Background
Shoulder pain is common among the adult population, but estimates have wide ranges for point (7–26%) and one-month prevalence (17–31%) [
1]. Although there are limited studies on the economic burden of shoulder pain, the cost is likely sizable with 250,000 rotator cuff repairs performed annually in the United States costing close to 3 billion USD [
2]. Shoulder disorders lead to lost time off work and are associated with poorer general health and poor mental health as evidenced by increased levels of depression, anxiety and disturbed sleep [
3‐
6].
It is unclear whether mechanical workplace exposures (such as repetitive movements, postures, or heavy loads) and levels of physical activity may predispose individuals to develop shoulder pain or specific disorders. Two recent systematic reviews concluded that there is limited-moderate evidence to suggest occupations requiring arm elevation and shoulder load are associated with shoulder disorders [
7,
8]. The Australian Institute of Health and Welfare (AIHW) investigated occupational-related shoulder injuries presenting to general practitioners (GP) and determined 13% were classified as “work-related” [
9]. Occupations that have a higher incidence of reported rotator cuff disorders are heavy labourers, those workers whose jobs required repetitive positioning in horizontal or above such as painters, and overhead athletes (e.g. swimming and throwing sports) [
10,
11].
Currently younger populations dominate the data concerning the association between occupational factors and shoulder pain reporting. However, a 2015 systematic review demonstrated that despite rotator cuff degeneration increasing linearly with age, shoulder pain peaks between 55 and 64 and does not continue in the same linear fashion past the common retirement age of 65 [
12]. While retirement itself may be responsible for the reduction in estimates, several studies have also shown a reduction of musculoskeletal pain reporting in older adults [
13,
14].
Psychological health status appears to be related to shoulder pain and disability. In a cross-sectional analysis depression and anxiety were correlated with shoulder pain intensity [
15]. It has also been reported that patients with shoulder disorders demonstrated a high prevalence of depression, anxiety and sleep disturbance when compared to a healthy control group [
6]. A population based longitudinal study demonstrated a bidirectional relationship between pain and depression [
16]. To this point there have been no longitudinal studies to suggest a direction for this relationship to shoulder pain.
Occupational, physical activity and psychosocial exposures may change during and after retirement age resulting in a reduction of shoulder complaints. Our aim is to determine the prevalence of shoulder pain in Australian adults and test the association with occupational factors, physical activity and mental health. In order to do this we will:
1.
Establish the point-prevalence estimates of shoulder pain in the young, and one-month prevalence estimates of the older adult population in Australia.
2.
Determine if occupational factors are associated with shoulder pain in young and older adult populations.
3.
Determine if levels of activity are associated with shoulder pain in young and older adult populations.
4.
Determine if depression or anxiety is associated with shoulder pain in young and older adult populations.
Methods
Study design, setting and participants
A cross-sectional analysis to assess the association between occupational, physical activity, mental health factors and self-reported shoulder pain. Data were accessed from a Western Australian pregnancy cohort study called the Raine Study [
17‐
19]. The primary study recruited 2868 mothers between May 1989 and November 1991 from the largest maternity hospital in Perth, Western Australia (King Edward Memorial Hospital).
Following birth, the children (Generation 2- Gen2) of these women were serially followed through to adulthood (27 years of age), with data points at years 1, 2, 3, 5, 8, 10, 13, 16, 20, 23 and 27. Of the 2262 eligible children, 1134 completed the Gen2-22 year follow-up between 2012 and 2014 at the University of Western Australia Centre for Sleep Science. The mothers and fathers have been labelled Generation 1 (Gen1) and had provided limited data at pre-natal and peri-natal time points as well as every follow-up Gen2 was involved with. At the Gen1-26 year follow-up the participants completed their first exclusive follow-up, where much more extensive data was provided.
All aspects of the Raine Study have been approved by the Human Ethics Commitees at King Edward Memorial Hospital, Princess Margaret Hospital, University of Westen Australia and Curtin Univeristy. This cross-sectional analysis study, analysing musculoskeletal data, was reviewed and received approval from the Murdoch University Human Research Ethics Committee (project number: 2019/238).
