Shoulder proprioceptive sense of force in chronic anterior instability: performance and clinical associations

Abstract

Background

Proprioception, our internal awareness of limb position and movement, is essential for shoulder stability through neuromuscular control. Following traumatic anterior instability (TAI), proprioceptive impairments may develop, potentially affecting upper limb function. These deficits vary across individuals. Previous studies have focused primarily on joint position sense and kinesthesia, while sense of force (SoF) deficits in chronic TAI remain largely unexplored. Moreover, clinical assessments are multifactorial and time-limited, and little is known about how SoF relates to common clinical measures. This cross-sectional study aimed to (1) determine whether chronic TAI is associated with shoulder SoF deficits compared with the contralateral side and a control group (CG) and (2) explore potential associations between SoF, maximal voluntary isometric strength (MVIC), and clinical questionnaires used in shoulder instability assessment.

Methods

Participants aged 18–40 with TAI underwent SoF assessment in both shoulders. A matched CG (n = 37) was tested only on the dominant arm. Proprioceptive accuracy (PA) during SoF testing, MVIC, and shoulder questionnaires (Western Ontario Shoulder Instability Index, Shoulder Instability–Return to Sport after Injury, Disabilities of the Arm, Shoulder and Hand (short version)) were collected for analysis.

Results

The TAI group included 37 participants (27% female, n = 10; 73% male, n = 27), subdivided into non-surgical (TAINS, n = 18) and surgical (TAIS, n = 19) subgroups. Analysis revealed a significant main effect of intensity on PA, F(1,36) = 26.835, P < .001, η²P = .427, but no significant main effect of group, F(1,36) = 0.545, P > .05, η²P = .015, nor intensity × group interaction, F(1,36) = 0.131, P > .05, η²P = .004. A significant main effect of side was observed, F(1,35) = 4.18, P < .05, η²P = .107, with higher PA score on the affected side (10% ± 7.6) than the unaffected side (8.5% ± 6.8; Cohen's d = 0.196). Only the model for the 20% external rotation condition was statistically significant (adjusted R² = 0.383; F(2,33) = 11.884,P < .001), with Disabilities of the Arm, Shoulder and Hand (short version) (β = 0.763, P < .001) and Shoulder Instability–Return to Sport after Injury score (β = −0.519, P = .002) emerging as significant predictors.

Conclusion

Chronic TAI subjects showed no difference in SoF performance compared with the CG. Associations between SoF, MVIC, and clinical questionnaire outcomes were minimal, emphasizing the importance of interpreting clinical assessments collectively rather than in isolation.

Proprioception, a critical component of the somatosensory system, refers to the sensory inputs that contribute to the awareness of body position and movement.³² Accordingly, it is commonly subdivided into joint position sense (JPS) (active and passive), kinesthesia, sense of force (SoF), and sense of velocity.³ Some researchers also include additional submodalities such as vibration, pressure, tension, and balance.³¹ Proprioception encompasses the perception, generation, anticipation, and reproduction of informations related to the body representation. By providing continuous neural feedback to both the central and peripheral nervous systems, proprioception plays a key role in neuromuscular control.³²

Among the various joints, the shoulder is particularly dependent on proprioceptive input due to its wide range of motion and relatively limited passive stability.²⁷ Traumatic shoulder dislocations frequently compromise both mechanical restraints and sensory structures such as mechanoreceptors and proprioceptors. Approximately 90% of these dislocations are anterior, with an annual incidence reaching up to 3%.^20,28

This is particularly concerning given that traumatic anterior dislocations affect both mechanical and sensorimotor components essential for maintaining dynamic shoulder stability. Several studies have shown that decreased sense of proprioception is associated with shoulder instability.^24,25 Given the growing evidence of proprioceptive deficits among individuals affected by traumatic anterior instability (TAI), clinical research has increasingly focused on proprioception outcome measures.⁴

