Rheumatology 2008;47:724730 Advance Access publication 14 April 2008

doi:10.1093/rheumatology/ken044

Course and prognosis of shoulder symptoms in general practice

M. L. Reilingh1, T. Kuijpers2, A. M. Tanja-Harfterkamp1 and D. A. van der Windt1,3
Objectives. To investigate the course and prognosis of shoulder pain in the first 6 months after presentation to the general practitioner. We separately studied patients with acute, subacute and chronic shoulder pain, as duration of symptoms at presentation has been shown to be the strongest predictor of outcome. Methods. A prospective cohort study with 6 months follow-up was carried out in The Netherlands, including 587 patients with a new episode of shoulder pain. Patients were categorized as having acute (symptoms <6 weeks), subacute (612 weeks) or chronic (>3 months) shoulder pain. The course of shoulder pain, functional disability and quality of life was analysed over 6 months. Patient and disease characteristics, including physical and psychosocial factors, were investigated as possible predictors of outcome using multivariable regression analyses. Results. Acute shoulder symptoms showed the most favourable course over 6 months follow-up, with larger pain reduction and improvement of functional disability. Patients with chronic shoulder symptoms showed the poorest results. The multivariable regression analysis showed that predictors of a better outcome at 6 months for acute shoulder pain were lower baseline disability scores and higher baseline pain intensity (explained variance 46%). Predictors of a better outcome for chronic shoulder pain were lower scores on pain catastrophizing and higher baseline pain intensity (explained variance 21%). Conclusions. The results indicate that, besides a different course of symptoms in patients presenting with acute or chronic shoulder pain, predictors of outcome may also differ with psychosocial factors being more important in chronic shoulder pain.
KEY
WORDS:

Shoulder pain, Disability, Psychosocial factors, General practice, Course, Prognosis, Prospective cohort study.

Introduction
Shoulder symptoms are a frequent problem in general practice, affecting between 7% and 34% of adults at any one time [1]. Not everyone consults the general practitioner (GP) for these symptoms. The annual consulting incidence in Dutch general practice for shoulder symptoms is estimated between 12 and 25/1000 persons-years [25]. Little is known about the pathophysiology and aetiology of shoulder disorders, although associations with obesity, age, female gender, physical work load and psychosocial factors have been proposed [68]. Various studies have contradicted the belief that shoulder pain is a benign and self-limiting problem. Only about 50% of all new episodes of shoulder pain presented in primary care show complete recovery within 6 months, after 1 yr this proportion increases to only 60% [912]. In the period 198795, the state of Washington (USA) each year accepted over 6000 work disability claims related to shoulder problems [13]. Information about the clinical course of shoulder symptoms after presentation in general practice is still limited. The medical literature on shoulder disorders is predominantly based on hospital surveys, although only a small proportion of shoulder patients in general practice (8%) are referred for a specialist opinion [10]. Yet, knowledge on the course may facilitate treatment decisions and may help to inform patients about their prognosis. Several prospective cohort studies have investigated neck, shoulder or upper limb problems in primary care populations [912, 1417]. The following factors were repeatedly identified as potential predictors of outcome: long duration of symptoms at baseline [10, 12, 1517], more intense pain [15, 16], history of symptoms [9, 12, 1517], musculoskeletal pain elsewhere [15, 17] and psychosocial factors [12, 16, 17]. The aim of our study was to investigate the course of shoulder symptoms during the first 6 months after visiting
1 2

the GP, in terms of pain, functional disability and quality of life. We made a distinction between patients with acute shoulder pain (duration of symptoms <6 weeks before consulting the GP), subacute shoulder pain (duration of symptoms between 6 and 12 weeks) and patients with chronic shoulder pain (duration of symptoms at least 3 months), as foregoing research showed that symptom duration is an important predictor of outcome, with acute shoulder pain having a better prognosis than chronic pain. The questions we addressed in our study were: (1) What is the course of shoulder symptoms during the first 6 months after visiting the GP, in terms of pain, functional disability and quality of life? (2) Is the prognosis different in patients with acute, subacute and chronic shoulder pain at consultation? (3) What are predictors of a better outcome after 6 months in patients presenting with either acute or chronic shoulder pain?

Methods Study population

This study is based on the results of a cohort study that was performed in 103 general practices in three geographical regions in The Netherlands (Amsterdam, Groningen and Maastricht) [18]. Patients were selected if they were !18 yrs, and had not consulted their GP or received any form of treatment for the afflicted shoulder in the preceding 3 months. Shoulder pain was characterized as pain in the deltoid and upper arm region, provoked or increased by movement in the shoulder joint. GPs were instructed to select consecutive patients. Sufficient knowledge of the Dutch language was required to complete written questionnaires. Exclusion criteria were acute trauma or systemic, physical or psychological conditions (i.e. fractures or luxation in the shoulder region; rheumatic disease; neoplasm; neurological or vascular disorders; dementia).