Data collection
The participants underwent a clinical assessment and a questionaire that included information about occupation, physical activity, mental health and musculoskeletal pain. The parents (Gen1) were invited to participate in the Raine Study Parent Assessment from 2016 to 2018, with 1098 completing the main questionnaire. The questionnaire included information about occupation, physical activity, mental health and musculoskeletal pain, as described in Tables
1 and
2.
Demographic variables
Participants’ age (in years), sex (female or male) and smoking status (yes/no) were collected via the participant questionnaire. Age groups were defined based on a retirement age of 65 for Gen1. Participants heights were measured with a Holtain Stadiometer and body weight with a Wedderburn Chair Scale. BMI was calculated using the standard equation of BMI = weight (kg)/height (m)2.
Table 1
Shoulder pain outcomes
Shoulder pain | Reporting of pain for Gen1 participants was from a one-month time period. Shoulder pain data were collected for participants via the Orebro Musculoskeletal Pain Questionaire (OMPQ) at the Gen1-26year follow-up. Participants were asked “ Please indicate the sites below in which you have had pain in the last month”, with options including left shoulder or right shoulder. | Reporting of pain for Gen2 participants was from a single point in time (point prevalence). Shoulder pain data were collected for participants via musculoskeletal pain questions within the Gen2-23 year follow-up questionnaire. Participants were asked “Do you currently have any body pain?”. For positive responses, the follow-up question asked “Where do you have pain?”, with options including left shoulder or right shoulder. |
Table 2
Potential predictor variables
Occupational Factors | | |
Employment Status | Participants were asked to answer the following question: “Which of the following best describes your current employment situation?”. Options were: employed full-time; employed part-time; employed, but away from work (e.g. long service leave); unemployed looking for full time work; unemployed looking for part time work; not in the labour force (retired, not looking for work, unable to work); do paid casual work; doing unpaid or voluntary work; other. | Participants were asked to answer the following question: “What are you doing now?”. Options were: studying full-time; studying part-time; an apprenticeship; working full-time; working part-time; looking for work; carer for my child; carer for a family member; other. Participants were then asked “do you currently have a full-time, part-time or casual job of any kind?”. The options were: no, do not have a job – not seeking work; no, do not havea job – actively seeking work; yes, do work for payment or profit; yes, do unpaid work in a family business; yes, do other unpaid work. |
Length of time in current occupation | For those participants that had reported they were currently working, they were asked to report how many years or months they had been working in their current occupation or job. |
Industry code | Participants were asked to report what industry do they work in for their current job. They were provided a list of industry codes. |
Work hours | Participants were asked to report how many hours per week they usually work in all (current) jobs: 1–15; 16–24; 25–34; 35–39; 40; 41–48; 49–55; more than 55. Those participants that reported being unemployed or retired where asked to list the main jobs that had in the past 5 years, the industry code and approximate years and months in that role. | Participants were asked to report how many hours per week they usually work in the last 7 days. Those participants that reported being unemployed or retired where asked to list the main jobs that had in the past 5 years, the industry code and approximate years and months in that role. |
Description of work | Participants were asked to indicate on a scale of 1 (not al all) to 10 (extremely) if their work was “heavy or monotonous”. Participants were also asked to select which statement best described the work they do for their current job: sedentary occupation (e.g. secretary – where you spend most of your time sitting); standing occupation (e.g. shop assistant, security guard, spend most of your time standing/walking but not intense physical effort; Physical work (e.g. plumber, nurse – a job that requires some physical effort incuding handling of heavy objects and use of tools); heavy manual work (e.g. bricklayer – a job that involves very vigorous physical activity including handling very heavy objects). |
Physical activity | International Physical Activity Questionnaire (IPAQ). The participants were asked to report the number of days and hours of vigorous, moderate and walking based physical activity in the last 7 days. Using these measures, the participants were classified as either low, moderate or vigorous levels of exercise. They were also asked to report how many hours they spent sitting on weekdays and weekends in the past 7 days. |
Depression and anxiety | Depression Anxiety Stress Scale (DASS21) [ 20]. Participants were asked to respond to 21 statements on a 0–3 scale. The sub scale scores were used to classify the patient as having normal, mild, moderate or severe symptoms of depression. We dichotomized this variable into ‘no depression’ symptoms and ‘depression’ symptoms. We followed the same process for the anxiety subscale. |
Statistical methods
Descriptive statistics of sample demographic data were based on means and standard deviations for normally distributed continuous data or medians and interquartile ranges for non-normally distributed continuous data. Prevalence rates were provided as percentages with 95% confidence intervals.