Proprioceptive accuracy (PA) refers to an individual's ability to detect, differentiate, or replicate limb positions, movements, or forces during conscious proprioceptive testing. However, findings across studies have been inconsistent, suggesting that PA is influenced by multiple factors such as pathology,² surgical status,²² time since injury,³³ and the type of proprioceptive testing used.¹⁴ In the context of TAI, most research has focused on JPS and kinesthesia, while the SoF, the ability to perceive, interpret, and reproduce force applied to or generated by a joint, remains largely unexplored.^4,23 Moreover, SoF may be linked to stiffness regulation, a key component of dynamic joint stability,⁹ and recent clinical protocols now provide a simple and practical way to assess it using a hand-held dynamometer (HHD).⁵ In contrast, sense of velocity, although also insufficiently studied, still requires more complex and less clinically feasible assessment methods.

In clinical settings, maximal voluntary isometric strength (MVIC) and patient-reported outcome measures (PROMs) are commonly used to assess shoulder function.^25,29 While general tools like the Disabilities of the Arm, Shoulder and Hand (short version) (QuickDASH) are widely used,¹¹ more specific questionnaires, such as the Western Ontario Shoulder Instability Index (WOSI), which evaluates symptoms and quality of life,³⁰ and the Shoulder Instability–Return to Sport after Injury (SIRSI), which assesses psychological readiness to return to sport,²³ have been developed and validated for TAI populations. Given the complexity of musculoskeletal conditions, a multifactorial approach that integrates sensorimotor performance, strength, and patient perception is essential.⁸ Nevertheless, no studies to date have investigated the potential relationships between PA during SoF testing, MVIC, and PROMs in individuals with TAI.

Therefore, the primary aim of this study is to provide preliminary data on SoF performance in individuals with TAI, comparing their affected arm to their unaffected arm and to the dominant arm of a control group (CG). As a secondary objective, the study will explore potential relationships between PA during SoF testing, MVIC, and PROMs, to better understand how these variables interact and contribute to functional outcomes.

Materials and methods

This study employed a cross-sectional matched control design.

Participants

Thirty-seven individuals with a history of at least 1 episode of traumatic anterior shoulder dislocation, resulting in a clinical diagnosis of TAI, were recruited from the Research Unit of Rehabilitation Sciences (Université libre de Bruxelles, Belgium), affiliated clinics, and the Faculty of Human Movement Sciences (Université libre de Bruxelles, Belgium), Both men and women aged 18 to 40 years were included, provided they met the inclusion and exclusion criteria outlined in Table I. A younger population was specifically selected to minimize the potential confounding effects of age when comparing the CG with TAI subjects, as PA is known to decline with aging.¹

Table I.

Eligibility criteria for traumatic anterior instability group.

	Inclusion Criteria	Exclusion Criteria
1	Age between 18 and 40 yr	Posterior or multidirectional shoulder instability
2	History of an Initial episode of acutely reduced or anterior shoulder dislocation	Generalized joint hypermobility (Beighton score >4)
3	Medical diagnosis of traumatic anterior shoulder instability confirmed by a physician	Non-traumatic shoulder dislocation

A matched CG of 37 pain-free participants was recruited based on age and gender. This convenience sample was selected through posters, email distribution lists, and face-to-face recruitment at the Université libre de Bruxelles, Belgium. Eligible participants were aged 18 to 40 years and had no diagnosed upper limb disorders in the past 6 months. Exclusion criteria included any history of shoulder dislocation, subluxation, upper quadrant surgery, or pain in the shoulder, neck, or upper limb within the 6 months preceding the study. The study received ethical approval from the Erasme ULB Ethical Committee (registration number B4062022000190), and all participants provided written informed consent prior to participation.