Department of General Practice, VU University Medical Center, Amsterdam, Dutch Institute for Healthcare Improvement CBO, Utrecht, The Netherlands and 3 Primary Care Musculoskeletal Research Centre, Keele University, Keele, UK. Submitted 21 August 2007; revised version accepted 18 January 2008. Correspondence to: D. A. van der Windt, EMGO Institute, VU University Medical Centre, Van der Boechorststraat 7, 1081 BT Amsterdam, The Netherlands. E-mail: dawm.vanderwindt@vumc.nl

Design
The GP informed the patients about the study, after which written consent was obtained according to the Declaration of Helsinki.
724

The Author 2008. Published by Oxford University Press on behalf of the British Society for Rheumatology. All rights reserved. For Permissions, please email: journals.permissions@oxfordjournals.org

Course and prognosis of shoulder symptoms The GP started treatment according to national Dutch guidelines for shoulder pain, which advise a stepwise approach (advice, medication, steroid injection or physiotherapy) in which the next treatment option is only considered for patients returning with persistent pain and disability [19]. Information was recorded about the treatment provided to the patients at the first consultation. Within 10 days after the consultation a baseline assessment was performed, consisting of a patient history, standardized physical examination and a questionnaire. A second questionnaire (first follow-up) was sent after 6 weeks, the third questionnaire (second follow-up) after 3 months and the fourth questionnaire (third follow-up) after 6 months. The study was approved by the Medical Ethics Committee of the VU University Medical Centre in Amsterdam.

725

Outcome measures and potential prognostic factors

In each questionnaire, the intensity of shoulder pain, functional disability and quality of life was measured. Pain was recorded by the patient on a numeric rating scale (010 points, 0 no pain; 10 very severe pain). Functional disability was measured using the 16-item Shoulder Disability Questionnaire (SDQ; 0100) [20]. Quality of life was measured by the EuroQol (EQ-5D; 01) [21, 22], covering five domains: mobility, self-care, usual activities, pain/discomfort and anxiety/depression. During the baseline assessment information was collected on a variety of potential prognostic factors. Sociodemographic variables included age, gender, educational level and work status. Characteristics of the shoulder pain problem included intensity of pain, an acute or gradual onset, previous episodes of shoulder pain and whether or not the dominant side was affected. The questionnaire also included questions on co-existing musculoskeletal pain at the neck, back and upper extremities. Causes of shoulder pain as perceived by the patient were categorized as unexpected movement, overuse due to unusual activities, overuse due to usual activities, accident or sports injury and were measured with questions answered by yes or no. Physical activity was measured with a single question (less/ equally/more active than others). We measured physical workload with a self-constructed scale of five questions (yes/no) concerning pushing and pulling, lifting weights, working with hands above shoulder level and the use of vibrating tools on at least 2 days a week (05). Repetitive movements, on at least 2 days a week, were measured with a single question answered with yes or no. The psychosocial work environment was assessed using the 27-item Job Content Questionnaire which measures all dimensions of the DemandControlSupport model [23]. Psychological factors were measured with widely used standardized questionnaires. Coping was assessed with the 43-item Pain Coping and Cognition List (PCCL) [24], consisting of the subdomains catastrophizing (16), coping with pain (16), internal (16) and external locus of control (16). Anxiety (024), depression (012), somatization (032) and distress (032), were measured with the 50-item Four-Dimensional Symptoms Questionnaire (4DSQ) [25, 26]. The questionnaire finally included a general one-item question regarding the presence (yes/no) of any psychological problems (e.g. worries, depressive symptoms, anxiety).

Predictors were selected based on evidence from the literature on their potential predictive value in patients with musculoskeletal pain (age, history of pain, dominant side affected, musculoskeletal pain elsewhere, work load factors and pain catastrophizing) [9, 10, 12, 1417, 27]. Another reason for selecting a potential predictor could be differences between patients with acute and chronic pain at baseline, indicating potential predictive importance of the factor. Univariable linear regression analyses were performed to examine the relationship between each of the potential predictors and change in pain intensity after 6 months. Predictors measured on a continuous (e.g. age) or interval scale (e.g. baseline pain intensity) were entered as continuous variables, which provides an estimate of the mean reduction in pain for each point increase on the scale. If a continuous variable showed a non-linear association with outcome it was dichotomized or divided into tertiles (low, medium, high scores). This was the case for external locus of control and somatization. Subsequently, predictors that were associated with the outcome (P < 0.30) were included simultaneously in a multiple linear regression model. Using a manual backward selection procedure, we sequentially excluded predictors with the lowest predictive value from the model until all predictors were significantly associated with outcome (Wald statistic P < 0.10) and further elimination resulted in a large deterioration of the explained variance of the model. The percentage of explained variance (R2) was calculated to give an indication of the predictive power of the final models.

