Experimental shoulder pain models do not validly replicate the clinical experience of shoulder pain

1 Introduction

People with shoulder pain often demonstrate impaired shoulder movement patterns, muscle activity and muscle strength [1], [2]. It is not known whether shoulder pain causes changes in muscle function, or whether impaired shoulder muscle function causes shoulder pain. Without knowing the causal direction, clinicians cannot make effective decisions to manage and rehabilitate patients with shoulder pain. For example, clinicians do not know whether performing rehabilitation exercises with pain is potentially detrimental to patients.

Pain induced under experimental conditions in  asymptomatic participants is a common method used to investigate the effect of pain on muscle function and local nociception induced by the injection of hypertonic saline is commonly used. Previous studies on the shoulder have induced pain by injecting hypertonic saline into the subacromial space [3], [4], [5] or into the supraspinatus muscle [3]. However, it is not known whether either of these models validly reproduces the clinical experience of shoulder pain.

Current literature suggests that people experience pain from nociception in the shoulder in the lateral proximal arm, which is aggravated during shoulder movements, passive shoulder impingement tests and isometric shoulder strength tests [6], [7]. People with clinical shoulder pain also describe the quality of pain across all domains of the McGill Pain Questionnaire [8]. An ideal model of experimental shoulder pain should reproduce the clinical experience in terms of the distribution, quality and behaviour of pain. The aim of this study was to determine whether injection of hypertonic saline into the subacromial space or supraspinatus constitutes such a model.

2 Materials and methods

2.2 Study design

The study design is illustrated in Fig. 1. Outcome data were collected on three separate testing sessions in all participants, with at least a 1-week rest period between sessions to “wash out” any residual effects of testing. Using a computer generated (Excel) randomised sequence, participants were randomly allocated to receive an injection of hypertonic saline either into the subacromial space or into the supraspinatus muscle. Subsequently, participants received two further injections in random order: an injection into the same site of isotonic saline that was matched in volume to the preceding injection of hypertonic saline; and an injection of hypertonic saline into the site that was not previously tested. The injection of isotonic saline served as the control condition to identify whether pain was due to chemical irritation from the hypertonic saline or to mechanical effects of the injection. Note that there were four test conditions but each participant only received three injections: all nine participants received hypertonic saline injection to each site, but isotonic saline injections were only given to a subset of participants for each site. This was done to minimise the total number of injections participants received. An investigator not involved in data collection performed all injections.

Fig. 1:

Study design.

3 Results

The mean (SD, range) volume of 6% hypertonic saline required to obtain an NRS score of 7 was 2.4 mL (0.7 mL, 1.7–4.0 mL) for subacromial injections and 2.2 mL (0.5 mL, 1.5–3.0 mL) for supraspinatus injections.

The distribution of pain is summarised in Fig. 2 and Table 2. All hypertonic subacromial injections produced pain in the anterolateral shoulder region. Hypertonic supraspinatus injections produced pain in the anterolateral region of the shoulder in eight out of nine participants. Hypertonic subacromial injections produced pain across a wider area compared with hypertonic supraspinatus injections. No participant experienced pain outside the areas shown in Fig. 2. Pain experienced during isotonic injections was localised to the site of the injection.

Fig. 2:

Composite body charts representing changes in pain distribution following injection of hypertonic saline. Coloured regions were formed by overlaying regions of pain for each participant. Darker colours indicate more participants reported pain in that area.

Following hypertonic saline injection into both the subacromial space and supraspinatus the most common response to pain provocation testing was a decrease in pain levels (Table 3). After each provocation test, the reported level of shoulder pain was substantially less than 7/10 in most participants.

The results of the MPQ are summarised in Table 4. There was no difference in the quality of pain reported from subacromial or supraspinatus hypertonic saline injections. The most common descriptor used for both conditions was “aching” (six for each condition) while five participants described the subacromial injection as “throbbing” compared with two for supraspinatus injection.

All participants experienced pain following hypertonic saline injection into the supraspinatus muscle and into the subacromial space. Pain scores over time are shown in Fig. 3. There were no significant differences between the amount of pain experienced from subacromial or supraspinatus hypertonic saline injections at any time point (p=0.157–0.916). Following hypertonic saline subacromial injection, all participants were pain-free by 10 min post injection. Following supraspinatus hypertonic saline injection, two out of the nine participants had an NRS score of 1 at 10 min post-injection.

Fig. 3:

Time-course of pain following injection of either hypertonic or isotonic saline in both supraspinatus and the subacromial space. Group median (error bars: interquartile range) are shown with larger markers. Individual data for the nine participants are shown with smaller markers.

Isotonic saline injections produced a mean (SD) peak NRS of 1.4 (1.1) at the supraspinatus site and 0.3 (0.5) at the subacromial space site. Isotonic saline injections at either site produced a shorter duration of pain than hypertonic saline injections.

