The effect of experimental shoulder pain on contralateral muscle force and activation

Abstract

Inhibition of rotator cuff activation and force after local experimental pain has been previously shown. Clinically, strength is often indexed to the uninvolved side in order to quantify deficits during injury and recovery. This study assessed the effect of experimental subacromial pain on contralateral shoulder external rotation (ER) force and activation. We hypothesized that subacromial pain would not affect contralateral shoulder external rotation force and voluntary activation (VA) of the contralateral infraspinatus because we believe that the effects of acute experimental pain will largely exert an ipsilateral, spinal segmental effect. Twenty healthy adults were tested. External rotation force and VA were tested while participants performed maximum voluntary isometric contractions of shoulder external rotation, during which a brief electrical stimulus was applied to the infraspinatus muscle at an intensity that maximized external rotation force. To elicit pain, 1.5 ml of 5% hypertonic saline was injected into the contralateral subacromial space. Mean (SD) pain immediately after injection was 6.3/10 (0.85) resulting in a 7.6% decline in contralateral external rotation force (p< 0.01) and 3.3% decline in infraspinatus muscle VA (p = 0.48). A subset of participants (n=9) showed a more substantial decline in both force (15.4%; p < 0.000001; d = −3.53) and VA (8.7%; p = 0.045; d = −0.98). Experimental shoulder pain had a modest effect on contralateral ER force and VA in a subset of participants; therefore, it may be important to index or compare strength bilaterally throughout rehabilitation to capture this bilateral effect as pain resolves.

INTRODUCTION

Shoulder pain in the general population has a point prevalence between 7–26% (Luime et al., 2004), and painful conditions of the shoulder have been shown to impair muscle function and to be related to outcomes (Andersen, Holtermann, Jorgensen, and Sjogaard, 2008; Celik, Sirmen, and Demirhan, 2011; Joensen, Couppe, and Bjordal, 2009; MacDermid et al, 2004; McClure et al, 2004; Merolla et al, 2010; Sokk et al, 2007). The shoulder girdle is heavily reliant on muscle action to maintain stability, and abnormal EMG activation patterns of muscles surrounding the shoulder have been well documented in patients with painful shoulder conditions (Bigliani and Levine, 1997; Chester, Smith, Hooper, and Dixon, 2010; Glousman et al, 1988; Graichen et al, 2001; Lucas, Rich, and Polus, 2010; Ludewig and Cook, 2000). Therefore, experiencing shoulder pain may adversely affect a patient’s ability to recruit and drive the firing rate of key muscle groups.

Voluntary activation (VA) or “central drive” of muscles during maximum voluntary isometric contractions has been well studied in both experimental (Bilodeau, 2006; Gandevia, 1998; Stackhouse et al, 2013; Stackhouse, Stapleton, Wagner, and McClure, 2010; Zarkou, Stackhouse, Binder-Macleod, and Lee, 2017) and clinical populations (Fitzgerald et al, 2004; Mizner et al, 2005; Stackhouse, Binder-Macleod, and Lee, 2005). Conceptually, VA is comprised of both motor unit recruitment and attained firing rate (Knight and Kamen, 2008). EMG is often used to assess muscle activation, however to interpret the degree of activation, normalization to a maximal voluntary isometric contraction (MVIC) is typically employed. If muscle activation is limited by pain, the use of an MVIC for normalization obscures the meaning of the derived value. Measurement of VA, using a superimposed electrical stimulation technique remedies this problem. We have previously applied this technique to directly assess the VA of the infraspinatus muscle (Stackhouse, Stapleton, Wagner, and McClure, 2010). Briefly, a maximal electrical stimulus is superimposed during a voluntary isometric contraction. If additional force is produced by the electrical stimulus, this represents incomplete VA of the muscle.

