Weakness in patients with subacromial pain syndrome is local and more pronounced in females

Abstract

Background:

Subacromial pain syndrome is the predominant cause of shoulder pain, accounting for approximately half of all shoulder complaints. This population presents with weakness of the involved shoulder. However, there is a gap in our understanding of how pain contributes to this weakness, and whether there are sex related differences.

Methods:

Regional and global isometric strength was tested at the involved shoulder joint and remote joints (uninvolved shoulder and both knees) in patients with subacromial pain syndrome. Data were collected before and after acute pain reduction with a subacromial injection.

Findings:

Patients demonstrated weakness at the involved shoulder while remote joints demonstrated normal strength. When compared to healthy controls, male patients were shown to exhibit greater levels of weakness than female patients at the involved shoulder, based on comparisons with sex-matched controls using z-scores. Pain reduction (through an anesthetic injection) had no influence on strength in the short-term.

Interpretation:

Weakness in patients appears to be sex dependent and is not resolved with reduction of pain. This calls into question the assumptions of the physiological causes of this weakness.

1. Introduction

Muscle weakness frequently accompanies both chronic and acute pain (Bank et al., 2013; Graven-Nielsen et al., 2002; Lund et al., 1991). While multiple hypotheses have emerged regarding the precise neurophysiological pathways associated with this association (Hodges & Tucker, 2011; Lund et al., 1991), it is generally accepted that weakness serves to protect damaged tissue from further injury. This is mainly achieved by overriding motor commands and inhibiting motor unit recruitment and force generation (Tucker et al., 2009).

At the shoulder, weakness of the rotator cuff muscles has been demonstrated in patients with impingement syndrome, also known as subacromial pain syndrome (SPS) (Leroux et al., 1994; MacDermid et al., 2004; McCabe et al., 2005; Saito et al., 2018; Tyler et al., 2005). Experimental models have demonstrated that pain can have a direct effect on shoulder strength. Healthy participants exposed to experimental subacromial pain demonstrate a substantial loss of strength (Stackhouse et al., 2013) and can have an inhibitory effect on muscle activity (Castelein et al., 2017). Patients with large rotator cuff tears, a condition which is thought to be a continuum of SPS, demonstrate large increases in shoulder strength minutes after experiencing pain reduction from a subacromial injection (Itoi et al., 1997; Kirschenbaum et al., 1993).

Inhibition of the supraspinatus and infraspinatus muscles is consistent with pain models, as the nociceptive stimulus associated with SPS arises from near or even within these muscles (Ben-Yishay et al., 1994; Park et al., 2008; Soifer et al., 1996), and thus inhibition of these muscles could serve as a protective mechanism. Although rotator cuff weakness may offer protection in the acute phases of this pathology, prolonged weakness could lead to further disease progression. Due to the mobility of the glenohumeral joint and its limited bony congruence, the dynamic action of the rotator cuff muscles play a crucial role in providing joint stability (Apreleva et al., 2000). Therefore, it is not surprising that weakness of the rotator cuff muscles is associated with an increase in superior translation of the humeral head in patients (Deutsch et al., 1996; Keener et al., 2009; Sahara et al., 2021), as well as experimentatal models of rotator cuff weekness involving induced muscle fatigue (Chen et al., 1999) and a suprascapular nerve block (San Juan et al., 2013).

When assessing strength deficits in patients with SPS, the symptomatic shoulder is either compared to the non-symptomatic side (McCabe et al., 2005; Tyler et al., 2005) or to healthy controls (Camargo et al., 2008; Leroux et al., 1994; Mattiello-Rosa et al., 2008). But to our knowledge, no study has assessed both on the same population. This is important, since acute pain has been shown to cause bilateral changes in motor unit recruitment (Schomburg et al., 2015) and cortical activation (Xiao et al., 2015) in animal models. There are also motor abnormalities noted at the non-involved arm including altered kinematic patterns (Hebert et al., 2002) and longer times to perform occupational tasks (Camargo et al., 2008). Beyond the upper extremity, persons with SPS also display larger center of pressure displacement while performing occupational tasks (Madeleine et al., 1999), suggesting that motor control abnormalities may extend to the lower extremities. Finally, female populations with pathologies of the rotator cuff suffer from higher levels of disability and pain compared to male populations both prior to and after receiving treatment (Razmjou et al., 2011). It is possible that shoulder weakness is more prevalent in female patients with SPS, contributing to the sex related differences in disability and pain. To our knowledge, no previous studies have assessed whether the development of weakness is dependent upon sex in a population with SPS.