Results Study population and follow-up

At baseline, 587 patients were interviewed and physically examined. At 6 weeks 487 (83%), at 3 months 517 (88%) and at 6 months 538 (92%) patients returned the postal questionnaire. The drop-outs at 6 weeks and 6 months were significantly younger than the responders (mean difference 4 yrs and 6 yrs, respectively). Additionally, drop-outs at 6 months more often showed an acute onset (49 vs 36%), and less repetitive movements in their work (26 vs 36%) in comparison with the responders. Table 1 lists the baseline characteristics of the participants separately for those with acute, subacute or chronic pain at baseline. The results show that patients with chronic shoulder pain reported higher pain intensity, more disability, had a higher somatization score, and more often reported a gradual onset of their shoulder pain compared with patients with subacute or acute shoulder pain. Patients with acute shoulder pain more often had a paid job, and slightly higher scores on external locus of control compared with those with subacute or chronic shoulder pain. These six factors were considered as potential predictors of outcome in further analyses (in addition to factors suggested by the literature).

Management of shoulder pain

At baseline, most patients (n 451, 77%) received a wait-and-see policy, paracetamol or an NSAID. Furthermore, 68 patients (12%) received an injection with a corticosteroid, 58 patients (10%) were referred for physiotherapy and 11 (2%) received other therapies. Table 2 lists the management of shoulder pain of the participants separately for those with acute, subacute or chronic pain at baseline. The proportion of patients receiving wait-and-see policy, paracetamol or NSAIDs was higher in patients with acute shoulder pain compared with those with chronic shoulder pain, while the proportion of physiotherapy referrals was higher in patients with chronic shoulder pain.

Analysis
Descriptive statistics were used to summarize characteristics of the study population. We present this information separately for patients with acute, subacute and chronic shoulder pain. The course of symptoms in terms of pain, disability and quality of life was also described separately for these three subgroups by plotting the mean scores at each point in time. Next, we analysed potential predictors of outcome separately for patients with acute and chronic shoulder pain. Absolute change in pain intensity between baseline and 6 months follow-up was used as outcome measure.

726

M. Reilingh et al.

TABLE 1. Baseline characteristics of patients with shoulder pain, separately for those with acute, subacute or chronic pain at baseline Acute (n 205) <6 weeks Demographic and work variables Age (yrs), mean (S.D.) Gender, n (% female) Educational level, n (%) Low Middle High Paid work Psychosocial work environment (n 350), mean (S.D.)a Decision authority Skill discretion Decision latitude Psychological demands Supervisor support Coworker support Disease characteristics Gradual onset, n (%) Previous episode, n (%) Pain intensity, mean (S.D.) Functional disability (SDQ 0100), mean (S.D.) Dominant side affected, n (%) Psychological factors PCCL, mean (S.D.)a Catastrophizing (16) Coping with pain (16) Internal locus of control (16) External locus of control (16) 4DSQ, mean (S.D.), mediana Distress (032) Depression (012) Anxiety (024) Somatization (032) Physical load Physical burden (05), mean (S.D.), median Repeated movement with arms or wrists, n (%) Perceived cause according to patient, n (%) Unexpected movement Overuse unusual activities Overuse usual activities Accident Sports injury Physical activity, n (%) >30 min/day walking or biking Regular long walk or cycling tour Pain elsewhere, n (%) Neck or upper extremities Low back
a

Subacute (n 139) 612 weeks 51.9 (13.9) 62 (55.4) 37 61 40 77 (26.6) (43.9) (28.8) (55.4) (1.77) (2.68) (3.99) (2.24) (2.22) (1.37)

Chronic (n 242) >3 months 52.9 (13.3) 127 (52.5) 98 93 47 134 (40.5) (38.4) (19.4) (55.4) (1.84) (2.80) (4.17) (2.57) (2.83) (2.16)

49.5 (14.7) 105 (51.2) 74 80 48 139 (36.1) (39.0) (23.4) (67.8) (1.74) (2.92) (4.07) (2.88) (2.44) (2.02)

9.4 15.4 24.8 13.2 12.2 12.4

9.2 15.4 24.6 12.9 11.1 12.3

9.4 15.1 24.6 12.3 11.3 12.2

78 (38.0) 119 (58.0) 4.4 (2.5) 56.8 (28.3) 120 (58.5) 2.2 3.0 3.4 3.4 2.1 0.2 0.4 2.9 (0.8) (1.0) (1.0) (0.9) (4.5), (1.4), (1.5), (3.6), 0.0 0.0 0.0 2.0

