Shoulder Pain in Persons With Tetraplegia and the Association With Force Application During Manual Wheelchair Propulsion

Abstract

Objective

To investigate the association between propulsion biomechanics, including force application and spatio-temporal characteristics, and shoulder pain in persons with tetraplegia.

Design

Cross-sectional, observational study.

Setting

Non-university research institution.

Participants

16 community dwelling, wheelchair dependent persons with a chronic tetraplegia between C4 and C7, with and without shoulder pain (age, 49.1±11.7 years; 94% men, 23.4±9.5 years past injury).

Interventions

Not applicable.

Main Outcome Measures

Force application and spatio-temporal characteristics of wheelchair propulsion on a treadmill (0.56 m/s, 10W and 0.83 m/s, 15W). Participants were stratified in groups with low, moderate, and high pain based on their Wheelchair User Shoulder Pain Index (WUSPI) score on the day of measurement.

Results

The mixed-effect multilevel analysis showed that wheelchair users with high levels of shoulder pain applied propulsion force more effectively (and with a lower medial component) and over a longer push angle, thus shortening the recovery time as compared with persons with low or moderate levels of shoulder pain.

Conclusions

In contrast with previous results from persons with a paraplegia, persons with tetraplegia and high levels of shoulder pain propel their wheelchair more optimal with regard to risk factors for shoulder pain. Our results therefore affirm that there is a different interaction of shoulder pain and propulsion biomechanics in persons with a tetraplegia which should be considered when further analyzing risk factors for shoulder pain in wheelchair users or applying literature results to different patient populations.

Manual wheelchair users are dependent on their upper extremities for all activities of daily living. Shoulder pain is very common in this population¹ and interferes with performing these daily activities. As a consequence, it reduces independency, participation in life, and thus quality of life.^2,3

Activities and participation in daily life may both strengthen but also stress the upper body and the shoulder. If the demand of the activity is higher than the capacity of the wheelchair user, it may increase the risk for overuse problems. Several wheelchair related tasks have been associated with a higher risk for shoulder problems: daily wheelchair propulsion,⁴ transfers in and out of the wheelchair,⁵ and pressure relief lifts for pressure injury prevention.⁶ Wheelchair propulsion is a predominant activity in manual wheelchair users and approximately 1400 pushes are done throughout a day.⁷ The repetitiveness may predispose tissue degeneration over time demonstrating the need to optimize the load on the shoulder. Based on ergonomic considerations and on results from previous studies,8, 9, 10, 11 recommendations for wheelchair propulsion evolved, which are published in the clinical practice guidelines for health care professionals.¹² In order to preserve upper limb function after SCI, wheelchair users are advised to (a) use long, smooth strokes that limit high effects on the pushrim and (b) allow the hand to drift down naturally, keeping it below the pushrim when not in actual contact with the part of the wheelchair. This semicircular pattern is recommended because it is associated with lower stroke frequency, greater time spent in the push phase and less angular joint velocity and acceleration.¹² These recommendations have been strengthened by studies, which analyzed the association between shoulder pain and propulsion biomechanics.13, 14, 15 Beirens et al showed that wheelchair users with severe shoulder pain propel with less smooth strokes, indicated by a higher rate of rise (ROR) and jerk.¹³ Jerk is the rate of change of force and represents a measure of smoothness of force exertion over a complete push phase of wheelchair propulsion. However, other researchers did not find a direct relation between the wheelchair users’ shoulder pain and spatio-temporal and kinetic parameters of submaximal wheelchair propulsion.16, 17, 18

The vast majority of studies analyzing the association between shoulder pain and wheelchair propulsion are based on data from manual wheelchair users with paraplegia or other health conditions such as spina bifida, cerebral palsy or lower limb amputation.8, 9, 10^,13,17 In this population, persons are not restricted in the use of the upper extremities. In persons with tetraplegia, however, certain muscles of the upper extremities are denervated. A reduced or missing triceps activation for example, which is present in persons with a lesion level of C7 and above, limits or hinders the individual to actively extend the arms resulting in shorter pushes. Thus, depending on the lesion level, persons with tetraplegia will move and propel their wheelchair differently, which will result in a different exposure and potentially different risk factors for shoulder pain. It is not known, however, whether the same association between shoulder pain and wheelchair propulsion characteristics exists as in persons with paraplegia. In order to optimally support and advise persons with tetraplegia, this lack in knowledge should be studied.