Proprioceptive outcomes

Glenohumeral-oriented proprioception was assessed using a low-level ipsilateral matching task.¹⁸ SoF, 1 of the conscious proprioception modalities, was especially targeted as no prior data exists for individuals with TAI.¹² Currently, there is no universally accepted gold standard for measuring upper limb proprioception.⁴ To address this gap, a clinically feasible protocol using a HHD (MusTec BioFET, Almere, Netherlands, 50 Hz) was recently developed to assess SoF in the upper limb.⁵ This study employed a modified version of this protocol and preliminary reliability assessment was carried out on 56 healthy subjects. Since no differences in reliability were initially reported between internal rotation (IR) and external rotation (ER), only IR was assessed. Intra-rater between-day reliability showed intraclass correlation coefficient values ranging from 0.64 [0.49, 0.75] at 20% MVIC in IR to 0.74 [0.63, 0.81]. The standard error of measurement ranged from 3% (at 50% IR) to 8% (at 50% IR), while the minimal detectable change (MDC) with 95% confidence ranged from 8% (at 50% IR) to 13% (at 20% IR). For the purpose of this study, a lower PA indicates better conscious shoulder proprioception.

Sense of force protocol with hand-held dynamometer

Testing position

The testing position was standardized across all measurements (Fig. 1). Participants lay supine on a table with the tested arm abducted to 90°, the shoulder in neutral rotation, and the elbow flexed to 90°. The arm was supported by the table to prevent horizontal abduction, while the nontested arm rested on the abdomen. The forearm was placed against the HHD, positioned 2 cm proximal to the ulnar styloid process on the ventral forearm for internal rotation. The HHD was secured using a double-sided hook-and-loop strap (3M HOOK & LOOP, Saint Paul, Minnesota, USA) attached to a fixed wall mount. The height of the device was individually adjusting according to each participant's forearm length. This testing position was chosen to replicate the apprehension test position, a clinical indicator of shoulder instability.²¹ Furthermore, it offers high clinical practicality, allowing participants to remain relaxed in a supine position while enabling clinicians to efficiently replicate the setup across evaluations. Throughout the evaluation, participants were not provided any visual feedback regarding the force they produced.

Figure 1.

Standardized testing position used during the sense of force protocol.

Maximal voluntary isometric strength testing

MVIC of the shoulder rotators muscles was conducted using the same standardized testing position as in the SoF protocol. Participants were instructed to gradually increase force from zero to their maximum effort over a 2-second period, then maintain maximal contraction for 5 seconds. Three trials were performed for each rotation, with a 60-second rest interval between trials to minimize fatigue.¹⁴

Force reproduction task

The highest MVIC value was used to determine the 2 targets intensities (20% MVIC, 50% MVIC) for the force reproduction task.¹⁴ Two intensities were selected based on previous studies on the upper limb showing that SoF performance may vary depending on contraction intensity.^14,19

The force reproduction task consisted of isometrics contractions divided into 2 phases: the target phase and the reproduction phase.¹⁴ In the target phase, participants were instructed to gradually match the target force within 2 seconds, guided by verbal cues from the assessor, and to maintain the contraction for 8 seconds.¹⁰ During this phase, participants were asked to focus on the sensation of force generated by the shoulder rotators. All target trials were completed before starting the reproduction phase.

In the reproduction phase, participants attempted to reproduce the target force within a 10-second window, without any feedback from the assessor. They verbally signaled when they believed they had reached the target force and maintaining the contraction until instructed to stop. The 3 seconds following the participant's signal were recorded for analysis. A 20-second rest period separated the target and reproduction phases, during which the testing arm remained in the same position. Three trials were performed at each contraction intensity level for both phases, with a 10-second rest between trials. The order of contraction intensity and direction was randomized, and a 30-second rest was provided between each target (Fig. 2). Participants with TAI were evaluated on both arms, in both IR and ER. In the CG, only the dominant arm in IR was assessed, as previous validation of the HHD-based SoF protocol showed no differences between dominant and non-dominant arms, nor between rotation directions, in healthy subjects.

Figure 2.

General description of the SoF protocol. R, rest period between 2 different targets; r, rest period between each of the 3 trials; MVIC, maximal voluntary isometric contraction; SoF, sense of force.