Given the gaps identified above, the aim of the present study was to perform a comprehensive study of weakness in patients with SPS. Our first hypothesis was that patients with SPS would demonstrate smaller peak torque values at the symptomatic shoulder and remote locations (contralateral shoulder, both knees), when compared with matched controls. Our second hypothesis was that upon pain reduction, patients experiencing SPS would demonstrate greater peak torque values at both shoulders and both knees relative to pre-injection values. Finally, our third hypothesis was that after controlling for pain duration and pain intensity, female patients would demonstrate smaller peak torque than male patients, when normalized to controls with the use of z-scores.

2. Methods

2.1. Participants

Informed consent was obtained from a sample of convenience of 20 patients (10 females) with unilateral SPS and 20 healthy age, sex, and arm dominance-matched controls (table 1). Prior to recruitment, patients met with a single treating orthopaedic surgeon and were already scheduled to receive a subacromial injection as part of their treatment plan following diagnosis of stage II impingement (Hawkins & Abrams, 1987). To be diagnosed with stage II impingement, patients must have been in pain for at least one month. Moreover, patients needed to demonstrate a positive sign to the treating physician on at least three of the following five manual tests: Hawkins-Kennedy, Neer, painful arc, empty can (Jobe), and painful external rotation resistance (Cotter et al., 2018). A diagnosis of stage II impingement was further confirmed by a positive subacromial injection test, meaning their pain was reduced during the manual tests in the post-injection testing relative to the pre-injection testing. Subjects were excluded if either shoulder had a history of glenohumeral osteoarthritis, humeral head fractures, glenohumeral arthroplasty, rotator cuff tears or repairs, joint laxity, or pain at rest (non-involved shoulder for patients & bilateral shoulders for controls). Additionally, subjects were excluded if either knee had a history of tibiofemoral osteoarthritis, total knee replacement, ACL tears or reconstruction or pain at rest.

Table 1.

Characteristics (means ± SDs) of patients with subacromial pain syndrome and controls.

	Patients	Controls
Sex (males/females)	10 / 10	10 / 10
Age (years)	51 ± 10	52 ± 10
Height (cm)	169 ± 10	171 ± 10
Weight (kg)	86 ± 18	80 ± 17
BMI	30 ± 5	27 ± 4
Dominant Arm (Right/Left)	16 / 4	16 / 4
Dominant Leg (Right/Left)	16 / 4	16 / 4
Injured Shoulder (Dominant/Non-dominant)	6 / 14	---
DASH	38.6 ± 16.8	2.6 ± 3.6*
Pre-injection VAS (0–10)	6.5 ± 2.6	0.0 ± 0.1*
Post-injection VAS (0–10)	29 ± 14	0,0 ± 0.1*

^*Denotes a significant difference between patients with SPS and controls (p≤0.05).

Self-reported measures, including information about health history, demographics, pain location, pain duration (in months), and Disability of the Arm, Shoulder and Hand (DASH) scores were obtained (Hudak et al., 1996). A Visual Analog Scale (VAS) was used to assess pain, once prior and once after the injection/rest period (Hjermstad et al., 2011).

