Acute Muscle Stretching and Shoulder Position Sense

Abstract

Context: Stretching is common among athletes as a potential method for injury prevention. Stretching-induced changes in the muscle spindle properties are a suggested mechanism, which may imply reduced proprioception after stretching; however, little is known of this association.

Objective: To evaluate whether acute stretching of the shoulder muscles affects position sense.

Design: A crossover design with subjects randomized to 3 groups.

Setting: A university human research laboratory.

Patients or Other Participants: Nine male (age = 24 ± 3 years) and 9 female (age = 21 ± 2 years) healthy volunteers.

Intervention(s): Stretching of shoulder (1) agonists or (2) antagonists or (3) nonstretching control.

Main Outcome Measure(s): We determined position sense acuity of the right shoulder before and after the interventions by having subjects attempt to reproduce arm positions of 15° and 30° (shoulder adduction) while starting at 45° to the sagittal plane. The outcome variables were response variability (variable error) and overall accuracy (absolute error).

Results: The relative change in variable error (ie, variable error after/variable error before) was not significantly different between the interventions ( P = .38). Similarly, no change in absolute error was found ( P = .76). Furthermore, no differences were noted regarding test sequence or the interaction of intervention × sequence for either variable error ( P = .73 and .53, respectively) or absolute error ( P = .71 and .67, respectively).

Conclusions: We found no effect on shoulder position sense after an acute bout of stretching of either agonist or antagonist shoulder muscles.

Flexibility training incorporating muscle stretching is recommended for athletes to improve joint range of motion and athletic performance. ¹ With regard to the latter, a general finding is that long-term stretching programs lead to enhanced performance, ^{2, 3} whereas diminishing effects on performance are often reported for acute episodes of stretching ^4–7 (see also review ⁸). Although the exact mechanisms behind the acute effects of stretching on performance are not clear, decreased musculotendinous stiffness or neural mechanisms (or both) leading to depression of muscle activation after acute stretching have been proposed. ^{4, 7}

Recent findings ^{9, 10} suggest that stretching, as a component of a long-term intervention program, can be effective in preventing injury during sporting events. In a recent review of the literature, Witvrouw et al ¹¹ proclaimed the benefits of warm-up stretching on injury prevention for high-intensity sporting events that entail bouncing and jumping (eg, soccer) but not for low-intensity activity (eg, jogging, cycling). In investigating the acute effects of stretching on Achilles tendon reflex activity, Rosenbaum and Hennig ¹² found a reduction in reflex force characteristics (eg, decreased peak force, electromyographic latency) and attributed this to improved muscle compliance, thus suggesting a reduced risk of injury. Repeated passive stretching of the triceps surae muscle resulted in a clear and immediate reduction in reflex sensitivity, which was attributed to reduced sensitivity of the muscle spindles to the repeated stretch. ¹³ Both skeletal muscle and the muscle spindle can change their mechanical tension after a contraction or stretch, a property referred to as thixotropy. ¹⁴ With regard to the spindles, this may have implications for the aforementioned decreased musculotendinous stiffness (as an explanation of the reduction in performance after acute stretching) and observed reduction in reflex activity. Because muscle spindle activity is associated with muscle stiffness ¹⁵ and acute stretching increases range of motion ¹⁶ and has been shown to reduce muscle stiffness and tension, ^{4, 12, 17, 18} it may be reasonable to assume a reduction in muscle spindle activity after a bout of stretching. Of importance is that the muscle spindles play an integral role in limb proprioception. Thus, it can be implied that proprioception is altered after acute stretching. However, reduced proprioceptive acuity may be attributed to joint injury in athletes during sporting events, ¹⁹ which may appear paradoxical to the above discussion of injury prevention because of increased compliance and reduced spindle activity after stretching. Investigating the effects of stretching on limb proprioception is thus warranted. To our knowledge, only 1 group has directly tested this. Larsen et al ²⁰ recently reported no effect on knee joint position sense after stretching. Whether this holds true for the upper extremity, which in comparison with the lower extremity has greater limb degrees of freedom and is more dexterously demanding, has not been investigated.