Study protocol

On arrival, participants in the TAI group completed an online questionnaire to record their sociodemographic data, arm dominance, injury status, and pain levels experienced during the previous. Symptoms and functional limitations were evaluated using the French versions of the WOSI,³⁰ the SIRSI,²³ and the QuickDASH.¹¹ The CG completed only the QuickDASH to confirm normal global shoulder function.

Participants then performed submaximal isometric contractions in the testing position for 3 minutes to warm up and familiarize themselves with the force reproduction tasks. After a 2-minute rest, MVIC were assessed. Following a 3-minute recovery period, participants performed 2 force reproduction tasks at target intensities of 20% and 50% of their MVIC, maintaining isometric contractions throughout.

Data reduction

Raw data from HHD were extracted into spreadsheets for analysis. Mean and standard deviation (SD) were calculated using Excel V.16.56 for Mac. To measure PA during SoF testing, relative error (RE) was used as the primary outcome measure. For each trial, the RE was calculated as the absolute difference between the targeted force (%MVIC) and the force actually produced by the participant. A sampling frequency of 50 Hz was sufficient for SoF testing, as a 3-s time window was used. The mean value over the 3 s was calculated, and the mean of the 3 trials was then used for analysis. RE was expressed as the percentage error between the theoretical target and the target identified by the participant as the expected target. Lower RE values indicated greater PA during the force reproduction task.

Sample size and statistical analysis

All statistical analyses were performed using JASP (Version 0.13.1, University of Amsterdam).

Based on preliminary reliability data with an MDC of 13%, a SD of 11, and an intraclass correlation coefficient of 0.7, and using an alpha level of 0.05 and a power of 0.9, a sample size of 23 subjects per group was required, accounting for a 10% attrition rate.

Repeated-measures analyses of variance (ANOVAs) were conducted to examine the effects of within- and between-subject factors on PA. Assumptions of normality and sphericity were assessed prior to analysis. Effect sizes were reported using partial eta squared (ηp²). Post hoc tests with Holm correction were used when appropriate to account for multiple comparisons. The significance level was set at α = 0.05 for all analyses.

First, a 2-way mixed ANOVA was conducted to compare the absolute PA between 2 groups (CG vs. pathological participants; between-subject factor) and 2 testing intensities (20% and 50% of maximal effort; within-subject factor), only in IR on their dominant or injured arm respectively.

A separate 4-way mixed ANOVA was then performed within the TAI group only to explore the effects of surgical status (operated vs. non-operated; between-subject factor), and 3 within-subject factors: contraction intensity (20% vs. 50% MVIC), movement direction (IR vs. ER), and side (affected vs. unaffected limb), on PA during SoF testing.

Finally, a backward stepwise linear regression was conducted to identify the most relevant predictors of PA. The initial model included the following clinical covariates: MVIC in IR and ER, QuickDASH, WOSI, and SIRSI scores. The strength of the association was assessed using the adjusted coefficient of determination (adjusted R²). Co-linearity diagnostics and normality of residuals were evaluated, including visual inspection of Q–Q plots. Non-significant predictors were sequentially excluded, and the final model retained only variables with a significant contribution to the outcome (P < .05). Model fit was confirmed by a significant F-test for the final regression model.⁶

Results

The TAI group included 37 participants (27% female, n = 10; 73% male, n = 27) with a mean age of 26 years (SD = 6). The CG also comprised 37 healthy participants (27% female, n = 10; 73% male, n = 27), with a mean age of 23 years (SD = 3). Within the TAI group, 18 participants had not undergone surgery (TAINS), while 19 had received surgical treatment (TAIS) (see Table II).

Table II.

Participant demographics and group characteristics.