96 (69.1) 63 (45.3) 4.7 (2.1) 59.1 (23.0) 89 (64.0) 2.2 2.9 3.4 3.1 2.5 0.2 0.4 2.8 (0.9) (0.9) (0.9) (0.9) (5.1), (1.3), (1.3), (3.9), 0.0 0.0 0.0 2.0

188 (77.7) 161 (66.5) 5.2 (2.2) 62.9 (22.7) 152 (62.8) 2.3 3.0 3.3 3.1 2.4 0.3 0.3 4.0 (0.9) (1.0) (0.9) (0.9) (4.1), (1.1), (1.0), (4.5), 0.0 0.0 0.0 3.0

1.35 (1.50), 1.00 138 (67.3) 12 32 43 10 6 (5.9) (15.6) (21) (4.9) (2.9)

1.19 (1.52), 0.00 75 (54.0) 9 24 36 4 6 (6.5) (17.3) (25.9) (2.9) (4.3)

1.18 (1.40), 1.00 170 (70.2) 12 39 59 19 17 (5.0) (16.1) (24.4) (7.9) (7.0)

70 (34.1) 87 (42.4) 86 (45.3) 44 (21.5)

53 (38.1) 72 (51.8) 56 (45.5) 31 (22.3)

93 (38.4) 116 (47.9) 131 (58.5) 64 (26.4)

The psychosocial work environment was only assessed in patients with paid work.

TABLE 2. Management of patients with shoulder pain at first consultation, separately for those with acute, subacute or chronic pain Acute (n 205) <6 weeks Management of shoulder pain, n (%) Wait-and-see policy, 174 (84.9) paracetamol or NSAIDs Injection with corticosteroid 22 (10.7) Physiotherapy 6 (2.9) Subacute (n 139) 612 weeks 107 (77.0) 12 (8.6) 18 (12.9) Chronic (n 242) >3 months 169 (69.8) 34 (14.0) 34 (14.0)

quality of life during follow-up. Mean scores were consistently lower among patients with chronic shoulder pain. For each outcome measure the differences between the three subgroups were statistically significant at each moment of follow-up.

Predictors of change in pain intensity

The results of the univariable regression analyses demonstrating the association of each putative predictor with change in pain intensity after 6 months are presented in Table 3, separately for patients with acute and chronic shoulder pain. Baseline shoulder disability, catastrophizing, somatization, paid work, baseline pain intensity, previous episode, gradual onset, repeated movements with arms or wrists and co-existing pain in the neck or upper extremities, were associated with change in pain intensity during follow-up in patients with acute shoulder pain at baseline. In those with chronic shoulder pain, baseline shoulder disability, age, pain catastrophizing, external locus of control, gradual onset and baseline pain intensity with chronic shoulder pain showed an univariable association (P 0.30) with outcome. These variables were selected for the multivariable regression analyses. Table 4 presents the variables included in the prediction models for persistent symptoms at 6 months after backward stepwise selection. Predictors of a better outcome at 6 months for acute shoulder pain were lower baseline disability scores and higher

Course. Figure 1 shows that patients with acute pain at baseline had the most favourable outcome over 6 months followup, with pain decreasing from a mean score of 4.3 (S.D. 2.6) at baseline to 1.3 (S.D. 2.2) at 6 months (mean pain reduction since baseline 70%). Patients with subacute pain at baseline (mean pain reduction 54%) or chronic pain at baseline (mean pain reduction 44%) showed a slower rate of recovery and maintained higher levels of pain at 6 months. The course of functional disability presented in Fig. 2 shows very similar patterns; patients with acute pain at baseline reported the largest decrease in shoulder disability after 6 months (mean reduction in functional disability 69%). The results of the EQ-5D (Fig. 3) show that there was little change in

Course and prognosis of shoulder symptoms

10 9 8 Pain intensity (010) 7 EQ-5D (01) 6 5 4 3 2 1 0 0 5 10 15 20 25 30 Follow-up time (weeks) Acute (06 weeks) Chronic (> 12 weeks) Subacute (612 weeks) 0 0 5 10 15 20 25 30 Follow-up time (weeks) Acute (0-6 weeks) Chronic (> 12 weeks) Subacute (612 weeks) 0.2 0.6 0.8 1

727

0.4

FIG. 1. Course of shoulder pain intensity over 6 months after GP consultation.

FIG. 3. Course of quality of life (EQ-5D) over 6 months after GP consultation for shoulder pain.

