A systematic review of measures of shoulder pain and functioning using the International classification of functioning, disability and health (ICF)

Shoulder pain is common in the general population; prevalence estimates range from 7 to 26 per cent [1]. The large range in the prevalence rates has been explained by the use of different definitions of the condition in the literature [1]. Pain in the neck or shoulder emerged as the most frequent work-related health complaint in a Norwegian cohort study, and diagnosed shoulder pain accounted for almost 18 per cent of all sick leave benefit claims in a Swedish survey [2, 3]. Shoulder pain is characterised by restricted and painful movement of the arm, which results in difficulties in performing movement-related activities [4‐6]. In recent decades, research has shown that psychological and social functioning may also be affected by shoulder pain; additionally, environmental factors may contribute to the development or persistence of the condition [7‐10].

Functional assessments are an important aspect of clinical decision making and research pertaining to patients with shoulder pain. A number of condition-specific measures are available for making these assessments, including standardised clinical examination methods, patient-reported questionnaires and composite scores [5, 6, 11‐14]. Whether the condition-specific symptoms should be limited to movement-related functions of the shoulder region or be expanded to include additional aspects of functioning, such as work, leisure activities and sleep quality has been debated [12, 15]. To make the assessments more comprehensive and to facilitate comparisons with other health conditions, some have advocated the inclusion of generic measures in the assessments [7, 13, 16]. Generic measures may focus on a specific function or broadly include the concept of general health [12]. So far, there are no commonly accepted guidelines for functional assessment in the area of shoulder pain. Given the increasing standards of health measurements, considerable research effort has been devoted to investigating the psychometric properties of the condition-specific measures [17‐24]. Although the content of such measures also needs to be considered, it often receives less attention [25].

With the approval of the International Classification of Functioning, Disability and Health (ICF) in 2001, a conceptual framework and classification is now available for content analysis of functional measures from a bio-psycho-social perspective [26]. The ICF is based on an integrative model that classifies functioning within the components of body functions (b), body structures (s), activities & participation (d) and environmental (e) and personal factors (not classified). The ICF classification provides categories of functioning and environmental factors that are arranged in a hierarchical fashion using an alphanumeric coding system. The initial letter refers to the component. This letter is followed by a numeric code that starts with the chapter number (e.g., Mobility, d4), which is followed by the second level (e.g., d445 Hand and arm use) and then the third level (e.g., d4452 Reaching). A fourth level of classification is also available when appropriate. The categories at a lower level are included in the higher level categories and chapters. Procedures have been established to classify the content of functional measures using ICF categories, regardless of their purpose, their extent and administration method [27, 28].

The ICF classification is comprehensive. Shorter lists of categories, known as ICF core sets, have been developed to describe the typical spectrum of problems in the functioning of patients with a specific health condition [29]. The core set development process was based on literature reviews, expert surveys and single quantitative and qualitative clinical studies. A review investigating commonalities across ICF core sets for musculoskeletal conditions found a large number of common categories for the conditions low back pain, osteoarthritis, osteoporosis, and rheumatoid arthritis; however, there were also unique categories associated with each particular condition [30]. As part of this core set development process, a literature review was conducted to analyse the content of measures for each of the musculoskeletal disorders [31]. Such a review based on a bio-psycho-social perspective on functioning has not been conducted for shoulder pain. The aims of this systematic literature review were to identify the most frequently addressed aspects of functioning in assessments of shoulder pain and provide an overview of the content of frequently used measures.

Using the ICF as a reference, we first identified and quantified the concepts included in frequently used measures of shoulder pain and functioning. The content of the measures was linked to 11 different ICF categories within 3 of 8 domains of body functions and structures, and 28 ICF categories within 8 of 9 domains of activitie s and participation. Environmental factors were scarcely addressed, accounting for only one category. The finding displays that the measures of shoulder pain cover a large number of concepts of daily activities and also some particular concepts of body functions.

As expected, the ICF category sensation of pain was highest ranked. Different concepts of pain were requested in both condition-specific single and multi-item measures and also in generic measures. This is consistent with previous recommendations to regard pain as a global construct measured by pain intensity and by interference with activities [59]. In a systematic literature review on prognostic factors in primary care populations of shoulder disorders, strong evidence was found that high pain intensity at baseline predicts a poor outcome [60]. The ICF categories mobility of joint, structures of the shoulder region and muscle power functions were ranked second, third and fourth, and in most cases linked from concepts in condition-specific measures. However, not all such concepts were common in the measures; the ICF category muscle endurance was not frequent above the 1% limit, although isometric muscle endurance has been proposed as a psycho-physiological measure for shoulder pain [61].

Sleep functions, classified in the ICF as a mental function, was the fifth most frequent ICF category. Concepts of sleep were included in many condition-specific and generic measures, whereas concepts linked to the less frequent ICF categories emotional functions and energy and drive were extracted from only a few measures. A study that included a community based population of subjects with chronic shoulder pain, found that the relation between pain and psychological health was dependent of level of disability [9]. Moreover, a previous review points to the influence of psychosocial and behavioural factors in chronic neck-and-shoulder pain [62]. According to the current finding, concepts of psychological health may be underestimated in commonly used measures of shoulder pain. However, one comprehensive measure on psychological functioning was found, the generic 4DSQ, which captured five different mental functions according to the ICF.