Variables	TAI	CG
Number of participants	37	37
Gender (M/F)	27 M/10F	27 M/10F
Age (yr, mean ± SD)	26 ± 6	23 ± 3
Dominant side (R/L)	36 R/1L	36 R/1L
Affected side (R/L)	17 R/20L
TAINS/TAIS ratio	18/19

TAI, traumatic anterior instability; CG, control group; TAINS, TAI without surgery; TAIS, TAI with surgery; M, male; F, female; R, right; L, left; SD, standard deviation.

Detailed clinical characteristics, including MVIC, PROMs, sport volume, and the statistical difference between the groups are presented in Table III. MVIC was normalized to body mass, which was measured using a scale. The CG was significantly less active than the TAI group (11 ± 6 vs. 6 ± 4, P < .001, effect size = 0.61), and the QuickDASH score was significantly lower in the CG compared with the TAI group (7 ± 9 vs. 36 ± 16, P < .001, effect size = 0.96) (see Table III for details).

Table III.

Clinical outcomes and group comparisons.

Variables	TAI	CG	P value
Weekly sport volume (h)	11 ± 6	6 ± 4	P < .001 ES:.61
QuickDASH	36 ± 16	7 ± 9	P < .001 ES:-.96
WOSI	20 ± 20
SIRSI	25 ± 20
Fmax affected//dominant CG IR (N/kg)	1.8 ± 0.5	2.2 ± 0.7	P < .01 ES:.52
Fmax affected/Unaffected IR (N/kg)	1.8 ± 0.5/2.1 ± 0.6		P < .001 ES:-.78
Fmax affected/Unaffected ER (N/kg)	2 ± 0.5/2.1 ± 0.6		P = .01 ES-0.43
Fmax Affected IR/ER (N/kg)	1.8 ± 0.5/2 ± 0.5		P < .023 ES:-.40
Fmax Unaffected IR/ER (N/kg)	2.1 ± 0.6//2.1 ± 0.6		P = .54 ES:.1

IR, internal rotation; ER, external rotation; TAI, traumatic anterior instability; CG, control group; Fmax, Maximal isometric force; WOSI, Western Ontario Shoulder Instability Index; SIRSI, Shoulder Instability–Return to Sport Index: QuickDASH, Disabilities of the Arm, Shoulder and Hand (short version); ES, effect size.

Traumatic anterior instability vs. control group analysis

A repeated two-way ANOVA was conducted to examine the effects of intensity (within-subject: 20% IR, 50% IR) and group (between-subject: CG, TAI subjects) on PA score (%) during SoF testing. The analysis, performed only in IR and on the injured arm (TAI) or dominant arm (CG), revealed a significant main effect of intensity on PA score, with a large effect size, F(1, 36) = 26.835, P < .001, η²P = .427. However, there was no significant main effect of group, F(1, 36) = 0.545, P > .05, η²P = .015, nor a significant interaction between intensity and group, F(1, 36) = 0.131, P > .05, η²P = .004.

Descriptive statistics (Table IV) revealed that at 20% intensity, the mean PA score was 12,8% (SD = 9.3) for the CG and 14,3% (SD = 9.7) for the TAI group. At 50% intensity, the means were 8,1% (SD = 5.9) for the CG and 8,8% (SD = 6.0) for the TAI group. Post hoc tests indicated a significant difference between 20% and 50% intensity conditions (mean difference 5,1% P < .001), with a large effect size (Cohen's d = 0.639).

Table IV.

Propriocetive accuracy (% error) during sense of force testing.

Intensity	TAIS		CG
	Injured side	Uninjured side	Dominant side
20% IR	14 ± 10	12 ± 9	13 ± 9
50% IR	9 ± 6	9 ± 5	8 ± 6
20% ER	14 ± 12	11 ± 9
50% ER	8 ± 6	6 ± 5

IR, internal rotation; ER, external rotation; TAIS, traumatic anterior instability (surgical); CG, control group.

Traumatic anterior instability within group analysis

The four-way mixed ANOVA revealed significant main effect of intensity on absolute force reproduction error, F(1,35) = 14.32, P < .001, η²P = .290. Post hoc comparisons revealed higher PA score at 20% MVIC (mean = 13.3% ± 0.096) compared to 50% MVIC (mean = 8.7% ± 0.058; P < .001), with a medium effect size (Cohen's d = 0.571, 95% confidence interval [0.235, 0.907]).