100 90 80 Shoulder disability (0100) 70 60 50 40 30 20 10 0 0 5 10 15 20 25 30 Follow-up time (weeks) Acute (06 weeks) Chronic (> 12 weeks) Subacute (612 weeks)

months, increasing to only 32% after 12 months [24]. Similarly, in a recent study on non-traumatic arm, neck and shoulder complaints [17], 25% of participants reported complete recovery after 6 months. These studies demonstrate a poor outcome of neck and shoulder symptoms, but included very heterogeneous populations. In a previous analysis of our cohort [18], we developed a prediction rule for shoulder pain with perceived recovery as outcome measure, and used the total cohort that was heterogeneous with respect to baseline symptom duration. For the current analysis, we decided to study the influence of potential prognostic factors in more detail, by looking at their effect on absolute changes in shoulder pain intensity, and by investigating prognostic factors in more homogeneous subgroups of patients. Given the strong evidence for the prognostic value of symptom duration we decided to stratify our analyses for acute, subacute and chronic pain at presentation, creating more homogeneous subgroups with different characteristics, and demonstrating a different prognosis.

FIG. 2. Course of shoulder pain-related disability (SDQ) over 6 months after GP consultation.

Prognostic factors
Previous prognostic cohort studies have showed that several disease characteristics (symptom duration, levels of disability and more intense pain) are indicators of a poor outcome of neck or shoulder complaints [9, 10, 12, 1417]. Our analyses confirmed the association between baseline levels of pain and disability, and changes in symptoms during follow-up. In our study, higher pain intensity at baseline was associated with larger reductions of pain at follow-up in both patients with acute or chronic shoulder pain. This finding may be explained by the fact that more pain at baseline leaves more room for improvement during follow-up [12]. It has previously been suggested that psychosocial factors such as worrying [12], somatization [17, 28], catastrophizing [28], distress [28, 29] and fear-avoidance beliefs [28, 30, 31] are likely to predict a poor outcome of painful musculoskeletal conditions. The association between psychosocial factors and musculoskeletal pain has mainly been established in patients with chronic pain syndromes [31, 32]. In our population, somatization was associated with poorer outcome in the univariable analyses, but the association was no longer significant in the multivariable model. More catastrophizing, however, was significantly related to smaller reductions of pain at follow-up in patients with chronic shoulder pain. We previously reported an association between catastrophizing thoughts and perceived recovery of shoulder pain at 3 months follow-up [33], but our analyses now show that in patients with chronic shoulder pain, catastrophizing is the

baseline pain intensity. The model shows, for example, that with each point increase in baseline pain intensity (scale 010) the mean change in pain at 6 months increased by 0.84 points (95% CI 0.71, 0.98). The explained variance was 46% at 6 months. Predictors of a better outcome at 6 months for chronic shoulder pain were lower scores on pain catastrophizing and higher baseline pain intensity scores. The model explained 21% of the variance in pain reduction at 6 months.

Discussion
The results of our prospective cohort study showed a mean pain reduction of 70% among patients presenting with acute shoulder pain, compared with 54% in those with subacute symptoms and 44% in patients with chronic pain at baseline. The course of functional disability presented very similar patterns. There was little change in quality of life during follow-up. Prognostic factors partly differed between patients with acute/chronic shoulder pain.

Course of shoulder symptoms

In a previous study on neck and shoulder complaints in general practice, 24% of patients reported complete recovery after 3

728

M. Reilingh et al.

TABLE 3. Predictors of change in pain intensity after 6 months: univariable associations Acute shoulder pain (n 190) Mean change (95% CI)a Age (per year older) Baseline disability score (SDQ, per point increase) Catastrophizing (PCCL, per point increase) External locus of control (PCCL, compared < 3) 34 >4 Somatization (4-DSQ, ! 5 vs < 5) Paid work (yes vs no) Gradual onset (vs sudden) Previous episode (yes vs no) Baseline pain intensity (per point increase) Dominant side affected (yes vs no) Repeated movement with arms or wrists (yes vs no) Pain in neck or upper extremities (yes vs no)
a

Significance

Chronic shoulder pain (n 224) Mean change (95% CI)a 0.02 (0.01, 0.05) 0.02 (0.00, 0.04) 0.26 (0.72, 0.20) 0.79 0.21 0.16 0.31 0.59 0.07 0.59 0.01 0.11 0.19 (1.60, 0.02) (0.92, 1.35) (1.01, 0.68) (1.07, 0.45) (1.51, 0.34) (0.75, 0.88) (0.43, 0.74) (0.78, 0.79) (0.72, 0.94) (0.95, 0.58)

Significance

0.01 (0.02, 0.04) 0.02 (0.00, 0.03) 1.00 (0.44, 1.57) 0.35 0.22 0.87 0.95 0.56 0.61 0.78 0.28 0.65 1.05 (0.63, 1.32) (0.89, 1.33) (0.23, 1.98) (1.85, 0.06) (1.43, 0.3) (0.25, 1.47) (0.65, 0.90) (0.58, 1.14) (1.53, 0.24) (0.21, 1.89)

0.50 0.01 0.01 0.49 0.70 0.12 0.04 0.21 0.16 0.00 0.52 0.15 0.02

0.30 0.04 0.26 0.06 0.71 0.71 0.42 0.21 0.87 0.00 0.99 0.79 0.63

Positive values indicate larger reductions in pain, negative values indicate an increase in pain (per point increase on the predictor).