We have previously shown that experimental subacromial pain elicits a decline in force and VA of the ipsilateral infraspinatus muscle (Stackhouse et al, 2013). Experimental pain has been induced by injecting both periarticular structures and muscle tissue and has been shown to alter muscle force and activation (Bandholm et al., 2008, Diederichsen et al., 2009, Stackhouse et al., 2013) and alter throwing performance (Wassinger, Sole, and Osborne, 2012). However, these studies have only examined muscle function ipsilaterally after the induction of pain. Across the course of clinical management for shoulder pain, shoulder strength is frequently measured and compared/indexed to the uninvolved side. We questioned therefore, if the inhibitory effect of acute, moderate intensity experimental pain was primarily local and confined to ipsilateral muscles, or if contralateral muscle groups might be affected by spinal or supraspinal inhibitory mechanisms. Similar mechanisms have been observed after unilateral ACL rupture where failure of voluntary activation of both the involved- and uninvolved-side quadriceps can be observed (Chmielewski, Stackhouse, Axe, and Snyder-Mackler, 2004; Williams et al, 2005). If bilateral deficits in force occur with unilateral shoulder pain, it may be important for the clinician to always measure the contralateral side during follow-up assessments as pain resolves. The purpose of this study was to assess the effect of experimental subacromial pain on the contralateral maximum shoulder external rotation force and infraspinatus voluntary activation. We hypothesized that subacromial shoulder pain would have no effect on contralateral voluntary activation (VA) and force.

METHODS

Study Participants

Twenty asymptomatic, healthy volunteers were recruited for the study (mean age 26.9 +/−5.2; 9 males, 11 females). A brief general health checklist was used to screen for current musculoskeletal and neurologic compromise. A brief physical examination, including isometric strength testing, instability testing and provocative maneuvers (Apprehension test, Neer’s test, Hawkin’s test, Jobe’s test, pain with resisted external rotation), was used to screen for potential shoulder pathology that could be an additional source of pain during the testing procedures. If a potential participant had a positive result on any test, they were excluded from participation. All study participants reviewed and signed informed consent forms that were approved by the University’s Institutional Review Board.

Electrode Placement and Force Testing

For the infraspinatus muscle, the motor point (i.e. point on skin that elicits the best motor response) was identified using a 1.5 cm point stimulator as previously described (Stackhouse, Stapleton, Wagner, and McClure, 2010), and two electrodes were applied so that only isolated external rotation without shoulder extension was produced with electrical stimulation. To assess external rotation force production, the non-dominant forearm was secured to a dynamometer (KIN-COM 500H dynamometer Chattecx Corp, Chattanooga, TN) with shoulder at 30° abduction, 0° flexion, 5° of internal rotation, 90° elbow flexion (Figure 1). External rotation force data during testing were sampled at 200 Hz and analyzed with custom-written software (LabView version 5.1, National Instruments, Austin, TX).

Figure 1.

Electrode placement for voluntary activation and isometric external rotation force assessment

(Reproduced from Stackhouse et al, Journal of Shoulder and Elbow Surgery 2010; 19(2):226 from Figure 1)

Experimental Session

To determine the stimulus intensity for the experimental procedure, the infraspinatus muscle was stimulated with a Grass S48 stimulator and the SIU8T stimulus isolation unit (West Warwick, RI, USA). The stimulation intensity was gradually increased until no further external rotation force was produced, indicating maximal recruitment from the electrical stimulus. The stimulus characteristics used for testing were a 3-pulse train (600-μs pulse duration) delivered at 50 pulses/s. The maximum voltage for participants ranged from 110 to 150 volts. To facilitate familiarization with the force dynamometer, participants were then given up to 3 attempts to practice a submaximal voluntary isometric contractions of an isolated external rotation movement while avoiding substitution with other motions (Gandevia, 1998; Knight and Kamen, 2008; Stackhouse, Binder-Macleod, and Lee, 2005; Stackhouse, Stapleton, Wagner, and McClure, 2010; Stackhouse et al, 2013; Zarkou, Stackhouse, Binder-Macleod, and Lee, 2017).

