Background
Shoulder pain is the third most common musculoskeletal disorder [
1]. Between 7 and 34% of adults experience shoulder pain at any one time [
2], and only half of new episodes of shoulder pain completely recover within the first six-months [
3,
4]. Shoulder pain can lead to functional limitations and negatively impacts health-related quality of life [
5,
6].
Data suggest that sensorimotor dysfunction could play an important role in shoulder pain [
7‐
10]. People with persistent shoulder pain exhibit findings consistent with the presence of central sensitization [
11], and neurophysiology studies have shown alteration in motor cortex function in people with shoulder problems [
12‐
14]. Proprioceptive impairments have been demonstrated in people with rotator cuff pathology and with shoulder instability [
15‐
18]. Furthermore, in healthy individuals, tactile acuity and motor imagery performance were associated with the physical performance of the shoulder [
19]. Most importantly, studies have shown that interventions targeting sensorimotor function, such as motor control retraining exercises [
20], proprioceptive exercise [
21,
22], mirror therapy [
23], and a combination of pain neuroscience education, tactile discrimination, and graded motor imagery [
24] may increase function and reduce shoulder pain intensity.
Sensorimotor impairments can manifest in a variety of ways and thus further developing our understanding of sensorimotor dysfunction in shoulder pain may improve current models of care. One manifestation of sensorimotor dysfunction is disturbed body perception. This is a complex phenomenon that is attributed to multiple factors; as Pazzaglia and Zantedeschi elegantly describe [
25]:
“knowledge of the body is filtered by perceptual information, recalibrated through predominantly innate stored information, and neurally mediated by direct sensory-motor information.” Evaluation of disturbed self-perception therefore represents a way of capturing the conscious correlate of disruption across a range of sensorimotor processes. In addition, as pain may be viewed as a reflection of an individual’s apprehension of threat to their bodily or existential integrity [
26], consciously-perceived body perception may be viewed as the basis for the appearance of pain.
Researchers have previously developed questionnaires to measure perceptual impairment in complex regional pain syndrome affecting the upper limb [
27,
28]; however, so far, no method to evaluate impaired body perception in people with shoulder pain is ready for clinical implementation. The Fremantle Back Awareness Questionnaire (FreBAQ) was developed to evaluate body perception specific to the back in people with chronic low-back pain [
29]. The initial development of the FreBAQ was based on the Galer and Jensen [
27] complex regional pain syndrome questionnaire, with the addition of items derived from the results of more up-to-date perceptual research [
29]. The FreBAQ has since been translated into Japanese [
30], Dutch [
31], and German [
32]. The questionnaire is composed of nine items investigating, neglect-like symptoms, reduced kinesthetic acuity, and perceived body shape and size. The FreBAQ score is related to pain intensity and disability and it has acceptable psychometric properties [
29,
30,
33]. A knee-specific body perception questionnaire, the Fremantle Knee Awareness Questionnaire - Japanese (FreKAQ-J), was also developed [
34]. The FreKAQ-J has acceptable psychometric properties and is also significantly associated with knee pain intensity and knee pain-related disability [
34].
We were interested in investigating whether people with persistent shoulder pain would also endorse symptoms consistent with body perceptual deficits and whether any perceptual impairment relates to clinical status. Changing the key wording of items to allow for use across other musculoskeletal pain issues is one of the methods available to modify a questionnaire, although thorough psychometric testing is essential before adopting the questionnaire into clinical practice. For example, the Keele STarT Back Screening Tool [
35‐
38] and the Fear-Avoidance Beliefs Questionnaire [
39‐
42] for low-back pain were modified for other musculoskeletal disorders including lumbar stenosis and knee, shoulder, and neck pain. Therefore, the aims of this study were to modify the FreBAQ-J by replacing ‘back’ with ‘shoulder’ to enable use for people with shoulder pain, determine whether people with shoulder pain report impairments in self-perception specific to the shoulder, assess the psychometrics of the scale, and investigate whether the scores on this scale relate to clinical status.
