Methods

Literature relevant to this review was identified using the following databases: PubMed (including Medline), EMBASE, PsycINFO, Cochrane Database of Systematic Reviews (CDSR), Cochrane Central Register of Controlled Trials (CCTR), Database of Abstracts of Reviews of Effectiveness (DARE) and Turning Research into Practice (TRIP). Retrieval was restricted by publication year (2006 to search date 28 July 2010) and English language, but not to any particular publication type. The basic search strategy used for PubMed (and subsequently translated into the specific search syntax required for the other databases) was: Shoulder∗ AND (occupation∗ OR workplace∗ OR industr∗ OR job∗ OR “work-related” OR employ∗) AND published in the last 5 years AND English [lang].

An additional PubMed search combined the term ‘shoulder∗’ with all citations appearing in a constructed search set of 47 journal titles oriented towards occupational diseases, such as the International Journal of Occupational and Environmental Health and Occupational & Environmental Medicine , was undertaken.

The focus of the article is on systematic reviews and reviews with extensive narrative. Where individual studies added new or specific information, these were included.

Classification

Classification systems are rules that define minimum criteria that are required to be met in order to establish the existence of a disorder. They help clinicians and researchers communicate about prevalence and incidence, the impact of disorders and prognosis. They may provide insight into some relevant aspect of current experience (pathology, pain and disability) or future course (response to treatment) . Ideally, such systems should be valid, reliable, easy to use and generalisable. Symptoms in the shoulder may arise from discrete pathological conditions, such as rotator cuff tendonitis. Alternatively, presentations may be non-specific or mixed, reflecting pain associated with other factors, which limit the ability to make a clear pathoanatomical diagnosis. There is a lack of consensus as to what constitutes a precise and rigorous case definition of a number of shoulder complaints, and the clinical assessment process is complicated by the indirect links between aetiology, pathology, diagnostic label and the subsequent impairment and disability. The area is also further complicated by a medicolegal environment, which can be acrimonious and occasionally adversarial, and, sometimes, a workers’ compensation system where access to financial support is determined in part by a legal (as well as a medical) process.

There is considerable uncertainty as to the classification of upper-limb disorders in general, and shoulder problems, more specifically. This classification is further complicated when issues of work-relatedness are considered. This lack of operational definition for these conditions is a frequently cited problem for epidemiological research and indeed for clinicians managing these issues. A number of frameworks have been proposed to address this issue. Van Eerd et al. found 27 different classification systems for work-related upper-extremity conditions . A more recent review of the classification and diagnosis of work-related upper-extremity conditions was published in 2009 . What emerged from this review confirmed that there is marked inconsistency in the terminology and classification of upper-limb conditions worldwide. The reviewers concluded that upper-limb conditions could be classified into three broad categories: (1) ‘specific conditions’ with evidence-based diagnostic criteria (such as rotator cuff syndrome), (2) ‘other specific conditions’ with no clearly defined diagnosis (such as acromioclavicular syndrome) and (3) ‘non-specific conditions’ characterised by pain, discomfort, fatigue, limited movement and loss of muscle power without a pattern allowing a specific diagnosis to be made. The authors argued that currently terminology used to describe work-related upper-limb conditions is deficient in the sense that it often fails to encapsulate an appropriate meaning consistent with all conditions. Terms sometimes define the condition by a single risk factor as the proposed disease mechanism (e.g., repetitive strain injury) or may be too restrictive in terms of anatomical regions. As a consequence, the labelling may provide little useful insight into the anatomy affected, disease severity, appropriate treatment or expected prognosis . They suggest a dynamic model with an emphasis on specific diagnosis, where conditions currently classified as ‘other specific conditions’ could move to the ‘specific conditions’ category as new evidence becomes available and consensus is reached as to their status. With respect to the shoulder, they propose that the following specific diagnoses be included: rotator cuff ‘syndrome’ (with specific diagnostic definitions including signs and symptoms of tendon inflammation, degeneration and rotator cuff tears), shoulder capsulitis (frozen shoulder) and ‘pain syndromes’ (including fibromyalgia). The category of ‘other specific conditions’ includes those conditions for which there are low prevalence rates, difficult or controversial diagnoses, difficulties in establishing work-relatedness, or where there is an unclear relationship to potential risk factors. It is proposed that this category includes bicipetal tendonitis, acromioclavicular syndrome, glenohumeral joint degenerative disease, subdeltoid bursitis, shoulder pain and scapulothoracic pain syndrome. Non-specific shoulder conditions characterised by pain, discomfort, fatigue, limited movement and loss of muscle power without a pattern allowing a specific diagnosis to be made are included in the schema as a separate category.

Of course, the classification of most soft-tissue disorders of the shoulder relies heavily on the clinical opinions of assessors using clinical assessment tools with limited sensitivity, specificity and reproducibility. This is a problem for both clinicians and their patients, whose clinical progress may be complicated by receiving conflicting diagnostic labels and advice. It is also a major issue for researchers, where misclassification may compromise trials, leading to uncertainty over multiple issues from pathology to causation.

