Background
Complaints of arm, neck and shoulder are common in Western societies [
1,
2]. In the Netherlands, the 12 months prevalence in the general population has been estimated at 31% for neck pain, 30% for shoulder pain, 11% for elbow pain and 18% for wrist or hand pain [
1]. The general practitioner (GP) is often consulted for these complaints [
1,
3,
4]. In Norway 45% of adults experiencing non-inflammatory musculoskeletal pain reported consulting a GP within 12 months [
3]. In persons with arm, neck and shoulder pain in the Netherlands this was about 30–40% [
1].
A multi-disciplinary consensus was recently reached in the Netherlands to define upper extremity musculoskeletal disorders, to help professionals classify patients unambiguously and to improve communication amongst health care workers [
5]. For the present study, we defined complaints as the symptoms for which a patient consults his/her GP, e.g. pain when active, pain in rest, tingling, stiffness, loss of strength, numbness, cold feeling in shoulder, arm or hand [
5,
6].
In the Netherlands, GPs are consulted 66 times annually per 1000 registered persons for a new complaint or new episode of neck or upper extremity complaints [
4]. Despite treatment of these complaints, many patients do not completely recover within 3, 6 or 12 months after the first consultation. Previous work in the present population of non-traumatic arm, neck and shoulder complaints showed that, 46% of the patients still reported non-recovery after 6 months [
7]. Similar results were found after 6 months in studies on shoulder pain [
8,
9]. Another study on neck and shoulder complaints reported 24% complete recovery after 3 months increasing to 32% after 12 months [
10]. In a study on elbow complaints 13% reported complete recovery and 24% much improvement at 3 months compared with 34% and 21%, respectively, at 12 months [
11].
Non-recovery in complaints of arm, neck and shoulder may be explained through the cognitive-behavioural oriented model for persistence of pain [
12]; this model has been validated in chronic low back pain. Here, kinesiophobia, (also known as fear of movement/(re)injury) may lead to avoidance behaviour resulting in hypervigilance to bodily sensations, followed by disability, disuse and depression which may lead to a vicious circle of fear and avoidance in patients experiencing pain. This is in contrast to non-catastrophizing patients in whom not pain-related fear but rather a rapid confrontation with daily activities is likely to occur, leading to faster recovery. In support of this model, studies on patients with chronic low back pain reported that patients with higher levels of pain-related fear, have higher scores on pain and disability [
12‐
15]. Furthermore, studies on acute low back pain and osteoarthritis in primary care have confirmed the relation between fear avoidance and disability [
14,
16,
17].
In contrast to low back pain, for non-traumatic complaints of the arm, neck and shoulder little is known about the degree of kinesiophobia as measured with the Tampa Scale for Kinesiophobia [
18] and its associated variables [
12,
17].
So far, no studies have investigated whether kinesiophobia remains stable during the transition period from new episode to chronic complaint. However, we expect the mean kinesiophobia scores to remain stable over time, because kinesiophobia was not specifically intervened upon. In addition, we expect that those variables of the fear-avoidance model involved in low back pain will also be associated with kinesiophobia in the case of non-traumatic arm, neck and shoulder complaints.
The aims of the present study were: (1) to examine the degree of kinesiophobia in patients with non-traumatic complaints of arm, neck and shoulder in general practice; (2) to establish whether the mean scores of kinesiophobia change over time in non-recovered patients; and (3) to evaluate variables associated with kinesiophobia in these patients at baseline.
Discussion
Recent studies in primary health care, reported mean TSK-AV scores similar to our study group; these populations consisted of patients with chronic neck pain [
30], osteoarthritis [
17], and acute low back pain [
16]. However, in secondary care two studies on chronic low back pain reported mean TSK-AV scores of 31.6 [SD: 7.2] [
19] and 33.8 [SD: 7.6] [
18]. In the non-recovered patients in our study group the mean TSK-AV score at 12 months follow-up was 26.1 [SD: 7.8].
A possible explanation for differences in mean kinesiophobia scores between primary care populations and patients with chronic complaints at other care levels, might be that fear is a predictive factor in developing chronic complaints. This would imply, that patients who develop chronic complaints more frequently have a higher baseline score compared to quick recoverers. In previous work in the present population of patients with non-traumatic arm, neck or shoulder complaints, we found a univariate relation (odds ratio 1.4; 1.0 to 2.0) of the TSK-AV score (higher than the median score) with non-recovery at 6 months [
7]. However, kinesiophobia did not contribute to the multivariate model on non-recovery [
7].