Discussion
The aims of the present study were to develop a shoulder-specific body perception questionnaire, to investigate whether people with shoulder pain experience shoulder-specific self-perception deficits, evaluate the psychometric properties of the scale and explore the scales relationship with clinical status. Our results suggested that the newly developed FreSHAQ-J has acceptable categorical order, unidimensionality, no misfitting items, and excellent test-retest reliability. The formation of a unidimensional scale of nine items suggests that the scores can be summed to obtain a measure of perceptual impairment. This scale does not produce bias by age or sex and clinical characteristics such as pain intensity, pain duration, or disability. The category rating scale functions as expected. We observed no floor or ceiling effects in the clinical sample as participants with the lowest or highest possible score were less than 15% of the sample, in fact no participant with shoulder pain recorded either the lowest or highest possible score. Altogether, the FreSHAQ-J showed sufficient psychometric properties to be used for assessing disturbed body perception in people with shoulder pain in a Japanese population.
This is the first study to suggest that people with shoulder pain endorse symptoms consistent with disrupted self-perception and adds some weight to the suggestion that distorted body perception should be targeted as part of the management of people with persistent shoulder pain. How we experience our body is a complex issue with various terms and taxonomies used to try and capture different aspects of body representation. Three main criteria are generally used to distinguish between the different ways the body is viewed and experienced. Firstly, distinction is made between short-term representations and longer-term, more stable representations. Secondly, body representations can be distinguished by their availability to consciousness. Some representations are quite implicit and operate largely outside of consciousness whereas others are unequivocally conscious representations. Lastly, distinction is made about functional role, particularly whether the representation is action orientated or perception orientated [
62]. A recent scoping review sought to map and examine the range of literature on functional representations of the body in people with musculoskeletal pain disorders [
63]. This group considered the three key domains of body representation pertinent to musculoskeletal pain, the perception of the body (how the body feels to the person), the perception of the space around the body and the sense of ownership of the body, and reviewed both implicit and explicit ways of assessing these domains [
63]. Under this taxonomy the FreSHAQ-J would be regarded as measure of explicit body perception, with reasonable face validity as it employs a subjective self-reported scale to measure a subjective consciously felt construct.
The mean availability of the FreSHAQ-J (− 1.21 logits) was consistent with those of the FreBAQ-J (− 0.88 logits) [
30] and the FreKAQ-J (− 0.92 logits) [
34]. The FreSHAQ-J, similar to the FreBAQ-J, covers moderate and high levels of distortion of body perception and is most suitable for use in people experiencing frequent perceptual dysfunction. In contrast, participants with lower levels of perceptual impairment were not successfully targeted by the FreSHAQ-J. The results suggest that from a psychometric perspective, the addition of new items is needed to assess more subtle impairments. However, it remains to be determined whether low levels of perceptual impairment do in fact impact on clinical status.
The person reliability index was 0.65. This is slightly below the acceptable level of 0.70 and lower than that seen in people with low back pain (FreBAQ-J person reliability index = 0.76) [
30] and in people with knee osteoarthritis (FreKAQ-J person reliability index = 0.81) [
34]. The person reliability index depends on how targeted the scale is. Low person reliability suggests that questionnaires have limited ability to distinguish distinct classes of perceptual impairment. Sensitivity to change in the questionnaire may also not be ideal. That is, the questionnaire may not be able to detect changes when the patients with high-level body perception disturbance experience improvement in body perception because the FreSHAQ-J does not include questions targeting low-level body perception disturbance. Again, adding scale items that are more easily endorsed and assess lower levels of disturbance should improve the tool. However, Cronbach’s alpha was 0.71, which is higher than the acceptable level of 0.7. There are some differences between Cronbach’s alpha and the person reliability index. Cronbach’s alpha is calculated using all scores, including the maximum and minimum scores, while the estimate of the person reliability index requires extrapolated values for extreme scores. This difference in computation may impact the difference seen between the Cronbach’s alpha and the person reliability index. A minimum value of 0.7 for the Person Reliability Index is recommended for population use and 0.85 for personal use [
52]. Therefore, the FreSHAQ-J may be better suited to group use, for example comparing body perception between people with two different conditions, than to use with individual patients.