TABLE 4. Predictors of change in pain intensity for patients with acute shoulder pain (R2 0.46) or chronic shoulder pain (R2 0.21) after 6 months follow-up: results of multiple linear regression analyses Acute shoulder pain (n 190) Mean change (95% CI)a Baseline disability score (per point increase) Catastrophizing (per point increase) Baseline pain intensity (per point increase)
a

Significance

Chronic shoulder pain (n 224) Mean change (95% CI)a 0.62 (1.03, 0.20) 0.65 (0.48, 0.81)

Significance

0.01 (0.03, 0.00) 0.84 (0.71, 0.98)

0.03 0.001 0.001 0.001

Positive values indicate larger reductions in pain, negative values indicate an increase in pain (per point increase on the predictor).

strongest predictor of change in pain intensity over 6 months follow-up, next to baseline levels of pain. Catastrophizing [34] is considered to be an ineffective coping strategy in which pain is perceived as overly destructive, and patients have a pessimistic view of their prognosis. The fact that catastrophizing was especially important in patients with chronic shoulder pain may indicate that catastrophizing is a consequence of pain, rather than a predictor of the development of chronic pain. Surprisingly, Table 3 also shows that in patients with acute pain, catastrophizing was associated with larger improvements of pain, even though the association was weak and not statistically significant in the multivariable model. Possibly, this is a chance finding. Alternatively, based on the assumption that the catastrophizing scale measures if patients are more strongly oriented towards pain stimuli, one may hypothesize that patients with high scores on catastrophizing avoid pain-provoking activities, which may actually be an effective strategy in the acute phase of a shoulder problem. Inadequate beliefs and attributions of pain, however, may become stronger when pain persists, or when recurrences of pain occur. If the negative influence of catastrophizing can be confirmed in other shoulder pain populations, future research might be aimed at the development and evaluation of interventions aimed at reducing such negative processes, from which patients with chronic shoulder pain in primary care may benefit.

Strengths and weaknesses

Our sample of shoulder pain patients is relatively large, allowing stratification of patients according to duration of their symptoms. Drop out among patients was low (8% at 6 months). Although drop-outs were younger, more often showed an acute onset and less repetitive movements in their work, the absolute differences in age, acute onset and repetitive movements were not large and it is unlikely that these differences have strongly influenced the reported associations between potential predictors and outcome. The prognostic model for patients presenting with acute shoulder pain explained 46% of the variance in pain scores at follow-up, but for chronic pain patients this was only 21%. This means that much of the variance in this patient group remains unexplained. Even though we included a wide variety of potential

predictors, we may have missed relevant predictors, such as social support or aspects of the psychosocial work environment [17, 35, 36]. We may also have been unable to measure important predictors with sufficient accuracy. Exposure to physical load, for example, was measured using a few simple questions, but has been shown to be associated with neck or shoulder pain in other cohorts [36, 37]. Further research is needed to identify relevant and preferably modifiable predictors of outcome, especially in patients with chronic shoulder symptoms. Studies carried out in occupational settings may be more suitable to address the importance of work-related factors in the prognosis of shoulder pain. We decided not to include treatment in our models, as we assumed that confounding by indication could influence our findings. Patients with more severe symptoms and thus, probably a poorer outcome, are more likely to receive more extensive treatment. Our results seem to confirm this assumption; the proportion of patients receiving wait-and-see policy or pain medication was higher in patients with acute pain compared with chronic pain, while the proportion of physiotherapy referrals was higher among those with chronic shoulder pain. However, the large majority of patients in each subgroup was treated by wait-and-see or medication in the first month, indicating that our GPs largely adhered to the Dutch practice guidelines for shoulder complaints, which recommend such a policy during the first 24 weeks after presentation [19]. This means that our subgroups were relatively homogeneous regarding treatment at baseline. Treatment variables were, indeed, not strongly associated with outcome nor strongly influenced the association of other predictors (data not shown).