Another study in general practice reported a small and only borderline significant effect of high fear avoidance predicting less future pain (at 3 and 12 months) and less functional disability (at 3 months) [
10]. In both studies, other psychosocial variables (such as worrying and somatization) were more important predictors of poor outcome than kinesiophobia [
7,
10].
A study in physiotherapy practice in these complaints, reported that high kinesiophobia, high catastrophizing and high somatization were predictors of non-recovery [
31]. Differences in the distribution of population characteristics may affect the importance of kinesiophobia as a predictor of outcome. At baseline, our population consisted of 58% women, compared to 71% in the study of Karels et al. [
31], duration of complaints less than 6 weeks: 50% vs 24%; 6 weeks-6 months:24% vs 41%; and more than 6 months:26% vs 35%; specific diagnosis (59% vs 36%).
Besides, distribution of population characteristics, the time period can play a role as well. Bot et al. [
10] reported that the psychosocial variables predictive of outcome at 12 months, are different from those predictive of outcome at 3 months; which was in line with the findings of both Boersma and Linton [
32] and van der Windt et al. [
33] who reported that associations of several psychological variables and outcome can be different in subgroups with a longer duration of complaints Therefore, no consistent conclusions can yet be drawn about the prognostic value of kinesiophobia and fear avoidance, in the outcome in non-traumatic arm, neck and shoulder complaints and its consequences for treatment.
Furthermore, van der Windt et al. reported that possible differences on scores may be due to the location of complaints as well. In their prognostic study in primary care low back pain patients scored higher on catastrophizing, distress and somatization, compared to patients with shoulder pain. However, scores on fear avoidance did not significantly differ [
33].
In our study population there was no change over time on mean TSK-AV scores in patients who did not recover from complaints of arm, neck or shoulder. However, no definite conclusion about the stability of scores in the transition from a new episode to chronic complaint, can be based solely on this result. Although all patients had a new episode of complaints for which they had not consulted their GP in the previous 6 months, at baseline 25.7% of them already reported that they had endured their symptoms for more than 6 months.
In an intervention study on treatment of kinesiophobia in 6 patients with chronic low back pain [
34], the scores in the 4-week baseline period also seemed stable. Here, kinesiophobia scores were only reduced by an exposure in vivo intervention (not during graded activity). In this chronic low back pain population influencing kinesiophobia seems to require specific treatment. Although the results of the latter study seem to be in line with our results, it should be noted that these low back pain patients were recruited in rehabilitation, with a median pain duration of 4 years, and had to have a relatively high score (> = 40) on the TSK to be included in the study [
34].
Additionally, at baseline we found no multivariate relation of duration of complaint with the TSK-AV score. Time did not explain differences in the degree of kinesiophobia in primary care.
In our total population, at baseline, positive associations were found between kinesiophobia and a high degree of catastrophizing, a high degree of disability, and comorbidity of musculoskeletal complaints. Based on the 'fear- avoidance model' [
12], the association between catastrophizing and the TSK score was expected, as also confirmed in patients with low back pain [
13,
19,
20] and patients with chronic musculoskeletal and neuropathic diagnoses [
20].
The association of disability, also part of the fear-avoidance model, has also been confirmed in other studies [
15,
16,
18,
20], some of which report on functional disability. Another study in patients with osteoarthritis [
17], reported on an association between kinesiophobia and functional limitations. Although, disability and functional limitations are not exactly the same, they are connected. In the present study, we reported on disability measured with the DASH, which, according to its developers focuses on physical function. According to the International Classification of Functioning (ICF), the DASH mainly focuses on disability at the level of activity limitations, which is the domain of rehabilitation therapy [
35].
A noteworthy finding was that the presence of musculoskeletal comorbidity was associated with a higher score on kinesiophobia. On closer inspection of the subgroup reporting musculoskeletal comorbidity (n = 330), we found that the majority also had low back pain, followed by a smaller group reporting osteoarthritis of hip or knee, and a few (n = 23) reporting comorbidity of arm, neck or shoulder. However, we have no information on the duration of this co-occurring musculoskeletal complaint. This raises the question, whether the higher score on kinesiophobia was mainly the effect of the concurrent chronic low back pain, or a previous negative experience in general. Although heterogeneity is the reality of the general practice population, we checked whether having co-morbidity modified the association between the variables in the final model and kinesiophobia. This was not the case.