Each subject completed a total of five separate trials (Baseline 1, Baseline 2, 0-minutes, 5-minutes, 10-minutes) for maximum force and voluntary activation assessment. Two of these trials (Baseline 1 and Baseline 2) were performed prior to a subacromial injection of hypertonic saline (pain stimulus), and three trials following the injection (0, 5, and 10-minutes). A five-minute rest period was given between all trials. For all trials, the subject was instructed to perform a maximum voluntary isometric contraction (MVIC) for external rotation. Once a plateau in voluntary force was reached, an electrical stimulus was applied to the muscle being tested. Electrical stimulation was delivered to the muscle at the subject specific intensity that maximized force using a train of three pulses of 600-μs duration delivered at 50 pulses/s. Immediately following the first trial, the subject relaxed, and when their force returned to 0, another electrical stimulus was applied once every 5 seconds, over the next 25 seconds, for a total of 5 resting stimuli. The purpose of these multiple resting stimuli was to establish a stable baseline for the force achieved during a resting burst stimulus (Stackhouse et al, 2013). The first two trials were performed to capture pain-free baseline performance. The last three trials (0-, 5-, and 10-minutes) were collected following a 1.5 ml(cc) injection of 5% hypertonic saline solution into the subacromial space contralateral to the shoulder being tested via a posterior approach by a fellowship trained shoulder surgeon, using vapocoolant spray to numb the injection site briefly. This injection was timed to occur at 3-min after the Baseline 2 trial, and allow for 2-minutes additional rest and for the subjects to reach a peak pain response to the injection (Stackhouse et al, 2013). Post-injection, each participant rated their pain every 30 seconds for the remainder of the session using a numeric pain rating scale (0=no pain and 10=worst imaginable pain). The third trial did not commence until the subject’s pain rating either peaked or reached a minimum of 6/10, typically within two minutes after the injection. Visual force feedback from a computer monitor and strong verbal encouragement (“push, push, push”) were given during all trials.

Data Analysis

Force data from MVICs were normalized to the highest force produced during any trial. VA was calculated for each trial as (1 – force augmentation from the stimulus during the isometric contraction/average of five resting stimuli), where a value of 1.0 is considered full activation (Figure 2). This calculation method is a common approach to quantification of VA (Gandevia, 1998; Knight and Kamen, 2008; Stackhouse, Binder-Macleod, and Lee, 2005; Stackhouse, Stapleton, Wagner, and McClure, 2010; Stackhouse et al, 2013). Means and standard deviations were calculated for pain rating, normalized force, and VA. Normalized force and VA data were checked for skewness (between −1 to 1 and not > 2 times the standard error), kurtosis (between −2 to 2 and not > 2 times the standard error), and sphericity (Mauchly’s W), and no violations were found. A one-way repeated measures ANOVA was used to determine if there were main effects for normalized force and VA. If a main effect was found, Bonferroni-corrected paired t-tests were used to determine if there were significant differences between specific time points. Simple linear regression was used to determine the relationship between changes in normalized force and VA between the average of the Baseline tests and immediately post-injection (0-minute time point). All statistical procedures were performed using SPSS software (IBM, Armonk, NY; v17) and significance was set at p ≤ 0.05. Effect sizes (Cohen’s d) were calculated for the difference between the average baseline trials and the 0-min trial for the normalized force and VA measures.

Figure 2.

Force-trace representing calculation of voluntary activation (VA). An electrical stimulus is superimposed when a maximal voluntary isometric contraction force plateaus and the force augmentation is calculated (5 N). When the subject relaxes, a second electrical stimulus is applied to calculate the force produced during a resting stimulus (17 N). VA was calculated for each trial as [1 - augmentation from the stimulus during the isometric contraction/average of five resting stimuli], where a value of 1.0 is considered full activation (VA = 0.71)

(Reproduced from Stackhouse et al, Journal of Shoulder and Elbow Surgery 2010; 19(2):226 from Figure 1)