Clinical usefulness
In The Netherlands, nearly every Dutch resident is registered with a GP. Patients first visit their GP before visiting a specialist in rheumatology or orthopaedics. This is comparable with, for instance, the British healthcare system. The participating practices, being situated in both rural and urban areas in various provinces, form a representative sample of Dutch GPs. Although we encouraged GPs to select every eligible patient, we do not know what proportion of patients was invited or whether this was

Course and prognosis of shoulder symptoms a selective sample. When asking GPs about their success in recruiting patients, they indicated that common reasons for not including patients concerned the exclusion criteria and lack of time or motivation to ask all patients during consultation hours. Patient characteristics at baseline and overall recovery rates (30% after 6 weeks and 54% after 6 months) were similar to those found in other studies carried out in primary care populations [911], which may strengthen the generalizability of our findings to other primary care patients with shoulder problems. The GPs in our study adhered to the stepwise approach recommended by the national guidelines, and offered injections or physiotherapy to only a few patients. Our findings indicate that GPs can take duration of symptoms at presentation into account, and do not necessarily have to apply the same stepwise approach to each patient with shoulder pain. A more targeted approach (with for example, advice regarding pain coping strategies, or more early referral for exercise) might lead to better patient outcomes in chronic shoulder pain patients, but this needs to be confirmed by randomized research. As yet, the results from our study may help GPs to provide patients with more accurate information on the expected course of their symptoms. By separately analysing acute, subacute and chronic shoulder pain, we were able to present prognostic information for more homogeneous subgroups of patients compared with previous cohorts.

729

Conclusion
The results of this study indicate that patients presenting with acute, subacute or chronic shoulder pain show different baseline characteristics, a different course of symptoms, and possibly, also different predictors of outcome after 6 months. Improvement of symptoms in patients presenting with acute shoulder pain was mainly predicted by the severity of pain and disability, whereas psychological factors may influence outcome in those with chronic shoulder pain. The importance of psychosocial factors in patients with chronic shoulder pain needs to be confirmed in further research, as this may provide input for the development of future interventions.

Rheumatology key messages

 Previous studies demonstrate poor outcome of shoulder symptoms, but have included heterogeneous populations.  Our study shows that both course and prognosis differs between patients presenting with either acute or chronic shoulder pain.

Acknowledgements
We would like to acknowledge the contribution of all patients and general practitioners who participated in this project. Funding: This study was supported by a grant (No. 904-65-09) from The Netherlands Organisation for Scientific Research (NWO), The Hague, The Netherlands. Disclosure statement: The authors have declared no conflicts of interest.

References
1 Luime JJ, Koes BW, Hendriksen IJ et al. Prevalence and incidence of shoulder pain in the general population; a systematic review. Scand J Rheumatol 2004;33:7381. 2 Bot SDM, van der Waal JM, Terwee CB et al. J. Incidence and prevalence of complaints at the neck and upper extremity in general practice. Ann Rheum Dis 2005;64:11823. 3 Okkes IM, Oskam SK, Lamberts H. Van klacht naar diagnose: episodegegevens uit de huisartspraktijk. Bussum: Coutinho, 1998 [in Dutch]. 4 Van der Linden MW, Westert GP, De Bakker DH, Schellevis, FG. Complaints and disorders in the population and general practice [In Dutch: Klachten en aandoeningen in de bevolking en in de huisartsenpraktijk]. Utrecht/Bilthoven: NIVEL/ RIVM, 2005.