Classification

Classification systems are rules that define minimum criteria that are required to be met in order to establish the existence of a disorder. They help clinicians and researchers communicate about prevalence and incidence, the impact of disorders and prognosis. They may provide insight into some relevant aspect of current experience (pathology, pain and disability) or future course (response to treatment) . Ideally, such systems should be valid, reliable, easy to use and generalisable. Symptoms in the shoulder may arise from discrete pathological conditions, such as rotator cuff tendonitis. Alternatively, presentations may be non-specific or mixed, reflecting pain associated with other factors, which limit the ability to make a clear pathoanatomical diagnosis. There is a lack of consensus as to what constitutes a precise and rigorous case definition of a number of shoulder complaints, and the clinical assessment process is complicated by the indirect links between aetiology, pathology, diagnostic label and the subsequent impairment and disability. The area is also further complicated by a medicolegal environment, which can be acrimonious and occasionally adversarial, and, sometimes, a workers’ compensation system where access to financial support is determined in part by a legal (as well as a medical) process.

There is considerable uncertainty as to the classification of upper-limb disorders in general, and shoulder problems, more specifically. This classification is further complicated when issues of work-relatedness are considered. This lack of operational definition for these conditions is a frequently cited problem for epidemiological research and indeed for clinicians managing these issues. A number of frameworks have been proposed to address this issue. Van Eerd et al. found 27 different classification systems for work-related upper-extremity conditions . A more recent review of the classification and diagnosis of work-related upper-extremity conditions was published in 2009 . What emerged from this review confirmed that there is marked inconsistency in the terminology and classification of upper-limb conditions worldwide. The reviewers concluded that upper-limb conditions could be classified into three broad categories: (1) ‘specific conditions’ with evidence-based diagnostic criteria (such as rotator cuff syndrome), (2) ‘other specific conditions’ with no clearly defined diagnosis (such as acromioclavicular syndrome) and (3) ‘non-specific conditions’ characterised by pain, discomfort, fatigue, limited movement and loss of muscle power without a pattern allowing a specific diagnosis to be made. The authors argued that currently terminology used to describe work-related upper-limb conditions is deficient in the sense that it often fails to encapsulate an appropriate meaning consistent with all conditions. Terms sometimes define the condition by a single risk factor as the proposed disease mechanism (e.g., repetitive strain injury) or may be too restrictive in terms of anatomical regions. As a consequence, the labelling may provide little useful insight into the anatomy affected, disease severity, appropriate treatment or expected prognosis . They suggest a dynamic model with an emphasis on specific diagnosis, where conditions currently classified as ‘other specific conditions’ could move to the ‘specific conditions’ category as new evidence becomes available and consensus is reached as to their status. With respect to the shoulder, they propose that the following specific diagnoses be included: rotator cuff ‘syndrome’ (with specific diagnostic definitions including signs and symptoms of tendon inflammation, degeneration and rotator cuff tears), shoulder capsulitis (frozen shoulder) and ‘pain syndromes’ (including fibromyalgia). The category of ‘other specific conditions’ includes those conditions for which there are low prevalence rates, difficult or controversial diagnoses, difficulties in establishing work-relatedness, or where there is an unclear relationship to potential risk factors. It is proposed that this category includes bicipetal tendonitis, acromioclavicular syndrome, glenohumeral joint degenerative disease, subdeltoid bursitis, shoulder pain and scapulothoracic pain syndrome. Non-specific shoulder conditions characterised by pain, discomfort, fatigue, limited movement and loss of muscle power without a pattern allowing a specific diagnosis to be made are included in the schema as a separate category.

Of course, the classification of most soft-tissue disorders of the shoulder relies heavily on the clinical opinions of assessors using clinical assessment tools with limited sensitivity, specificity and reproducibility. This is a problem for both clinicians and their patients, whose clinical progress may be complicated by receiving conflicting diagnostic labels and advice. It is also a major issue for researchers, where misclassification may compromise trials, leading to uncertainty over multiple issues from pathology to causation.

Causality

It is estimated that 65–70% of all shoulder pain is due to rotator cuff disease. This is associated with overloading, instability of the glenohumeral and acromioclavicular joint, muscle imbalance due to adverse anatomical features, cuff degeneration with ageing, ischaemia and musculoskeletal diseases resulting in wasting of the cuff muscles. Adhesive capsulitis accounts for nearly 2% of cases of shoulder pain and is associated with female sex, older age, shoulder trauma, surgery, diabetes, stroke, thyroid disease and cardio-respiratory disorders .

There is increasing recognition that there are multiple risk factors for developing shoulder pain. Other than the well-recognised associations listed above, occupation has been extensively studied and factors associated with this domain extensively reported, despite its limited overall role in the primary causation of shoulder pain in the community. The more recent work in this area is reported below. However, literature is emerging, which focusses across both work and non-work domains. Leijon et al. recently reported on different rates of shoulder pain in a cohort study of 1095 individuals clustered into 11 groups with different combinations of working and living conditions and followed up for 5 years . They found several groups to have increased burdens of musculoskeletal pain, including those in “onerous human service jobs,” “free agents,” those with a “family burden” and those “mentally stretched,” as well as the “physically strained” group. They argue that such analyses may assist in the development of broad-based prevention strategies, as opposed to the more common modification of the single-risk-factor approach. Cultural factors have also been the subject of some recent studies. Madan et al. found higher rates of back, neck and arm pain in UK white manual workers compared with workers from the Indian subcontinent. They hypothesised that societal beliefs about such illness and illness causation influence its occurrence .

