1 INTRODUCTION

Musculoskeletal pain results in significant economic impact at societal and individual levels. In the USA, the overall economic impact of all musculoskeletal diseases was estimated to be $980.1 billion annually between 2012 and 2014.¹ Meanwhile, Australia reported expenses related to musculoskeletal pain in the health system, excluding indirect costs, of $160.3 billion between 2015 and 2016.² Musculoskeletal pain includes a wide range of conditions, and among them, low back pain and osteoarthritis represent the leading global causes of disability, which are assessed as the number of Years Lived with Disability (YLD).³

The economic and disability burden caused by musculoskeletal pain may, disproportionately, affect rural populations compared to their urban counterparts. This urban–rural inequality is frequently associated with reduced access to health services,⁴ low socioeconomic and educational status,⁵ higher levels of physical inactivity, and socio-cultural factors. Although numerous studies have individually reported the prevalence of musculoskeletal pain in rural and urban populations, no study has systematically synthesised and compared the prevalence of shoulder, neck, hip, knee and overall musculoskeletal pain in rural and urban populations. Examining the prevalence of musculoskeletal pain according to rurality is essential to inform a better distribution of healthcare resources and identify vulnerable disadvantageous populations. Therefore, this systematic review aimed to appraise and summarise the literature on the global prevalence of musculoskeletal pain in rural compared to urban populations. Furthermore, we aimed to separately assess and compare the prevalence of back, knee, hip, shoulder and neck pain as well as the prevalence of adults seeking treatment for musculoskeletal pain living in rural and urban communities. The study hypothesis is that the prevalence of musculoskeletal pain will be higher in rural populations compared to those residing in urban areas.

2 MATERIALS AND METHODS

A review protocol was prospectively registered in The International Prospective Register of Systematic Reviews – PROSPERO-(CRD42017082167) (Appendix S1), with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guiding the report of the review (Prisma 2020 abstract checklist – Appendix S2, Prisma 2020 checklist – Appendix S3 and Prisma-S checklist – Appendix S4). The Australian New South Wales Country Women's Association (GL) was involved in all stages of this review. The study PECOS strategy included participants (adults with acute or chronic musculoskeletal pain), intervention/exposure (rural populations as per the studies' definitions), comparator (urban populations as per the studies' definitions), outcome [prevalence of pain in the general population (e.g. recall period of ≤3 months before the survey) and seeking treatment] and studies (observational studies). More information is available in the Table S1.

3 RESULTS

We identified 9869 records through database searching, and after excluding 1548 duplicates, 6918 titles and 642 abstracts not meeting eligibility criteria, 746 full-text studies were screened (Figure 1). We contacted 20 authors for additional data with one author¹⁰ providing the required data to be included in the meta-analysis. The study flowchart describes the reason for each full-text exclusion. During each stage of title, abstract and full-text screening, one reviewer (PD) performed screening of the studies and a second reviewer (JAM) screened a random sample of 200 titles with 90%, 82% and 89% of agreement, respectively, for each stage. Disagreements were resolved by consensus. One reviewer (CMC) performed an update of the full-text screening and all data extraction.

A total of 42 studies^10-51 were included in the systematic review and meta-analysis, with a total of 489 439 people, with 42% of the total population representing rural adults from 24 countries. Of these countries, seven were classified as high-income (Austria, Canada, Japan, Korea, Scotland, Spain and the USA), nine as upper-middle-income (Brazil, China, Ecuador, Guatemala, Mexico, Russia, South Africa, Suriname and Turkey) and eight as lower-middle-income countries (Bangladesh, Ghana, India, Indonesia, Iran, Nigeria, Pakistan and the Solomon Islands) according to the World Bank classification.⁵² The most common type of study design was cross-sectional, 85% of included studies; meanwhile, 15% were longitudinal cohorts.

The characteristics of the included studies reported one study with a sample of females²⁸ and five studies did not report the male–female ratio,^{10, 19, 32, 37, 39} Most of the studies included populations aged 18 years and older (range 14 and 99), and nine studies (23%) included populations aged 40 years and older. Only 13 studies (31%) used a standardised questionnaire such as the COPCORD (Table S3). The World Health Organisation/International League of Associations for Rheumatology developed the core questionnaire of COPCORD to record musculoskeletal symptoms such as pain, disability, swelling, stiffness and restricted motion in joints and/or musculoskeletal soft tissues in the last 7 days (current) and/or any time in the past.⁵³ The prevalence of musculoskeletal pain in 21 countries shows that Nigeria, Scotland and Turkey reported the highest prevalence of musculoskeletal pain (Figure 2). All outcomes assessed were rated as very low-certainty evidence according to GRADE (Table S4 and Figure S4). The studies' scores using the modified NOS tool ranged from 0.37 to 1.00, with higher scores indicating a lower risk of bias, and an average of 0.76, which represents an overall good quality for the included studies (Table S5). Detailed data used for all analyses can be found in Table S5.