4 Discussion

The results of this study indicate no difference between the experimental shoulder pain protocols investigated. Injection of hypertonic saline into the subacromial space and into supraspinatus muscle consistently reproduced the location of clinical shoulder pain i.e. in the lateral shoulder and proximal upper arm. However, both protocols failed to replicate the quality of the clinical experience and the behaviour of clinical shoulder pain in response to provocation tests. Indeed, pain provocation tests generally alleviated rather than provoked experimental shoulder pain. These results indicate that pain induced by local shoulder nociception produced by hypertonic saline injection has significant limitations as a model of clinical shoulder pain.

The pattern of experimental shoulder pain induced in the current study is consistent with the distribution of clinical shoulder pain. An investigation of 285 patients who had posterosuperior rotator cuff tear found that the majority (86%) reported pain in the anterolateral region of the shoulder and upper arm which rarely extended past the elbow [23]. In patients with various structural deficits at the shoulder, the most common location of pain was the anterolateral shoulder and upper arm [24], [25]. The pattern of pain induced in the current study is broadly consistent with these clinical findings.

The results of this study support the findings of previous research reporting pain distribution following hypertonic saline injection into the subacromial space [4], [26], where all participants experienced pain over the lateral proximal arm. However, in these previous studies a small proportion of participants also reported pain distal to the elbow. No previous studies have described pain distribution following injection of hypertonic saline into the supraspinatus muscle.

The provocation tests performed in the current study represent a typical range of tests used to elicit clinical shoulder pain. To validly reflect the clinical experience, an experimental model should be associated with increased induced shoulder pain during provocation testing. However, following injection of hypertonic saline, the most common experience reported by participants in the current study was a decrease in shoulder pain intensity regardless of whether the provocation test was performed passively, dynamically, or isometrically. A total of 54 provocation tests were performed and an increase in experimental shoulder pain intensity occurred in only five of these tests (two following hypertonic saline injection into the subacromial space and three following injection into supraspinatus). These results indicate that neither of the experimental models in the current study validly replicated the clinical experience of shoulder pain with respect to pain behaviour.

This is the only study to investigate the experimental pain response to typically provocative manoeuvres following hypertonic saline injection into the subacromial space and supraspinatus. However, our finding that there was no change or, more commonly, a decrease in experimental pain intensity following typical tests known to increase shoulder pain in patients is consistent with a previous study at the shoulder investigating the experimental pain response to provocation following hypertonic saline injection into the acromioclavicular joint [26]. In this previous study, provocation testing did not change experimental pain intensity in the majority of subjects.

There were no differences in the quality of pain experienced following hypertonic saline injection into either the subacromial space or supraspinatus. The score on the evaluative component of the MPQ, considered a measure of the intensity of the pain experience [21], was approximately 23% of the highest possible score for both experimental shoulder pain conditions. Most words chosen were in the sensory component of the MPQ with a rating of approximately 19% of the highest possible value for both conditions. Low pain descriptor scores were recorded in the miscellaneous (approximately 8% of the highest possible score) and affective components (less than 3% of the highest possible score) of the MPQ for both the conditions.

Quality of pain reported during clinical shoulder pain has been investigated in one study [8], which found higher ratings in the affective component of the MPQ (35.5% of total MPQ score) compared with the current study (supraspinatus injection site 2.1% of total MPQ score; subacromial injection site 0.7% of total MPQ score). As affective descriptors of pain are a measure of emotional distress [21] it is not surprising that short duration, chemically-induced experimental pain is unable to replicate the emotional state associated with clinical shoulder pain.

Isotonic saline injections produced either no pain or very mild pain. When participants did report mild pain, it was located in the immediate vicinity of the injection site with no pain referral. These results indicate that the pain felt during the experimental pain protocols is more likely due to the chemical irritation from the hypertonic saline rather than the mechanical effects of the volume of saline injected.

In the current study highly consistent results were demonstrated in our sample of nine participants. Both experimental shoulder pain protocols investigated consistently reproduced the most common location of clinical shoulder pain, scored similarly on the MPQ and, in over 90% of tests designed to provoke clinical shoulder pain, did not result in an increase in experimental pain intensity. As stated above, it is probably unsurprising that these commonly used experimental shoulder pain models are unable to replicate the quality of the clinical pain experience. However, it is concerning that they do not replicate clinical pain behaviour. Not only did experimental shoulder pain not increase following provocation testing, the majority of participants in this study reported a decrease in shoulder pain. The validity of experimental pain models which produce the opposite response to provocation testing to clinical pain must be questioned. It has to be concluded, therefore, that pain induced by local shoulder nociception produced by hypertonic saline injection has significant limitations as a model of clinical shoulder pain.