The aim of this preliminary study is to investigate the association between propulsion biomechanics, including force application and spatio-temporal characteristics, and shoulder pain in individuals with tetraplegia. Following the clinical guidelines and results from persons with a paraplegia, we hypothesize that persons with higher levels of shoulder pain apply higher maximal total force to the pushrim and propulsion force is applied less smoothly (higher ROR and jerk). Regarding spatio-temporal characteristics, we hypothesize that wheelchair user with higher levels of shoulder pain have a higher push frequency, apply the force over a shorter push angle and spend less time in the push phase relative to the whole cycle time.

Methods

This study had a cross-sectional, observational design. Ethics approval was obtained from the Ethics Committee of Northwest and Central Switzerland (project ID: 2015-192) and all participants provided written informed consent prior to participation.

Participants

The inclusion criteria for study participation were (1) age between 16 and 85 years, (2) chronic tetraplegia between C4 and C8, and (3) manual wheelchair dependency for ADL, either with or without supportive propulsion. The exclusion criteria were (1) major trauma of the upper extremity in the past year, (2) history of shoulder surgery, and (3) contraindications to MRI. Suitable participants (fulfillment of inclusion criteria, n=174) were identified in the database of the Swiss Spinal Cord Injury Cohort. The study information and a questionnaire were sent to these persons. The questionnaire included questions about the duration and intensity (numerical rating scale 0-10) of shoulder pain in the past 3 months and the fulfillment of the in/exclusion criteria. From the eligible persons who were willing to participate (n=28), a sample of 20 participants was selected for study participation. The selection was based on the intensity of shoulder pain and aimed at having a group of participants with a broad range of shoulder pain intensity.

Data collection

Participants were invited for 1 testing session to the biomechanical laboratory. Upon arrival they completed the Wheelchair User Shoulder Pain Index¹⁹ (WUSPI) to determine their shoulder pain. The WUSPI is a validated 15-item questionnaire assessing self-reported pain experienced in the past week during functional ADLs. Responses are scaled from a minimum of 0 (no pain) to a maximum of 10 (worst pain). A few participants were not able to perform all activities assessed by the WUSPI; therefore, we calculated a performance-corrected (PC-WUSPI) score.²⁰ Participants were then asked to propel their personal wheelchair on a motorized treadmill^a. A SmartWheel^b (24 inch) was fitted on the personal wheelchair of the participant to capture 3d forces and moments (Fx, Fy, Fz, Mx, My, Mz). The SmartWheel was fitted on the side where shoulder pain was highest. If the participants did not report pain, the SmartWheel was fitted on the dominant side. Prior to wheelchair propulsion on the treadmill, rolling resistance was calculated via a drag test.²¹ Rolling resistance was used to determine the extra weight needed, acting via a pulley system on the wheelchair, in order to set given power output conditions.²² After familiarization, participants performed 2 predefined speed/power output conditions for 1 min each in a randomized order: 0.56 m/s, 10W and 0.83 m/s, 15W. Speed and power output were standardized to have the same conditions for all participants in order to make results comparable. The speed and power output conditions were chosen based on the capabilities of manual wheelchair users with tetraplegia and to mimic the range of previously reported everyday conditions.^23,24 Data were collected from the last 30 s of each condition.

Data analysis

Based on the 50th and 75th percentile of the PC-WUSPI score of the investigated sample, participants were divided into 3 shoulder pain groups¹³: low pain group (LPG; PC-WUSPI <7.5), moderate pain group (MPG; PC-WUSPI 7.5–35.5), and high pain group (HPG; PC-WUSPI >35.5).

Kinetic data were collected at 240 Hz and filtered with a fourth order Butterworth filter with a cutoff frequency of 10 Hz.²⁵ The data were segmented into propulsion cycles consisting of a hand contact phase (total force >2N) and a recovery phase. The push phase was further segmented into initial contact, propulsion and release phase.²⁶

Initial contact and release phase can have a braking torque around the axle, below the threshold of -2 Nm, and propulsion phase a propulsive torque above the threshold of 2 Nm. Thresholds were defined based on the noise level of the measurement wheel.

Propulsion biomechanics were calculated using a custom written MATLAB^c script were normalized to cycle duration and averaged over all complete pushes measured during the trial. Outcome variables are listed in table 1.

Table 1.