Our purpose was to investigate whether passive stretching of the shoulder muscles, as is often done in connection with sport-related activities, affects shoulder position sense (a submodality of proprioception). Although proprioceptive signals from agonist and antagonist muscles around a joint contribute together to the sensation of limb position and movement, ^{21, 22} the information from the muscles being stretched during a movement (ie, the antagonist muscles) appears most important ²³ (see review ²⁴). Thus, in our experimental design, the effects on position sense of stretching of the agonist and the antagonist muscles were tested separately on different days.

METHODS

Subjects

Eighteen subjects, 9 men (age = 24 ± 3 years, height = 178.2 ± 6.1 cm, mass = 78.8 ± 10.3 kg) and 9 women (age = 21 ± 2 years, height = 165.0 ± 3.8 cm, mass = 66.6 ± 12.5 kg) participated in the study. All subjects were right handed and had no current musculoskeletal problems in the right shoulder. As an inclusion criterion, a screening test for shoulder movements was performed to assure that subjects were free of joint movement abnormalities, eg, bony restrictions to motion, which would contraindicate stretching.

Subjects were tested for position sense of the right shoulder before and after 2 different stretch procedures and before and after a control procedure. Most of the subjects were involved in recreational sports, and all had previous experience with stretching. The study was approved by the ethical committee and was performed after obtaining informed written consent from each subject.

Study Design

We used a randomized, prospective 3 × 3 crossover design. Thus, subjects appeared on 3 occasions, for which 3 different test conditions were implemented. These conditions were passive stretching of agonist muscles (AG-STRETCH), passive stretching of antagonist muscles (ANTAG-STRETCH), and a control procedure (CON) (see below for detailed explanations). The time between test occasions was on average 7.2 days, with a minimum of 2 days. Subjects were randomized to group A, B, or C, as indicated by the sequence of the test conditions (Table 1).

Table 1. Study Design*.

Position Sense Testing

Position sense tests were performed immediately before and after the intervention. A detailed description of the testing apparatus (Figure 1) was previously presented. ²⁵ Headphones were used to eliminate auditory cues and to administer prerecorded verbal instructions. A receiver beneath a rig for the arm continuously collected data of rig movement relative to a fixed electromagnetic transmitter (FASTRAK electromagnetic tracking system; Polhemus, Colchester, VT). From a starting position of 45° to the sagittal plane, horizontal movements to the target positions of 15° and 30° (shoulder adduction) were performed randomly (see Figure 1B). We chose the 2 target positions because it was previously suggested that position-matching acuity for movement extent relies on different cues and may, therefore, depend in part on different mechanisms. ²⁶ Another consideration was that the target positions should not be in an extreme joint position, in which joint receptors are activated. From the starting position, the subject actively moved the arm until a command to stop was given. The rig was locked, and the arm remained at the target position for 5 seconds. Then the subject actively returned to the starting position. Next, the subject actively moved the arm, attempting to reproduce the target position. When the subject considered the arm to be at the target position, he or she pressed the switch in the left hand. For each target position, 6 trials were performed, for a total of 12 trials per test. A training session was given to each subject immediately before the first position sense test on each occasion. This consisted of the subject performing the same procedure as for the testing but with a reduced number of trials, in order to confirm that he or she was fully informed and accustomed to the testing. During this session, the subject was trained to perform the horizontal movements at an angular velocity of about 10°·s ⁻¹. The device used for position sense testing was fully automated, such that instructions on arm movements and data acquisition and calculation of errors were controlled by 2 separate computers. Therefore, with virtually no interaction between subject and investigator, our position sense data were not vulnerable to the human error that can occur with goniometric measurements.

Figure 1. A, Schematic of a blindfolded subject seated in the testing device with the right arm resting on the motorized rig. B, Overhead view depicting the starting (45° to the sagittal plane) and the target positions (30° and 15° to the sagittal plane) of the position sense test.

Muscle Stretching

Before the first position sense test, the subject underwent a training session of the stretch procedure on the left arm, guided by the investigator.