2.2. Protocol

Patient and control groups were tested at two time points. Patients were first tested immediately prior to receiving a subacromial injection, and again fifteen minutes after receiving the injection. Subacromial injections for patient participants were administered utilizing a 23 gauge needle and an anterolateral approach. All injections were performed by the same physician (MS), with over 25 years of clinical experience with this procedure. The subacromial injection consisted of both an anesthetic (6 cc 0.5% Marcaine with Epinephrine) and corticosteroid agent (1 cc DepoMedrol). Following the injection, patient participants were given a 15-minute adjustment period during which time they were asked to move their arm to help disperse the agents within the subacromial bursa. For control participants, the two time points were separated by a fifteen-minute rest period.

2.3. Strength Measurement

Strength at the involved shoulder, contralateral shoulder, ipsilateral knee and contralateral knee was assessed utilizing a maximum voluntary isometric contraction (MVIC) protocol. While we could have used the hip or ankle as the distal joint, the selection of the knee was one of convenience, based on our experimental setup. Force data were sampled at 1000 Hz from a 100 pound uniaxial load cell (omega.com) and data acquisition unit running LabView (ni.com). For each joint, the peak force produced out of the series of three trials was used for data analysis.

The MVICs at the shoulder joint were performed against resisted sagittal plane flexion of the upper extremity, with the shoulder in 90 degrees of flexion, the elbow in full extension and the forearm in neutral pronation/supination. Participants were in a seated position and the trunk was stabilized with straps (figure 1a). The load cell was mounted to a metal testing frame and positioned just distal to the medial and lateral epicondyles of the humerus. The MVICs at the knee joint were performed against resisted sagittal plane extension, with the hip and knee joints flexed to ninety degrees and ankle joint in a neutral position. Participants were in a seated position and the trunk and thighs were stabilized with straps (Figure 1b). The load cell was mounted to a metal testing frame and positioned four inches proximal to the medial and lateral malleoli of the ankle.

Figure 1.

Experimental set-up used for peak torque (strength) measurements during the subacromial pain study. A) Shoulder flexion. B) Knee extension.

Each joint was tested three times, with thirty seconds of rest in between each trial. The testing of sides was randomized and counterbalanced between participants. Participants were verbally encouraged by the investigator during each muscle contraction and instructed to continue the contraction for five seconds. Feedback about performance was not given to participants.

2.4. Data Analysis

For each joint, the peak force was converted to a peak torque score (Nm). To calculate peak torque, peak force was multiplied by the moment arm (measured from the center of the joint to the middle of the point of contact with the load cell). To assess the influence of the subacromial injection, the change in peak torque scores was calculated for each joint by subtracting peak pre-injection torque scores from peak post-injection torque scores.

For the secondary analysis exploring the influence of sex, we created standardized z-scores for each of the four joints: involved shoulder, contralateral shoulder, ipsilateral knee, and contralateral knee. Standardized z-scores were computed for each female patient using the equation below:

$$\text{Standardized Female Z}-\text{Score: }\frac{{\text{X}}_{\text{i}.\text{F}}-{\overline{\text{X}}}_{\text{Controls.F}}}{{\sigma }_{\overline{\text{X}}.\text{F}}}$$ (1)

_{${\overline{\text{X}}}_{\text{Controls.F}}$} is the mean peak torque for the female control participants, ${\sigma }_{\overline{\text{X}}.\text{F}}$ is the standard deviation of peak torque scores for the female control participants, and ${\overline{\text{X}}}_{\text{i}.\text{F}}$ is the Peak Torque score for an individual female patient participant. Likewise, standardized z-scores were computed for each male patient using the equation below.

$$\text{Standardized Male Z}-\text{ Score: }\frac{{\text{X}}_{\text{i}.\text{M}}-{\overline{\text{X}}}_{\text{Controls.M}}}{{\sigma }_{\overline{\text{X}}.\text{M}}}$$ (2)

${\overline{\text{X}}}_{\text{Controls.M}}$ is the mean peak torque for the male control participants, ${\sigma }_{\overline{\text{X}}\cdot \text{M}}$ is the standard deviation of peak torque scores for the male control participants, and ${\overline{\text{X}}}_{\text{i}.\text{M}}$ is the peak torque score for an individual male patient participant.