Outcome variables of wheelchair propulsion biomechanics, including force application and spatio-temporal characteristics

Outcome Variable	Abbreviation, Unit	Description
Maximal total force	Ftotmax, N	Average of all maximal values of total force
Maximal tangential force	Ftanmax, N	Average of all maximal values of tangential force
Maximal medial force	Fmedmax, N	Average of all maximal values of medial force
Mean fraction of effective force	FEFmean, %	Average of mean FEF, FEF = Ftan2/Ftotal2
Rate of rise of force	ROR, N/s	ROR = maximum of $\frac{d\left(Ftotal\right)}{dt}$
Jerk	Jerk, N2/s	Monotonous increasing jerk: value of change of force13,27; natural logarithm of the area under the curve of squared derivate of total force. $Jerk=log\int {\left(\frac{d\left(Ftotal\right)}{dt}\right)}^2$
Maximal moment	Mmax, Nm	Average of maximal values of propulsive moment
Push frequency	Freq, pushes/min	Number of pushes per minute
Push angle	Angle, °	Average of angle of contact phase
Cycle time	Cycle, s	Average of cycle time, including contact and recovery time
Contact time	Contabs, s	Average of contact time, including initial contact, propulsion and release time
Recovery time	Recabs, s	Average of recovery time
Relative contact time	Contrel, %	Contact time relative to cycle time
Relative recovery time	Recrel, %	Recovery time relative to cycle time
Relative initial contact time	ICrel, %	Initial contact time relative to cycle time
Relative propulsion time	Proprel, %	Propulsion time relative to cycle time
Relative release time	Relrel, %	Release time relative to cycle time

Statistical analysis

To compare participant characteristics between shoulder pain group, a 1-way analysis of variance was performed. Bonferroni post hoc tests were done when a significant difference between shoulder pain groups was found. A mixed-effect multilevel analysis was used to identify the association between the dependent variables (propulsion biomechanics) and shoulder pain group when controlling for sex, lesion level, TSI, age, body weight, and height. These covariables were selected based on their association with shoulder load or shoulder pain.

The 2 trials (0.56 m/s, 10W and 0.83 m/s, 15W) were used as a grouping variable to account for between-subject variability. Bonferroni post hoc tests were performed when a significant difference was found between shoulder pain groups. All statistical analyses were performed with Stata, version 16^d.

Results

The characteristics of the analyzed manual wheelchair users with tetraplegia are presented in table 2. Because of technical issues, only data of 16 out of the 20 measured participants could be included into the analysis. Grouping into shoulder pain groups, based on the PC-WUSPI, resulted in 6 individuals in the LPG, 6 in the MPG, and 4 in the HPG. There were no statistically significant differences in participant characteristics between the pain groups, except for level of shoulder pain (table 2).

Table 2.

Participant characteristics (mean ± standard deviation) of all participants, and stratified by shoulder pain group

	Overall	LPG	MPG	HPG	Group Diff
Number	16	6	6	4
Sex	15m, 1f	6m	6m	3m, 1f
Lesion level	2 C4, 7 C5, 5 C6, 2 C7	2 C5, 4 C6	2 C4, 4 C5	1 C5, 1 C6, 2 C7
Age (years)	49.1±11.7	48.7±15.9	48.3±10.9	51.0±7.6
Weight (kg)	74.2±12.1	71.0±10.2	76.7±12.7	75.3±16.1
Height (m)	1.78±0.07	1.77±0.06	1.81±0.07	1.76±0.08
TSI (years)	23.4±9.5	21.8±11.6	22.2±4.9	27.5±12.4
PC-WUSPI	28.3±30.3	2.2±2.9	23.7±11.8	74.2±11.5	*,†,‡

Abbreviations: TSI, time since injury; Group diff, significant difference after Bonferroni corrections for multiple comparisons (P<.05) of

^⁎LPG vs MPG.

^†LPG vs HPG.

^‡MPG vs HPG.

The mean values of all wheelchair propulsion characteristics are listed in table 3 and in appendix 1.

Table 3.

Unadjusted mean ± standard deviation of wheelchair propulsion biomechanics, presented for each group