Stretching of Agonist Muscles (AG-STRETCH)

The stretching was performed on the right arm as self-stretching, according to Evjenth and Hamberg, ²⁷ in the direction of horizontal abduction (Figure 2A). The exercise was designed to stretch the biceps brachii, anterior deltoid, coracobrachial, and pectoralis major muscles. ²⁷ These muscles are activated in horizontal arm adduction, congruent with the movement direction from the starting to the target positions in the position sense test: hence, the designation of agonist.

Figure 2. A, AG-STRETCH: stretching in the direction of horizontal abduction. B, ANTAG-STRETCH: stretching in the direction of horizontal adduction.

Stretching of Antagonistic Muscles (ANTAG-STRETCH)

The stretching was performed on the right arm as self-stretching in the direction of horizontal adduction (Figure 2B). The exercise was designed to stretch the muscles of the posterior aspect of the shoulder and arm (eg, the posterior deltoid). These muscles restrict horizontal arm adduction, congruent with the movement direction from the starting to the target positions in the position sense test: hence, the designation of antagonist.

The contract-relax stretch method was used as follows:

1. The subject performed a 5-second submaximal isometric contraction in the direction of horizontal arm abduction (for ANTAG-STRETCH) or adduction (for AG-STRETCH), followed by 2 to 3 seconds of relaxation.

2. The subject stretched in the direction of horizontal adduction (for ANTAG-STRETCH) or abduction (for AG-STRETCH). The endpoint was defined as a subjective stretch sensation similar to that felt when stretching after training but not to the extent of inducing pain. The area of stretch sensation corresponded to the posterior (for ANTAG-STRETCH) or anterior (for AG-STRETCH) aspect of the shoulder and arm. The arm was kept at the endpoint for 20 seconds. ¹

3. Steps 1 and 2 were repeated twice more.

Control Procedure

In the control condition, subjects sat for 5 minutes in a chair with the arms on the elbow rests.

Data Handling and Statistical Analyses

The difference between the reproduced position and the target position was determined for each trial (algebraic error). To assess response variability, the standard deviation (population based ²⁸) of algebraic errors, that is, the variable error, was calculated for each position sense test before and after each test condition separately. Variable error was determined after detrending the data for each subject's test series (least square means were computed for each target position ²⁹). To estimate the overall accuracy of the position sense, the absolute error (the mean of the absolute values of the algebraic errors) was determined. In order to normalize the data, we calculated the relative change from before to after each intervention, that is, the posttest/pretest ratios of the outcome variables, for each target position. These relative change data were tested with a multivariate repeated-measures analysis of variance (version 13; SPSS Inc, Chicago, IL) in order to determine differences between the interventions (within-subjects variable) and the sequence of the interventions (between-subjects variables); subsequent univariate analyses were performed for each target position. Also, 95% confidence intervals were determined for these relative change data. Whenever the sphericity assumption was not met, the Huynh-Feldt correction was used. All multivariate tests were performed using the Wilks lambda. For all analyses, the level of significance was set at P < .05.

RESULTS

Response Variability (Variable Error)

The mean variable errors for each target position before and after the test conditions and the relative change are reported in Table 2. We noted no overall difference between interventions (F _4,50 = 1.07, P = .38) or test sequences (F _4,24 = 0.50, P = .73) or in the interaction of intervention × sequence (F _8,50 = 0.89, P = .53) in the variable error relative change from before to after the interventions. For comparisons between 15° and 30°, paired t tests performed for the pretesting data of each respective intervention (ie, the “Before” column in Table 2) revealed that variable error either tended to be or was significantly higher for 30° (range of P = .09–.007). Univariate analyses of target positions showed no significant differences between interventions or test sequences and no interaction between interventions and test sequences (P > .05).

Table 2. Variable Error on Position Sense Tests.

Overall Accuracy (Absolute Error)

We analyzed absolute error to estimate the overall accuracy of the position sense tests. Table 3 shows the mean absolute error for each target position before and after the test conditions and the relative change. We found no overall difference between interventions (F _4,50 = 0.47, P = .76) or test sequences (F _4,24 = 0.54, P = .71) or in the interaction of intervention × sequence (F _8,50 = 0.72, P = .67) in absolute error relative change from before to after the interventions. For comparisons between 15° and 30°, paired t tests performed for the pretesting data of each respective intervention (ie, the “Before” column in Table 3) revealed that absolute error tended to be higher for 30° (range of P = .20 to .05). Univariate analyses of target positions showed no significant differences between interventions or test sequences or interaction between interventions and test sequences (P > .05).