Several of the predominant concepts in measures of shoulder pain and functioning, were in the activities and participation component. Ten ICF categories belonged to mobility functions and five each to self-care and domestic life. Hand and arm use and lifting and carrying were both among the five highest ranked activities and participation categories. Concepts linked to these two ICF categories were extracted from almost all the condition-specific multi-item measures (see Additional file 2). This demonstrates that task orientated movements of the upper-extremity is in the core of the assessment of shoulder pain. The high ranking of the ICF category remunerative employment, was consistent with the high numbers reporting work-relatedness of their shoulder disorder in a previous epidemiological study [2]. Work-related concepts were addressed in a majority of the multi-item condition-specific measures, although the UCLA, SPADI and SDQ did not address any concepts of work. In a recent review of concepts in vocational rehabilitation measures, a number of work-related concepts were extracted [63]. One of the commonly used vocational measures, the JCQ was also identified in the current review [54]. Its comprehensiveness indicates that assessments of work need to capture several different functional domains.

Previous research shows that also social functioning may be affected by shoulder pain [7‐10]. Family-, informal social- and intimate relationship, all appeared among the lower ranked ICF categories and these concepts were included in only one condition-specific measure, the DASH. Although the SF-36 contains a social subscale, none of its concepts were linked to the ICF category interpersonal interactions and relationships[33]. This indicates that the SF-36 requests social relationships in a more general way and not as specific interpersonal interactions.

Products or substances for personal consumption that appeared with a relative frequency of 8.8%, was the only environmental factor above the 1% criteria. This finding reflects that the impact of the environment on functioning is not sufficiently taken into consideration in the assessments of shoulder pain. According to the ICF, the environment contains a large number of physical, social and attitudinal factors which may limit or facilitate functioning. Although some previous research has been devoted to identify risk factors in the workplace environment, the significance of external factors has scarcely been addressed within the shoulder pain research [64].

Concepts measured in different musculoskeletal disorders were identified in a previous review, and of particular interest for the current study was low back pain [31]. Although there were large similarities between the content of the shoulder pain and low back pain measures, some differences emerged. The comparisons showed that the measures of shoulder pain contained a higher number of concepts within self-care and domestic life, whereas the low back pain measures contained a higher number of environmental factor concepts, concerning support and relationships to persons and the attitudes of health professionals.

This review identified 44 condition-specific and 15 generic measures in use to assess functioning in patients with shoulder pain. When comparing the content of the single- and multi-item measures we found that the former requested only pain and movement related functions, whereas the latter included a wide range of body functions and structures, and activities. The wide-ranging DASH and the ASES were linked to 23 and 16 ICF categories respectively, whereas the Constant was linked to 7 categories and the Simple Shoulder Test (SST) and SPADI to 6 categories each (see Additional file 2). These comparisons, using the ICF as a framework, disclose both the similarities and differences in content of measures that all aim to assess aspects of functioning in patient with shoulder pain.

The variation in the type and number of concepts in the condition-specific measures might reflect disparate views on disability among developers of measures. Some of the measures, such as the SPADI and the Oxford Shoulder Score (OSS) were developed to capture joint-specific concepts and to avoid the influence of co-morbidity [41, 65]. On the contrary, the DASH aims at capture disability, defined as difficulty in doing activities in any domain of life [39]. Due to the complexity of the disability of shoulder pain, and the narrow content of many condition-specific measures, it has been recommended to supplement the condition-specific measures with the generic SF-36 [7, 13, 16]. However, as demonstrated in the current study, the SF-36 includes few additional concepts to those requested in the most wide-ranging condition-specific measures. Clarifying the content is of great importance for selecting the most appropriate measures in clinical work and in research, although the choice of a measure is also dependent on the purpose, patient population and the psychometric properties. In our opinion, use of a wide-ranging condition-specific measure may enhance the quality of assessments in many clinical situations. The wide-ranging (Quick-) DASH and the ASES were found to be among the most extensively investigated measures according to measurement properties in a recent review [24].

The current review had some limitations that should be noted. Meaningful concepts in the measures referring to personal factors in the ICF, such as fear avoidance and coping strategies were not reported. The updated linking rules enable the identification of personal factors, but they are still not classified in the ICF [28]. For 10 measures identified in the study, the content was linked in previous studies (32, 32). The commonly used SF-36 was analysed using the first version of the ICF linking rules [27]. Use of the updated linking rules may have given a somewhat different result [28]. For interpretation of the results, it is of importance that a particular ICF category was reported only once for each measure. As such, the content overview of the measures provides information on the breadth of each measure rather than their depth.

Using the ICF as a reference, a total of 40 second-level categories was used to classify the content of condition-specific and generic measures of shoulder pain. The most frequently addressed concepts were pain, movement-related body functions and structures, sleep, hand and arm use; self-care, household tasks, work and employment, and leisure activities. Concepts of psycho-social functioning and environmental factors were less frequently addressed. Commonly used condition-specific measures showed a large variation in content; the DASH and the ASES were linked to more than twice as many ICF categories as the Constant, SST and SPADI. These large differences demonstrate the importance of clarifying the content to select the most appropriate measure both in research and in clinical work. For clinical situations, we propose use of a wide-ranging condition specific measure that conceptualizes assessments of shoulder pain from a bio-psycho-social perspective. Further research is needed to investigate whether patient-reported problems in functioning are captured in the commonly used condition-specific and generic measures.