A significant main effect of side was also observed, F(1,35) = 4.18, P < .05, η²P = .107, with higher PA score on the affected side (mean = 10.1% ± 7.6) than on the unaffected side (mean = 8.5% ± 6.8); P < .05, Cohen's d = 0.196, 95% confidence interval [–0.004, 0.396]).

No significant main effect of direction was detected, F(1,35) = 2.63, P > .05), nor any effect of group (TAINS or TAIS) (F(1,35) = 0.48, P > .05). All 2- and 3-way interactions, including those involving group, were non-significant (all P > .05, η²P < .08).

Predictors of proprioceptive accuracy during sense of force testing in traumatic anterior instability subject

Preliminary analyses for conditions 20% IR, 50% IR, and 50% ER revealed no significant predictive power for the initial set of variables (all P > .05, Adjusted R² < .05) (Table V).

Table V.

Summary of dependent variables tested in the regression analyses.

Dependent variable	F	P value ANOVA	Adjusted R2
20% IR	0.60	>.05	−0.060
20% ER	5.21	.001∗	0.38
50% IR	0.90	>.05	−0.014
50% ER	0.80	>.05	−0.030

IR, internal rotation; ER, external rotation; ANOVA, analysis of variance.

^∗Indicates statistical significance of the model for the corresponding dependent variable.

Consequently, only the results for the 20% ER condition, which met the criteria for statistical significance, are detailed hereafter (Table VI).

Table VI.

Stepwise analysis for 20% ER.

Model	F	P value ANOVA	Adjusted R2
M0	5.21	.001∗	0.38
M1	6.66	<.001∗	0.39
M2	8.85	<.001∗	0.40
M3	11.88	<.001∗	0.38

ER, external rotation; ANOVA, analysis of variance.

M3 is the final model.

^∗Indicates statistical significance.

A backward stepwise linear regression was conducted to identify predictors for the 20% ER condition. The initial full model (M0), which included 5 predictors (maximal isometric force in IR, maximal isometric force in ER, QuickDash score, WOSI score, SIRSI score), was statistically significant (F(5,30) = 5.209, P = .001).

The final model (M3) was selected for its parsimony and statistical robustness, explaining 38.3% of the adjusted variance (adjusted R² = 0.383; F(2,33) = 11.884, P < .001). QuickDASH score was identified as a strong positive predictor (β = 0.763, P < .001), while SIRSI score emerged as a significant negative predictor (β = −0.519, P = .002). Co-linearity diagnostics were good, with variation inflation factor values of 1.426 for both variables, indicating no redundancy between predictors. Residual analysis confirmed that the model assumptions were met (Mean Residual ≈0).

Discussion

This study found no significant differences in PA during SoF testing between individuals with shoulder TAI and the CG. The between-group analysis was limited to the injured arm in the TAI group and the dominant arm in the CG, both assessed in IR. Within the TAI group, PA scores did not differ between participants who had undergone surgical stabilization and those treated non-operatively, and no differences were observed between IR and ER.

However, we observed a significant side-to-side difference, with the unaffected shoulder showing slightly better PA (ie, lower error scores).

Additionally, both groups exhibited higher PA score during SoF testing at 20% of MVIC compared to 50% MVIC. Of all linear regressions tested (20% and 50% ER/IR), only the 20% ER model was statistically significant, with the QuickDASH score and the SIRSI score emerging as the main predictors with a moderate association (adjusted R² = 0.375).

To our knowledge, our study is the first to report PA results specifically during SoF testing with TAI subjects.

Proprioceptive accuracy between traumatic anterior instability group and control group

Our findings diverge slightly from previous research, as we found no significant difference between TAI group and CG. Indeed, previous studies have reported differences in PA score such as active^24,37 and passive^17,36 JPS and threshold to detection of passive motion, between TAI patients and CG, particularly prior to surgery.