2.5. Statistical Analysis

SPSS version 22 (IBM, Chicago, IL) was used for all statistical analysis. Values of p < 0.05 were regarded as statistically significant for all analysis. Following conventional ANOVA logic, interaction effects were evaluated before proceeding to main effects.

To determine if there were strength differences between patient and control participants prior to treatment (hypothesis 1), a single two-way mixed model ANOVA was used. The dependent variable was peak torque. The between-subject effect was group (patient and control) and the within-subject effect was joint (involved shoulder, contralateral shoulder, ipsilateral knee, and contralateral knee). An a priori interaction contrast was also run for peak torque scores to assess whether the differences between torque scores for patient and control participants were different between the involved shoulder and remote joints. The a priori interaction contrast was coded as a group (1, −1) by joint (3, −1, −1, −1) interaction. To provide additional confirmation on the presence of weakness at the involved shoulder, an a priori paired t-test was run comparing patient’s involved shoulder to patient’s non-involved shoulder.

To determine if a pain reducing treatment influenced strength (hypothesis 2) a single two-way mixed model ANOVA was used. The dependent variable was change in peak torque after injection (or time for controls). The between-subject effect was group (patient and control) and the within-subject effect was joint (involved shoulder, contralateral shoulder, ipsilateral knee, and contralateral knee). An a priori interaction contrast was also run for change in peak torque to assess whether the differences between changes scores for patient and control participants were different between the involved shoulder and remote joints. The a priori interaction contrast was coded as a group (1, −1) by joint (3, −1, −1, −1) interaction.

For the secondary analysis (hypothesis 3), we ran a total of four hierarchical multiple regression models, one for each joint (involved shoulder, non-involved shoulder, ipsilateral knee, contralateral knee) on patient data only. Standardized z-scores for torque were the dependent variable. Pain duration (in months), pain intensity (based on the pre-injection VAS score) and sex were the independent variables. Pain duration and pain intensity were added to the model first, while sex was added to the model later to see if it increased the predictive power.

3. Results

3.1. Pre-injection – Patients vs Controls

The a priori interaction comparison was significant (p<0.01), revealing that differences between patients and controls peak torques were location dependent (involved joint versus remote joints). Based on this finding, pairwise comparisons were conducted. Peak torques at the involved shoulder were significantly smaller (p=0.04) in the patient population (mean = 34.9 Nm) versus the control population (mean = 48.5 Nm) (figure 2). There were no significant differences between groups at the remote joints, neither when the remote joints were pooled into one composite score (p=0.92) nor when individual joints were analyzed: contralateral shoulder (p=0.68), ipsilateral knee (p=0.75), and contralateral knee (p=0.79). Additionally, the a priori paired t-test revealed that peak torques were significantly smaller (p<0.001) at patients’ involved (mean = 34.9 Nm) versus contralateral (mean = 48.7 Nm) shoulder.

Figure 2.

Pre-injection peak torque (strength) scores (means ± SEMs) for patients with SPS and controls. Analyzed joints include the involved shoulder (IS), contralateral shoulder (CS), ipsilateral knee (IK), contralateral knee (CK), and average of the CS, IK, and CK joints (Remote). * Denotes a significant between-group difference. + Denotes a significant within-group difference.

3.2. Post-injection vs. Pre-injection

For the change in peak torque after treatment, the a priori interaction comparison was non-significant (p=0.46), revealing that differences between patient and control populations did not depend upon location (involved shoulder versus remote joints). We next looked at the ANOVA interaction, where the group*joint interaction (p=0.69) was also found to be non-significant. Finally, we looked at the main effects. Neither the main effect of joint (p=0.76) nor group (p=0.76) was significant (figure 3).

Figure 3.

Change in peak torque (strength) scores (means ± SEMs) for patients with SPS and controls following the injection/rest period. Analyzed joints include the involved shoulder (IS), contralateral shoulder (CS), ipsilateral knee (IK), contralateral knee (CK), and average of the CS, IK, and CK joints (Remote).