	LPG	MPG	HPG	Group Diff, P Value
Ftotmax, N	58.1±13.4	63.6±13.2	53.6±13.1
Ftanmax, N	37.2±4.7	40.3±12.8	34.6±8.0
Fmedmax, N	12.1±9.4	14.9±12.7	6.4±4.2	.002,* .001‡
FEFmean, %	35.2±14.7	40.8±15.9	40.7±5.8	.003,† .007‡
Jerk, N2/s	11.0±0.6	11.4±0.9	11.2±0.8
ROR, N/s	536.3±210.4	636.2±320.3	578.0±346.4
Mmax, Nm	9.6±1.2	10.4±3.3	8.9±2.1
Freq, push/min	50.4±9.2	46.6±6.9	55.0±14.7
Angle, °	81.2±9.7	84.7±8.5	85.6±21.9	.001,* <.001†
Cycle, s	1.2±0.2	1.2±0.2	1.1±0.3
Contabs, s	0.7±0.1	0.7±0.1	0.7±0.3
Recabs, s	0.5±0.1	0.5±0.1	0.4±0.1	<.001†
Contrel, %	59.0±4.5	56.7±5.0	62.3±5.6
Recrel, %	41.0±4.5	43.3±5.0	37.7±5.6
ICrel, %	3.8±3.4	2.1±2.1	4.8±1.9
Proprel, %	94.6±4.3	97.5±2.4	92.4±4.3	.040*
Relrel, %	1.5±1.5	0.3±0.4	2.8±3.2

Abbreviations: Cont_abs, contact time; Cont_rel, relative contact time; Freq, push frequency; Ftan_max, maximal tangential force; Ftot_max, maximal total force; IC_rel, relative initial contact time; M_max, maximal moment; Prop_rel, relative propulsion time; Rec_abs, recovery time; Rec_rel, relative recovery time; Rel_rel, relative release time

Group diff, significant difference after Bonferroni corrections for multiple comparisons (P<.05) of

^⁎LPG vs MPG.

^†LPG vs HPG.

^‡MPG vs HPG.

Spatio-temporal characteristics

Push angle, absolute recovery time, and relative propulsion time were significantly associated with shoulder pain group (table 3). Individuals in the LPG applied the force over a statistically significant lower push angle (81.2±9.7°) than individuals of both MPG (84.7±8.5°, P=.001) and HPG (85.6±21.9°, P=.001). Individuals in the LPG had a statistically significant longer recovery time (0.5±0.1 s) than individuals in the HPG (0.4±0.1 s, P<.001) and had a significantly shorter relative propulsion time (94.6%±4.3%) than individuals in the MPG (97.5%±2.4%, P=.040). Associations between pushrim biomechanics and the included covariables can be found in appendix 2.

Force application

Total applied force to the push rim for both measured conditions and stratified by shoulder pain group is displayed in figure 1. Medial applied force and effectiveness were significantly associated with shoulder pain group (table 3). Individuals in the MPG applied statistically significant higher maximal medial force (Fmed_max) to the pushrim (14.9±12.7 Nm) compared with individuals in both the LPG (12.1±9.4 Nm, P=.002) and HPG (6.4±4.2 Nm, P=.001). Fraction of effective force (FEF)_mean of the HPG (40.7%±5.8%) was statistically significant different than FEF_mean of both LPG (35.2%±14.7%, P=.003) and MPG (40.8%±15.9%, P=.007).

Fig 1.

Total force for the 2 conditions: 0.56 m/s, 10W, and 0.83 m/s, 15W. Data represent mean values cycle normalized to 101 points, stratified by the 3 pain groups: LPG, MPG, and HPG.

Discussion

The current study examined propulsion biomechanics during wheelchair propulsion at fixed power output and speed on a motorized treadmill and associations with shoulder pain in individuals with tetraplegia. On the basis of previous findings in persons with paraplegia, we expected that persons with high levels of shoulder pain would propel their wheelchair the least conform the guidelines on optimal wheelchair propulsion developed to reduce the risk for shoulder pain. This hypothesis was not met.

Spatio-temporal characteristics

Regarding spatio-temporal characteristics, we found associations between levels of shoulder pain and recovery time, relative propulsion time, and push angle (table 3).

Persons with high levels of shoulder pain spent significantly less time in the recovery phase (0.4 s) than persons with medial levels of shoulder pain (0.5 s). The clinical practice guidelines advise the wheelchair users to use a semicircular propulsion style because it is associated with a greater time spent in the contact phase relative to the recovery time. Following the reasoning of the guidelines, persons with high levels of shoulder pain have a more optimal timing of force application, because they spend less time in the recovery phase.