This difference may be explained by the high level of sports activity in the TAI group and the significant difference in physical activity levels compared with the CG, which may have contributed to improved PA. Indeed, recent evidence suggests that active exercise can enhance PA following specific training programs.³⁸ In addition, PA has been reported to be higher in elite athletes compared with CGs. Taken together, these findings suggest that PA can be improved through training, which is consistent with preliminary results previously reported for JPS.³⁴ However, these preliminary observations concerning SoF require further confirmation has the cross-sectional design of this study do not allow strong conclusion.

Proprioceptive accuracy within traumatic anterior instability group

This study reported a significant asymmetry between the affected and unaffected limbs. This finding should be considered preliminary, as the MDC for HHD SoF testing in IR has been reported to range from 8% to 13%, while the difference observed in the present study was less than 2% on average. Consistent with results from this study, patients with TAI have shown impaired proprioception in the affected limb compared to the uninjured side.^24,39

Following a traumatic dislocation, passive structures such as the joint capsule and ligaments may be damaged, leading to reduced sensory input to the central nervous system and potentially contributing to recurrent instability.²⁶ Surgical stabilization can restore tension in these passive tissues, which may enhance sensorimotor function and proprioception, particularly when followed by appropriate rehabilitation.²⁵ However, this recovery process may take several months, as some studies focusing on JPS and kinesthesia have reported persistent proprioceptive deficits at 8 months³⁵ post-surgery, but not beyond 10 months.⁷

In this study sample, the mean time since the first dislocation in the TAI group was 55 months, and the mean time since surgery was 44 months. Although this may partially explain the absence of differences reported in this study between TAIS and TAINS, it raises the question of whether SoF deficits may persist long after surgery compare to the unaffected side.

Muscle spindles and golgi tendon organs likely play a predominant proprioceptive role during voluntary activation of the muscle, especially in mid-range due to the relative laxity of the joint capsule at these angles and the larger role of changes in muscle length.³⁴ In this study, SoF testing was performed in a neutral rotation position, suggesting that PA score was unlikely to be influenced by capsuloligamentous input, which is more active at end-range positions.³⁴ This testing condition could explain the absence of differences between TAINS and TAIS participants, as both subgroups engaged in high levels of sport activity and were assessed under similar neutral joint conditions.^15,16

Study findings also align with previous reports in healthy populations,⁵ showing that higher contraction intensities during SoF testing are associated with improved PA score at the shoulder joint during isometric IR and ER. Greater contraction intensity likely leads to increased motor unit recruitment and greater tension and changes in muscle spindle and golgi tendon organ activity.²⁶ These factors may enhance PA by providing a richer afferent signal to the central nervous system.

Predictors of proprioceptive accuracy during sense of force testing

Finally, our regression analyses aimed to determine whether PA score during SoF testing could be explained by MVIC or PROMs. Only the 20% ER model reached significance, with the QuickDASH and SIRSI score as significant predictors. This suggests that greater perceived disability is associated with poorer PA at low intensity, possibly because such tasks provide less sensory input to the central nervous system,¹³ making deficits more apparent and highlighting a potential link between PROMs and conscious proprioceptive assessment. Nevertheless, the lack of significant predictors in the other tested conditions (20IR, 50IR, and 50ER), despite similar testing parameters, suggests that the findings for the 20ER condition should be interpreted with caution. This inconsistency across conditions underscores the need for further validation before drawing definitive conclusions regarding the generalizability of these predictors. Moreover, the significant model explained only 38% of the variance, suggesting that other unmeasured factors contribute to the PA score. It is also possible that conscious proprioceptive testing fails to capture the broader functioning of the sensorimotor system¹⁸ or detect subtler relationships with strength and other modalities. A recent study reported improvements in both PA score and QuickDASH scores following a proprioceptive training program.¹³ The authors hypothesized that enhanced PA during JPS testing could lead to reduced pain and improved function. However, their statistical analysis was not designed to assess the association between proprioceptive measures and PROMs. Further research is needed to better understand the complex interaction between clinical outcomes surrounding TAI subject.

