%)	−4.6% (−23.6 to 7.6)	.59	0.7
Elbow hyperextension	8 (18%)	6 (27%)	−9.5% (−32.0 to 10.1)	.52	0.65	10 (22%)	5 (23%)	−0.5% (−23.4 to 18.4)	1.0	1.0
VAS	2.1 ± 2.4	0.3 ± 0.9	1.8 (1.0 to 2.6)	<.001	<0.001	0.4 ± 1.4	0.0 ± 0.0	0.4 (−0.0 to 0.8)	.06	0.16
IR	66.1° ± 12.4	75.3° ± 8.6	−7.9 (−12.5 to −3.2)	.002	0.008	76.3° ± 11.7	75.9° ± 8	0.1 (−4.8 to 5.0)	.43	0.6
GIRD	22 (49%)	3 (14%)	35.3% (11.1 to 51.9)	.007	0.028	3 (7%)	2 (9%)	−2.4% (−21.7 to 10.5)	1.0	1.0
ER1	78.5° ± 9.7	81.1° ± 12.9	−2.7 (−9.0 to 3.6)	.4	0.59	78.1 ± 11.1	79.3 ± 12.6	−1.2 (−7.6 to 5.2)	.7	0.78
ER2	106.7° ± 12.5	125° ± 19.5	−18.3 (−27.6 to −9.0)	<.001	0.003	106.9 ± 12.2	123.2 ± 19.1	−16.3 (−25.4 to −7.2)	<.001	0.006
Dyskinesis	9 (20%)	1 (4%)	11.0% (−7.9 to 24.8)	.312	0.5	10 (22%)	7 (32%)	−11.8% (−34.6 to 8.9)	.36	0.59
STAM 1	6 (13%)	0 (0%)	11.1% (−5.0 to 23.5)	.018	0.045	7 (16%)	5 (23%)	−7.2% (−29.3 to 11.1)	.51	0.64
STAM 2	3 (6%)	1 (5%)	2.1% (−15.7 to 13.9)	1	1	3 (6%)	2 (9%)	−2.4% (−21.7 to 10.5)	.59	0.7
O'Brien	13 (29%)	2 (9%)	11.0% (−7.9 to 24.8)	.117	0.27	7 (16%)	1 (4%)	11.0% (−7.9 to 24.8)	.26	0.48
Pain ABER	7 (16%)	1 (5%)	11.0% (−7.9 to 24.8)	.256	0.48	8 (18%)	2 (9%)	8.7% (−11.9 to 23.7)	.48	0.63

SG, study group; CG, control group; CI, confidence interval; BH, Benjamini-Hochberg; VAS, visual analog scale; IR, internal rotation; GIRD, glenohumeral internal rotation deficit; ER1, external rotation with the arm at the side; ER2, external rotation with the shoulder abducted at 90°; STAM, scapulothoracic abnormal motion; ABER, abduction-external rotation.

Values are n (%). P values calculated with Fisher’s exact test. Benjamini-Hochberg q was used to control the false discovery rate. For binary variables, the difference is the absolute risk difference in percentage points, calculated as SG minus CG with Newcombe 95% CI. For continuous variables, the difference is the mean difference, calculated as SG minus CG with Welch 95% CI.

Statistically significant results are indicated in bold.

Interobserver reliability was calculated for both scapula dyskinesis (κ = 0.7745) and STAM types (κ = 0.6888), and a substantial agreement beyond chance was observed between the 2 examiners. An interobserver disagreement was found in the 7% (9/132) and the 10% (13/132) of the shoulders, for dyskinesis and STAM types, respectively. Interobserver reliability for inclinometer-based range-of-motion measurements was good, with ICC values ≥ 0.75 and lower 95% confidence limits also exceeding 0.75.

STAM

The prevalence of STAM type 1 in the dominant shoulder was 11.1% in athletes (5/45; 95% CI, 4.8%-23.5%) and 0% in controls (0/21; 95% CI, 0%-15.5%). STAM type 2 events were uncommon: dominant-shoulder STAM type 2 was observed in 6.7% of athletes (3/45; 95% CI, 2.3%-17.9%) and 4.8% of controls (1/21; 95% CI, 0.8%-22.7%). These wide CIs support cautious interpretation of subtype-specific findings.

Scapular dyskinesis was observed in 20% of both dominant and non-dominant shoulders in athletes, with no significant difference between athletes and controls. However, a significant difference emerged when STAM types were considered separately: STAM type 1 was more prevalent in the dominant shoulder of athletes than in controls (q = 0.045) regardless of sport type. STAM type 2 was associated with generalized joint hyperlaxity and higher Beighton scores in univariate analysis and remained associated after adjustment for dominance, age, and case-control status (q = 0.04) across both groups. No specific factor associated with scapular dyskinesis was identified in the univariate or multivariate regression analyses.