In contradiction to our expectation, wheelchair users with high levels of shoulder pain applied the force over a larger push angle. A larger push angle increases the distance over which the propulsion forces can be distributed, which reduces the magnitude of the peak forces and decreases push frequency.²⁸ Thus, the studied persons with high levels of shoulder pain apply the forces more optimal with regard to the risk factor push angle. Previous cross-sectional studies in persons with predominantly paraplegia or other health conditions not restricting upper extremity use did not find any association between shoulder pain and push angle.^16,17 However, when re-analyzing the same population after 4-6 months, Briley et al reported that manual wheelchair users with increasing shoulder pain adapted their propulsion characteristics.¹⁵ They changed their propulsion toward higher contact angle, decreased peak propulsion torque, and increased peak propulsion torque variability. These findings relate to the protective response theory which proposes that for tasks that provoke painful symptoms the nervous system searches for movement patterns that are less painful.^29,30 Whether all the associations found in the present study are related to such protective changes due to pain, or are the underlying cause of pain cannot be determined with this cross-sectional study. In order to better understand the interaction of shoulder pain and wheelchair propulsion, longitudinal studies are needed.

We did not find any association between shoulder pain and push frequency. Based on ergonomic considerations, frequency of a task is, next to duration and load of the task, related to the risk of an injury.³¹ Boninger et al showed in their study on wheelchair users with paraplegia that push frequency is correlated with the risk of median nerve injury.⁸ The current study on persons with tetraplegia (table 3) as well as the study of Rice et al did not find such association.¹⁷ A more informative measure for estimating the contribution of push frequency to the risk for shoulder pain would be quantifying push frequency in real life instead of in lab conditions. The quantification of push frequency over the day or during bouts of mobility represents the real exposure of the wheelchair user. In contrast to measures of force application, push frequency and other spatio-temporal characteristics of wheelchair propulsion can be quantified with wearable sensors. Such real-life measures should be used in future studies to find risk factors for shoulder pain and to determine the exposure of manual wheelchair users to these risk factors.

Force application

Regrading force application, we found associations between levels of shoulder pain and effectiveness of force application (FEF) and medial applied force (Fmed_max, table 3). Both factors were most optimal in persons with high levels of shoulder pain. They applied propulsion force more effectively than persons with low levels of shoulder pain, which is driven by a low medial directed force (table 3). This might be related to the participant characteristics of the studied population. The group of persons with high shoulder pain included 2 participants with a lesion level of C7, which have an active grip function and are thus able to grab the rim of the wheelchair. Persons without active grip function can only apply propulsion force while producing friction on the rim, which increases medial applied force.

Contrary to our expectation we did not find any association between level of shoulder pain and total applied force or the smoothness thereof (ROR and jerk). Participants with high levels of shoulder pain even applied the lowest maximal forces to the rim (table 3, fig 1). However, the differences between groups of shoulder pain were not statistically significant. ROR describes the force effect on the pushrim and is thus a measure of smoothness of force application. Individuals who apply the propulsion force with a high ROR reach the peak of the force during each push within a shorter time. This can affect the development of shoulder pathology. In persons with paraplegia, high ROR has been associated with high levels of shoulder pain¹³ or medial nerve injury.⁸ In the current study on persons with tetraplegia, no association between ROR and shoulder pain was found. Also another study, including wheelchair athletes with a lesion level up to C6 and persons with other health conditions, did not find a relation between peak ROR and shoulder pain groups.¹⁶ This is an indication that force application and the relation with shoulder pain is different in persons with and without impairment of the upper extremity. The same discrepancy can be found in jerk. Jerk reflects the amount of force variation throughout the push phase and is beside ROR another measure of smoothness of force application.¹³ In persons with paraplegia, persons with high levels of shoulder pain clearly applied the force with more variation.¹³ In the current study jerk was similar for all participants and no association with shoulder pain was found (table 3).

Limitations

To find a causal relation between force application and shoulder pain, a longitudinal study with more participants should be conducted. The present cross-sectional study can only find current associations in a limited number of participants. The small sample size may not allow for adequate representation of the variability that exists within the subpopulation of pain groups. Overall, the participants are a representative sample of persons with SCI living in Switzerland regarding age and have a slightly higher TSI.³² However, women are underrepresented in the studied population. Because women have higher odds for having shoulder pain,³³ this group must be better represented in a subsequent longitudinal study.

As mentioned before, a further limitation of the present study is that manual wheelchair propulsion is studied in a lab setting and the real exposure of the wheelchair user in daily life is not considered. For the detailed analysis of force application, wheelchair propulsion under controlled laboratory conditions is still the criterion standard. For quantifying the exposure of wheelchair user to other risk factors, such as number of pushes per day, push frequency or timing of the push phase, real life measures with wearable sensors should be conducted.