Furthermore, most psychosocial variables remained in the final multivariate model. Pearson correlation coefficients between the variables included in the final model ranged from -0.21 to 0.47, of which the highest was for distress and catastrophizing. Thus, distress, somatization and social support, do not measure the same thing, and each variable has its independent association with kinesiophobia. However, catastrophizing showed the strongest association with kinesiophobia.
Because in the present study the area of possible complaints was extensive (compared with studies on e.g. low back pain) we also included location of complaints as a variable. The results show that complaints involving the shoulder were positively related to kinesiophobia; we have no clear explanation for this finding. A possible explanation may be that the shoulder is a large and central joint (compared to elbow, wrist and hand) providing stability and mobility in many stances and movements of the whole upper extremity. However, this was not confirmed by additional analyses in which we compared mean disability scores. Further, we did not find more musculoskeletal comorbidity among patients with shoulder complaints. Besides a true association, this association may partly be explained by a larger group size and accompanying smaller confidence intervals and smaller p-values compared with, e.g., complaints located at the elbow.
The present study has some limitations that need to be addressed. First, the questions on the TSK-AV relate to 'pain', whereas our patients reported on 'complaints' (as defined in the introduction) and not exclusively on pain. However, 675 (99%) patients reported pain when active and/or in rest, and only 4 patients, without pain, reported on tingling. Therefore, our results will also hold when excluding these 4 latter patients. Besides, although the cognitive-behavioural oriented model was developed for persistence of pain, the concept of avoidance behaviour may also be applicable in patients reporting other complaints, such as tingling. Nevertheless, no definite conclusions can be drawn on this matter. In patients with chronic fatigue syndrome however, fear- avoidance has also been reported [
36]. After our inclusion period had started in September 2001, we found reports on an adjusted Tampa scale for patients with chronic fatigue syndrome where the term 'pain' had been replaced by 'symptoms' [
36]. In the present study, replacing 'pain' by another term might have been a better option.
Since its development in chronic and later acute low back pain patients, the TSK has also been introduced in other populations (e.g. chronic fatigue syndrome [
36], osteoarthritis [
17], chronic neck pain [
30], pain- free people [
37]). In patients with non-traumatic arm, neck and shoulder complaints, the TSK has been used as a possible predictor in prognostic studies [
7,
11,
31], and as outcome measure in randomised clinical trials in chronic neck pain [
30]. So far, no studies have reported on the psychometric properties of the TSK in arm, neck and shoulder complaints. Although our mean score seems comparable to those in other primary care populations, and associated variables seem in line with other studies, future studies on psychometric properties need to confirm whether the TSK is a valid measurement instrument in this particular population.
Another limitation is that we used one simple question to give an indication of 'health locus of control', instead of using a validated multi-item questionnaire; therefore, the strength of the association should be interpreted with caution. However, the negative direction of the association was as expected, i.e. a higher degree of kinesiophobia was associated with less health locus of control.
Despite also being part of the fear avoidance model, we did not measure depression in our patients. Although we did include questions on several other psychosocial variables, we considered that the questions of the 4DSQ depression scale (e.g.: "During the past week, did you feel that life was meaningless?" "Did you feel that life is not worth while?" "Did you feel that you would be better off if you were dead?") were less appropriate in our population with new non-traumatic arm, neck and shoulder complaints. In the validation study of the 4DSQ, Terluin et al. reported that the applicability of their depression scale was limited in an unselected sample of primary care patients (n = 2,127) because of the low mean scores on depression, due to the relatively low prevalence of depressive disorders; they further concluded that the distress score that was measured, gives an indication of psychosocial dysfunctioning in general, including mild depressive symptoms [
22]. However, including depression in our study would have yielded some additional information.
Authors' contributions
AF collected the data, performed the analyses with TvD and wrote the article. AF had full access to all the study data and had the final responsibility for the decision to submit for publication.
TvD entered part of the data, performed the analyses with AF, and wrote the first draft of the article.
SMAB-Z was the principal investigator of the study and contributed to the design and coordination of the study and interpretation of the results.
RMDB performed part of the analyses.
JANV is an expert in the field of musculoskeletal complaints and contributed to the interpretation of the results.
BWK is an expert in the field of epidemiology and contributed to the design and coordination of the study and the interpretation of the results.
HSM is an expert in the field of musculoskeletal complaints and epidemiology and contributed to the design of the study and the interpretation of the results.
All authors declare that they have participated in the writing and editing of the manuscript and that they have read and approved the final version.