There were no items showing excessive infit or outfit, and no item bias was observed. Unidimensionality is a key prerequisite for summing a set of items [
55,
64,
65]. The present study demonstrated that the scale showed unidimensionality, suggesting that distorted body perception, including three dimensions (neglect-like symptoms, reduced kinesthetic acuity, and perceived body shape and size), was the only construct in people with shoulder pain. Therefore, the FreSHAQ-J allows meaningful comparisons of the degree of body perceptual distortion.
Item hierarchy in the FreSHAQ-J was similar but not identical to that in the FreBAQ-J. The greatest discrepancy between the FreSHAQ J and FreBAQ J was found in item 3 (I feel as if my sore shoulder sometimes moves involuntarily, without my control). Participants with shoulder pain (0.01 logits) endorsed Item 3 to a higher degree than did the participants with low-back pain (0.91 logits). Item 3 was based on symptoms of motor neglect frequently observed in participants with complex regional pain syndrome. Although motor deficits are commonly found in participants with low-back pain [
66‐
68], motor neglect may be observed to a higher degree in participants with limb pain than with low-back pain, and it is certainly plausible that involuntary movement is a more common percept in the limbs than in the trunk.
The ICC of 0.84 (95% CI 0.70–0.92) showed that the FreSHAQ-J has excellent test-retest reliability, confirming the findings of the previous FreBAQ-J study with people with low-back pain (0.81, 95% CI 0.67–0.89) [
30] and the FreKAQ-J study assessing people with knee osteoarthritis (0.76, 95% CI 0.52–0.89) [
34].
The marked differences in the FreSHAQ-J total scores between the shoulder pain group (mean 9.1) and the control group (mean 2.5) and the relationship with disability and pain catastrophizing, support previous suggestions that perceptual impairment is associated with clinical status in persistent pain [
69]. Conversely, unlike the FreBAQ [
29,
30,
33] and the FreKAQ [
34], this study did not show a relationship between body perceptual disturbance and pain intensity during motion when we corrected for multiple comparisons. Pain intensity in participants with shoulder pain (52.9 ± 21.3 mm) was higher than that in participants with low-back pain (49.1 ± 27.1 mm) [
30] and knee osteoarthritis (43.5 ± 24.1 months) [
34], whereas the FreSHAQ-J scores in participants with shoulder pain (9.1 ± 5.0) were relatively lower than those in participants with low-back pain (11.7 ± 6.4) or knee osteoarthritis (12.4 ± 7.6). These differences may have influenced the relationship between pain intensity and body perceptual impairment and suggest different impacts of body perceptual dysfunction across different clinical presentations.
Limitations of this study need to be considered. First, the sample size of this study is relatively small. Previous studies suggest that small sample size (
n < 100) particularly impact on category disorder, targeting, and misfitting items [
44]. One hundred and twelve participants contributed data to the Rasch analysis and this may impact on the confidence in some of the results. Some of the issues reported above regarding targeting and misfitting may be less apparent in testing in larger samples. Second, the duration of pain in subjects with shoulder pain (12.8 ± 13.4 months) was shorter than that in subjects with low-back pain (88.8 ± 106.8 months) [
30] and knee osteoarthritis (57.7 ± 88.4 months) [
34]. Different results to those observed here may be seen in shoulder pain populations with greater chronicity. Third, the sample was derived from an orthopedic outpatient clinic and participants were referred into the study by an orthopedic shoulder surgeon. Further research, particularly including people from primary care, is needed to evaluate the generalizability of the reported findings. Fourth, we did not lodge and lock our protocol and statistical analysis plan prior to data collection. When we commenced this study, such practice was uncommon in our field, but now it is recommended, and is among those at the forefront of this push [
70]. Failure to do this clearly represents a shortcoming in transparency and reporting. Lastly, we chose pain intensity as the factor to determine clinical stability for the test-retest part of the study as pain related variables were key inclusion criteria and our previous work has shown consistent relationships between pain intensity and body perception. This relationship was not upheld in this study and an alternate measure such as global perceived effect may be more useful in future investigations.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.