5 Van der Windt DAWM, Koes BW, De Jong BA, Bouter LM. Shoulder disorders in general practice: incidence, patients characteristics, and management. Ann Rheum Dis 1995;54:95964. 6 Walker-Bone KE, Palmer KT, Reading I, Cooper C. Soft-tissue rheumatic disorders of the neck and upper limb: prevalence and risk factors. Semin Arthritis Rheum 2003;33:185203. 7 Van der Windt DAWM, Thomas E, Pope DP et al. Occupational risk factors for shoulder pain: a systematic review. Occup Environ Med 2000;57:43342. 8 Walker-Bone K, Cooper C. Hard work never hurt anyone: or did it? A review of occupational associations with soft tissue musculoskeletal disorders of the neck and upper limb. Ann Rheum Dis 2005;64:13916. 9 Croft P, Pope D, Silman A. The clinical course of shoulder pain: prospective cohort study in primary care. Primary Care Rheumatology Society Shoulder Study Group. Br Med J 1996;313:6012. 10 Van der Windt DAWM, Koes BW, Boeke AJ, Deville W, De Jong BA, Bouter LM. Shoulder disorders in general practice: prognostic indicators of outcome. Br J Gen Pract 1996;46:51923. 11 Winters JC, Sobel JS, Groenier KH, Arendzen JH, Meyboom-de Jong B. The longterm course of shoulder complaints: a prospective study in general practice. Rheumatology 1999;38:1603. 12 Bot SDM, van der Waal JM, Terwee CB et al. Predictors of outcome in neck and shoulder symptoms: a cohort study in general practice. Spine 2005;30:45970. 13 Silverstein B, Welp E, Nelson N, Kalat J. Claims incidence of work-related disorders of the upper extremities: Washington State, 1987 through 1995. Am J Public Health 1998;88:182733. 14 Luime JJ, Koes BW, Miedem HS, Verhaar JAN, Burdorf A. High incidence and recurrence of shoulder and neck pain in nursing home employees was demonstrated during a 2-year follow-up. J Clin Epidemiol 2005;58:40713. 15 Hoving JL, de Vet HC, Twisk JWR et al. Prognostic factors for neck pain in general practice. Pain 2004;110:63945. 16 Kjellman G, Skargen E, Oberg B. Prognostic factors for perceived pain and function at one-year follow-up in primary care patients with neck pain. Disabil Rehabil 2002;24:36470. 17 Feleus A, Bierma-Zeinstra MA, Miedema HS et al. Prognostic indicators for nonrecovery of non-traumatic complaints at arm, neck and shoulder in general practice6 months follow up. Rheumatology 2007;46:16976. 18 Kuijpers T, van der Windt DAWM, Boeke AJ et al. Clinical prediction rules for the prognosis of shoulder pain in general practice. Pain 2006; 120:27685. 19 Winters JC, Jongh AC, van der Windt DA et al. Practice guideline Shoulder complaints (NHG-Standaard Schouderklachten). Huisarts Wet 1999;42:22231 [in Dutch]. 20 Van der Heijden GJMG, Leffers P, Bouter LM. Shoulder disability questionnaire design and responsiveness of a functional status measure. J Clin Epidemiol 2000;53:2938. 21 EuroQol Group. EuroQol - a new facility for the measurement of health-related quality of life. The EuroQol Group. Health Policy 1990;16:199208. 22 Dolan P. Modeling valuations for EuroQol health states. Med Care 1997; 35:1095108. 23 Karasek RA, Pieper CF, Schwart JE. Job content questionnaire and users guide: revision 1. Los Angeles: USCLA, 1985. 24 Berg SGM, Vlaeyen JWS, Ter Kuil MM, Spinhoven P, Van Breukelen G, KoleSnijders AMJ. Instruments for measuring chronic pain, part 2. Pain coping and cognition list [In Dutch: Meetinstrumenten chronische pijn, deel 2. Pijn coping en cognitie lijst]. Maastricht: Pijn Kennis Centrum, 2001. 25 Terluin B, van Marwijk HW, Ader HJ et al. The Four-Dimensional Symptom Questionnaire (4DSQ): a validation study of a multidimensional self-report questionnaire to assess distress, depression, anxiety and somatization. BMC Psychiatry 2006;6:34. 26 Terluin B, Van Rhenen W, Schaufeli W, De Haan M. The FourDimensional Symptom Questionnaire (4DSQ): measuring distress and other mental health problems in a working population. Work Stress 2004; 18:187207. 27 Kuijpers T, van der Windt DAWM, van der Heijden GJMG, Bouter LM. Systematic review of prognostic cohort studies on shoulder disorders. Pain 2004;109:42031. 28 Pincus T, Burton AK, Vogel S, Field AP. A systemic review of psychological factors as predictors of chronicity/disability in prospective cohorts of low back pain. Spine 2002;27:E10920. 29 Macfarlane GJ, Hunt IM, Silman AJ. Predictors of chronic shoulder pain: a population based prospective study. J Rheumatol 1998;25:16125. 30 Vlaeyen JWS, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. Pain 2000;85:31732. 31 Crombez G, Vlaeyen JWS, Heuts PHTG, Lysens S. Fear of pain is more disabling than pain itself. Evidence on the role of pain-related fear in chronic back pain disability. Pain 1999;80:32940. 32 Turk DC, Okifuji A. Psychological factors in chronic pain: evolution and revolution. J Consult Clin Psychol 2002;70:67890. 33 Van der Windt DAWM, Kuijpers T, Jellema P, Van der Heijden GJ, Bouter LM. Do psychological factors predict outcome in both low back pain and shoulder pain? Ann Rheum Dis 2007;66:3139. 34 Sullivan MJL, Bishop S, Pivik J. The pain catastrophising scale: development and validation. Psychol Assess 1995;7:52432.

730

M. Reilingh et al.
37 Harkness EF, Macfarlane GJ, Nahit ES, Silman AJ, McBeth J. Mechanical and psychosocial factors predict new onset shoulder pain: a prospective cohort study of newly employed workers. J Occup Environ Med 2003;60:8507.

35 Van den Heuvel SG, Van der Beek AJ, Blatter BM, Hoogendoorn WE, Bongers PM. Psychosocial work characteristics in relation to neck and upper limb symptoms. Pain 2004;114:4753. 36 Walker-Bone K, Palmer KT, Reading I, Cooper C. Soft tissue rheumatic disorders of the neck and upper limb: prevalence and risk factors. Arthritis Rheum 2003;33:185203.

Copyright of Rheumatology is the property of Oxford University Press / UK and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use.