3.1 Prevalence of musculoskeletal pain

A total of 24 of the 42 included studies reported the prevalence of musculoskeletal pain including any rheumatic or joint pain. Results showed that globally, rural populations had 26% higher odds of reporting musculoskeletal pain compared to urban populations (mean odds ratio [OR] = 1.26, 95% CI = 1.08–1.47; n = 302 911) (Figure 3).

The 3-month mean prevalence of musculoskeletal pain was 35% and 30% for rural and urban populations, respectively. The definitions of musculoskeletal pain were variable: 16 studies (67%) reported acute or chronic musculoskeletal pain,^{13, 16-18, 20-23, 26, 30, 31, 33, 37, 38, 43, 49} four studies (17%) reported any rheumatic pain,^{19, 27, 34, 42} three studies (13%) reported any joint pain,^{11, 35, 45} and one study reported any kind of body pain (4%). Three studies showed substantially higher odds of musculoskeletal pain in rural populations in China (OR = 3.5, 95% CI = 3.1–3.9),⁴² Iran (OR = 2.8, 95% CI = 2.5–3.1),¹⁷ and Bangladesh (OR = 2.6, 95% CI = 1.8–3.7).²⁷ The certainty of the evidence was very low according to GRADE due to limitations in study design, high heterogeneity and indirectness in the representation of the outcome.

3.5 Prevalence of shoulder pain

Nine studies reported the prevalence of shoulder pain, with rural populations showing 42% higher odds of reporting shoulder pain compared to urban populations (OR = 1.42, 95% CI = 1.06–1.90; n = 63 575) (Figure 4). The 3-month mean prevalence was 17.2% and 12.5% for rural and urban populations, respectively. Definitions of pain were consistent across the studies and the certainty of evidence was very low according to GRADE due to limitations in study design, high heterogeneity and publication bias.

3.6 Prevalence of hip pain

Seven studies reported the prevalence of hip pain, with rural populations showing 62% higher odds of reporting hip pain compared to urban populations (OR = 1.62, 95% CI = 1.23–2.15; n = 56 251) (Figure 5). The 3-month mean prevalence of hip pain was 5.3% and 3.7% for rural and urban populations, respectively. Definitions of pain were consistent across all the studies, and very low evidence was found through GRADE due to limitations in study design, high heterogeneity and publication bias.

3.7 Prevalence of seeking treatment for musculoskeletal pain

Four studies reported the prevalence of people seeking treatment for musculoskeletal pain, with rural populations showing 24% lower odds of seeking treatment compared to urban populations, although this difference was not statistically significant (OR = 0.76, 95% CI = 0.55–1.03; n = 16 520) (Figure 6). The mean prevalence was 31% and 37% for rural and urban populations, respectively. Very low-certainty evidence was found through GRADE due to limitations in study design, high heterogeneity and publication bias.

4 DISCUSSION

Data from 42 observational studies conducted worldwide, with a total population of 489 439 people from 24 different countries were included in a random-effects meta-analysis that provided evidence that hip, shoulder and overall musculoskeletal pain were statistically more prevalent in rural populations compared to urban populations. Our results showed very-low certainty evidence of any difference in prevalence according to GRADE due to high heterogeneity, limitations in study design and publication bias. Back and knee pain, but not neck pain, were more prevalent in rural populations compared to urban populations, although not statistically significant. Additionally, the meta-analysis showed that adults in rural areas were less likely to seek treatment for musculoskeletal conditions than their urban counterparts, although not statistically significant.

To our knowledge, this is the first systematic review with a meta-analysis approach that summarised and reported on a direct comparison of the global prevalence of musculoskeletal pain between observational studies that included rural and urban populations. Some of the strengths of this study include the prospectively registered protocol, the use of a sensitive search strategy through eight different databases, the use of a robust tool to assess the methodological quality of the included studies and the use of GRADE to assess the certainty of the evidence.