Conclusions

The expected associations between shoulder pain and force application in wheelchair propulsion, based on clinical guidelines and previous results from persons with paraplegia, were not found in persons with tetraplegia. On the contrary, persons with high levels of shoulder pain applied propulsion force more effectively (and with a lower medial force component) and over a longer push angle, thus shortening the recovery time. Following these results, persons with high levels of shoulder pain propel their wheelchair more optimal as compared with persons with low to moderate levels of pain.

Our results therefore affirm that there is a different interaction of shoulder pain and propulsion biomechanics in persons with a tetraplegia which should be considered when further analyzing risk factors for shoulder pain in wheelchair users or applying literature results to different patient populations. To get insight into risk factors for shoulder pain in persons with tetraplegia and to optimally advise them on safe wheelchair propulsion, longitudinal studies on the association between shoulder pain and wheelchair propulsion in a larger group of persons with tetraplegia are needed.

Suppliers

• a.Bonte Technology BV.
• b.SmartWheel; Three Rivers Holdings, Inc.
• c.MATLAB; MathWorks, Inc.
• d.Stata, version 16; StataCorp LP.

Appendix 1. Unadjusted Mean ± Standard Deviation of the Dependent Variables of Both Wheelchair Propulsion Conditions: A: 0.56 m/s, 10W, B: 0.83 m/s, 15W

	Cond	Overall	LPG	MPG	HPG
Ftotmax, N	A	55.1±12.4	54.9±13.2	56.0±10.0	54.2±17.5
	B	63.6±13.5	62.8±14.2	71.2±11.9	53.1±9.5
Ftanmax, N	A	36.2±4.7	35.4±4.0	37.4±10.1	35.6±11.5
	B	39.5±10.9	39.7±5.1	43.2±15.5	33.6±3.7
Fmedmax, N	A	10.0±8.7	10.6±9.7	12.4±9.9	5.7±4.8
	B	13.5±11.8	14.36±9.9	17.3±15.5	7.1±4.0
FEFmean, %	A	38.4±14.1	34.7±14.3	41.4±18.4	39.4±6.8
	B	39.5±12.9	35.9±17.5	40.3±14.8	42.0±5.4
Jerk, N2/s	A	11.0±0.6	10.8±0.3	11.1±0.6	11.2±1.0
	B	11.5±0.8	11.3±0.8	11.8±1.0	11.3±0.6
ROR, N/s	A	513.4±242.3	465.9±112.2	518.9±226.2	576.3±421.4
	B	672.0±322.6	642.0±295.0	753.6±376.1	579.7±320.0
Mmax, Nm	A	9.3±2.1	9.1±1.0	9.6±2.6	9.2±3.0
	B	10.1±2.8	10.2±1.3	11.1±4.0	8.6±1.0
Freq,	A	44.5±8.2	46.1±8.1	41.4±3.4	47.0±13.3
pushes/min	B	56.5±9.0	57.0±6.8	51.8±5.4	63.0±12.4
Angle, °	A	81.6±15.3	77.8±11.1	82.4±9.0	85.9±27.9
	B	86.3±10.5	86.2±5.0	87.0±8.1	85.3±18.4
Cycle, s	A	1.3±0.2	1.3±0.2	1.4±0.1	1.3±0.4
	B	1.0±0.1	1.0±0.1	1.1±0.1	1.0±0.2
Contabs, s	A	0.8±0.2	0.8±0.1	0.8±0.1	0.9±0.3
	B	0.6±0.1	0.6±0.0	0.6±0.1	0.6±0.1
Recabs, s	A	0.5±0.1	0.5±0.1	0.6±0.1	0.5±0.1
	B	0.5±0.1	0.5±0.1	0.5±0.1	0.4±0.1
Contrel, %	A	61.6±4.3	61.1±3.1	60.1±3.8	64.3±6.06
	B	56.0±4.9	55.8±4.5	53.2±3.3	60.2±5.1
Recrel, %	A	38.4±4.3	38.9±3.2	39.9±3.8	35.7±6.0
	B	44.0±4.9	44.2±4.5	46.8±3.3	39.7±5.1
ICrel, %	A	2.8±2.4	3.5±3.2	1.4±0.9	4.0±2.0
	B	4.1±3.0	4.3±4.2	2.9±2.8	5.6±1.6
Proprel, %	A	96.0±3.4	94.5±4.7	98.2±1.2	95.0±2.1
	B	94.3±4.7	94.9±4.4	96.8±3.1	89.8±4.5
Relrel, %	A	1.1±1.3	2.0±1.8	0.4±0.5	1.0±0.6
	B	1.7±2.7	0.9±0.7	0.3±0.4	4.6±3.4