3.3. Sex Differences

For the involved shoulder, the associated regression model incorporating only pain duration and pain intensity was non-significant (p=0.31), however the addition of sex to the model significantly improved the prediction of standardized z-scores (p=0.03). Therefore a simpler model incorporating only the main effects of sex was examined to determine how sex predicted z-scores at the involved shoulder (table 2). Sex alone predicted 27% of the variance in z-scores (p=0.02), and female patients z-scores were significantly smaller than males (β = −1.49, p=0.02). Moreover, a follow-up one sample t-test revealed z-scores for males were not significantly different from zero at the involved shoulder (z-score=−0.69, p=0.10) while z-scores for females were significantly smaller than zero (z-score=−2.18, p<0.001) (figure 4).

Table 2.

Regression results for standardized peak torque (strength) z-scores for male and female patients. B is the unstandardized beta, and represents the slope of the line between the independent and dependent variables. SE B is the standard error for the unstandardized beta.

	Involved Shoulder		Contralateral Shoulder		Ipsilateral Knee		Contralateral Knee
	B	SE B	B	SE B	B	SE B	B	SE B
1st Model
Pain Duration (months)	0.01	0.01	0.00	0.01	0.00	0.01	0.00	0.01
Pain Intensity (VAS 0–10)	−0.06	0.13	0.14	0.14	−0.08	0.08	−0.06	0.08
R2	0.13		005		015		013
2nd Model
Pain Duration (months)	0.01	0.01	0.00	0.01	0.00	0.01	0.00	0.01
Pain Intensity (VAS 0–10)	−0.12	012	0.09	0.14	−0.11	0.08	−010	0.07
Sex (0 = male; 1 = female)	−1.45*	0.61	−1.22	0.76	−0.77	0.39	−0.77	0.38
ΔR2	0.23*		015		016		018
3rd Model
Constant	−0.69	0.41	0.61	0.46	018	0.27	013	0.26
Sex (0 = male; 1 = female)	−1.49	0.58	−1.22	0.65	−0.75	0.39	−0.75	0.36
R2	0.27*		016		017		019

^*Denotes significance at p < 0.05

Figure 4.

Peak torque z-scores for male and female patients at the involved shoulder. Circles represent individual scores while diamonds represent group averages. + Denotes a significant difference between male and female patients. * Denotes a significant difference from a one-sample test value of 0.

For each of the three remote joints (contralateral shoulder, ipsilateral knee, contralateral knee), the associated regression models incorporating only pain duration and pain intensity were non-significant (all p>0.05). Neither did the addition of sex to the models significantly improve the prediction of standardized z-scores (all p>0.05). To increase power, simpler models incorporating only the main effects of sex were examined to determine how sex predicted z-scores at each remote joint (table 2). The simpler models were also found to be non-significant (all p>0.05) (table 2).

4. Discussion

Our first hypothesis, that patients with SPS would demonstrate smaller peak torque compared to controls at the symptomatic shoulder as well as remote joints (contralateral shoulder, bilateral knees), was partially supported. Our results demonstrated that patients had significantly smaller peak torque values at the involved shoulder only (figure 2). For the involved shoulder, patients demonstrated substantial weakness when compared to both their healthy side, as well as the control group (28% deficit for both comparisons).

Previous strength testing of populations with SPS have utilized a variety of test positions including external rotation, internal rotation, abduction, and elevation motions among others, as well as a variety of protocols, including isometric and isokinetic, making it difficult to directly compared the results of the present study to previous studies in terms of magnitudes of weakness. Nonetheless our results are consistent with a large body of literature demonstrating weakness at the involved shoulder of patients with SPS in a variety of positions and during a variety of protocols (Camargo et al., 2008; Leroux et al., 1994; MacDermid et al., 2004; McCabe et al., 2005; Tyler et al., 2005; Warner et al., 1990).