A diagnosis of isolated STAM type 1 or type 2 was not associated with increased pain, higher GIRD rates, or lower Kerlan-Jobe or ASES scores.

Sport differences

The outcomes for the 3 types of overhead sports are summarized in Table III. Basketball players experienced significantly less pain in the dominant shoulder compared to water polo (q < 0.003) and CrossFit (q = 0.03). Water polo athletes showed decreased ER2 in both the dominant and non-dominant shoulder compared to basketball (q = 0.05 and 0.007, respectively) and to CrossFit (q = 0.069 and 0.007) athletes. Kerlan-Jobe score was lower in water polo and CrossFit athletes compared to basketball (q = 0.006 and < 0.001, respectively). Additionally, water polo athletes had a higher rate of inferior capsule hyperlaxity (Gagey test) compared with basketball (q < 0.001) and CrossFit (q = 0.013). GIRD was more frequently diagnosed in water polo athletes vs. CrossFit (q = 0.028), although it was not correlated with any other study variables. Supplementary files were added to reports test statistic, degrees of freedom where applicable, and effect size (Supplementary Table S1 for the SG vs. CG comparisons from Table II, Supplementary Table S2 for comparisons across sports from Table III).

Table III.

Study group characteristics by sport

Variables	Water polo	95% CI	Basketball	95% CI	CrossFit	95% CI	Unadjusted P value	Benjamini-Hochberg q
Age	25.5 ± 6.8	21.7 to 29.2	25.5 ± 5.4	22.6 to 28.4	33.6 ± 6.6	30.0 to 37.6	.009	0.005
Beighton	3.5 ± 2.7	2.0 to 5.1	1.9 ± 1.7	1.1 to 2.8	1.7 ± 2.4	0.3 to 3.1	.072	0.19
Agonism age	10.2 ± 2.9	8.6 to 11.8	7 ± 1.6	6.2 to 7.8	12.1 ± 11.4	7.3 to 20.5	.024	0.078
N° training	6.9 ± 0.6	6.6 to 7.3	7.0 ± 0.0	7.0 to 7.0	5.1 ± 0.3	4.9 to 5.2	<.001	<0.001
Gagey dominant	11 (73%)	48.0 to 89.1%	0 (0%)	0.0 to 19.4%	3 (21%)	7.6 to 47.6%	<.001	<0.001
Gagey non-dominant	9 (60.0%)	35.7 to 80.2%	0 (0%)	0.0 to 19.4%	2 (14%)	4.0 to 39.9%	<.001	0.003
Sulcus dominant	0 (0%)	0.0 to 20.4%	1 (6%)	1.1 to 28.3%	0 (0%)	0.0 to 21.5%	.4	0.59
Sulcus non-dominant	0 (0%)	0.0 to 20.4%	1 (6%)	1.1 to 28.3%	0 (0%)	0.0 to 21.5%	.4	0.59
Coudane Walch dominant	1 (7%)	1.2 to 29.8%	1 (6%)	1.1 to 28.3%	2 (14%)	4.0 to 39.9%	.69	0.78
Coudane Walch non-dominant	0 (0%)	0.0 to 20.4%	2 (12%)	3.5 to 36.0%	0 (0.0%)	0.0 to 21.5%	.15	0.32
Elbow hyperextension dominant	3 (20%)	7.0 to 45.2%	3 (19%)	6.6 to 43.0%	2 (14%)	4.0 to 39.9%	.91	1.0
Elbow hyperextension non-dominant	3 (20%)	7.0 to 45.2%	4 (25%)	10.2 to 49.5%	3 (21%)	7.6 to 47.6%	.94	1.0
VAS dominant shoulder	3.6 ± 2.4	2.3 to 4.9	0.4 ± 1.2	−0.2 to 1.1	2.4 ± 2.4	1.0 to 3.7	<.001	0.003
VAS non-dominant shoulder	0.0 ± 0.0	0.0 to 0.0	0.2 ± 0.8	−0.2 to 0.6	1.1 ± 2.2	−0.2 to 2.4	.08	0.2
IR dominant	60.7° ± 7.5	59.2 to 62.1	70.6° ± 8.5	66.1 to 75.2	70.1° ± 11.8	63.2 to 76.8	.003	0.014
IR non-dominant	76.7° ± 11.6	70.2 to 83.1	79.4° ± 11.2	73.4 to 85.4	71.4° ± 11.7	64.7 to 78.2	.5	0.64
GIRD dominant	11 (73%)	48.0 to 89.1%	8 (50%)	28.0 to 72.0%	3 (21%)	7.6 to 47.6%	.02	0.05
GIRD non-dominant	0 (0%)	0.0 to 20.4%	1 (6%)	1.1 to 28.3%	2 (14%)	4.0 to 39.9%	.3	0.55
ER1 dominant	79° ± 9.7	73.6 to 84.4	78.4° ± 11.5	72.3 to 84.6	77.9° ± 8.0	73.2 to 82.5	.95	1.0
ER1 non-dominant	75.3° ± 11.9	68.8 to 81.9	80.3° ± 12.8	73.5 to 87.2	78.6° ± 7.7	74.1 to 83.0	.46	0.62
ER2 dominant	100° ± 11.2	93.8 to 106.2	111° ± 12.8	103.8 to 117.4	109° ± 11.4	102.7 to 115.9	.036	0.1
ER2 non-dominant	97.7° ± 10.7	91.8 to 103.6	113° ± 12.5	106.1 to 119.5	110° ± 6.8	106.1 to 113.9	<.001	0.004
STAM dominant	2 (13%)	7.0 to 45.2%	4 (25%)	10.2 to 49.5%	1 (7%)	1.3 to 31.5%	.39	0.59
STAM non-dominant	3 (20%)	10.9 to 52.0%	1 (6%)	1.1 to 28.3%	5 (36%)	16.3 to 61.2%	.13	0.29
O'Brien dominant	4 (27%)	7.0 to 45.2%	3 (19%)	1.1 to 28.3%	6 (43%)	7.6 to 47.6%	.34	0.59
O'Brien non-dominant	3 (20%)	7.0 to 45.2%	1 (6%)	1.1 to 28.3%	3 (21%)	7.6 to 47.6%	.44	0.6
Pain ABER dominant	1 (7%)	1.2 to 29.8%	4 (25%)	10.2 to 49.5%	2 (14%)	4.0 to 39.9%	.37	0.59
Pain ABER non-dominant	4 (27%)	10.9 to 52.0%	0 (0%)	0.0 to 19.4%	4 (29%)	11.7 to 54.6%	.07	0.19
ASES	86.6 ± 13.3	79.2 to 94.0	100.0 ± 0.0	100.0 to 100.0	83 ± 4.5	nan to nan	.002	0.008
Kerlan-Jobe	92.3 ± 6.7	88.7 to 96.0	98.1 ± 3.6	96.2 to 100.0	67.6 ± 16.9	57.9 to 77.4	<.001	<0.001