Within group univariate categorical comparisons were done using chi squared tests. The association of demographic predictor variables with shoulder pain was assessed using logistic regression models including interaction with age and sex for Gen 1 and Gen 2 respectively. Results were summarized using odds ratios (OR) and 95% confidence intervals (CI). Data were analysed using IBM SPSS statistics for Mac (version 24; IBM Corp., Armonk, NY).
Discussion
The aim of this research was to determine the prevalence of shoulder pain and its association with three key domains: occupational factors, physical activity and mental health. In the Gen2 sample population almost twice as many females reported pain compared to males and more than four times as many females reported bilateral shoulder pain as their male counterparts. Around a third of adults aged 40–80 reported the presence of shoulder pain in the last month, with no significant difference between females and males. Within the older cohort there were higher rates of self-reported shoulder pain in those that reported their occupation involved physical or heavy manual work. This same association was not demonstrated in the younger cohort. Both samples had associations between depression or anxiety and shoulder pain reporting, but neither population had associations between physical activity level and shoulder pain.
The reporting of shoulder pain within the older adult generation was highest between the ages of 45 and 65, though these results cannot be considered significant as the confidence intervals overlapped. However, within Gen1 those over 65 were less likely to report shoulder pain than those under. These results are in line with research suggesting that shoulder pain prevalence steadily increases with age until 65 and then either remains stable or slightly decreases [
12]. According to previous research those people still working past the age of 65 continue to report shoulder pain [
21]. Therefore, the likely mechanism for this result is retirement and subsequent cessation of exposure to occupational factors that either cause or aggrevate shoulder pain.
Recent research has indicated that glenohumeral joint hypermobility (GJH) is an observable finding in female athletes across several varied sporting disciplines both recreational and elite. Features commonly associated with GJH include localised pain, ligamentous sprains, dislocations, and subluxations. Female athletes with GJH appear to have a threefold increase in shoulder concerns comparison to athletes without GJH [
22]. Although a direct causal relationship is not inferred with respect to GJH and the current studies populations. Glenohumeral joint hypermobility may be a potential contributing factor that could account for the significant difference in reported shoulder pain between females and males in the younger generation in this study. Additionally, findings indicate that normalised general joint laxity in growing children (9 to 15 yrs) is greater in girls than boys and increases with age [
23]. Findings of this nature may contribute to the reported difference in bilateral shoulder pain observed within the younger female population.
Our study investigated the relationship between shoulder pain and physical occupational factors in both generations. When considering shoulder pain and physical occupational factors the exposure-response relationship is not well understood [
24]. However, several key occupational shoulder pain risk factors have been identified including repetitive motion, heavy loading, forceful actions, vibrational tasks and working with elevated arms [
24‐
26]. The finding of this study suggest occupation may play a role in the development of shoulder pain. Research indicates that cumulative exposure to key factors result in fatigue failure over time which may represent an important etiological developmental factor associated with shoulder pain [
24,
26]. Cumulative exposure over time may account for the findings observed in Gen2 as the younger population simply has not been exposed to the physical demands of their roles for long enough to develop shoulder pain. The cross-sectional nature of this analyses reduces the ability to infer causation however, key factors could predict shoulder pain in a working adult population.