The proportion of work-related shoulder pain caused by identifiable clinical conditions, such as rotator cuff disease, differs from the background rates in the general community. It appears that the proportion of shoulder pain cases with non-specific shoulder pain is far higher in the working population. For example, a Finnish study compared the rates of shoulder pain with a specific diagnosis versus non-specific shoulder pain in a population of adults, who had worked in the last 12 months. They found the rates of non-specific shoulder pain to be approximately 6 times greater (12%) than shoulder pain with a specific diagnosis (2%). They also found distinct differences in the determinants of the two groups, arguing that this suggests that the two are distinct entities, rather than the non-specific pain being a precursor of a more severe, clinically specific disease. Specifically, they argued that non-specific symptoms without clinical findings seemed to be associated more strongly with psychological factors and personality traits, rather than biologic, metabolic and biochemical features .

The occupational associations and risk factors for the development of shoulder pain have been the subject of a huge number of studies and systematic reviews . Potential risk factors relate to physical factors and include heavy workload, awkward postures , working above shoulder height , carrying weights on one shoulder , repetitive movements , vibration, pushing and pulling and duration of employment. There is evidence that cumulative intensive shoulder work, particularly incorporating combinations of exposures, is associated with significantly increased rates of prevalence of shoulder disorders . The exposures that are suspected of causing or aggravating these problems are complex and difficult to measure . Recently, however, there have been some attempts made to measure workload more objectively . Much of the work on associations has been cross-sectional (therefore reporting associations rather than risk factors) and often used self-reported data. A recent prospective study, however, has showed that occupational physical loading increases the risk of a subsequent clinical shoulder disorder. In this Finnish study, 909 participants were re-examined 20 years after a national survey, which had gathered data on occupational exposures. The risk of developing a chronic, physician-diagnosed shoulder disorder was increased by 80–150% when workers were consistently exposed to a combination of heavy lifting, working in awkward postures, work involving vibration or repetitive movements . In a recent review for the Health and Safety Executive, Burton et al. reported that more recent epidemiological studies involving longitudinal designs also suggest an association between physical exposures and the development of upper-limb disorders, but they report the effect size to be rather modest and largely confined to intense exposures. The primary outcome investigated in relevant studies (primary causation, symptom expression or symptom modification) is inconsistent across studies and remains a subject of debate .

Many studies have assessed psychosocial risk and found associations less consistent than those with physical factors. These studies are also difficult to perform and assessment methods vary considerably . Psychosocial factors studied include pain associated with high psychological demands, poor control at work, poor social support and job dissatisfaction . Newly employed workers were the subject of a study by Nahit et al. , with follow-up 12 months later . One-fifth (20%) of 1081 workers in their study reported shoulder pain, and this was associated with individuals perceiving their job as stressful most of the time. At 12 months, psychological distress (odds ratio (OR) 2.1), job demands, support from colleagues and satisfaction with work (OR between 1.4 and 1.7) were all associated with increased odds of reported pain. Exposure to monotonous work , psychological distress , low work control, poor relationships with colleagues and individual psychological factors, such as anxiety and depression , have all been suggested as factors associated with shoulder pain in various studies.

Recently, a task force commissioned by the European League against Rheumatism (EULAR) examined the strength of evidence in published review articles relating to the topic of workplace psychosocial factors and musculoskeletal pain . With respect to shoulder/neck pain, they concluded that the most consistent findings were with high job demands (four reviews positive out of six) and low job demands (two reviews positive out of three). They were critical of the reviews in the sense that they found they often evaluated different bodies of evidence (according to their search criteria, year of review, the quality assessment of the studies included and the combination of risk factors included in the studies) and varied in whether explicit criteria for making conclusions based on the strength of evidence were stated. The review pointed out the relative lack of longitudinal studies in this area.

Although the burden of reported shoulder pain with no specific diagnosis far outweighs that with a specific diagnosis, it appears that there is evidence that the latter group is still associated with a number of the work-related risk factors reported earlier. For example, rotator cuff syndrome has been found to be associated with forceful exertion and shoulder flexion , repetitive movements of the shoulder, working with the hand above the shoulder and high psychosocial job demand . Adhesive capsulitis is associated with exposure to overhead work, work involving lifting weights, poor workplace support from colleagues and supervisors and psychological morbidity .

The pathophysiological link between work and pain is an area of current research. Recent work in this area includes demonstrating that shoulder pain is associated with vasodilatation in the trapezius muscle but not with muscle activity . The authors of this work postulate that the interaction between blood vessels and nociceptors may be important in the activation of muscle nociceptors in people with chronic shoulder and neck pain. Autonomic imbalance or increased activity of the sympathetic–adrenal medullary system in patients with chronic local or regional pain is also under investigation. The physiology of neck–shoulder pain and trapezius myalgia was the focus of a previous review in this series .