Abbreviations: Cont_abs, contact time; Cont_rel, relative contact time; Freq, push frequency; Ftan_max, maximal tangential force; Ftot_max, maximal total force; IC_rel, relative initial contact time; M_max, maximal moment; Prop_rel, relative propulsion time; Rec_abs, recovery time; Rec_rel, relative recovery time; Rel_rel, relative release time

Appendix 2. Significant Results of Mixed-effect Multilevel Analysis to Identify the Association Between the Dependent Variables (Propulsion Biomechanics) and Group of Shoulder Pain (LPG, MPG, HPG), When Controlling for Sex (Men, Women), Body Weight (kg), Time since Injury (TSI, years), Age (years), Height (m), and Lesion level (C4-C7)

	Wald χ2	P Value, χ2		β	SE β	P Value	Bonf Corr, P Value
Ftotmax	37.15	0.00	LPG
			MPG	−0.63	6.30	0.92
			HPG	−14.79	9.14	0.11
			Men
			Women	18.69	15.56	0.23
			Weight	0.41	0.19	0.03*
			TSI	0.38	0.50	0.44
			Height	−36.45	32.02	0.26
			Age	0.16	0.33	0.64
			C4
			C5	−10.92	6.08	0.07
			C6	−12.51	8.78	0.15
			C7	−12.01	12.24	0.33
Ftanmax	84.54	0.00	LPG
			MPG	−5.38	3.42	0.12
			HPG	5.22	4.96	0.29
			Men
			Women	−7.90	8.45	0.35
			Weight	0.06	0.10	0.57
			TSI	−0.69	1.27	0.01*
			Height	45.45	17.38	0.01*
			Age	0.64	0.18	0.00*
			C4
			C5	−14.92	3.30	0.00*
			C6	−16.50	4.77	0.00*
			C7	−23.28	6.65	0.00*
Fmedmax	29.35	0.00	LPG
			MPG	17.43	5.21	0.00*	LPG: 0.00, HPG: 0.00
			HPG	−18.44	8.18	0.02*
			Men
			Women	12.58	14.03	0.37
			Weight	0.34	0.16	0.03*
			TSI	0.65	0.44	0.14
			Height	−87.93	28.17	0.00*
			Age	−0.53	0.30	0.07
			C4
			C5	8.81	5.53	0.11
			C6	21.55	7.54	0.00*
			C7	31.25	10.80	0.00*
FEFmean	29.79	0.00	LPG
			MPG	−0.03	0.07	0.66
			HPG	0.36	0.11	0.00*	LPG: 0.00, MPG: 0.01
			Men
			Women	−0.41	0.19	0.03*
			Weight	−0.01	0.00	0.00*
			TSI	−0.02	0.01	0.00*
			Height	1.11	0.37	0.00*
			Age	0.01	0.00	0.00*
			C4
			C5	0.07	0.07	0.38
			C6	−0.15	0.10	0.14
			C7	−0.29	0.14	0.04*
ROR	23.54	0.01	LPG
			MPG	132.03	154.33	0.39
			HPG	−400.89	232.80	0.09
			Men
			Women	1092.33	397.48	0.01*
			Weight	3.16	4.66	0.50
			TSI	17.91	12.56	0.16
			Height	−568.22	808.69	0.48
			Age	−3.59	8.40	0.67
			C4
			C5	23.81	155.99	0.88
			C6	61.62	218.87	0.78
			C7	160.12	309.24	0.61
Mmax	84.54	0.00	LPG
			MPG	−1.38	0.88	0.12
			HPG	1.34	1.28	0.29
			Men
			Women	−2.03	2.17	0.35
			Weight	0.01	0.01	0.57
			TSI	−0.18	0.07	0.01*
			Height	11.69	4.47	0.01*
			Age	0.16	0.04	0.00*
			C4
			C5	−3.84	0.85	0.00*
			C6	−4.24	1.23	0.00*
			C7	−5.99	1.71	0.00*
Freq	25.56	0.00	LPG
			MPG	−8.02	5.37	0.14
			HPG	−17.20	7.79	0.03*
			Men
			Women	35.49	13.25	0.01*
			Weight	0.12	0.16	0.45
			TSI	0.64	0.42	0.13
			Height	−1.70	27.27	0.95