Upon visual inspection of the frequency distribution on the force loss from baseline to the 0-min time point, we had a bimodal distribution with a peak of 8 subjects in the 0–5% force loss bin and another peak of 6 subjects in the 10–15% force loss bin. Thus, we choose to divide our data into two subsets: those with ≥10% decline in force and those with < 10% decline in force. The 10% threshold was chosen based of the bimodal distribution and on our own internal laboratory reliability data (Harrington et al, 2013) and that of Cools et al. (2016), which showed that a 10% change in force is beyond the standard error of measurement and the minimal detectable change (90% confidence interval) and is slightly more than double their reported asymmetry of isometric shoulder external rotation force, and thus would represent a true change. Normalized force and VA data of the subsets were checked for skewness (between −1 to 1 and not > 2 times the standard error), kurtosis (between −2 to 2 and not > 2 times the standard error), and sphericity (Mauchly’s W), and no violations were found. The subsets were compared for baseline characteristics of age, sex, body mass index, normalized baseline force, baseline VA, normalized force at 0-min, VA at 0-min, and Pain rating at 0-min using independent two-tailed t-tests with a significance value set at ≤ 0.05.

RESULTS

The results for normalized force, VA and pain across all five trials for the infraspinatus muscle are shown graphically in Figure 3. The two baseline trials were not significantly different (paired t-test) and therefore were averaged to represent normal baseline force and activation for subsequent comparisons. Mean (SD) initial pain from the injection was 6.3 (0.8) out of 10 and corresponded with a 7.6% (8.3) decrease in contralateral external rotation force with an effect size (d) of −1.13. The initial pain was associated with a 3.3% (13.9) decrease in VA with an effect size (d) of −0.33. Repeated measures ANOVA showed a main effect for normalized force (p < 0.01) but not for VA (p = 0.44). Subsequent paired t-tests revealed significant declines in force (p < 0.01) at each of the three post-injection time points when compared to baseline. Additionally, only 11% of the variation in force could be explained by a change in VA from baseline to immediately post-injection (r² = 0.11; p = 0.15). Pain diminished over the 10-minute period to a mean of 0.6/10 (1.23), however force remained depressed by 7.3% compared to the average of the normalized baseline forces. A subset (n = 9 out of 20) showed a decline in external rotation muscle force by ≥ 10% at the 0-minute time point. Please see Table 1 for baseline mean (SD) comparisons between the subsets with ≥ 10% force loss and < 10% force loss on normalized force, VA, and sex. When comparing force loss at the 0-min time point, the ≥10% force loss subset had a mean (SD) decrease in force of 15.4% (6.1), whereas the < 10% force loss subset had a decrease of 1.4% (3.1; p < 0.000001; d = −3.53). The mean decrease in VA was for the >10% force loss subset was 8.7% (14.6), whereas the < 10% force loss subset was 0.03% (12.1; p = 0.045; d = −0.98). The pain ratings were similar between subsets (Table 1).

Figure 3.

Normalized external rotation force and infraspinatus voluntary activation ratios (left vertical axis) before and after subacromial injection of hypertonic saline. Pain is shown according to the scale on the right vertical axis. The injection time point occurred 3-minutes after Baseline 2 measurements and about 2-minutes for peak pain to develop. Error bars represent standard error.

Table 1.

Comparison Between Force Loss Subsets

	Age	BMI	Normalized Force Baseline	VA Baseline	Normalized Force 0-min	VA 0-min	Pain 0-min
≥10% force loss	28.8 (5.9)	23.3 (2.8)	0.950 (0.024)	0.861 (0.07)	0.796 (0.056)	0.774 (0.101)	6.3 (1.0)
<10% force loss	25.5 (4.2)	23.1 (2.2)	0.969 (0.024)	0.876 (0.078)	0.955 (0.034)	0.880 (0.115)	6.2 (0.8)
independent t-test p-value	0.16	0.86	0.09	0.67	0.0000003	0.045	0.40
Effect size (Cohen’s d)					−3.53	−0.98	0.17