Practical implication

The most clinically relevant finding is the feasibility of applying the SoF protocol in a TAI population. All participants were able to complete the testing without discomfort, supporting its practical use in routine assessment. Given the lack of correlation observed between low-level proprioceptive measures, PROMs and MVIC, SoF testing should be considered a complementary dimension of clinical testing rather than an interchangeable one, providing distinct information on neuromuscular function. Using the contralateral arm as a reference may serve as a pragmatic baseline in clinical settings. However, considering the potential presence of bilateral deficits, pre-injury or pre-surgery data would offer a more accurate interpretation when available. Finally, contraction intensity during SoF testing appeared to influence performance, suggesting that clinicians should vary force levels when assessing or training SoF in order to stimulate the somatosensory system through different loading demands.

Strengths and limitations

This study is the first to report SoF data in individuals with TAI. Because proprioceptive performance depends on both the assessed modality and the underlying pathology, examining all components of proprioception is important. Focusing on a single outcome may seem restrictive, but this study represents the first application of an SoF protocol using a HHD in TAI, and the targeted approach minimized procedural burden. Analytical testing in the supine position may raise questions regarding ecological validity. However, this design was chosen to allow for easy replication of testing conditions across sessions and to help patients feel relaxed during the initial assessment with the clinician. To streamline the testing procedure, this study compared only the injured arm in the TAI group and the dominant arm in the CG, both assessed in IR. Although this choice may be considered a limitation, it was based on initial data from healthy participants showing no differences between arms or between rotation directions. The analysis of the impact of rotation was therefore limited to the TAI group and confirmed the trend observed in healthy participants, with no difference in PA scores between rotations. However, the absence of data for the non-dominant arm in the CG prevented comparison with the unaffected arm in the TAI group, which limits the ability to assess a potential effect of training volume on SoF performance. As no differences were observed between the affected arm and the dominant arm, this issue should be investigated in future studies.

Given the sample size and the stepwise approach, these findings should be considered exploratory and require validation in a larger independent cohort. Although established guidelines⁶ suggest a ratio of 15 to 30 subjects per predictor for stable coefficient estimation, the initial inclusion of 5 variables in model M0 resulted in a lower ratio of approximately 7.4:1. This initial configuration may increase the risk of overfitting; however, the final model (M3) effectively mitigates this by retaining only 2 significant predictors. This reduction provides a more acceptable subject-to-variable ratio of 18.5:1, thereby strengthening the reliability of the observed associations. Results from this study highlight promising avenues for future research, specifically the adoption of more comprehensive analytical frameworks. Utilizing multivariate analyses that integrate both clinical and proprioceptive variables will be essential to enhance the robustness and clinical relevance of findings in this field.

Methodologically, the cross-sectional observational design precludes causal inference. Substantial heterogeneity within the TAI group, may have limited the detection of associations between key clinical measures. While the sample of 37 participants partially mitigated this issue, limited statistical power cannot be excluded. Finally, the population consisted solely of chronic TAI patients (≥9 months post-injury; mean 55 months), restricting generalizability to the acute stage (<6 months).

Conclusion

This cross-sectional study found no significant differences in SoF testing between the affected arm and a CG. However, a PA deficit during SoF testing may persist for several months following a traumatic anterior dislocation. Surgical history did not appear to influence PA in individuals with chronic TAI by testing them in mid-range of motion. A significant association was observed with both the QuickDASH and SIRSI scores in the 20% ER condition. These preliminary results suggest a potential link between perceived disability, psychological readiness to return to activity, and PA at low contraction intensities, though these findings remain exploratory given the limited sample size.

Disclaimers:

Funding: No funding was disclosed by the authors.

Conflicts of interest: The 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.