Table 3. Absolute Error on Position Sense Tests.

DISCUSSION

Acute stretching of the shoulder muscles did not affect shoulder position sense. This result pertained to both the response variability (variable error) and the overall accuracy (absolute error) of the position sense test. Subjects performed agonist and antagonist muscle stretches on different occasions to allow us to test whether the outcome would differ because of the relative contributions for encoding position sense. ^{21, 22} However, because neither intervention yielded a significant change in position sense acuity, further elaboration on agonist/ antagonist contribution is not possible at this time.

As noted in the introduction, the muscle spindles are regarded as important contributors to proprioceptive acuity. Therefore, the absence of stretching effect on shoulder position sense could mean that the stretching did not appreciably change the spindle firing characteristics. Thus, the question could arise as to whether the stretching intervention was enough to achieve the reduced tension and stiffness presumed to be important attributes in poststretch injury prevention. It is not possible to ascertain this because we did not directly measure stiffness. However, it is of interest to point out that acute reductions of tension and stiffness of the muscle-tendon unit lasting up to 1.5 hours were demonstrated after bouts of stretching similar to those used in the present study. ^{4, 12, 17} Also important is that the amount of stretching in the study, that is, the stretch duration and the number of repetitions, is recommended as adequate for achieving muscle-tendon unit lengthening and is, therefore, a common protocol in the field ^{4, 30–32} (see reviews ^{1, 33}). Also, all participating subjects had previous experience with stretching in connection with physical activity, which made it easier for them to comply with the instructions to control the intensity of stretching as demonstrated by the investigator (see “Methods” section).

It could also be that our position sense test was not sensitive enough to detect a change that may have occurred in the muscle spindle activity induced by stretching. This brings up the issue of whether our choice of proprioceptive test was the right one. Proprioception is currently acknowledged as a complex entity encompassing several different types of sensations, eg, position sense, velocity sense, movement detection, and force (see review ³⁴). Of these, position sense is most commonly tested in clinical and experimental settings. We previously showed a reduction in position sense accuracy after low-level activity to fatigue. ²⁵ This reduction was attributed to altered muscle spindle activity that was secondary to fatigue-induced metabolite accumulation in the muscle. Based on this experience, we used the position-matching test in the present study. It could be that one of the other submodalities of proprioception would have been more sensitive to our stretching maneuvers. Therefore, before we can generalize about the effects of stretching on proprioception, further investigation is required.

Ours is the only study to our knowledge that has directly tested the effects of stretching on proprioception for the upper extremity. Larsen et al ²⁰ recently reported no change in knee joint position sense after static stretching of the quadriceps and hamstring muscles. Although the reader's initial impression could be that their finding and ours are in corroboration, it is important to point out that the reported outcome variables for the studies were not the same. Larsen et al ²⁰ relied on constant error, whereas our interpretation was based on variable and absolute error. Constant error represents the response bias, that is, the subject's tendency to overestimate or underestimate the perceived stimuli. Constant error is labile; it drifts over time, ³⁵ depends on aftereffects, ³⁶ and is easily modified by feedback. ³⁷ However, variable error does not change easily. To a large extent, it reflects the signal-to-noise ratio in a system and, thus, the limitation for information transfer. Accordingly, a variability measure seems best suited for measuring proprioceptive acuity. ^{29, 38} Absolute error is the most common outcome variable of position sense tests in clinical studies but may be hard to interpret because it is a composite of constant error and variable error. Therefore, in contrast to the study by Larsen and colleagues, ²⁰ our use of variable error, multiple trials, and more than 1 target position suggests a more robust experimental approach, albeit both groups reached the same conclusion of no effect of stretching on position sense.

In summary, we found no effect on shoulder position sense after a bout of stretching of the shoulder muscles with a protocol similar to that recommended and performed within a warm-up procedure. A consideration would be to examine other submodalities of proprioception (eg, movement detection) in order to generalize our findings in the context of the effects of stretching on proprioception.