The limitations of this review are mostly related to the included studies, such as the high between-study heterogeneity in the analyses performed, which could be partly explained due to the use of different types of surveys (e.g. internet-based, via email, self-reported, interview) and questionnaires used to assess pain. Overall, only five studies (12%) with a Newcastle Ottawa score <5 (indicative of high risk of bias),⁵⁴ were included in the Systematic review. For example, only three studies (12%) for the musculoskeletal pain outcome,^{31, 35, 51} one study (11%) for shoulder outcome,⁴⁸ one study (6%) for knee outcome,³² and one study (4%) for low back pain outcome.³¹ The most common reason for low-certainty scores was related to the representativeness of the sample (e.g. some studies restricted the inclusion criteria to specific age groups), lower than 30% response/completion rate, higher than 20% gender difference, sampling bias and lack of statistical adjustments or stratification of the data by age or other risk factors.

Another potential cause of high between-study heterogeneity is the use of differences in the definitions of rurality across the studies. The notions of ‘rural’ and ‘rurality’ lack globally consistent definitions. It is unrealistic to expect one unique definition to cater to the needs of all researchers across varied disciplines, nations and policy contexts. We acknowledge these concepts to be adaptable, versatile and capable of multiple interpretations. Within this review, only five out of 42 studies (12%) reported the definition of rurality, with three studies^{10, 37, 44} using population size and density as a factor to define rurality. Two studies^{33, 45} reported their national guidelines to define rurality.

Recall bias could also have influenced the results of the individual studies, which are subject to desirable responses; however, this is a common methodological flaw in most observational studies. Although most of the studies in our systematic review included populations from different countries and regions, we cannot discard the possibility of case overlapping (i.e. the same participant could be included in two studies). Potential researcher bias could be present in the results of the risk of bias and GRADE assessment as only one researcher completed these processes.

Only one systematic review was found focused on the prevalence of low back pain in rural and urban populations. Hoy et al. reported in 2012⁵⁵ the global prevalence of low back pain of 31.9% and 30.7% in rural and urban communities, respectively, with differences in prevalence estimates between populations being similar to those observed in the current review (12.4% in rural compared to 11.1% in urban populations). Differences in the magnitude of prevalence rates could be explained by the variations in the prevalence period and case definition used in the data analysis of both reviews. Furthermore, the current review only included studies that performed a direct comparison between rural and urban populations, and also included specific conditions (e.g. knee pain, shoulder pain, hip pain and neck pain) unlike the aforementioned study.⁵⁵ Similar prevalence rates were reported in a systematic analysis that estimated the global age-standardised prevalence rate of low back pain in 2020 was 7.46%, although rural and urban rates were not reported.⁵⁶

Three studies from China,⁴² Iran¹⁷ and Bangladesh²⁷ contributed to substantially high odds of musculoskeletal pain in rural communities (two of them used the COPCORD survey^{17, 42}). Zeng et al. 2015⁴² reported epidemiological data from rural Beijing in 1987 to urban Shantou in 2012 using COPCORD. The authors hypothesised that the increase in the social economy and living standards, lower labour intensity and development of health services could explain their findings. Davatchi et al. 2009¹⁷ highlighted the lower educational level, harder manual work and perceptions and beliefs in the rural populations of Iran, could explain their findings. However, their results were the highest ever seen in the Asia Pacific League of Associations for Rheumatology countries. Kabir et al. 2003²⁷ investigated urban–rural differences and indicated higher reporting of health problems, lack of awareness, lower educational level, accessibility and affordability of health services in rural populations as hypotheses for their findings.

Our results suggest that people living in rural areas may be less likely to seek care for musculoskeletal pain compared to those living in urban areas. This might be due to the current challenges in healthcare services and limited access to optimal care in rural populations compared to urban populations.⁴⁴ In Spain, 22% of residents of rural homes reported difficulties accessing health services, compared to only 7.4% of residents living in metropolitan homes.⁵ Additionally, the World Health Organisation estimated a worldwide shortage of 17.4 million health workers in 2013.⁵⁷ Rural areas are most neglected in terms of the number of health workers: while 34% of urban populations experience health workforce shortage, 52% of rural populations worldwide cannot access health services due to scarcity of health-related workers.⁵⁸ In Australia, the number of healthcare workers (allied health practitioners) per 100 000 people, decrease according to the level of the remoteness of geographical areas: while in major cities there were 429 allied health practitioners per 100 000 people in 2018, in very remote areas there were only 217.²