DISCUSSION

In this study, we aimed to explore whether experimental subacromial pain altered MVIC force generation and VA of the side contralateral to pain induction via some central mechanism. Our primary finding was that although contralateral shoulder force output was modestly decreased (7.6%), VA was not reduced significantly (3.3%) and the change in MVIC force was not significantly explained by a change in VA (11%). This suggests that a substantial reduction in central drive did not occur for contralateral shoulder muscles when examining our total sample. Unfortunately in our study, the effect on VA was small (d = 0.24) and we were underpowered (power = 0.27) to detect this small effect. Based on our unpublished reliability data with this technique, we believe a change in VA of at least 7% is necessary to represent a true change. With our sample size of 20 and the obtained standard deviation (13.9), the power to detect a 7% change in VA was 0.70.

To our knowledge, the effect of subacromial pain on muscle force and VA at remote sites has not been previously studied. In our earlier work testing the ipsilateral infraspinatus muscle, subacromial pain resulted in an average 33% decrease in force and 23% decrease in VA and a majority of the variation in force (~78%) from baseline to immediately post-injection could be explained by the variation in VA (Stackhouse et al, 2013). The findings from the current study related to strength and activation of the contralateral infraspinatus may be important for clinicians and researchers who often compare or index the involved side strength to the uninvolved side. Because force declined on the contralateral side, we suggest that monitoring force bilaterally over the course of rehabilitation could be important. That is, as pain symptoms resolve, the contralateral force may also improve in some people.

It is our belief that the experimental pain (moderate intensity, mean rating = 6.3/10) experienced in this study did cause the observed modest reduction in force because Farshad et al. (2012) showed no decrement in abduction and external rotation forces to injections (5 ml of physiologic saline or 5 ml of lidocaine) in the subacromial space. The volume of the injections in the Farshad study were 3.3 times the volume that we injected (1.5 ml) and they only analyzed data from those subjects who demonstrated that the injected fluid was located in the subacromial space by comparing US imaging before and after the injection. Because of the imaging performed, Farshad et al. (2012) force measurements occurred approximately 45-minutes after the injection, which is a difference from the timing used in our study. Therefore, we cannot state for certain, that only pain and not the mechanical change in space as a result of our 1.5 ml injection affected the force and VA measurements. We, however, still contend that subjects in Farshad et al. (2012) would still have the injected fluid in the subacromial space at that timepoint, as it was just confirmed via imaging prior to the strength assessments.

After initially examining the frequency distribution data on the force decline from Baseline to 0-min, we observed a bimodal distribution as described in the Methods section. Nine of 20 (45%) participants showed a decrement in strength of 10% or greater (mean = 15.4%) with a concomitant change in VA of 7.8% on the contralateral (non-painful) side. The potential mechanism of this contralateral inhibition is unclear, as the array of sensory communication between sensory and motor areas is complex. Supraspinally, there are both intrahemispheric and interhemispheric pathways for inhibition of the primary motor cortex (Di Lazzaro et al, 1999; Swayne, Rothwell, and Rosenkranz, 2006; Tsutsumi et al, 2012) that can be activated by the pain experience. A spinal mechanism of contralateral muscle inhibition is unlikely because nociceptive fibers are not believed to synapse directly in the contralateral gray matter at the same segmental level (Heise and Kayalioglu, 2009), so it is our assumption that any contralateral reduction in strength and activation observed in this study came from a supraspinal mechanism. The finding that some people may have bilateral deficits is a phenomenon that has been observed for the quadriceps femoris after acute and subacute anterior cruciate ligament (ACL) rupture (Chmielewski, Stackhouse, Axe, and Snyder-Mackler, 2004; Urbach, Nebelung, Weiler, and Awiszus, 1999; Williams et al, 2005). Chmielewski, Stackhouse, Axe, and Snyder-Mackler (2004) reported on a consecutive sample of 100 people with confirmed isolated ACL rupture without clinically-detectable effusion, and 21% of the sample showed a bilateral strength and VA deficit.