CI, confidence interval; VAS, visual analog scale; IR, internal rotation; GIRD, glenohumeral internal rotation deficit; ER1, external rotation with the arm at the side; ER2, external rotation with the shoulder abducted at 90°; STAM, scapulothoracic abnormal motion; ABER, abduction-external rotation; ASES, American Shoulder and Elbow Surgeons; SD, standard deviation; ANOVA, analysis of variance.

Values are mean ± SD for continuous variables and n (%) for categorical variables. CrossFit, n = 14. P values compare the 3 sports. ANOVA was used for continuous variables; chi-square test was used for categorical variables. Benjamini-Hochberg q was used to control the false discovery rate. For continuous variables, 95% CIs are confidence intervals around the group mean using the t distribution. For binary variables, 95% CIs are Wilson confidence intervals around the group proportion. STAM rows were treated as combined STAM positivity for the corresponding side, consistent with the uploaded table counts.

Statistically significant results are indicated in bold.

Hyperlaxity

The Beighton score was positively correlated with all specific hyperlaxity shoulder tests (Sulcus P = .008, Gagey P = .003, Coudane Walch P = .028) and with elbow hyperextension (q < 0.001). However, the specific shoulder laxity tests were not correlated with each other. Only the Beighton score was correlated to STAM 2 (q = 0.04) in the dominant and non-dominant arm across both groups at the regression analysis. Isolated directional hyperlaxity (sulcus, Gagey, and Coudane Walch tests) does not influence STAM 2. Selective shoulder hyperlaxity was not correlated with VAS, Kerlan-Jobe score, ASES score, or any specific shoulder pathology. Only the Gagey test was associated with a higher level of shoulder pain (q = 0.046), but this result was likely linked to the throwing motion typical of water polo athletes.

Sport performance

Lower Kerlan-Jobe and ASES scores, as expected, were both associated with pain (q < 0.001). ASES was also associated with decreased internal rotation (q = 0.031), while Kerlan-Jobe score was associated with reduced external rotation (q = 0.04). VAS was associated with positive O'Brien test (q = 0.004) and reduced internal rotation (q = 0.017). Neither GIRD nor STAM affected pain level and the performance indicators taken in consideration.