Limited studies have found emerging evidence for motor abnormalities across the noninvolved shoulder of persons with SPS, including changes in time to peak torque or acceleration (Camargo et al., 2010; Mattiello-Rosa et al., 2008). Moreover, while a large number of studies has assessed strength across the non-involved shoulder, few have statistically compared this shoulder to a control population. To our knowledge, only one study has found peak torque deficits in the non-involved shoulder (Camargo et al., 2008). The present study found no evidence for weakness at the non-involved shoulder or either knee. These results have important implications for the utility of using the contralateral side as a control for experimental studies.

Our second hypothesis, that pain reduction (via an anesthetic injection) would result in increased strength across the involved shoulder and remote joints of patients with SPS, was not supported. No significant changes to peak torque were observed across any joints following the injection (figure 3). In the long-term, patients with SPS have been shown to experience strength gains across the involved shoulder after a variety of interventions that successfully reduced their pain (McClure et al., 2004; Viswanath et al., 2013; Yu et al., 2006), however it is unclear whether strength gains are possible after the acute reduction in pain. Similar to the present study, Park et al. (2008) failed to find isometric strength gains across the involved shoulder in a population of 153 patients with SPS thirty minutes after a subacromial injection. Interestingly, they also tested a population with rotator cuff tears, finding that strength gains did occur. A number of other studies have shown acute strength gains after a pain relieving injection in populations with rotator cuff tears or mixed rotator cuff tear and SPS populations (Ben-Yishay et al., 1994; Itoi et al., 1997; Kirschenbaum et al., 1993). Collectively our results paired with previous studies suggest that patients with SPS may not have the same motor responses to acute pain removal as patients with rotator cuff tears, despite the fact that SPS and rotator cuff tears are viewed as a continuum of diseases. Unlike patients with rotator cuff tears, in order for patients with SPS to regain strength, additional time or interventions may be needed in addition to a subacromial injection. The lack of changes to strength could also be due to the position that we used. A flexion task such as the one we used required contributions from both the supraspinatus and deltoid. However it is possible that a different motion that required more torque generation to come from the impaired rotator cuff muscles, such as external rotation, could show greater inhibition and changes.

Our third hypothesis, that after controlling for pain duration and pain intensity, female patients would demonstrate greater weakness than male patients, was supported. Our results revealed significantly greater z-scores at the involved shoulder in female compared to male patients, indicating that female patients experienced greater weakness than male patients. It is important to point out that because we are using z-scores, this difference in not in absolute strength values, but when compared to controls. Interestingly, pain intensity and pain duration did not demonstrate a significant correlation with z-scores, suggesting that greater strength deficits were not present in those with greater pain intensity or prolonged pain. To our knowledge, no previous studies have evaluated whether sex differences contribute differentially to motor responses in SPS. Similar to our study however, Miller et al. (2016) found that females with rotator cuff tears exhibited greater normalized strength deficits than males. The association between weakness and sex hold promise for future research on shoulder pain, as female patients generally report greater physical limitations than male patients both before and after treatment. (Razmjou et al., 2011)

There are several limitations to the present study. Only one position was utilized in this study to assess strength. Patients were recruited and tested right when they were diagnosed. Since they weren’t planning on taking part in a study that day, we wanted to minimize the time of the experiment to help with recruitment. It is possible that other positions such as external rotation would yield different conclusions about the influence of pain on rotator cuff strength. Also, there may be a difference between the acute and prolonged reduction of pain, therefore future studies are needed to look at the long-term effects of pain reduction on strength.

Conclusions

The present study found that patients with SPS demonstrate weakness, as measured by smaller peak torques, across the involved shoulder while remote joints appear to demonstrate normal strength. Female patients were shown to exhibit greater levels of weakness than male patients at the involved shoulder. Moreover, pain reduction (through an anesthetic injection) had no influence on strength in the short-term. Further studies are required to investigate which characteristics are associated with the development of weakness as well as the influence of sex on motor responses to treatment.

Highlights.

• Patients with subacromial pain syndrome demonstrate substantial weakness in their involved shoulder

• Shoulder weakness in this population is not resolved after pain reduction

• Female patients exhibit greater levels of weakness than male patients