The Global Alliance for Musculoskeletal Health highlighted that people in rural settings are a priority and vulnerable group for education about musculoskeletal health.⁵⁹ Telehealth has been shown to be effective in managing musculoskeletal pain in rural Australian communities⁶⁰ as well as for a wide range of general health conditions in low- and middle-income countries.⁶¹ The relationship between the social determinants of health and musculoskeletal pain has been highlighted with particular factors such as education status, socioeconomic status and occupational factors.⁶² While it is apparent that social determinants, including socioeconomic status, underlie the health disparities reported, it is essential to delve deeper into this issue to understand the nuances and complexities involved. Further exploration should include engaging participants from a variety of backgrounds, conducting screenings for social health factors, incorporating outcome variables relevant to equity, and ensuring sufficient statistical power for subgroup analyses.⁶³

Low-value care interventions for musculoskeletal pain are also associated with the burden of disease; however, it is unknown whether its prevalence is higher in rural populations. The implementation of preventive measures and health education programs for musculoskeletal pain has been suggested to be the key strategy to tackle the burden of musculoskeletal pain in rural populations.²¹ Self-management strategies are believed to be a cost-effective approach to addressing the increasing burden of non-communicable diseases.²⁴ These strategies should also incorporate advice and support on physical activity participation among people with musculoskeletal pain.⁴⁵ Little is known, however, about the acceptance, accessibility and uptake of self-management strategies in rural communities.

Mass media/social marketing campaigns have been suggested as a potential strategy to change population behaviour and promote self-management. However, their likelihood of success is low for regular behaviours such as food selection, exposure to sunlight and physical activity.⁶⁴ Media advocacy campaigns, which influence the portrayal of a public health issue by news and entertainment outlets, also serve as a potential supplementary approach to traditional media campaigns; however, their effects in rural and remote communities are unknown.⁶⁴

As for the funding of resources/services for musculoskeletal pain care, it is generally recognised as insufficient. The increasing burden of musculoskeletal conditions requires greater investment in care resources and services. A more holistic approach including prevention, early detection and comprehensive treatment strategies should be prioritised. Additionally, funding should support research and workforce training to improve the overall quality of care and patient outcomes.⁵⁹

Future research is needed to identify genetic, ethnic, social, demographic, environmental or biological factors associated with epidemiological profiles of musculoskeletal pain inherent to each region or country. The development of a unique and standardised tool to assess and monitor levels of musculoskeletal pain could also allow homogeneity and comparison of pain-related outcomes across different communities globally. This study highlights the lack of observational data (i.e. except for government reports) comparing the prevalence of rural and urban musculoskeletal pain in the Australian general population. Addressing this gap would include lobbying for funding to conduct large-scale epidemiological studies in this field. Collaboration among universities, hospitals and research institutions to gather data and conduct large-scale research, especially in the rural healthcare sector, is warranted.

Very low-certainty evidence showed that rural communities worldwide have a higher prevalence of musculoskeletal pain, hip and shoulder pain compared to urban communities. Very low-certainty evidence showed no statistical difference in the prevalence of neck, back, knee pain and on seeking treatment for musculoskeletal pain of people living in rural and urban areas. Strategies aimed to reduce the burden of musculoskeletal pain should consider the specific needs and limited access to quality evidence-based care for musculoskeletal pain of rural populations.

AUTHOR CONTRIBUTIONS

Carlos I. Mesa-Castrillon: Conceptualization; methodology; software; data curation; formal analysis; validation; investigation; visualization; project administration; writing – review and editing; writing – original draft. Paula R. Beckenkamp: Methodology; software; writing – review and editing; writing – original draft; supervision. Manuela Ferreira: Conceptualization; investigation; writing – review and editing; validation; supervision. Milena Simic: Supervision; validation; writing – review and editing; investigation; conceptualization. Phillip R. Davis: Conceptualization; investigation; validation; formal analysis; software; methodology. Antonio Michell: Conceptualization; methodology; software; writing – review and editing; formal analysis. Evangelos Pappas: Conceptualization; investigation; writing – review and editing; methodology. Georgina Luscombe: Conceptualization; investigation; writing – review and editing; methodology. Marcos De Noronha: Conceptualization; investigation; writing – review and editing; methodology. Paulo Ferreira: Conceptualization; investigation; writing – original draft; writing – review and editing; validation; formal analysis; supervision; project administration; methodology; software; visualization; data curation.

FUNDING INFORMATION

The authors declare no specific funding was received from any bodies in the public, commercial or not-for-profit sectors to carry out the work described in this article.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

Ethical approval was not sought for this systematic review, as the data analysed were publicly available from published articles.

STUDY PROTOCOL DEVIATIONS

We have included a co-author (GL) who is an active member of our research group and participated in the development of the current review due to her expertise, and networking with rural communities.