The IPAQ is a validated self-reporting instrument for scrutinising physical activity and inactivity [
27]. Participants physical activity was assessed over a 7-day period in two general areas either low, moderate, or vigorous walking-based activities or the number of hours spent sitting. While it is widely recognized that increased physical activity is essential for promoting good health [
28], lowering the risk of disease development [
29], and reducing premature mortality [
30], our findings indicate that physical activity does not show a significant association with shoulder pain. Unfortunately, the physical activity reported did not represent any occupational specific demands performed in the 7-day period. It has been reported that single question measures such as the IPAQ can significant under-report sedentary time in comparison to selected device measures such as accelerometers, inclinometers, and pedometers [
31]. It is conceivable that the lack of association between physical activity and shoulder pain identified in this study may be contributed to the sensitivity of the questionnaire employed. It has been identified that reporting accuracy is enhanced by implementing multi-item questionnaires, logs/diaries in comparison to single item questions [
31]. Questionnaires intended to investigate the type and frequency of upper extremity activity could possibly identify an association between physical activity and shoulder pain.
The association of depression and anxiety with shoulder pain was investigated in the fourth objective. This produced the clearest association with the younger generation being almost twice, and the older population being more than twice as likely to report shoulder pain when also being classified has having depression or anxiety. This is in line with previous research demonstrating that patients with shoulder pain were more likely to report depression and anxiety when compared with a healthy cohort [
6]. Our findings add to the existing knowledge by demonstrating this association across multiple generations in a population based sample. Though the direction of this association is not established, a bi-directional relationship has been demonstrated to exist between general pain and depression [
16]. It is possible that those suffering from psychological distress are increasingly reporting shoulder pain and visa versa. A longitudinal analyses would need to be performed to further establish the strength of the relationship.
Strengths and limitations
The samples are community-dwelling and predominantly Caucasian (85%), but are considered to be representative of the Western Australians of the same age [
18]. The representativeness of Gen1 mothers compared to the Western Australian population has been investigated at six time points, including the Gen1-26 year follow-up. There were only small differences between the Gen1 mothers and the 2016 Western Australian Population Census data of women 55–64 years of age [
19]. This strengthens the external validity of the results. The Gen1 sample has an age range that crosses retirement age, allowing for group comparisons of pre and post retirement age.
We recognise that this study has several key limitations. The Gen1 cohort may not entirely represent the general population of females as the participants were all mothers. Shoulder pain was self-reported with no anatomical definition or body diagram provided to the participants. This was because the question had been asked through the Orebro questionnaire. We cannot be certain whether this would have resulted in an under or overestimation of shoulder pain prevalence. Furthermore, there were not data that could be linked directly to the shoulder pain question regarding duration of pain, intensity of pain, disability, and number of episodes. This could again result in participants reporting short-lived pain that may or not be disabling in nature and potentially inflating the rates of reported shoulder pain.
We acknowledge a limitation in the collection of pain data, where theGen1 cohort were asked to report shoulder pain experienced in the last month, with Gen2 asked to report if they were currently experiencing shoulder pain. While there is evidence to indicate that recollection of pain may be accurate when reported over a one month period [
32], we are unable to make direct comparisons between generations. The relationships between predictor variables and prevalence estimates could be influenced by the different recollection periods between the generations.
Our study included an extensive list of potential predictor variables, but we acknowledge that many other potential confounders may influence shoulder pain estimates. Future studies could investigate the association with co-morbidities, sleep duration, and pain at more than one site. Finally, the small number of Gen1 participants in the over 65 age group limited the power of this analysis and resulted in wide confidence intervals for the odds ratios.
Acknowledgements
Bruce Walker was the primary supervisor of the research student (CH) and involved in planning the study. We would like to acknowledge the Raine Study participants and their families for their ongoing participation in the study and the Raine Study team for study co-ordination and data collection. We also thank the NHMRC for their long term contribution to funding the study over the last 30 years. The core management of the Raine Study isfunded by The University of Western Australia, Curtin University, Telethon Kids Institute, Women and Infants Research Foundation, Edith Cowan University, Murdoch University, The University of Notre Dame Australia and the Raine Medical Research Foundation.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.