Discussion

STAM types

The most relevant finding of our study was that isolated STAM type 1 was more frequently diagnosed in the dominant shoulder of professional overhead athletes than in the control group. Conversely, the overall presence of STAM was similar between the dominant and non-dominant shoulders in both professional overhead athletes and controls. Because of the cross-sectional design, these findings should not be interpreted as evidence that overhead sport causes STAM type 1 or that generalized joint hyperlaxity causes STAM type 2.

A high prevalence of scapula dyskinesis in asymptomatic individuals has already been reported in several studies; however, its clinical significance remains unclear. For example, in professional baseball players, scapula dyskinesis was not associated with a higher incidence of shoulder injuries during the season,³⁴ whereas other studies have correlated it with shoulder pain.¹⁴

Our study highlights an observed difference between STAM type 1, which was more frequently detected in the dominant shoulder of overhead athletes, and STAM type 2, which was more commonly observed in individuals with general joint hyperlaxity, independently of the shoulder activity level. One possible explanation is that repeated overhead activity may be associated with adaptation involving the pectoralis minor and upper trapezius. Conversely, generalized joint hyperlaxity may be associated with altered neuromuscular control or reduced serratus anterior activation, as previously reported for both the lower^16,35 and upper limbs.²³ However, these mechanisms were not directly measured in the present study and remain hypothetical.

Alternative explanations could account for our findings, including pre-existing scapular morphology, athlete-selection effects, unmeasured neuromuscular adaptations, and thoracic posture. Moreover, the role of the pectoralis minor remains uncertain, for example, when considering the rate of STAM reported in patients who have undergone the Latarjet procedure.^3,6 Further studies are needed to evaluate the impact of different muscular structures, thoracic kyphosis,³¹ and objective scapular motion patterns.

In summary, our findings show associations rather than causal relationships. STAM type 1 was observed more frequently in the dominant shoulder of overhead athletes, whereas STAM type 2 was associated with generalized joint hyperlaxity independently of sport type and activity level. These findings should be interpreted as exploratory associations, because pre-existing scapular morphology, training selection effects, unmeasured neuromuscular adaptations, and thoracic posture may also contribute to the observed patterns.

Differences among overhead sports

Water polo, basketball, and CrossFit were selected to represent 3 different overhead-loading profiles: repeated dominant-arm throwing in ABER, multidirectional intermittent overhead activity, and bilateral overhead resistance loading. Other overhead sports, including volleyball, baseball, tennis, and handball, were not included because access to complete professional teams during the study period was not available. This selection increases internal consistency within each sport group but may limit external validity to the broader overhead-athlete population.

We included 3 different types of overhead sports to investigate whether STAM prevalence varies according to the specific athletic gesture involved. Different overhead sports are known to produce distinct movement patterns and sport-specific adaptation in joint laxity and muscle activation.^1,15,41

Water polo is a contact sport which has been associated with selective shoulder hyperlaxity and altered muscle activation, featuring a throwing motion in the ABER position and bilateral overhead activity during swimming.^27,40 Basketball is a contact sport associated with a lower risk of GIRD and shoulder neuromuscular adaptation due to less repetitive overhead motions;^12,38 nevertheless, a 28.7% of scapula dyskinesis has been reported in this type of athletes.³² CrossFit is a non-contact sport, characterized by different shoulder kinematics, increased internal rotation strength, and a decreased external rotation, creating a high injury risk.³⁶

We did not observe any specific differences in STAM frequency or type according to the sport played. However, in our series, water polo and CrossFit were associated with a higher dominant shoulder pain and lower Kerlan-Jobe score compared to basketball, as expected. This supports the hypothesis that overactivation of the pectoralis minor and upper trapezius is likely linked to muscular training rather than to the specific type of sport, as already observed in swimmers.²⁵ Some authors have suggested an association between STAM and GIRD in overhead athletes.⁵ In our cohort, GIRD was more frequently observed in water polo athletes due to the throwing activity, but no clear association with STAM was found. However, the absence of statistically significant differences in the subgroup analyses should not be interpreted as evidence of no association, because these analyses were not independently powered and may be subject to type II error.