			Age	−0.15	0.28	0.58
			C4
			C5	−3.77	5.18	0.47
			C6	−5.93	7.48	0.43
			C7	11.41	10.42	0.27
Angle	57.36	0.00	LPG				MPG: 0.00, HPG: 0.00
			MPG	20.34	5.48	0.00*
			HPG	30.80	8.07	0.00*
			Men
			Women	−41.92	13.75	0.00*
			Weight	−0.24	0.16	0.13
			TSI	−0.59	0.44	0.18
			Height	−68.03	28.18	0.02*
			Age	−0.03	0.29	0.92
			C4
			C5	5.30	5.38	0.33
			C6	17.59	7.69	0.02*
			C7	−5.13	10.77	0.63
Cycle	16.95	0.08	LPG
			MPG	0.20	0.13	0.14
			HPG	0.42	0.19	0.03*
			Men
			Women	−0.66	0.33	0.04*
			Weight	−0.00	0.00	0.38
			TSI	−0.01	0.01	0.24
			Height	0.07	0.68	0.92
			Age	0.00	0.01	0.67
			C4
			C5	0.07	0.13	0.57
			C6	0.18	0.19	0.34
			C7	−0.24	0.26	0.30
Recabs	100.21	0.00	LPG
			MPG	0.06	0.03	0.06
			HPG	0.19	0.05	0.00*	LPG: 0.00
			Men
			Women	−0.32	0.08	0.00*
			Weight	−0.00	0.000	0.02*
			TSI	−0.01	0.00	0.00*
			Height	0.60	0.17	0.00*
			Age	0.00	0.00	0.02*
			C4
			C5	0.02	0.03	0.51
			C6	0.04	0.05	0.44
			C7	−0.18	0.06	0.00*
Contrel	33.65	0.00	LPG
			MPG	0.78	2.57	0.76
			HPG	−2.83	3.73	0.45
			Men
			Women	5.24	6.34	0.41
			Weight	0.07	0.08	0.35
			TSI	0.40	0.20	0.05*
			Height	−41.19	13.05	0.00*
			Age	−0.22	0.13	0.10
			C4
			C5	0.51	2.48	0.84
			C6	2.59	3.58	0.47
			C7	7.05	4.99	0.16
Recrel	33.65	0.00	LPG
			MPG	−0.78	2.57	0.76
			HPG	2.83	3.73	0.45
			Men
			Women	−5.24	6.34	0.41
			Weight	−0.07	0.08	0.35
			TSI	−0.40	0.20	0.05*
			Height	41.17	13.05	0.00*
			Age	0.22	0.13	0.10
			C4
			C5	−0.51	2.48	0.84
			C6	−2.59	3.58	0.47
			C7	−7.05	4.99	0.16
ICrel	26.94	0.00	LPG
			MPG	−2.53	1.43	0.08
			HPG	−3.09	2.15	0.15
			Men
			Women	3.85	3.67	0.29
			Weight	0.06	0.04	0.19
			TSI	0.24	0.12	0.04*
			Height	−10.62	7.47	0.16
			Age	−0.06	0.08	0.46
			C4
			C5	−1.26	1.44	0.38
			C6	−1.79	2.03	0.38
			C7	0.98	2.86	0.73
Proprel	44.57	0.00	LPG
			MPG	4.55	1.84	0.01*	LPG: 0.04
			HPG	3.02	2.69	0.26
			Men
			Women	−1.29	4.59	0.78
			Weight	−0.14	0.05	0.01*
			TSI	−0.32	0.1	0.03*
			Height	13.72	9.41	0.15
			Age	0.08	0.10	0.39
			C4
			C5	0.40	1.79	0.82
			C6	1.81	2.57	0.48
			C7	−2.80	3.60	0.44
Relrel	31.38	0.00	LPG
			MPG	−2.02	1.02	0.05*
			HPG	0.06	1.48	0.97
			Men
			Women	−2.55	2.52	0.31
			Weight	0.08	0.03	0.01*
			TSI	0.08	0.08	0.34
			Height	−3.14	5.20	0.55
			Age	−0.02	0.05	0.64
			C4
			C5	0.88	0.99	0.37
			C6	−0.00	1.42	0.99
			C7	1.93	1.99	0.33

Abbreviations: Cont_rel, relative contact time; Freq, push frequency; Ftan_max, maximal tangential force; Ftot_max, maximal total force; IC_rel, relative initial contact time; M_max, maximal moment; Prop_rel, relative propulsion time; Rec_abs, recovery time; Rec_rel, relative recovery time; Rel_rel, relative release time.