It is common during rehabilitation to compare force to the unaffected side as a baseline or goal, particularly in an injured athlete or patient with high functional demand (Kuhn, 2006; Toole et al, 2017). This allows the clinician to have a benchmark for assessing progress during rehabilitation and ultimately the decision to return to play or work. For example, the Multicenter Orthopedic Outcomes Network Group recommends recovery of the injured side quadriceps strength to ≥ 85% of the contralateral side strength for return to sport after ACL reconstructions (http://www.ucdenver.edu/academics/colleges/medicine/sportsmed/cusm_pt/Documents/MOON%20CU%20ACL%20Protocol.pdf) and the Delaware-Oslo ACL cohort group recommends >90% (Grindem et al, 2016). Based on the results of the current study, comparison to the contralateral side seems reasonable to guide the decision process of when to return to activity, particularly if there is minimal pain in the affected extremity. Because 45% of our sample experienced greater than or equal to a 10% reduction in contralateral force while under acute pain, it may be postulated that continued assessment of the contralateral side may be necessary as pain resolves over the course of rehabilitation to provide a more accurate index for side-to-side strength comparisons. Further, much less is known about the restoration of muscle strength symmetry in the upper extremity for shoulder conditions and how it relates to outcomes or risk of future problems. This manuscript is an exploration into potential changes in shoulder external rotation MVIC force and voluntary activation of the contralateral shoulder/side during acute experimental shoulder pain and should serve to spur research in clinical populations.

There are a few limitations to this study that must be acknowledged. This experiment simulated acute pain of short duration, which at best could translate to being similar to acute bursitis, tendinopathy, tears, or other trauma about the shoulder (Bertin, Behier, Noel, and Leroux, 2003; Dougados, Le Henanff, Logeart, and Ravaud, 2007; Lombardi et al, 2008; McClure, Michener, and Karduna, 2006; McClure, et al, 2004; Reilingh, Kuijpers, Tanja-Harfterkamp, and van der Windt, 2008; Unger et al, 2019). Symptomatic patients, particularly those with chronic pain, may show a different response because of the well-documented phenomena of altered pain processing via peripheral and central sensitization and diminished central pain modulation (Woolf, 2011). For example, a recent systematic review with meta-analysis in people with lateral epicondylalgia found the presence of secondary mechanical hyperalgesia (i.e. a sign of central sensitization) and that contralateral wrist extension strength was diminished compared to healthy controls (standardized mean effect size = 0.74) (Heales, Lim, Hodges, and Vicenzino, 2014). Also, we only sampled the infraspinatus muscle for VA and it is possible that changes could have occurred in other muscles groups. Likewise, we did not assess electromyographic activity, which may have been altered despite only modest changes in force and no significant changes in VA. Changes in electromyographic activity have previously been used to suggest alternative neural strategies in the presence of shoulder pain. (Bandholm et al, 2008; Diederichsen et al, 2009; Ludewig and Cook, 2000). Another major limitation of this study was the decision to not include a separate control group that received an injection of physiologic saline. We justified this choice by using a double baseline procedure to show stability of our measures and the prior finding that 5 ml injections (3.3 times the volume injected in our study) of physiologic saline and 5 ml injections of lidocaine, which were all confirmed via US imaging, did not produce detectable losses of MVIC force (Farshad et al, 2012). As noted above in the discussion, we were underpowered to detect the small, observed decrement in VA, however the estimated number of participants (n=112) needed to show a significant difference was not readily feasible to collect and shows the limits of our VA detection method. Future studies should include the study of other muscle groups and continue to explore strength impairments bilaterally in patients with painful shoulder conditions.

CONCLUSION

In conclusion, experimental shoulder pain of moderate intensity has only a small effect on contralateral infraspinatus muscle force and does not have a significant effect on contralateral infraspinatus muscle VA. There also may be a subset for whom a contralateral shoulder muscle response is more significant and; therefore, may be important to index or compare strength bilaterally throughout the rehabilitation to capture the bilateral effect as pain resolves. The results of this study should be interpreted with caution, due to the lack of a control injection group, and should be used to spur further investigations into bilateral force and VA effects in experimental pain and clinical conditions.