Hyperlaxity

We observed a similar rate of selective unidirectional shoulder hyperlaxity in both the study and control group. Inferior laxity diagnosed with Gagey test, anterior laxity diagnosed with Coudane Walch test, and sulcus sign in neutral rotation were observed in the 33%, 8%, and 3% of dominant shoulder in athletes and in 5%, 14%, and 9% in the control group, respectively. Although these tests were associated with higher Beigthon score and general hyperlaxity, they were not intercorrelated. This indicates that isolated unidirectional hyperlaxity is relatively common and independent of activity level. Inferior capsule hyperlaxity, as diagnosed by Gagey test, was more frequent in the dominant arm of water polo athletes, likely due to the throwing gesture. Moreover, general hyperlaxity was associated to STAM type 2 independently from the type of sport activity.

Clinical implications

Identifying scapular dyskinesis in overhead athletes requires the recognition of distinct patterns of altered shoulder girdle kinematics. However, in the present cohort, STAM subtype classification was not associated with pain, GIRD, ASES score, or Kerlan-Jobe score. Therefore, although subtype classification may be useful for descriptive and epidemiologic purposes, its current clinical utility remains uncertain. Individualized training programs based on STAM subtype should be considered hypothesis-generating rather than evidence-based until supported by prospective interventional studies. Future studies including symptomatic patients, as well as longitudinal studies, could clarify the impact of STAM subtype classification on injury prevention, treatment selection, and performance.

Limitations

Our study has several limitations. First, scapular motion was clinically assessed and, although specific tests and multiple observers were used, the diagnosis of STAM types remains subjected to error. No 3-dimensional computer-based analysis was performed, as suggested by previous studies.^7,26 This limitation is particularly relevant because the primary aim of the present study was to distinguish STAM subtypes; therefore, clinical sub-classification may not fully capture the complexity of 3-dimensional scapular kinematics. Moreover, examiners were not blinded to sport type. This may have introduced expectation bias. However, all participants were evaluated according to a standardized protocol using predefined criteria, which likely reduced variability in assessment. Another limitation is the potential circularity of STAM subtype classification. Because the clinical classification incorporates scapular position, pectoralis minor tenderness, and SFRT findings, subsequent interpretations linking STAM type 1 to pectoralis minor involvement or STAM type 2 to serratus anterior dysfunction are partly embedded within the diagnostic framework itself. These mechanistic interpretations should therefore be considered hypothesis-generating. Second, all athletes were evaluated during the competition season to reflect a realistic scenario. However, both training and competition can affect scapular motion, even in the short term. In our series, each subject was assessed only once during the season, and consequently, our findings should be interpreted as associations rather than causal or prognostic relationships. Third, the exclusion of the female population from both the study and the control groups prevents us from generalizing our results and conclusions to the overall population, limiting them to males only. Further studies will be necessary to confirm our findings in the female population. Fourth, the control group was composed of healthy volunteer medical students, most of whom were either inactive or recreationally active in sports with minimal upper-limb involvement. This may introduce selection bias, healthy-volunteer bias, and an activity-level mismatch compared with professional athletes. Finally, the limited sample size, especially the small number of STAM type 2 events, increases the risk of unstable estimates, type II error, and wide CIs. Although Firth-penalized regression was used to reduce small-sample bias, the study was not powered to establish distinct biological etiologies for STAM subtypes. The sample size calculation was based on an expected prevalence of 40% derived from previous literature, whereas the observed prevalence in our cohort was approximately 20%. This lower event rate reduced the effective number of observations for this variable and may have limited the power to detect small between-group differences. Moreover, the subgroup analyses were limited by smaller sample sizes and were not independently powered. Therefore, these results should be interpreted cautiously and regarded as exploratory and hypothesis-generating. Larger studies are needed to better evaluate the clinical role of different type of STAM among professional athletes.

Conclusion

STAM type 1 was more common in the dominant shoulder of professional overhead athletes. In contrast, STAM type 2 was observed similarly in dominant and non-dominant shoulders in both the study and control groups and was associated with generalized joint hyperlaxity. STAM types 1 and 2 were not associated with pain, GIRD, ASES score, or Kerlan-Jobe score in this cohort. Therefore, these findings support epidemiologic differences between STAM subtypes, but do not establish causal mechanisms, distinct biological etiologies, or definite clinical relevance.

Acknowledgments

The authors would like to thank Circolo Nautico Posillipo, Napoli Basketball, and CrossFit Napoli.

Disclaimers

Funding: No funding was disclosed by the authors.

Conflicts of interest: Andrea Cozzolino and Roberto de Giovanni report that they received support for attending meetings by Enovis Lima and Johnson & Johnson Italia. Andrea Cozzolino received payment for lectures, and educational events from Stryker Italia s.r.l. for topics not related to the subject of the present study. Any additional authors, their immediate families, and any research foundations with which they are affiliated have not received any financial payments or other benefits from any commercial entity related to the subject of this article.